|
low support values are confirmed in the aLRT analysis. In Bayesian and distance analyses, the
frog is basal to the chicken, but these sequences are again at the base of a biologically irrelevant
teleost/mammal clade. However, unlike distance and ML methods, Bayesian posterior
probabilities strongly support this relationship. Bayesian posterior probabilities traditionally
overestimate support values for a variety of reasons (Cummings et al., 2003: Lewis, pers.
comm..) and can give incorrect clades strong posterior probabilities. This may explain why this
biologically unexpected relationship is strongly supported only in the Bayesian trees. Therefore,
ML analyses (supported by bootstrap and aLRT scores) show the most probable scenario for
COX-1 evolution in the gnathostomes: a basal sCOX, followed by a teleost COX-1 clade (as
described above) sister to a tetrapod COX-1 clade.
The COX-2 groups within the gnathostomes mirrors the same general trends as the COX-
1 clade. Teleosts consistently form a well-supported, monophyletic group. The COX-2a
sequences from zebrafish and trout group together to form the most basal members of this clade.
The COX-2b sequences group with the non-duplicate COX-2 sequences from the
acanthopterygiians to form the other teleost COX-2 clade. Sister to the teleosts, the tetrapods
form another COX-2 clade in the gnathostomes. In this tetrapod COX-2 clade, frog and chicken
sequences form the most basal (although not well-supported) clade, with the mammals as a sister
clade. This biologically relevant relationship between COX-2 sequences is predicted in all
analyses.
Basal to the COX-1 and COX-2 clades in the gnathostomes are two clades that are
consistently and robustly supported in all phylogenetic analyses: a urochordate COX clade and a
cephalochordate COX clade. The cephalochordate clade consists of the two COX sequences
obtained in this study (COXc and COXd) and the urochordate clade consists of the two
LIST OF TABLES
Table page
2-1. Degenerate primers used in PCR ........................................................ ............. 64
2-2. Amphioxus COX specific primers used in PCR..................................... ............... 65
2-3. Hagfish COX specific primers used in PCR........................................... ............... 66
2-4. Lamprey COX specific primers used in PCR ............................................ ............ .... 67
2-5. Killifish COX specific primers used in PCR.................. ...... ...............68
2-6. Sculpin COX specific primers used in PCR ................. .......................................... 69
2-7. Sequences included in phylogenetic analyses......................................... ............... 70
2-8. LRTs between amino acid based models of COX evolution.................. .....................72
2-9. LRTs between tree topologies with altered positions of hCOX and 1COX .......................73
Figure 2-11. Alternate likelihood ratio test (aLRT) scores for branch support of COX sequences.
The most likely tree is shown and the topology, branch lengths, and model of
evolution are the same as in Figure 2-9. Scores represent the gain in likelihood when
collapsing a particular branch and are independent of bootstrap scores (Felsenstein,
1985). Scores were interpreted using a conservative estimate of the minimum of
parametric chi-squared and non-parametric Shimodaira-Hasegawa-like (SH) support
value interpretations (Anisimova and Gascuel, 2006). Analysis was performed using
PhyML (Guindon and Gascuel, 2003). Support values of interest to the present study
are enclosed in boxes.
BIOGRAPHICAL SKETCH
Justin Chase Havird was born in 1984, in Valdosta, Georgia to Dawn and Kurt
Havird. Justin has one younger brother, Joshua. The Havirds soon moved to Lake City,
Florida, where Justin spent the majority of his childhood and adolescence. Justin
attended Columbia High School in Lake City and graduated with top honors in 2002.
Justin then attended the University of Florida in Gainesville where he majored in zoology
after an initial interest in aerospace engineering. While at UF, Justin was heavily
involved in the university's band program, playing trumpet in the marching, basketball,
volleyball, concert, and symphonic bands for several years. As an undergraduate, Justin
worked in the laboratory of Dr. David H. Evans for a number of years, investigating ion
transporters in the gills of fishes. His senior thesis was titled "Neuronal Nitric Oxide
Synthase in the Gill of the Euryhaline Killifish, Fundulus heteroclitus." Justin graduated
summa cum laude from UF in 2006 with a B.S. in zoology.
In 2006, Justin entered graduate school at the University of Florida in the
Department of Zoology. Continuing to work with Dr. Evans, Justin spent three summers
at the Mount Desert Island Biological Laboratory in Salisbury Cove, Maine where he
collaborated with other researchers investigating fish gill physiology. During his
graduate studies, Justin has earned several awards and grants, including a GIAR from
Sigma Xi. Justin has also presented his research at scientific meetings and conferences.
While earning his M.S., Justin also designed experiments and instructed students in a
senior level animal physiology course for 2 years. Justin has also collaborated with other
scientists on a wide range of marine life science projects, most notably a revision of the
cobitid genus Leid,11/ I'h1/li /////\ in SE Asia with Dr. Larry Page. After earning his
COX-la TK-WCPYVDFHVPRNEEERKPS---------------------TEL 452
COX-lb SR-TCPYVAFSVPAEEEPGRNDGKERS--------------SEL 598
COX-2 VRGPCPVASFYVPDVKETGSMTINSSTSHSRDSNINPTVILKERTSEL 610
Figure 2-1. Continued.
ACKNOWLEDGMENTS
I most of all thank my supervisory committee chair, Dr. David H. Evans. Although
David was nearing retirement and had resolved to not accept any new students, he welcomed me
as a graduate student without any hesitation. He has guided me in developing research
strategies, implementing those strategies, and interpreting the results. He has always supported
my projects and none of them would have been accomplished without his enthusiasm.
I also thank my supervisory committee members for their commitment to my research
and their willingness to share their expertise. Dr. Michael Miyamoto contributed heavily to the
phylogenetic aspects of my project and encouraged me to be thorough in my analysis. When I
started this project I knew very little about generating meaningful evolutionary relationships.
Michael selflessly took the time necessary to make sure I understood how to develop a
biologically relevant reconstruction of evolutionary changes that took place over 600 million
years ago. He also put his own projects aside while I used his computers for months at a time.
Finally, his guidance was absolutely necessary in interpreting the conflicting results of my
analyses. Dr. Marty Cohn was a great influence on my research and prompted me to think in an
integrative and comparative framework. He generously allowed me to use his lab space. His
expertise working with evolutionarily interesting chordates was invaluable. His students also
graciously helped me learn many of the techniques used in this research.
I am also indebted to several colleagues for their assistance. Dr. Keith Choe pioneered
and taught me many of the molecular techniques now commonly used in the Evans' lab that were
critical to this research. He spent countless hours with me in multiple laboratories
troubleshooting problems, explaining theory, and developing this research. He also encouraged
me to work with the hagfish, lamprey, and amphioxus and continually helped me develop new
ideas and strategies. Dr. Kelly Hyndman was also constantly supportive of this research. She
Prostaglandins were first extracted from semen, prostate (hence the name prostaglandin),
and seminal vesicles by Goldblatt and von Euler in the 1930s (von Euler, 1936) and were shown
to affect blood pressure. The reaction that catalyzes Prostaglandin G2 synthesis was first
characterized in the 1970s (Hamberg and Samuelsson, 1973; Hamberg et al., 1974), but it wasn't
until 1988 that cyclooxygenase (COX) was first described as the enzyme responsible for
endoperoxide synthesis by three separate groups (DeWitt and Smith, 1988; Merlie et al., 1988;
Yokoyama et al., 1988), all of which described a COX from sheep seminal vesicles which was
later named COX-1. However, before COX-1 was isolated there was early evidence that
multiple COX forms existed, based on prostaglandin activity in the rat brain and spleen (Flower
and Vane, 1972). In the early 1990s, a second COX form was cloned from mouse and chicken
fibroblast cell cultures and was named COX-2 (Kujubu et al., 1991; Xie etal., 1991; O'Banion et
al., 1992). This discovery would later prove to be of major importance in the field of
prostaglandin biology due to the development of drugs that could selectively inhibit the different
forms of cyclooxygenase.
Both forms catalyze the reactions that form prostaglandins from arachidonic acid, but
early studies suggested that COX-1 was constitutively expressed in many tissues, while COX-2
was an inducible isoform that was expressed in response to physiological stimuli (Funk, 2001).
Because basal expression of COX-2 was low in some cell types and COX-2 was found to be
expressed much more so than COX-1 in response to mitogens and cytokines, it was assumed that
COX-1 was constitutively expressed at basal levels and served a housekeeping function and that
COX-2 was inducibly expressed to serve in inflammatory functions (Kujubu et al., 1991; Xie et
al., 1991; O'Banion et al., 1992). However, this was later shown to be an oversimplification
(Funk, 2001). Notably, it was shown that COX-2 is constitutively expressed in the mammalian
COX-la MKRANLSNYFLETNHCEQNISDNYHFITLITIL-SRES--KSLASFPSGLTFCCWGA 54
COX-lb MRSPVLGPVCALLLL-LREPAC------------------RGDEVTSDTVNPCCY-- 36
COX-2 MYRFTFA-VF---LLALGVLVC--------------------- EGGNPCC--- 25
N6 8 : ................ ..............
COX-la HVLLTQVLIPVVITLVRTQECVCDSVQIT-TNVTAL LASMETTALSRSSGPEFVRS 110
COX-lb --FPCQHWGVCVRYGEDKYECDCTRTGYTGGNCTVPEFWSRVHQFL-KPS-PEVLHY 9
COX-2 -SEPCQNRGVCTALGTDNYECDCTRTGYHGHNCTTPEFLTWVKISL-KPS-PNTVHY ,79
...................................................... .......... RY.U ...................................
COX-la RTRRSSSSSPTSIGCWHLVNN-SFLRGTVMRLVETVRSDLIPSPPTYNTKYGYLNWE 166
COX-lb LTHFN-------WLWDIINH-TFLRDVLMRMVLTVRSNLIPSPPTYNSKYDYLSWE 138
COX-2 .LTHF.K....-......-.GFWNLNSSF.F.RDAIMR TSRSHLIDSPPTFNADYGYKSWE 129
*N144
COX-la SYYNISYYTRLLPPVPEDCPLPMGTKGRPDLPDPKVLTERFFRRKTFRPDPQGANLM 223
COX-lb SYYNLSYYTRILPPVPKDCPTPLGVKGKAGLPDPELLVERLLKRRTFRPDPQGSNLM 195
COX-2 AYSNLSYYTRTLPPVPEDCPTPMGVVGKKELPDAKLLAEKLFMRRQFIPDPQGTSLM 186
*Q203/H207
COX-la FAFMAQHFTHQFFKTDHELQGGFT LGVDAGNIYGDNLAKQHHLRLHKDGKLKY 280
COX-lb FAFFAQHFTHQFFKTYNRMGVGFT ALAH VDAGHVYGDNLQRQLKLRLHKDGKLKY 252
COX-2 FAFFAQHFTHQFFKSDMKKGPAFT ATG VDLNHVYGGSMERQHKLRLRQDGKLKY 243
--------------*R277----
COX-la QIVNGETYPPTTSEAPVHMMYPEDVPPEKRLAIGQEVFGLLPGLTMYATI LREHNR 337
COX-lb QLVDGQIYPPSVVDAPVKMSYPPGVPPEAQMAICQEVYGLLPGLGMFATL LREHNR 309
COX-2 QVLDGEVYPPTVKEVGADMHYPPHVPESHRFAVGHEAFGLVPGLMMYATILREHNR 300
COX-la LCDILKAEHPTWDDEQLFQTTRLIVIGEIINIIIEEYVQQLSGYQLKLKFDPTLLFN 394
COX-lb VCDELKAEHPTWDDEQLFQTARFVIIGETIKIVIEEYVQQLSGYLLQLKFDPALLFN 366
COX-2 VCDJLKEVHPDWDDERLFQTTRLILIGETIKIVIEDYVQHLSGYNFKLKFDPELLFN 357
S* *Y385/H388 *N410
COX-la ERFQYSNRIALEFCHLYHWHSLMPDSFLIDGDDIPYSQFFYNTSILMHYGVEKLVDA 451
COX-lb SNFQYGNRIALEFSQLYHWHPLMPESFLINGDELPYKRFLFNNTVLTHYGIENLVTA 423
COX-2 QRFQYQNRIASEFNTLYHWHPLMPDSFHIEEKDYSYKEFVFNTSVVTEHGIGNLVES 414
Figure 2-2. Amino acid alignment of the three forms of COX sequenced from the gills of the
longhorn sculpin (Myoxocephalus octodecemspinosus): COX-la, COX-lb, and COX-
2. All sequences were obtained during this study. Amino acids that share at least
60% identity across sequences are highlighted. *Indicates amino acids (numbered
based on ovine COX-1) involved in COX function: active site of COX (tyrosine-385,
histidine-388, and serine-530), substrate binding site (arginine-120), N-glycosylation
sites (asparagine-68, 144, and 410), sites for peroxidase activity (glutamine-203 and
histidine-207), and the two sites which define conformational differences in channels
between COX-1 and COX-2 in mammals (arginine-513 and valine-523) (Yang et al.,
2002; Ishikawa et al., 2007). The black boxes indicate the two domains that define
the haem-binding sites and the dashed box indicates the membrane binding domain
(Kulmacz et al., 2003; Ishikawa et al., 2007). The solid lines represent disulfide
bonds, with the beginning and end of the longest bond represented by arrows
(Kulmacz et al., 2003). The dashed line represents the arginine-277 loop region,
which has been shown to differ between COX-1 and COX-2 forms in mammals
(Yang et al., 2003; Kulmacz et al., 2003). The numbers to the right of the sequences
indicate the amino acid number of each sequence.
sequence identity to COX-2 as well (Yang et al., 2002). Furthermore, sea squirts subphylumm
Urochordata) possess two forms of COX (named COXa and COXb) that do not correspond to
the COX-1 or COX-2 of vertebrates and represent an ancestral origin of COX in the chordates
(Jarving et al., 2004).
The current view of the evolution of cyclooxygenase in the chordates represents an
interesting but incomplete account of this gene family, both from a phylogenetic and functional
standpoint. Therefore, the first goal of this study was to pinpoint the origin of COX-1 and COX-
2 in the chordates by searching for COX forms in the sea lamprey (Petromyzon marinus),
Atlantic hagfish (Myxine glutinosa), and amphioxus (Branchiostoma lancelolatum). A second
goal was to confirm the findings of Ishikawa and colleagues by searching for the 3 forms of
COX in the euryhaline killifish (Fundulus heteroclitus) and longhorn sculpin (Myoxocephalus
octodecemspinosus). It was hypothesized that both teleosts would posses COX-la, COX-lb, and
COX-2 since both are in the Acanthopterygii clade of Teleostei and all other acanthopterygiians
posses COX-la, COX-lb, and COX-2. Here, I report the cloning of 9 novel COX sequences
from the species mentioned above. Sequence and phylogenetic analyses suggest that ancestral
COX forms may have had similar functions as COX-1 and COX-2 in mammals, which likely
originated with the origin of the vertebrates, although the hagfish and lamprey may have COX-1
and COX-2 forms that represent novel lineages.
Materials and Methods
Animals and Holding Conditions
All procedures were approved prior to beginning the experiment by the University of
Florida Institutional Animal Care and Use Committee (IACUC). Euryhaline killifish (Fundulus
heteroclitus) were captured from Northeast Creek near the Mount Desert Island Biological
Laboratory (MDIBL), Salisbury Cove, Maine using minnow traps. They were transported to the
taught me the localization techniques used here and also helped with many aspects of retrieving
and analyzing genetic data. Finally, she was always receptive to critical discussions of my
research and tackled every circumstance in a friendly, often contagious manner.
I am very thankful to my family and friends for their continued support of my education
and research. My entire family has been very interested in the research I have undertaken and
when they were not interested they feigned interest in order to make me happy. My parents,
Kurt and Dawn Havird, have encouraged my desire to be a scientist, and have been
overwhelmingly supportive during my academic career; often in the form of food, clean laundry,
or money. My friends have also been very supportive of my research interests: fellow scientists
have empathized with the typical problems associated with research and non-scientists have
given me the confidence to persevere.
I also thank the following people for providing help with this project in particular, other
related projects, or the daily support needed to become a successful student: Dr. James B.
Claiborne, Rachel Rose, Dr. GuangJun Zhang, Donovan German, Justin Catches, Andrew
Diamanduros, Sara Takeuchi, Jim Stidham, Curtis Lanier, Julia Curtis-Burnes, Oriana Galardi-
Este, Rebecca Kreh, Dr. Susan Edwards, Gini Luchini, Makesha Foster, Emily Cornwell, Dr.
Larry Page, Pete Ryschkewitsch, Frank Davis, Diana Davis, Cathy Moore, Karen Pallone, Vitrell
Sherif, Dr. David Reed, Dr. Paul Lewis, and the laboratories of Dr. Lou Guillette and Dr. David
Julian.
My research was primarily funded by National Science Foundation grants IBN-0089943
and IOB-0519579 to David Evans. A grant-in-aid of research to Justin Havird from Sigma Xi
also supported a portion of this research. Finally, initial support was provided in part by a
University Scholars award to Justin Havird.
Table 2-1. Degenerate primers used in PCR
Name
CH COX Fl
CH COX F2
CH COX F2.25
CH COX F2.5
CH COX F3
CH COX F5
CH COX R1
CH COX R2
CH COX R3
CH COX-1 F4
CH COX-1 F5
CH COX-1 R2
CH COX-la Fl
CH COX-la F2
CH COX-la F3
CH COX-la F4
CHCOX-laF5
CHCOX-la R1
CH COX-la R2
CH COX-la R3
CH COX-lb Fl
CH COX-lb RO
CH COX-lb R3
Orientation
Sense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Sense
Antisense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Antisense
Antisense
Nucleotide sequence (5' to 3')
ATG GAC GAC TAC CAG TGY GAY TGY AC
GAT GTT TGC ATT TTT CGC TCA RCA YTT YAC
TGG CGT GGA CCT AGG TCA NRT NTA YGG
CGA GCT GCG GTT CCA TAA ARA YGG NAA RYT
CCA CTA TAT GGC TGC GGG ARC AYA AYM G
CAT GTG GAA TTC CAT CAC CTG TAY CAY TGG CA
CCC CGA AGG TGG ATG GYT TCC AVY A
GCA TCA GCG GGT GCC ART GRT A
TCC TCT TTT AGT ATG TCG CAG ACT CKR TTR TGY TC
TGC CAG ACA GCA TCC ACA THG AYG GNG A
CAG CAG ACA ATG CGC AGG NCA RAT HGG
TGG GTG TAT GTT GTG TCC TCC NCC DAT YTG
CCC CAC CAA CCT ACA ATA CCA ART AYG GNT A
AAA GTG CTG ACT GAR MGN TTY TT
GGA CTA ATC TGA TGT TCG CGT TYA TGG CNC A
AGC GGA AGG CGG CTT YAC NAA NGC
CAA ATA GTA AAT GGG GAR AYN TAY CC
ATA GGG AAT GTC GTC NCC RTC DAT
GGG GCA CGT TTT CCG GRT ANA YCA T
ACC TCA GTG ACA GTG GGA GGR TAN RTY TC
CGG ATG GGT GTG GGC TTY ACN AAR GC
GGA GAA GGG AGC ACC CAT YTC NAC CAT
CCA GCG TCC GCG CDA TYT CYT C
for MEGA 4. A complete deletion option was also selected. After comparing results between
ML, ME, and Bayesian Phylogenetics (BP), lamprey and hagfish sequences were identified as
possible rouges and another distance analysis with identical parameters (but with a = 0.742) was
performed excluding these two sequences.
A maximum likelihood phylogeny (ML) was generated using the program PhyML
(Guindon and Gascuel, 2003). An initial phylogeny was generated using WAG analysis, 1000
bootstrap replicates, 8 substitution rate categories, and a gamma shape parameter with an
estimated a. Again, the analysis was repeated with the rouge lamprey and hagfish sequences
removed. To evaluate support for monophyletic groups, an approximate likelihood ratio test
(aLRT, Anisimova and Gascuel, 2006) was performed both including and excluding the lamprey
and hagfish sequences. This test used the minimum of Chi-Squared and Shimodaira-Hasegawa
(SH) support values.
Finally, A Bayesian phylogenetics (BP) analysis was also performed using an initial run
with 2 million generations and a final run with 10 million generations using the program
MrBayes 3.1.2. Once again, analyses were performed with and without hagfish and lamprey
sequences. Each run consisted of one cold and three heated chains (T= 0.2). The WAG + F
model was used with an estimated a. After discarding the first 10% of each run as burning,
posterior probabilities were calculated based on trees sampled every 500 generations.
Due to inconsistencies in tree topologies between the three methods used above,
alternative topologies were generated in which the positions of the hagfish and lamprey
sequences were forced to be included in different clades (Figure 2-15). These alternative
topologies were generated and viewed using TreeView 1.6.6. Alternative topologies were
Vio, C. P., Cespedes, C., Gallardo, P., and Masferrer, J. L. (1997). Renal identification of
cyclooxygenase-2 in a subset of thick ascending limb cells. Hypertension. 30, 687-692.
van-Lanschot, J. J., Mealy, K., Jacobs, D. O., Evans, D. A., and Wilmore, D. W. (1991).
Splenectomy attenuates the inappropriate diuresis associated with tumor necrosis factor
administration. Surg. Gynecol. Obstet. 172, 293-297.
von Euler, U. S. (1936). On the specific vaso-dilating and plain muscle stimulating substances
from accessory genital glands in man and certain animals (prostaglandin and
vesiglandin). J. Physiol. 88, 213-234.
Wald, H., Scherzer, P., Rubinger, D., and Popovtzer, M. M. (1990). Effect of indomethacin
in vivo and PGE2 in vitro on mTAL Na-K-ATPase of the rat kidney. Pluge'i \ Arch. 415,
648-650.
Wang, H., Ma, W. G., Tejada, L., Zhang, H., Morrow, J. D., Das, S. K., and Dey, S. K.
(2004). Rescue of female infertility from the loss of cyclooxygenase-2 by compensatory
up-regulation of cyclooxygenase-1 is a function of genetic makeup. J. Biol. Chem. 279,
10649-1058.
Whelan, S. and Goldman, N. (2001). A general empirical model of protein evolution derived
from multiple protein families using a maximum-likelihood approach. Mol. Biol. Evol.
18, 691-699.
Willingale, H. L., Gardiner, N. J., McLymont, N., Giblett, S., and Grubb, B. D. (1997).
Prostanoids synthesized by cyclooxygenase isoforms in rat spinal cord and their
contribution to the development of neuronal hyperexcitability. Br. J. Pharmacol. 122,
1593-1604.
Woolf, C. J., Allchorne, A., Safieh-Garabedian, B., and Poole, S. (1997). Cytokines, nerve
growth factor and inflammatory hyperalgesia: the contribution of tumour necrosis factor.
Br. J. Pharmacol. 121, 417-424
Wood, C. M. and Laurent, P. (2003). Na+ versus Cf transport in the intact killifish after rapid
salinity transfer. Biochim. Biophys. Acta. 1618, 106-119.
Wong, E., Bayly, C., Waterman, H. L., Riendeau, D. and Mancini, J. A. (1997). Conversion
of prostaglandin G/H synthase-1 into an enzyme sensitive to PGHS-2-selective inhibitors
by a double His513 to Arg and Ile523 to Val mutation. J. Biol. Chem. 272, 9280-9286.
Xie, W., Chipman, J., Robertson, D., Erikson, R. and Simmons, D. (1991). Expression of a
mitogen-responsive gene encoding prostaglandin synthase is regulated by mRNA
splicing. Proc. Natl. Acad. Sci. USA 88, 2692-2696.
hCOX with COX-1 and COX-2 proteins from various vertebrates (Figure 2-5) shows that hCOX
has an insertion of amino acids near the C-terminal; characteristic of COX-2 proteins in
vertebrates. However, there are several regions where COX-1 differs from COX-2 but hCOX
shows no identity to COX-1 or COX-2 amino acids (indicated as sites with a + over them in
Figure 2-5).
Two putative incomplete COX sequences were obtained from amphioxus. These
sequences, referred to here as COXc and COXd, are most identical (81%) to each other, but also
share notable identity with hCOX (66%), 1COX (67%), vertebrate COX-1 (> 60%), and
vertebrate COX-2 (> 60%). They are the least identical to Ciona COXa (48%) and COXb
(55%). Based on an alignment of non-mammalian, non-teleost COX proteins (Figure 2-3), it
appears that amino acids critical to COX function are conserved in COXc and COXd. However,
the first haem-binding domain seems to be very different in the amphioxus forms. This is also
the case for Ciona COX forms and COXa, COXb, COXc, and COXd all share notable identity in
this region as shown by alignment (Figure 2-6).
Phylogenetic analyses
A phylogeny (Figure 2-7) generated under a distance optimality criterion with 1000
bootstrap replicates using VMEGA4 shows predicted COX clades. COX-la and COX-lb forms
from teleosts form monophyletic sister clades. The novel COX-1 forms from the killifish and
sculpin group together with other teleosts in these clades as expected based on their designation
here. Eutherian COX-1 forms group together in a monophyletic clade sister to the teleost COX-1
clade. The COX-1 proteins from Gallus, Monodelphis, and Xenopus are located at the base of
this monophyletic vertebrate COX-1 clade, with sCOX being the most basal. Vertebrate COX-2
forms group together in another monophyletic clade that is sister to the vertebrate COX-1 clade.
There are two monophyletic clades within the COX-2 group: a teleost clade and a tetrapod clade.
Diao, H. L., Zhu, H., Ma, H., Tan, H. N., Cong, J., Su, R. W., and Yang, Z. M. (2007). Rat
ovulation, implantation and decidualization are severely compromised by COX-2
inhibitors. Front. Biosci. 12, 3333-3342.
Crollius, H. R. and Wessenbach, J. (2005). Fish genomics and biology. Genome Res. 15,
1675-1682.
Cummings, M. P., Handley, S. A., Myers, D. S., Reed, D. L., Rokas, A., and Winka, K.
(2003). Comparing bootstrap and posterior probability values in the four-taxon case. Syst.
Biol. 52, 477-487.
DeWitt, D. L. and Smith, W. L. (1988). Primary structure of prostaglandin G/H synthase from
sheep vesicular gland determined from the complementary DNA sequence. Proc. Natl.
Acad. Sci. USA. 85, 1412-1416.
Escalante, B. A., Ferreri, N. R., Dunn, C. E., and McGiff, J. C. (1994). Cytokines affect ion
transport in primary cultured thick ascending limb of Henle's loop cells. Am. J. Physiol.
266, C1568-1576.
Evans, D. H., Rose, R. E., Roeser, J. M. and Stidham, J. D. (2004). NaCl transport across the
opercular epithelium of Fundulus heteroclitus is inhibited by an endothelin to NO,
superoxide, and prostanoid signaling axis. Am. J. Physiol. 286, R560-R568.
Evans, D. H., Piermarini, P. M. and Choe, K. P. (2005). The multifunctional fish gill:
dominant site of gas exchange, osmoregulation, acid-base regulation, and excretion of
nitrogenous waste. Physiol. Rev. 85, 97-177.
Felsenstein, J. (1985). Confidence limits on phylogenies: An approach using the bootstrap.
Evolution. 39, 783-791.
Ferreri, N. R., Schwartzman, M., Ibraham, N. G., Chander, P. N., and McGiff, J. C. (1984).
Arachidonic acid metabolism in a cell suspension isolated from rabbit renal outer
medulla. J. Pharmacol. Exp. Ther. 231, 441-448.
Ferreri, N. R., An, S. J., and McGiff, J. C. (1999). Cyclooxygenase-2 expression and
function in the medullary thick ascending limb. Am. J. Physiol. 277, F360-368.
Ferreira, S. H. (1972). Prostaglandins, aspirin-like drugs and hyperalgesia. Nature 240, 200-
203.
Flippin, J. L., Huggett, D., and Foran, C. M. (2007). Changes in the timing of reproduction
following chronic exposure to ibuprofen in Japanese medaka, Oryzias latipes. Aquat.
Toxicol. 81, 73-78.
2-10. Evolutionary history of COX sequences excluding sequences from the hagfish
(hCOX) and lamprey (1COX) inferred using a maximum likelihood optimality
criteria n ....... ......... ......................................................................... 9 2
2-11. Alternate likelihood ratio test (aLRT) scores for branch support of COX sequences.......94
2-12. Alternate likelihood ratio test (aLRT) scores for branch support of COX sequences
excluding hagfish (hCOX) and lamprey (1COX) sequences..............................96
2-13. Evolutionary history of COX sequences inferred using a Bayesian optimality
c rite rio n ................... .......................................................... ................ 9 7
2-14. Evolutionary history of COX sequences excluding hagfish (hCOX) and lamprey
(1COX) sequences inferred using a Bayesian optimality criterion ..................................99
2-15. The 21 alternate tree topologies generated to test the effects of relocating the hagfish
(hCOX) and lamprey (1COX) sequences on the likelihood score (Ln L) ......................100
2-16. Unrooted phylogeny showing the most likely scenario for COX evolution in the
chordates based on several other analyses ............................................ ...............104
3-1. Euryhaline killifish (Fundulus heteroclitus) COX-lb mRNA expression following
procedures outlined in Choe et al. (2006) ............................ .................................... 113
3-2. Longhorn sculpin (Myoxocephalus octodecemspinosus) COX-la and COX-2 mRNA
expression following procedures outlined in Choe et al. (2006)...................................114
3-3. Atlantic hagfish (Myxine glutinosa) COX (hCOX) mRNA expression following
procedures outlined in Choe et al. (2006) .................................................................... 115
3-4. Sea lamprey (Petromyzon marinus) COX (1COX) mRNA expression in various
tissues following procedures outlined in Choe et al. (2006) .................................. 116
LIST OF FIGURES
Figure page
1-1. Cascade of reactions depicting the conversion of arachidonic acid to the primary
prostaglandins via catalyzation with either cyclooxygenase (COX) form ......................35
1-2. General phylogeny representing the different COX forms in the fishes .........................36
1-3. Accepted evolution of ancestral chordates (Pough, et. al, 2005)...................................37
1-4. Evolution of COX forms in the chordates, based on Jarving et al. (2004).....................38
2-1. Amino acid alignment of the forms of COX sequenced from the gills of the
euryhaline killifish (Fundulus heteroclitus)..................... .... ......................... 74
2-2. Amino acid alignment of the three forms of COX sequenced from the gills of the
longhorn sculpin (Myoxocephalus octodecemspinosus): COX-la, COX-lb, and
C O X -2 ................... ........................................................... ................ 7 6
2-3. Amino acid alignment of the forms of COX sequenced from the gills of the lamprey
(Petromyzon marinus), hagfish (Myxine glutinosa), and amphioxus (Branchiostoma
lancelolatum), along with sCOX from the dogfish (Squalus acanthias, Accession
#A AL37727) ............................................................... .... ...... ......... 78
2-4. Amino acid alignment of 1COX sequenced from the gills of the lamprey (Petromyzon
marinus), with COX-1 and COX-2 forms from vertebrates (Accession # NP_032995,
NP_035328, XP_425326, XP_422297, AAS21313, and NP_705942).............................80
2-5. Amino acid alignment of 1COX sequenced from the gills of the lamprey (Petromyzon
marinus), with COX-1 and COX-2 forms from vertebrates (Accession # NP_032995,
NP_035328, XP_425326, XP_422297, AAS21313, and NP_705942).............................82
2-6. Amino acid alignment of COXc and COXd sequences from the lancelet
(Branchiostoma lancelolatum) with COXa and COXb sequences from the sea squirt
(C io n a in testing a lis)............................................................................................... .. 8 5
2-7. Evolutionary history of COX sequences inferred using the minimum evolution
method with a distance optimality criterion ............. ............................................. 87
2-8. Evolutionary history of COX sequences excluding hagfish (hCOX) and lamprey
(1COX) sequences inferred using the minimum evolution method with a distance
optim ality criterion ........ ......................................................................... ....... .. .... 89
2-9. Evolutionary history of COX sequences inferred using a maximum likelihood
optim ality criterion ........ ......................................................................... ....... .. .... 90
e) f)
g) h) .
L mi m
i)j)..
-,.
Figure 2-15. Continued.
101
their respective sculpin counterparts. However, COX-lb from the sculpin seems to have both N
and C-terminal insertions. Disulfide bonds near the n-terminus of sculpin COX-la also seem to
be altered due to the substitution of other amino acids for cysteine. It also appears that some
functional amino acid residues near the C-terminus of sculpin COX-la are not conserved with
the other forms. These include the substitution of Leu for Arg at position 513, Leu for Val at
position 523, and Phe for Ser at position 530.
Lamprey, hagfish, and amphioxus COX forms
The one incomplete putative COX protein obtained from the gills of the lamprey shares
near equal identity (63%) with Mus COX-1 and COX-2. The gene, referred to here as lamprey
COX (1COX) (Yang et al., 2002) shares the most identity (70%) with a form from the lancelet
(COXc), but shares notable identity (64%) with the form from the hagfish as well. Based on an
alignment of COX forms sequenced from select non-mammalian and non-teleost chordates
(Figure 2-3), the partial COX form sequenced from the gills of the lamprey has the predicted
histidine site (His-207) critical for peroxidase activity as well as conserved haem-binding
domains. However, the membrane binding regions near the Arg-277 loop that differ between
COX-1 and COX-2 forms appear to be different from both COX-1 and COX-2 in the lamprey
form. This is seen when the lamprey form is aligned with COX-1 and COX-2 forms from
vertebrates (Figure 2-4).
The putative COX protein from the gills of the hagfish shares slightly more identity with
Mus COX-2 than Mus COX-1 (63% versus 60%). The gene, referred to here as hCOX, shares
the most identity (> 60%) with vertebrate COX-2 forms, but also shares notable identity with
1COX (64%) and forms from amphioxus (> 60%). An alignment of hCOX with other non-
mammalian, non-teleost COX proteins (Figure 2-3) shows that all important amino acid residues
and binding sites critical to cyclooxygenase activity are conserved in hCOX. An alignment of
Figure 2-7. Evolutionary history of COX sequences inferred using the minimum evolution
method with a distance optimality criterion. The optimal tree with the sum of branch
length = 6.63322594 is shown. The percentage of replicate trees in which the
associated taxa clustered together in the bootstrap test (1000 replicates) is shown next
to the branches (Felsenstein, 1985). The tree is drawn to scale, with branch lengths in
the same units as those of the evolutionary distances used to infer the phylogenetic
tree. The evolutionary distances were computed using the JTT matrix-based method
(Jones et al., 1992) and are in the units of the number of amino acid substitutions per
site (scale bar = 0.05 amino acid substitutions per site). The rate variation among sites
was modeled with a gamma distribution (shape parameter = 0.737). The ME tree was
searched using the Close-Neighbor-Interchange (CNI) algorithm at a search level of
1. The Neighbor-Joining algorithm was used to generate the initial tree. All positions
containing gaps and missing data were eliminated from the dataset (Complete
deletion option). There were a total of 147 positions in the final dataset. Phylogenetic
analyses were conducted in MEGA4 (Tamura et al., 2007). Sequences enclosed in a
box are novel and reported here for the first time.
COX1 Mus
COX1 G.gal
COX1 Danio
1COX
COX2 Mus
COX2 G.gal
COX2 F.het
COX1 Mus
COX1 G.gal
COX1 Danio
1COX
COX2 Mus
COX2 G.gal
COX2 F.het
COX1 Mus
COX1 G.gal
COX1 Danio
1COX
COX2 Mus
COX2 G.gal
COX2 F.het
COX1 Mus
COX1 G.gal
COX1 Danio
1COX
COX2 Mus
COX2 G.gal
COX2 F.het
COX1 Mus
COX1 G.gal
COX1 Danio
1COX
COX2 Mus
COX2 G.gal
COX2 F.het
*H207
HQFFKTSGKMGPGFT ALGH VDLGHIYGDNLERQYHLRLFKDGKLKYQVLDG
HQFFKTSGKMGRGFT ALGH VDLGHLYGDNLQRQHQLRLFQDGKLKFQVVNG
HQFFKTHNRVGLGFT GLGH VDAGHIYGDSLDRQLELRLHKDGKLKYQVLNG
HQFFNTDPVRGPAFT ALGH VDLNHIYGGTLERQHQLRLFKDGKLKFQMIDG
HQFFKTDHKRGPGFT GLGH VDLNHIYGETLDRQHKLRLFKDGKLKYQVIGG
HQFFKTDHKKGPGFTCAYGH VDLNHIYGETLERQLKLRLRKDGKLKYQMIDG
HQFFKSDMKNGPAFT AKGH VDLGHIYGENLEKQHKLRLFKDGKLKYTMVNG
-------- --*R 77----
EVYPPSVEQASVLMRYPPGVPPERQMAVGQEVFGLLPGLMLFSTIILREHNRV
EVYPPSVTEVPVHMVYPPAIPKEKQLAMGQEVFGLLPGLCMYATLLREHNRV
DIYPPTVLHAQVKMSYPPSVPPEQQLAIGQEVFGLLPGLGMYATLLREHNRV
EAYPPVVRDAPVHMVYPEHVPASLRFAVGHEVYGLLPGLLVYATVILREHNRV
EVYPPTVKDTQVEMIYPPHIPENLQFAVGQEVFGLVPGLMMYATIILREHNRV
EMYPPTVKDTQAEMIYPPHVPEHLQFSVGQEVFGLVPGLMMYATI LREHNRV
EVYPPLVKDVGVEMHYPPHVPDSQRFAVGHEAFGLVPGLLMYATI LREHNRV
7D jLKEEHPTWDDEQLFQTTRLILIGETIKIVIEEYVQHLSGYFLQLKFDPEL
:DILKQEHPTWGDEQLFQTARLILIGETIKIVIEDYVQHLSGYFLNLKFDPEL
:EILKQEHPTWGDEQLFQTARLIIIGETIRIVIEEYVQHLSGYRLKLHFDPTL
:DJLHARHPRWDDERLFQTARLILTGETMKIVIEEYVQHLSGYNFHLKFDPTL
:DILKQEHPEWGDEQLFQTSRLILIGETIKIVIEDYVQHLSGYHFKLKFDPEL
:DJLKQEHPEWDDEQLFQTTRLILIGETIKIVIEDYVQHLSGYHFKLKFDPEL
LDLKGVHPDWDDERLFQTTRLILIGETIKIVIEDYVQHLSGYHFKLKFDPEL
*Y385/H388 *N410
LFRAQFQYRNRIAMEFNHLYHWHPLMPNSFQVGSQEYSYEQFLFNTSMLVDYG
LFEQQFQYRNRIAVEFNQLYHWHALMPDSFTIQGQEYSYEQFLYNTSMLMDYG
LFNSQFQYQNRISVEFNQLYHWHPLMPDSFYIDGDHIQYSKFIFNTSILTHYG
LFGVNFQYSNRMSLEFNHLYHWHPLMPDSLLIDGRNYSYDEFLFNPGLLADKK
LFNQQFQYQNRIASEFNTLYHWHPLLPDTFNIEDQEYSFKQFLYNNSILLEHG
LFNQRFQYQNRIAAEFNTLYHWHPLLPDTFQIHNQEYTFQQFLYNNSIMLEHG
LFNQRFQYQNRIASEFNTLYHWHPLMPDSFHIEEKDYSYKEFVFNTSVVTEHG
VEALVDAFSRQRAGRIGGGRNFDYHVLHVAVDVIKESREMRLQPFNEYRKRFG
VEALAESFSMQTAGRIGGGQNINANVLGVAVGVIEESRQLRLQPFNEYRKRFG
LEKLVEAFSIQPAGQIGGGHNIHPVVSGVAERVIVESRELRLQPFNEYRKRFN
LMPLVRSFMRQRAGTVSGGRNINKNLLHVATSIIEHGRTLRLQSLNQYRHRFN
LTQFVESFTRQIAGRVAGGRNVPIAVQAVAKASIDQSREMKYQSLNEYRKRFS
LSHMVKSFSKQSAGRVAGGKNVPAAVQKVAKASIDQSRQMRYQSLNEYRKRFM
INNLVDSFSKQIAGRVAGGRNVPGPIMYVAIKSIENSRKMRYQSLNAYRKRFS
Figure 2-4. Amino acid alignment ICOX sequenced from the gills of the lamprey (Petromyzon
marinus), with COX-1 and COX-2 forms from vertebrates (Accession # NP 032995,
NP 035328, XP425326, XP422297, AAS21313, and NP_705942). Amino acids
that share at least 60% identity across sequences are highlighted. *Indicates amino
acids (numbered based on ovine COX-1) involved in COX function: active site of
COX (tyrosine-385 and histidine-388), N-glycosylation site (asparagine-410), and
sites for peroxidase activity (glutamine-203 and histidine-207) (Yang et al., 2002;
Ishikawa et al., 2007). The black boxes indicate the two domains that define the
haem-binding sites (Ishikawa et al., 2007). The dashed line represents the arginine-
277 loop region, which has been shown to differ between COX-1 and COX-2 forms
in mammals (Yang et al., 2003; Kulmacz et al., 2003). The numbers to the right of
the sequences indicate the amino acid number of each sequence.
1COX---
18S--
Figure 3-4. Sea lamprey (Petromyzon marinus) COX (1COX) mRNA expression in various
tissues following procedures outlined in Choe et al. (2006). Expression was constant
across tissues for the ribosomal protein 18S. COX (1COX) expression was most
abundant in the heart (labeled as H), followed by the intestine (I), stomach (S), gills
(G), kidney (K), muscle (M), and brain (B)
G M
B H S I K
CQOXd BrarLosboina
COxa Crona
COX2 GaRlus-
COxq Xerpus
CGO2 Ovi
COX2 Manodel*his
COX2 Cuiis
COX2 Pru
COX, MiVS
cox2 aa oi
cCX20 riasw
COX2-a OrLshynd s
COX2 Ga$terWisV
COX2tb Dan
COX2b OQnXWhyrthvS
COx2 Tw"raod
COX Oryzias
CoX2 Fridukis
G0X2 WOiaOMMlu
1COX Petronmyrmn
SCOX Squalus j
COX1 Bos
COX1 M~nidelpis
COXl Monodflophis
COX1 laindel s
coxi Ws
COX1 Rautus
COX1 Orycolagus
COX HTna'
Coxi Cams
COXi Crdas
COXIa Tetaodrn
COX 1a Galserosiem
COX 1 a Myoocephaks
COXa Funrdulus
COX1 Danio
COX Oncorwhhtus
COX1b Funailus
COX b Tetrankn
COXlb Gasierole-us
COXib Myonxccihakus
.U*i it .jr,:iia
Cephalochordata
Urochordata
2a
2b/2
la
1
lb
Vertebrata-2
Vertebrata-1
Figure 2-16. Unrooted phylogeny showing the most likely scenario for COX evolution in the
chordates based on several other analyses. Clades of interest are highlighted on the
right and sequences that represent unclear placements are highlighted in red boxes.
process in the lizard Pogona vitticeps. Inhibition of COX-1 and COX-2 resulted in no increased
heart rate during heating (Seebacher and Franklin, 2003). Furthermore, administration of
Prostaglandin F2a and Prostaglandin 12, but not Thromboxane B2, caused an increased heart rate
(Seebacher and Franklin, 2003). However, this effect was not seen in crocodiles (Crocodylus
porosus) (Seebacher and Franklin, 2004) or the lizard Phrynocephalusprzewalskii (Liu et al.,
2006), suggesting that thermoregulatory functions of COX via changes in heart rates may vary
between closely related species.
Interestingly, COX was also implicated in tail regeneration in the house lizard. Twenty
days after tail removal (the time period associated with tissue differentiation), an increase in
cyclooxygenase activity occurred along with the appearance of endogenous Prostaglandin E2,
which may signal a cascade resulting in tissue differentiation (Jayadeep et al., 1995). Finally,
indomethacin was used in the turtle bladder to show that prostaglandins likely do not play a role
in the inhibition ofNa+ or H+ transport due to high intracellular calcium (Arruda, 1982). COX
function in reptiles obviously needs further study as many functions associated with COX in
mammals have yet to be investigated in reptiles and there have been no COX targeted
sequencing efforts.
Birds
Cyclooxygenase and prostaglandin function have been studied somewhat in model avian
species, specifically chickens, where there is support that COX is involved in similar processes
as in mammals, (e.g., reproduction). COX-2 (but not COX-1) and prostaglandin activity was
shown to increase with administration of transforming growth factor in granulosa cells of white
leghorn hen follicles, suggesting that COX-2 and not COX-1 plays a role in granulosa cell
proliferation during follicular development (Li et al., 1996). It was also shown that
indomethacin reduced the proliferation of granulosa cells (Jin et al., 2006). When using COX-1
Furthermore, a similar up-regulation in sheep may reduce progesterone levels to the point that
pregnancy cannot be maintained (McLaren et al., 1996). However, it was recently shown that
during the course of pregnancy there is a gradual change in COX expression, with COX-1
dominating during early pregnancy and COX-2 dominating in later stages (Johnson et al., 2006).
It was concluded that while COX-2 is expressed at high levels in late stages of pregnancy, it is
not up-regulated suddenly to induce labor (Johnson et al., 2006). COX inhibitors can halt
premature labor by inhibiting prostaglandin production (Sawdy et al., 1997). However, as in
treatment strategies that use COX inhibitors for other problems, there are negative side effects
associated with the kidneys: renal clearance and glomerular filtration were reduced during the
first weeks of life in babies that were exposed to COX inhibitors to stop premature birth
(Allegaert et al., 2006). As in other cases, COX-2 selective inhibitors were thought to eliminate
these negative side effects but have not proven more effective than non-specific NSAIDs (Olson
2005).
Nervous System
COX-1 and COX-2 are both expressed in the mammalian nervous system and may
participate in nerve transmission, sensory processing, thermoregulation, and pain effects. As in
other systems, the use of COX inhibitors to treat neural diseases (such as Alzheimer's) is
prevalent, but is accompanied by undesired side effects that are not prevented by using COX-2
selective inhibitors.
The brain is one of the few organs that constitutively expresses COX-2 and contains high
amounts of COX-2 in the cortex, hippocampus, hypothalamus, and spinal cord (Breder et al.,
1995), whereas COX-1 is abundantly expressed in the forebrain, where it may be involved in
sensory information processing (Yamagata et al., 1993). COX-2 is expressed in both neuronal
TABLE OF CONTENTS
page
A CK N O W LED G M EN T S ................................................................. ........... ............. .....
L IST O F TA B LE S ......... ..... ............. ................................................................... 8
LIST O F FIG U RE S ................................................................. 9
A B S T R A C T ........................................... ................................................................. 1 1
CHAPTER
1 GENERAL INTRODUCTION ............................... ......................... 13
Introduction ............................ ... ............................. ...................13
Cyclooxygenase Evolution ............................................................. ...................... 16
M a m m a ls .........................................................................................................................1 7
F ish e s ................... ...................1...................8..........
O th er V erteb rates ................................................................................ 19
O th er C h o rd ate s ......................................................................................................... 19
Cyclooxygenase Function in M ammals................................. ................... 20
G gastrointestinal Tract .......................................................................... 20
T h e K id n ey ................................................................2 1
P late lets ................... ...................2...................3.........
R e p ro d u ctio n ............................................................................................................. 2 3
N ervou s Sy stem ....................................... ................................................... ............... 25
Cyclooxygenase Function in Non-Mammals ......................................... ............... 27
Fishes ................. .................................................................27
Osmoregulation .................................. ..... ....................... 27
R production .............................................................29
D differential functions of novel form s ............. .. ......................... ....................30
O their functions .................................................... .... ..................... 30
A m p h ib ia n s ................................................................................................................ 3 1
Reptiles ...... .................................. ......... ........ ....... ....... ............... 31
B ird s ...........................................................................3 2
Stu dy O v erview ................................................................33
2 CYCLOOXYGENASES IN THE ANCESTRAL CHORDATES: SEQUENCE AND
PHYLOGENETIC ANALYSES ............................ ........ 39
Introduction ............................................. ..................39
M materials an d M eth o d s ...........................................................................................................4 1
Animals and Holding Conditions ................................................41
Tissue Collection ............... .... .. ......... ... .. ................ ................ 42
Reverse Transcription, Primer Design, PCR, Cloning, and Sequencing......................43
5' and 3' Rapid Amplification of cDNA Ends (RACE)..................................................45
6
sequences from Ciona (COXa and COXb). The urochordate COX forms always forms the base
of the entire tree and the cephalochordate COX forms always group as a sister clade to the
gnathostome COX clade. This study is the first to show this biologically relevant relationship
between COX forms in these "ancestral" chordates.
The hagfish (hCOX) and lamprey (1COX) sequences have alternate phylogenetic
positions depending on the technique used to reconstruct their evolutionary histories. Also,
when these sequences are removed from analysis, bootstrap, aLRT, and posterior probability
scores supporting COX clades increase. Because of their uncertain placement in COX evolution,
these sequences are designated as "rogues" and analyses excluding them likely show a more
accurate representation of COX evolution. In distance analysis, both hCOX and 1COX are
placed outside the gnathostome COX clade and form two separate COX lineages between the
cephalochordate and gnathostome clades. In ML analysis, 1COX remains basal but hCOX is
placed at the base of the gnathostome COX-2 clade. In BP analysis, hCOX remains in the COX-
2 clade but 1COX is placed within the cephalochordate clade. Support values for hCOX and
1COX placement are weak in all but BP analyses. Clearly, the placement of hCOX and 1COX in
COX evolution is not clear and without further evidence, placing them with any degree of
certainty becomes difficult. In an attempt to find the most likely placement of these sequences,
21 different phylogenies were generated based on the most likely tree with hCOX and 1COX
placed in variable positions. The trees were then evaluated in a maximum likelihood framework.
It is important to note that although Ln L scores were generated for each topology, other
biological factors may influence which topology or type of topology is actually the most likely.
For example, placing hCOX in the COX-2 clade and 1COX at the base of the tree (technically the
most likely tree) implies that another COX lineage exists for all vertebrates (the lineage
kidney, where it regulates blood flow and ion transport in the Loop of Henle (Harris and Breyer,
2001) and plays a role in cell survival in medullary interstitial cells during dehydration (Yang et
al., 2002; Yang, 2003).
Despite their potential functional differences in different tissues, COX-1 and COX-2 are
biochemically very similar. Both are inhibited by nonsteroid anti-inflammatory drugs
(NSAIDs), such as aspirin, due to competition with arachidonic acid for the active sites of the
enzymes or irreversible acetylation (Vane, 1971). Both proteins in humans are -600 amino acids
and share about 63% sequence similarity with each other. However, the proteins originate from
different genes on different chromosomes and encode different mRNA transcripts. COX-1 comes
from a much larger gene (22 kilobases) than COX-2 (8 kilobases). The amino acid residues that
form the binding and catalytic sites (the functionally important sites) as well as the adjacent
amino acids are all identical in human COX-1 and COX-2, except for two substitutions. At
positions 434 and 523, valine is substituted for isoleucine in COX-2 when compared to COX-1.
These substitutions have been implicated as the most likely reason behind the biochemical
differences between COX-1 and COX-2. COX-1 appears to be more selective in terms of the
fatty acids it will accept as substrates than COX-2 (Otto and Smith, 1995). COX-2 selective
inhibitors have offered an insight into the biochemistry behind the different forms. These
inhibitors can differentiate between COX-1 and COX-2 with over 1000 fold specificity
(Griswold and Adams, 1996). The rationale behind these selective inhibitors is that the smaller
size of valine on position 523 in COX-2 allows the inhibitor to bind to a side pocket in COX-2,
but the larger isoleucine on position 523 of COX-1 appears to prevent this binding. When this
position is altered in both forms to the alternate amino acid, the selectivity for inhibitors is
reversed (Wong et al., 1997; Gierse et al., 1996; Guo et al., 1996), suggesting that this residue
COXa Ciona
COXb Ciona
COXc
COXd
COXa Ciona
COXb Ciona
COXc
COXd
*H207
HMFFKTDPMKGMPYQ-GDQ-VDLSQIYGHGEKRQHELRSHVNGKLKVSLVDGH
HQFFKTNTIKGMPFQIGEH3VDLSHVYGHTIQRQHELRSHIDGKLKVFETNGE
HQFFKTDFKKGAGRTIGDH3VDLSHIYGETVERQHQLRSFTDGKLKFQRVEGE
HQFFKTDFKKGPGRTISDH4VDMSHVYGETVERQRQLRSFQDGKLKYQLVDGE
------------* 77----
EFPPLSNQTTANMSNINLLPQEYQFVFGHQGFSLMPTFLIWSTI LREHNRIC
VFPPLTESANVTMSGEKLM-RGRKFAIGHPGFGAFPSFFVIATLLREHNRVC
VYPPSLADAPVHMIYPPYVPEGKRFAIGHEFFGLLPGLFVYSTVILREHNRVC
AFPPSLQDASVHMIYPPDVPEKKRFALGHEFFGLLPGLFVWATVILREHNRVC
COXa Cional D[IKEENPAWDDERIFQTARLVLTGETIKVVIEDYVQHLSGFHYKLLYDPELV 385
COXb Cional DLKDLHPDWDDERLFQTARLILTGETLKIIVEDYVQHVSGFHFQLSYDPEIL 332
COXcI DMKELHPDWDDERLFQTARLILTGETINIIINEYVQHLSGYNFDLFWDPELL 159
COXd DJMKDLHPDWDDERLYQTARLILISETINIVIGEYVQHLAGKNFQLFWDPELL 159
COXa Ciona QGGSHSFHNQIHVEFQL 402
COXb Ciona HKSTFSYNNQIHAEFHI 349
COXc FSDQFQYQNRIFVEFNH 176
COXd FEEQXQYQNSIFVEFNH 176
Figure 2-6. Amino acid alignment of COXc and COXd sequences from the lancelet
(Branchiostoma lancelolatum) with COXa and COXb sequences from the sea squirt
(Ciona intestinalis). Amino acids that share at least 60% identity across sequences
are highlighted. *Indicates the amino acid (numbered based on ovine COX-1)
involved in peroxidase activity (histidine-207) (Yang et al., 2002; Ishikawa et al.,
2007). The black boxes indicate the two domains that define the haem-binding sites
(Ishikawa et al., 2007). The dashed line represents the arginine-277 loop region,
which has been shown to differ between COX-1 and COX-2 forms in mammals
(Yang et al., 2003; Kulmacz et al., 2003). The numbers to the right of the sequences
indicate the amino acid number of each sequence.
Figure 2-10. Evolutionary history of COX sequences excluding sequences from the hagfish
(hCOX) and lamprey (1COX) inferred using a maximum likelihood optimality
criterion. The most likely tree with an Ln L score = -24242.30059 is shown. The
number of replicate trees in which the associated taxa clustered together in the
bootstrap test (1000 replicates) is shown next to the branches (Felsenstein, 1985). The
tree is drawn to scale, with branch lengths in the same units as those of the
evolutionary distances used to infer the phylogenetic tree. The evolutionary distances
were computed using the WAG matrix-based method (Whelan and Goldman, 2001)
and are in the units of the number of amino acid substitutions per site (scale bar = 0.1
amino acid substitutions per site). The rate variation among sites was modeled with a
gamma distribution (shape parameter estimated = 0.742). Phylogenetic analyses were
conducted using PhyML (Guindon and Gascuel, 2003). Sequence names are
abbreviated from Figure 2-9.
2008 Justin Chase Havird
be harder to clone and sequence due to accumulated mutations over evolutionary time, which
would render the sequences unidentifiable and possibly non-functional. However, this is not the
case in the other vertebrates, as COX-1 and COX-2 seem to each have critical functionality.
Using Southern blotting techniques (Southern, 1975) to identify COX DNA sequences in the
DNA of hagfish, lamprey, and dogfish could immediately reveal the presence of two forms of
COX in these animals. If two forms are present, it could be concluded that they likely represent
COX-1 and COX-2 forms, and thus the origin of these forms would coincide with the vertebrate
lineage (as predicted in the current phylogenetic analysis). However, if only one form of COX
was found the results would suggest a more recent origin of COX-1 and COX-2. This technique
could be applied to a wide diversity of chordates to determine how many COX variants they
contain.
Additional Taxa
In this study, a comparative evolutionary approach was taken by sampling a wide range
of chordates, most of which had no previous record of COX genes. However, there are still
many evolutionarily interesting chordates that have not been examined for COX genes. While
COX genes are known from multiple mammals and teleosts, other chordate groups have only
been minimally sampled. Most notably is the lack of any reported COX genes from any reptile.
COX has also only been documented from a single avian species (Gallus gallus), a single
elasmobranch species (Squalus acanthias), and a single marsupial (Monodelphis domestica.
These groups represent major lineages within the vertebrates and may contain novel forms of
COX. The results of this and other recent studies also warrant further investigation within
certain chordate groups. For example, evolutionarily ancestral fishes such as the sturgeon,
paddlefish, and bowfin could likely contain a single COX-1 and COX-2 gene, while the teleosts
studied here have multiple COX-1 or COX-2 genes. Ishikawa and Herschman (2007)
q) c := r) c
u~tot
s) t)}
_____ ______ Cm~ltianu m^
hillashwites
u)l
|- ---- .. ----- L-2--
L*"raxlhumar
Figur 2-5.
^. -I103
)) t> ,nvs
ua :*Brram
SQ ~n ic, _co! l?
coooor*au
Figure~ 2-15.Contnued
103 .r- =
I r ^ ':,*.'*
__________________ ~ BLJ| O.K'S. BE~tfifct`
L ^ f "- _F,.,.
J ^ f
l nh __ ^ ^
___________~r ""**_____
I-f C ^ ^ ^ ^ ^ ^ ^ f
^ -------------------- ^ 1
Ui "2 ^NL RI-,~I
I- ------- ^^^^ = ^ ^h.^.
-- 0 X L .----------
1~ ~ 1 TeE -tt-s --
---^----^ |i I&'l-iWU~i ^^^^^^^^^^^^ ^
^^|},-.K1
Figur I--lj- C01CS'Oida
103
Precambrian Paleozoic Mesozoic Cenozoic
Cambrian Ordovician Silurian Devonian Carboniferous Permian Triassic Jurassic Cretaceous Tertiary
600 mya 543 490 443 417 354 290 251 206 144 65
To
tetrapods
-Teleostei
Petromyzontoidea
Myxinoidea
Cephalochordata
*Urochordata
Figure 1-3 Accepted evolution of ancestral chordates (Pough, et. al, 2005). Black lines show
relationships only; they do not indicate times of divergence nor the unrecorded
presence of taxa in the fossil record. Bars shaded red indicate ranges of time when
the taxon is known to be present. The subphylum Vertebrata includes the
Myxinoidea hagfishess), Petromyzontoidea (lampreys), Teleostei (teleosts), and
tetrapods. There is some controversy over whether the Urochordates (sea squirts) or
the Cephalochordates (lancelets) represent the most basal chordate lineage. Also, it
has historically been suggested that the hagfish and lamprey constitute a
monophyletic group (the Agnathans or Cyclostomes).
Flower, R. J. and Vane, J. R. (1972). Inhibition of prostaglandin synthetase in brain
explains the anti-pyretic activity of paracetamol (4-acetamidophenol). Nature 240,
410-411.
Forster, W. and Parratt, J. R. (1996). The case for low-dose aspirin for the prevention of
myocardial infarction: But how low is low? Cardiovasc. Drugs Ther. 10, 727-734.
Funk, C. D. (2001). Prostaglandins and leukotrienes: advances in eicosanoid biology.
Science 294, 1871-1875.
Furlan, R. L., Macari, M., Malheiros, E. B., Secato, E. R., and Guerreiro, J. R. (1998).
Effect of indomethacin on hyperthermia induced by heat stress in broiler chickens. Int. J
Biometeorol. 42, 73-76.
Gardiner, N. J., Giblett, S., and Grubb, B. D. (1997). Cyclooxygenases in rat spinal cord:
selective induction of COX-2 during peripheral inflammation. Br. J. Pharmacol. 120,
71P.
Gibb, W. and Sun, M. (1996). Localization of prostaglandin H synthase type 2 protein and
mRNA in term human fetal membranes and decidua. J. Endocrinol. 150, 497-503.
Gierse, J. K., McDonald, J. J., Hauser, S. D., Rangwala, S. H., Koboldt, C. M. and
Seibert, K. (1996). A single amino acid difference between cyclooxygenase-1 (COX-1)
and -2 (COX-2) reverses the selectivity of COX-2 specific inhibitors. J. Biol. Chem. 271,
15810-15814.
Gobbetti, A. and Zerani, M. (2002). Pituitary adenylate cyclase-activating polypeptide
induces testicular testosterone synthesis through PGE(2) mediation in crested newt,
Triturus carnifex. J. Exp. Zool. 293, 73-80.
Goetz, F. W., Duman, P., Ranjan, M. and Herman, C. A. (1989). Prostaglandin F and E
synthesis by specific tissue components of the brook trout (Salvelinusfontinalis) ovary. J.
Exp. Zool. 250, 196-205.
Goetz, F. W., Hsu, S.-Y. and Selover, A. (1991). Stimulation of prostaglandin synthesis in fish
follicles by a phorbol ester and calcium ionophore. J. Exp. Zool. 259, 355-364.
Goudet, G., Mugnier, S., Callebaut, I., and Monget, P. (2007). Phylogenetic analysis and
identification of pseudogenes reveal a progressive loss of zona pellucida genes during
evolution of vertebrates. Biol. Reprod. Epub. Ahead ofPrint.
Griswold, D. E. and Adams, J. L. (1996). Constitutive cyclooxygenase (COX-1) and
inducible cyclooxygenase (COX-2): rationale for selective inhibition and progress
to date. Med. Res. Rev. 16, 181-206.
Table 2-5. Killifish COX specific primers used in PCR
Name
Fh COX-lb 3' Fl
Fh COX-lb 3' F2
Fh COX-lb 3' F3
Fh COX-lb 5' R1
Fh COX-lb 5' R2
Fh COX-lb 5' R3
Fh COX-la 3' Fl
Fh COX-la 3' F2
Fh COX-la 3' F3
Fh COX-la 5' R1
Fh COX-la 5' R2
Fh COX-la 5' R3
Orientation
Sense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Nucleotide sequence (5' to 3')
CTT CAC CGA TAG CGA GGA GAT AG
AGC TCT ACG GTG ACA TCG ACA CT
GGG TAA CCC CAT ATG TTC TCC AC
CCC ATG TGA GAG AGC CTT AGT GA
CAG CCT CTC AAA CAA CAC CTG AG
GTA GTA GGA TTC CCA GCT GAG GTA GT
GGT GTA CCC CGA AGG TTT CC
GAG TAC GTG CAG CAC CTG AGC
AGT ACA CCA ATC GCA TCG CC
GAT GAT AAG TCT GGC GGT CTG G
GAT GGT GGC ATA CAC GGT GA
TTT ATG AAG CCG AAG CTG GTG
COXb Ciona
- COXa Ciona
sCOX Squalus
- COX1 Xenopus
COXI Gallus
COXi Oryctolag a
1Fl.98COX1 Homo
00COX1 Canis
J 49?X1 Bos
COX1b Monodelphis
1 1. COXlaMyoxocephalusl
COXla Gasterosteus
COX1a Tetraodon
COX Myxine
COX2 Monodelphis
COX2 Oryctolagus
aOX2 Canis
53COX2 Homo
0.94 1 L po i 2 Bos
1.01 1 L COX2 Ovi
1C9X2 Mus
COX2 Rattus
COX2 Gallus
0.91
COX2 Xenopus
0.9 COX2b Danio
0COX2b Oncorhynchus
1.00 COX2 Tetraodon
COX2 Fundulus
S 1.0 COX2 Oryzias
1. OX2 Gasterosteus
-COX2 Myoxocephalus
SCOX2a Danio
SCOX2a Oncorhynchus
[COX Petromyzon
0.93 COXe Branchiostoma
0 COXd Branchiostoma
0.1
Figure 2-13. Evolutionary history of COX sequences inferred using a Bayesian optimality
criterion. The consensus tree is shown. The posterior probabilities (10 million
generations) are shown next to the branches. Trees were sampled every 500
generations with a burnin of 10% (401 trees). The tree is drawn to scale, with branch
lengths in the same units as those of the evolutionary distances used to infer the
phylogenetic tree. The evolutionary distances were computed using the WAG matrix-
based method (Whelan and Goldman, 2001) and are in the units of the number of
amino acid substitutions per site (scale bar = 0.1 amino acid substitutions per site).
The rate variation among sites was modeled with a gamma distribution (shape
parameter estimated = 0.672275). Phylogenetic analyses were conducted using
MrBayes 3.1.2. Sequences enclosed in a rectangle are novel and reported here for the
first time.
Figure 2-12. Alternate likelihood ratio test (aLRT) scores for branch support of COX sequences
excluding hagfish (hCOX) and lamprey (1COX) sequences. The most likely tree is
shown and the topology, branch lengths, and model of evolution are the same as in
Figure 2-11. Scores represent the gain in likelihood when collapsing a particular
branch and are independent of bootstrap scores (Felsenstein, 1985). Scores were
interpreted using a conservative estimate of the minimum of parametric chi-squared
and non-parametric Shimodaira-Hasegawa-like (SH) support value interpretations
(Anisimova and Gascuel, 2006). Analysis was performed using PhyML (Guindon
and Gascuel, 2003). Support values of interest to the present study are enclosed in
boxes.
and further studies will likely indicate more similarities and differences with COX functions in
mammals.
Amphibians
As in other vertebrates, there is evidence that COX is involved in ovulation and
reproduction in amphibians. Indomethacin was shown to inhibit oocyte ovulation in the frog
(Rana temporaria), while Prostaglandin F2u was shown to promote ovulation (Skoblina et al.,
1997). This suggests another mediator may be influencing PGF2, production in the frog, but this
contrast between seemingly related proteins has not been confirmed in other species. COX may
also function in testosterone synthesis via PGE2 in the crested newt, Triturus carnifex (Gobbetti
and Zerani, 2002). Ion and water transport have also been studied and non-specific COX
inhibitors reduce short circuit current and voltage potential across the frog skin, bladder, and
corneal epithelium (Shakhmatova et al., 1997; Carrasquer and Li, 2002). As in reptiles, fever
and thermoregulatory changes are behaviorally induced in amphibians. Injection of
lippopolysaccarides (LPS) normally causes a behaviorally induced fever in toads. However,
when indomethacin is administered, the fever induced by LPS is completely blocked in the toad
(Bufoparacnemis) (Bicego et al., 2002). This indicates that the COX pathway plays a role in the
behaviorally induced fever caused by LPS and that this response has an ancient origin that may
be conserved in all tetrapods.
Reptiles
Cyclooxygenase has not been studied extensively in the lizards, crocodilians, snakes, and
turtles although some data suggest prominent roles in thermoregulation. Reptiles undergo
behavioral thermoregulation to adjust their body temperatures, and this process is amplified by
changes in heart rate. Specifically, an increase in heart rate (heart-rate hysteresis) accompanies
an increase in temperature. Prostaglandins and cyclooxygenase were shown to play a role in this
over COX-lb in teleosts based on conserved amino acid sequences between predicted COX-lb
forms in Gasterosteus aculeatus, Oryzias latipes, and Tetraodon nigroviridis based on genomic
data and a COX-lb found in killifish earlier in this study (see below). All degenerate primers
were designed using the online program Consensus-Degenerate Hybrid Oligonucleotide Primers
(CODEHOP) (Rose et al., 1998).
Initial PCR reactions were performed on 1/20th of a reverse transcriptase reaction with a
TaKaRa Ex TaqTM Hot Start DNA Polymerase Kit (Takara Bio Inc., Japan) in a PCR Express
thermocycler (ThermoHybaid, Franklin, MA) with the following parameters: initial denaturing at
95 C for 5 minutes, then 35 cycles of 30 seconds at 95 OC, 30 seconds at 60 OC, and 1.5 minutes
at 72 C. A final elongation step of 7 minutes at 72 C was performed for each PCR and then the
reaction was held at 4 OC. Initial PCR products were visualized by ethidium bromide staining on
1-2% agarose gels and then ligated into PCR4-TOPO vectors and transformed into TOP 10
chemically competent cells using a TOPO TA Cloning Kit (Invitrogen) following the
manufacturer's protocol. Plasmids were then isolated using a High Pure Plasmid Isolation Kit
(Roche, Germany) following the manufacturer's protocol and plasmid DNA was sequenced in
both directions at the Marine DNA Sequencing Center at the MDIBL.
After initial fragments were sequenced, Primer Express software (Applied Biosystems,
Foster City, CA) was used to design specific primers based on the initial fragments (Tables 2-2,
2-3, 2-4, 2-5, and 2-6). These specific primers, along with the original degenerate primers, were
used to extend COX sequences. PCR was completed as above with slight modifications to
annealing temperature and elongation time for each reaction, as well as nested PCR reactions in
some cases to increase specificity. Products were visualized, cloned, and sequenced as above.
(hypertension, renal ischemia, and GI toxicity) associated with chronic NSAID use have been
known for some time, the discovery of COX-2 selective inhibitors suggested a treatment for
inflammation without causing disruption of normal COX function (thought to be mediated by
COX-1). The role of COX-2 in the kidney has reduced the supposed power of these new drugs
and negative side effects associated with COX-2 selective inhibitors are now well documented
(see Harris and Breyer, 2006 for a review). Furthermore, in 2005 the Food and Drug
Administration (FDA) required additional warning labeling of celecoxib to highlight side effects
and required Pfizer to remove valdecoxib from the market, both of which are selective COX-2
inhibitors.
Platelets
Interestingly, COX-1 seems to be the only form expressed in mammalian platelets.
COX-2 was not detected in human platelets using western blotting techniques (Reiter et al.,
2001) or in canine platelets using northern blotting techniques (Kay-Mugford et al., 2000). In
the platelet, COX-1 is involved in platelet aggregation (clotting) through the production of
Thromboxane A2 (TXA2). When inhibited with aspirin, COX-1 mediated production of TXA2 is
irreversibly inhibited, causing reduced platelet aggregation (blood thinning). This effect occurs
about an hour after oral ingestion of even low doses of aspirin and can inhibit platelet function
for several days after a dose (Vane et al., 1998; Vane, 1971). Inhibition of COX-1 by resveratrol
and other components of red wine is responsible for the anti-platelet activity associated with the
cardiovascular benefits of red wine consumption (Szewczuk and Penning, 2004).
Reproduction
Cyclooxygenase and the prostaglandins it produces play critical roles in almost every
aspect of reproduction in mammals, although its role in pregnancy has been studied the most
evaluated using Ln L scores generated by maximum likelihood criterion with online execution in
PhyML (Guindon et al., 2005).
Results
Molecular Identification of Cyclooxygenases
Branchiostoma
Initial PCR reactions using the degenerate primers CH COX F3 and CH COX R2 (Table
2-1) amplified two 532 bp products from the lancelet (Branchiostoma lancelolatum). These two
products (referred to hereafter as COXc and COXd) were found to share the most identity with
COXa and COXb from the tunicate. COXc was 50.9% identical to COXa from the tunicate and
COXd was 58.8% identical to COXb from the tunicate. All attempts to extend these sequences
using specific primers designed against these sequences (Table 2-2) were unsuccessful. After
searching for open reading frames, two putative 177 amino acid proteins were predicted from the
lancelet.
Myxine
Initial PCR reactions using the degenerate primers CH COX Fl, CH COX F2, CH COX
R1, and CH COX R2 (Table 2-1) were used to amplify a single 1454 bp product from the hagfish
(Myxine glutinosa). This product (hereafter referred to as hCOX) was found to be 63.8%
identical to COX-1 of the zebrafish (Danio rerio). RACE primers (5' and 3', Table 2-3) were
designed against this initial product and used with kit primers to completely sequence the 2416
bp product. Searching for open reading frames yielded a putative 610 amino acid protein.
Petromyzon
Initial PCR reactions using degenerate primers CH COX F3 and CH COX R1 (Table 2-1)
amplified a single 541 bp product from gill tissue of the lamprey (Petromyzon marinus). This
product (hereafter referred to as 1COX) was later extended using degenerate primers (Table 2-1)
5' and 3' Rapid Amplification of cDNA Ends (RACE)
After initial sequences were extended with specific primers, the 5' and 3' ends were
sequenced using 5' and 3' RACE. Following the manufacturer's protocols, a GeneracerTM Kit
(Invitrogen) was used to make 5' and 3' cDNA for hagfish, killifish, and sculpin. Initial
sequence data was used to make 5' and 3' RACE specific primers for each species (those labeled
with 5' or 3' in Tables 2-2, 2-3, 2-4, 2-5, and 2-6). These primers, along with kit primers that
anneal to the ends of the 5' and 3' cDNA, were used in touchdown PCR reactions with the
following parameters: 5 cycles of 30 seconds at 94 OC and 1.5 minutes at 72 C; 5 cycles of 30
seconds at 94 C and 1.5 minutes at 70 OC; and 25 cycles of 30 seconds at 94 OC, 30 seconds at
60 C, and 1.5 minutes at 72 C. As above, a final elongation step of 7 minutes at 72 C
terminated the reaction, which was held at 4 OC. For some reactions, annealing temperatures and
elongation times were slightly altered and nested PCR reactions were performed for some
reactions using specific primers (Tables 2-2, 2-3, 2-4, 2-5, and 2-6) to increase specificity.
Products were visualized, cloned, and sequenced as above.
Sequence Analysis
Initial sequences from degenerate primers, sequence extensions, and 5' and 3' RACE
were assembled using GeneTools software (BioTools Inc., Edmonton, Alberta) and the resulting
sequences were searched for open reading frames. The predicted amino acids were aligned with
other COX proteins using PepTools software (BioTools Inc.) for phylogenetic analysis. Multiple
alignments were also generated using ClustalW to search for conserved protein domains across
COX sequences. Alignments were manually adjusted and the N and C-terminal areas were
realigned using Microsoft Word based on previous COX alignments (Ishikawa et al., 2007; Yang
et al., 2002; Kulmacz et al., 2003). For the novel sequences, COX forms were aligned with
Table 2-7. Continued.
# Name
45 COX2 Mus
46 COX1 Oryctolagus
47 COX2 Oryctolagus
48 COX1 Ovis
49 COX2 Ovis
50 sCOX Squalus
Organism
Mus musculus (house mouse)
Oryctolagus cuniculus (rabbit)
Oryctolagus cuniculus (rabbit)
Ovis aries (sheep)
Ovis aries (sheep)
Squalus acanthias (spiny dogfish)
Size (AA)
604
606
604
600
603
593
CHAPTER 1
GENERAL INTRODUCTION
Introduction
Cyclooxygenase (COX) is the enzyme responsible for the rate limiting step that converts
arachidonic acid to Prostaglandin G2 and Prostaglandin H2. Arachidonic acid is an essential fatty
acid required by most mammals and can be obtained by the conversion of linoleic acid found in
plants (occurring in herbivores). Some animals cannot convert linoleic acid into arachidonic acid
and therefore must obtain it from other animals (obligate carnivores). COX oxidizes arachidonic
acid to make the hydroperoxy endoperoxide Prostaglandin G2, which is then reduced to form the
hydroxy endoperoxide Prostaglandin H2. Prostaglandin H2 can then be converted to the primary
prostanoids (Prostaglandin 12, Prostaglandin D2, Prostaglandin E2, Prostaglandin F2u, and
Thromboxane A2) by a variety of enzymatic and non-enzymatic pathways. Therefore, the
expression of COX can directly influence the amount of prostaglandins that are synthesized due
to the arachidonic acid cascade (Vane et al., 1998, Fig. 1-1).
Prostaglandins are found in vertebrates, some invertebrates, and possibly even in plants
(Rowley et al., 2005; Ryan, 2000). In mammals, they are expressed in virtually all tissues and
have a myriad of physiological functions, but they mainly regulate vascular tone in smooth
muscle by acting as vasodilators and vasoconstrictors. In addition, prostaglandins are key
regulators of such diverse processes as salt and water homeostasis in the mammalian kidney
(Harris and Breyer, 2001), protection of stomach lining (Kargman et al., 1996), platelet
aggregation in blood (F6rster and Parratt, 1996), birth (Challis et al., 2002), and central nervous
system function (Hopkins, 2007). They also may play prominent roles in pathological states
such as fever, inflammation, and cancer (Vane et al., 1998).
COX-la FSHQPAGQIGGGHNSHAVVLKVAEMVIRESRETRVQPFNEYRKKFNLQPYTSFYDLT 508
COX-lb FSRQVAGQIGGGFNINAAVTKVSVLTIKESRKLRMQPFNEYRKRFNLKPYTSFREFT 480
COX-2 FTNQIAGRVAGGRNVPGPIMYVAIKSIENSRKMRYQSLNAYRKRFSMKPYSSFEDMT 471
*R513 *V523 *S530
COX-la GDIEMAKGLEELYGDIDAVEFYPGLMLEKTLPTRIFGESMLEMGAPLFPERPVGKPH 565
COX-lb DNEEIARELEEFYGDVDALEFYPGLLLERTREGSIFGESMVEMGAPFSLKGLLGNPI 537
COX-2 GEKEMAAILEEFYGHVDAVELYPGLLVEKPRSNAIFGETMVEMGAPFSLKGLMGNPI 528
COX-la LLPV-LEAQHLWGRDGFQYRNFHSEETGVPQHQVVSIRGL-PCSPKGRRNQTNKSI- 620
COX-Ib CSPVYWKPSTFGGKVGF---DIVNSAT---LKKLVCLNTR-TCSYVAFRVPTEEQLK 587
COX-2 CSPEYWKPSTFGGSEGF---NIVNTAS---LQRLVCNNVQGPCPVASFSGPDVKDSG 579
COX-la -----------------------YTL 622
COX-lb TGNDDSKTRT------------ DEL 600
COX-2 SMIINSSTSNSDINPTVILKERTTEL 605
Figure 2-2. Continued.
S+ + + *Q203/H207
COX1 Mus KDCPTPMGTKGKKQLPDVQLLAQQLLLRREFIPAPQGTNILFAFFAQHFTHQF 211
COX1 Danio NDCPTPMGTKGKIKLPDPKLLVEKFMLRRNFRLDPQGTNLMFAFFAQHFTHQF 209
COX1 Ovis RDCPTPMDTKGKKQLPDAEFLSRRFLLRRKFIPDPQSTNLMFAFFAQHFTHQF 209
hCOX QGCPTPMGRTGKKELPDAQKLAERFLLRRTFIPDPQGSNLLFAFFAQHFTHQF 199
COX2 Mus DDCPTPMGVKGNKELPDSKEVLEKVLLRREFIPDPQGSNMMFAFFAQHFTHQF 195
COX2 F.het EDCPTPMGVVGKKELPDVKVLAEKLLVRRRFIPDPQGTSLMFAFFAQHFTHQF 200
COX2 Ovis DDCPTPMGVKGRKELPDSKEVVKKVLLRRKFIPDPQGTNLMFAFFAQHFTHQF 194
+
COX1 Mus FKTSGKMGPGFT( ALGH3VDLGHIYGDNLERQYHLRLFKDGKLKYQVLDGEVY 264
COX1 Danio FKTHNRVGLGFTKGLGH3VDAGHIYGDSLDRQLELRLHKDGKLKYQVLNGDIY 262
COX1 Ovis FKTSGKMGPGFTCALGH3VDLGHIYGDNLERQYQLRLFKDGKLKYQMLNGEVY 262
hCOX FKTDFKRGPGFTCALGH3VDLSHIYGDTLDKQHKLRLHNNGKLKFQMIDGEVF 252
COX2 Mus FKTDHKRGPGFTrGLGH3VDLNHIYGETLDRQHKLRLFKDGKLKYQVIGGEVY 248
COX2 F.het FKSDMKNGPAFT[AKGH3VDLGHIYGENLEKQHKLRLFKDGKLKYTMVNGEVY 253
COX2 Ovis FKTDIERGPAFTk KM3VDLSHVYGESLERQHNRRLFKDGKMKYQMINGEMY 247
--+-----+---- *R277 +--
COX1 Mus PPSVEQASVLMRYPPGVPPERQMAVGQEVFGLLPGLMLFSTILREHNRVCD-j 317
COX1 Danio PPTVLHAQVKMSYPPSVPPEQQLAIGQEVFGLLPGLGMYATLILREHNRVCEr 315
COX1 Ovis PPSVEEAPVLMHYPRGIPPQSQMAVGQEVFGLLPGLMLYATIILREHNRVCD- 315
hCOX PPLVSEAPVDMIYPPHVPEAARFAVGHEAFGLVPGLMLYGTLLREHNRVCDV 305
COX2 Mus PPTVKDTQVEMIYPPHIPENLQFAVGQEVFGLVPGLMMYATI4LREHNRVCDj 301
COX2 F.het PPLVKDVGVEMHYPPHVPDSQRFAVGHEAFGLVPGLLMYATIrLREHNRVCD 306
COX2 Ovis PPTVKDTQVEMIYPPHIPEHLKFAVGQEVFGLVPGLMMYATILREHNRVCD 300
+
COX1 Mus LKEEHPTWDDEQLFQTTRLILIGETIKIVIEEYVQHLSGYFLQLKFDPELLFR 370
COX1 Danio LKQEHPTWGDEQLFQTARLIIIGETIRIVIEEYVQHLSGYRLKLHFDPTLLFN 368
COX1 Ovis LKAEHPTWGDEQLFQTARLILIGETIKIVIEEYVQQLSGYFLQLKFDPELLFG 368
hCOX LDVDHPEWDDERIFQTARLILIGETIKIVIEDYVQHLSGYHFKLKFDPKLLFG 358
COX2 Mus LKQEHPEWGDEQLFQTSRLILIGETIKIVIEDYVQHLSGYHFKLKFDPELLFN 354
COX2 F.het LKGVHPDWDDERLFQTTRLILIGETIKIVIEDYVQHLSGYHFKLKFDPELLFN 359
COX2 Ovis LKQEHPEWGDEQLFQTSRLILIGETIKIVIEDYVQHLSGYHFKLKFDPELLFN 353
+ *Y385/H388 + *N410 + ++
COX1 Mus AQFQYRNRIAMEFNHLYHWHPLMPNSFQVGSQEYSYEQFLFNTSMLVDYGVEA 423
COX1 Danio SQFQYQNRISVEFNQLYHWHPLMPDSFYIDGDHIQYSKFIFNTSILTHYGLEK 421
COX1 Ovis AQFQYRNRIAMEFNQLYHWHPLMPDSFRVGPQDYSYEQFLFNTSMLVDYGVEA 421
hCOX QPFQYSNRISVEFNHLYHWHGLNPDAFRVGTQEYQYSQFLFNNTILLNHGVRG 411
COX2 Mus QQFQYQNRIASEFNTLYHWHPLLPDTFNIEDQEYSFKQFLYNNSILLEHGLTQ 407
COX2 F.het QRFQYQNRIASEFNTLYHWHPLMPDSFHIEEKDYSYKEFVFNTSVVTEHGINN 412
COX2 Ovis QQFQYQNRIAAEFNTLYHWHPLLPDVFQIDGQEYNYQQFIYNNSVLLEHGVTQ 406
++ + + + ++ + +
COX1 Mus LVDAFSRQRAGRIGGGRNFDYHVLHVAVDVIKESREMRLQPFNEYRKRFGLKP 476
COX1 Danio LVEAFSIQPAGQIGGGHNIHPVVSGVAERVIVESRELRLQPFNEYRKRFNLKP 474
COX1 Ovis LVDAFSRQPAGRIGGGRNIDHHILHVAVDVIKESRVLRLQPFNEYRKRFGMKP 474
hCOX LLEAFNVQQAGRIGGGQNIHGALLHVATASIKHGRKMRFQSLNQYRKQFGLQP 464
COX2 Mus FVESFTRQIAGRVAGGRNVPIAVQAVAKASIDQSREMKYQSLNEYRKRFSLKP 460
COX2 F.het LVDSFSKQIAGRVAGGRNVPGPIMYVAIKSIENSRKMRYQSLNAYRKRFSMKP 465
COX2 Ovis FVESFTRQIAGRVAGRRNLPAAVEKVSKASLDQSREMKYQSFNEYRKRFLLKP 459
*V523
COX1 Mus YTSFQELTGEKEMAAELEELYGDIDALEFYPGLLLEKCQPNSIFGESMIEMGA 529
COX1 Danio YTSFAELTGEQEMSKELEELYGHIDAMEFYPALLLEKTRPGAVFGESMVEMGA 527
COX1 Ovis YTSFQELTGEKEMAAELEELYGDIDALEFYPGLLLEKCHPNSIFGESMIEMGA 527
hCOX YQSFEQLTGETEMAADLAELYSDINAMEFYLGLMVEKPRQGALFGETMVEAGA 517
COX2 Mus YTSFEELTGEKEMAAELKALYSDIDVMELYPALLVEKPRPDAIFGETMVELGA 513
COX2 F.het YTSFEDLTGEKEMAAILEELYGDVDAVELYPGLLVEKPRTNAIFGETMVEMGA 518
COX2 Ovis YESFEELTGEKEMAAELEALYGDIDAMELYPALLVEKPAPDAIFGETMVEAGA 512
Figure 2-5. Continued
Sorbera, L. A., Asturiano, J. F., Carrillo, M., and Zanuy, S. (2001). Effects of
polyunsaturated fatty acids and prostaglandins on oocyte maturation in a marine teleost,
the European sea bass (Dicentrarchus labrax). Biol. Reprod. 64, 382-389.
Southern, E. M. (1975). Detection of specific sequences among DNA fragments separated by
gel electrophoresis. JMolBiol. 98, 503-517.
Spaziani, E. P., Lantz, M. E., Benoit, R. R., and O'Brien, W. F. (1996). The induction of
cyclooxygenase-2 (COX-2) in intact human amnion tissue by interleukin-4.
Prostaglandins. 51, 215-223
Stamatakis, A. (2006). RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses
with thousands of taxa and mixed models. Bioinformatics. 22, 2688-2690.
Steiner, A. A., Rudaya, A. Y., Robbins, J. R., Dragic, A. S., Langenbach, R., and
Romanovsky, A. A. (2005). Expanding the febrigenic role of cyclooxygenase-2 to the
previously overlooked responses. Am. J. Physiol. Regul. Integr. Comp. Physiol. 289,
R1253-1257.
Stokes, J. B. and Kokko, J. P. (1977). Inhibition of sodium transport by prostaglandin E2 across
the isolated, perfused rabbit collecting tubule. J. Clin. Invest. 59, 1099-1104.
Szewczuk, L. M. and Penning, T. M. (2004). Mechanism-based inactivation of COX-1 by red
wine m-hydroquinones: a structure-activity relationship study. J. Nat. Prod. 67, 1777-
1782.
Tamura, K., Dudley, J., Nei, M., and Kumar, S. (2007). MEGA4: Molecular Evolutionary
Genetics Analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596-1599.
Trautman, M. S., Edwin, S. S., Collmer, D., Dudley, D. J., Simmons, D., and Mitchell, M.
D. (1996). Prostaglandin H synthase-2 in human gestational tissues: regulation in amnion.
Placenta 17, 239-245.
Urick, M. E. and Johnson, P. A. (2006). Cyclooxygenase 1 and 2 mRNA and protein
expression in the Gallus domesticus model of ovarian cancer. Gynecol. Oncol. 103, 673-
678.
Vane, J. R. (1971). Inhibition of prostaglandin synthesis as a mechanism of action for the
aspirin-like drugs. Nature 231, 232-235.
Vane, J. R. and Botting, R. M. (1994). Biological properties of cyclooxygenase products. In
LipidMediators, ed. F Cunningham, pp. 61-97. London: Academic.
Vane, J. R., Bakhle, Y. S. and Botting, R. M. (1998). Cyclooxygenases 1 and 2. Annu. Rev.
Pharmacol. Toxicol. 38, 97-120.
COX1 Mus
COX1 Danio
COX1 Ovis
hCOX
COX2 Mus
COX2 F.het
COX2 Ovis
COX1 Mus
COX1 Danio
COX1 Ovis
hCOX
COX2 Mus
COX2 F.het
COX2 Ovis
COX1 Mus
COX1 Danio
COX1 Ovis
hCOX
COX2 Mus
COX2 F.het
COX2 Ovis
++ +
MSRRSLSLWFPPLLLLLLPPTPSVLLADPGVPSPVNPCCYYPCQNQGVCVRFG
MRECNFLLKWTVI--LLLSVSFCAGEESPTSSNTANPCCYYPCQNQGICVRYG
MSRQSISLRFPLL--LLLLSPSPVFSADPGAPAPVNPCCYYPCQHQGICVRFG
MTSEVFVVLCVAVVFA-GAAAAD------------DPCCGSPCENKGVCVSVG
MLFRAVLLCAALGLSQAA-----------------NPCCSNPCQNRGECMSTG
MNRITFAVFLLALCFSFHKEVLG------------NACCSEPCQNRGVCTAMG
MLARALLLCAAV-VCGAA-----------------NPCCSHPCQNRGVCMSVG
---- +*N68 ,...+...........+..... ..................
LDNYQCDCTRTGYSGPNCTIPEIWTWLRNSLRPSPSFTHFLLTHGYWLWEFVN:
LERYECDCTRTGYYGENCTIPLWTRVYRLLKPSPNVVHYILTHFDWLWDLINi
LDRYQCDCTRTGYSGPNCTIPEIWTWLRTTLRPSPSFIHFLLTHGRWLWDFVN:
FEDYECDCTRTGYYGSNCTYPETWTWIVNMLKPLPDTVHYFLTHFAPFWSLVN:
FDQYKCDCTRTGFYGENCTTP FLTRIKLLLKPTPNTVHYILTHFKGVWNIVNi
SDSYECDCTRTGYRGQNCTTPEFLTWIKISLKPTPNTVHYILTHFKGFWNIIN:
FDQYKCDCTRTGFYGENCTTPEFLTRIKLLLKPTPDTVHYI.THFKGVWNIVN
+ ++ + *R120+ *N144 +
.A'Tn'REV'LM~A7LfLTVRSNLIPSPPTYNSAHDYISWESFSNVSYYTRILPSVP
R-SFLRDWLMRK LTVRANLIPSPPTYNSRYDYLNWEAYSNITYYTRILPPVP
-ATFIRDTLMRL LTVRSNLIPSPPTYNIAHDYISWESFSNVSYYTRILPSVP
kAAFLRDRVMRHVLMSRAHMVSSPPTYNADYGYKSWEAYSNLSYFTRLLPPVP
NIPFLRSLIMKY LTSRSYLIDSPPTYNVHYGYKSWEAFSNLSYYTRALPPVA
SISFLRDAIMKY LTSRSHMIDSPPTYNADYGYKSWEAYSNLSYYTRALPPVP
ISF.LRNMIMR.LTSRSHLIESPPTYNVHYSYKSWEAFSNLSYYTRALPPVP
Figure 2-5. Amino acid alignment of ICOX sequenced from the gills of the lamprey
(Petromyzon marinus), with COX-1 and COX-2 forms from vertebrates (Accession #
NP 032995, NP035328, XP_425326, XP_422297, AAS21313, and NP705942).
Amino acids that share at least 60% identity across sequences are highlighted.
*Indicates amino acids (numbered based on ovine COX-1) involved in COX
function: active site of COX (tyrosine-385, histidine-388, and serine-530), substrate
binding site (arginine-120), N-glycosylation sites (asparagine-68, 144, and 410), sites
for peroxidase activity (glutamine-203 and histidine-207), and the two sites which
define conformational differences in channels between COX-1 and COX-2 in
mammals (arginine-513 and valine-523) (Yang et al., 2002; Ishikawa et al., 2007).
+Indicates amino acids unique to hCOX. The black boxes indicate the two domains
that define the haem-binding sites and the dashed box indicates the membrane
binding domain (Kulmacz et al., 2003; Ishikawa et al., 2007). The solid lines
represent disulfide bonds, with the beginning and end of the longest bond represented
by arrows (Kulmacz et al., 2003). The dashed line represents the arginine-277 loop
region, which has been shown to differ between COX-1 and COX-2 forms in
mammals (Yang et al., 2003; Kulmacz et al., 2003). The numbers to the right of the
sequences indicate the amino acid number of each sequence.
COXb Ciona
COXa Ciona
COX2 Monod
QOX2 Canis
COX2 Xenop
Gallu
COX2a Onco
COX2a Dani
IX2b Dani
OX2b Onco
COX2 Tetra
,SY2 Myoxo
COX a Tetr
4 /Ud COX1a Myox
691 COX a Oryz
L 9 COX1a Fund
997
8 COX1b Tetr
COX b Myox
1000
I OXl b Gast
954
1 dCOX1b Fund
681 675
65 COXlb Oryz
946
1000 2 COX1 Danio
1000 724
COX1 Oncor
COX1 Xenop
COX1 Gallu
COX1 Monod
6G9 COXlb2 Mon
1000
COX1b Mono
981 COX1 Rattu
Mus
Oox1 Oryct
693DX1 Homo
80COX1 Canis
7Q 1 Ovis
Cu(1 Bos
COXd Branc
979
COXc Branc
0.1
EVOLUTION OF CYCLOOXYGENASE IN THE CHORDATES
By
JUSTIN CHASE HAVIRD
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
UNIVERSITY OF FLORIDA
2008
PAGE 1
EVOLUTION OF CYCLOOXYGENASE IN THE CHORDATES By JUSTIN CHASE HAVIRD A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORI DA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2008 1
PAGE 2
2008 Justin Chase Havird 2
PAGE 3
To the teachers who inspired, directed, and challenged me. My high school marine biology teacher, Larry Joye, first introduced me to the fa scinating world of life in the water and since then my endeavors have never strayed far from it. Dr. David Evans enthusiastically welcomed me into his lab as an undergraduate, although I ha d no former research expe rience. What started as a part-time project soon evolved into an academic career as I was given the opportunity to conduct original research. During this time, Dr. Keith Choe was an invaluable asset and inspiration. Growing from this original partnership, I conti nued working with David on my masters research and he has always supported my academic and research endeavors; I cannot envision a better advisor. 3
PAGE 4
ACKNOWLEDGMENTS I most of all thank my supervisory comm ittee chair, Dr. David H. Evans. Although David was nearing retirement and had resolved to not accept any new students, he welcomed me as a graduate student without any hesitation. He has guided me in developing research strategies, implementing those stra tegies, and interpreting the resu lts. He has always supported my projects and none of them would have been accomplished without his enthusiasm. I also thank my supervisory committee memb ers for their commitment to my research and their willingness to share thei r expertise. Dr. Mi chael Miyamoto contri buted heavily to the phylogenetic aspects of my project and encouraged me to be thorough in my analysis. When I started this project I knew very little about generating meaningful evolutionary relationships. Michael selflessly took the time necessary to make sure I understood how to develop a biologically relevant reconstr uction of evolutionary change s that took place over 600 million years ago. He also put his own projects aside while I used his computers for months at a time. Finally, his guidance was absolutely necessary in interpreting the conflicting results of my analyses. Dr. Marty Cohn was a great influence on my research and prompted me to think in an integrative and comparative framework. He gener ously allowed me to use his lab space. His expertise working with evolutionarily interestin g chordates was invaluable. His students also graciously helped me learn many of the techniques used in this research. I am also indebted to seve ral colleagues for thei r assistance. Dr. Keith Choe pioneered and taught me many of the molecu lar techniques now commonly used in the Evans lab that were critical to this research. He spent coun tless hours with me in multiple laboratories troubleshooting problems, explaining theory, and developing this re search. He also encouraged me to work with the hagfish, lamprey, and amph ioxus and continually helped me develop new ideas and strategies. Dr. Kelly Hyndman was also constantly supportive of this research. She 4
PAGE 5
taught me the localization techniqu es used here and also helped with many aspects of retrieving and analyzing genetic data. Finally, she was al ways receptive to critical discussions of my research and tackled every circumstance in a friendly, often contagious manner. I am very thankful to my family and friends for their continued support of my education and research. My entire family has been very interested in the resear ch I have undertaken and when they were not interested they feigned inte rest in order to make me happy. My parents, Kurt and Dawn Havird, have encouraged my desire to be a scientist, and have been overwhelmingly supportive during my academic caree r; often in the form of food, clean laundry, or money. My friends have also been very suppor tive of my research interests: fellow scientists have empathized with the typical problems asso ciated with research a nd non-scientists have given me the confidence to persevere. I also thank the following pe ople for providing help with this project in particular, other related projects, or the daily support needed to become a successful student: Dr. James B. Claiborne, Rachel Rose, Dr. GuangJun Zhang, Donovan German, Justin Catches, Andrew Diamanduros, Sara Takeuchi, Jim Stidham, Curtis Lanier, Julia Curtis-Burnes, Oriana GalardiEste, Rebecca Kreh, Dr. Susan Edwards, Gini Luchini, Makesha Foster, Emily Cornwell, Dr. Larry Page, Pete Ryschkewitsch, Frank Davis, Di ana Davis, Cathy Moore, Karen Pallone, Vitrell Sherif, Dr. David Reed, Dr. Paul Lewis, and th e laboratories of Dr. Lou Guillette and Dr. David Julian. My research was primarily funded by Na tional Science Foundation grants IBN-0089943 and IOB-0519579 to David Evans. A grant-in-aid of research to Justin Havird from Sigma Xi also supported a portion of this research. Fina lly, initial support was provided in part by a University Scholars award to Justin Havird. 5
PAGE 6
TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES.........................................................................................................................9 ABSTRACT...................................................................................................................................11 CHAPTER 1 GENERAL INTRODUCTION..............................................................................................13 Introduction................................................................................................................... ..........13 Cyclooxygenase Evolution.....................................................................................................16 Mammals.........................................................................................................................17 Fishes...............................................................................................................................18 Other Vertebrates.............................................................................................................19 Other Chordates...............................................................................................................19 Cyclooxygenase Function in Mammals..................................................................................20 Gastrointestinal Tract......................................................................................................20 The Kidney......................................................................................................................21 Platelets............................................................................................................................23 Reproduction...................................................................................................................23 Nervous System...............................................................................................................25 Cyclooxygenase Function in Non-Mammals.........................................................................27 Fishes...............................................................................................................................27 Osmoregulation........................................................................................................27 Reproduction............................................................................................................29 Differential functions of novel forms.......................................................................30 Other functions.........................................................................................................30 Amphibians..................................................................................................................... .31 Reptiles............................................................................................................................31 Birds................................................................................................................................32 Study Overview................................................................................................................. .....33 2 CYCLOOXYGENASES IN THE ANCE STRAL CHORDATES: SEQUENCE AND PHYLOGENETIC ANALYSES............................................................................................39 Introduction................................................................................................................... ..........39 Materials and Methods...........................................................................................................41 Animals and Holding Conditions....................................................................................41 Tissue Collection.............................................................................................................42 Reverse Transcription, Primer Desi gn, PCR, Cloning, and Sequencing.........................43 5 and 3 Rapid Amplification of cDNA Ends (RACE)..................................................45 6
PAGE 7
Sequence Analysis...........................................................................................................45 Phylogenetic Analysis.....................................................................................................46 Results.....................................................................................................................................48 Molecular Identification of Cyclooxygenases.................................................................48 Branchiostoma .........................................................................................................48 Myxine ......................................................................................................................48 Petromyzon ...............................................................................................................48 Fundulus ...................................................................................................................49 Myoxocephalus .........................................................................................................49 Sequence Analyses..........................................................................................................50 Teleosts COX-1a, COX-1b, and COX-2..................................................................50 Lamprey, hagfish, and amphioxus COX forms........................................................52 Phylogenetic analyses..............................................................................................53 Discussion...............................................................................................................................56 Novel COX Forms...........................................................................................................57 Phylogenetic Analyses.....................................................................................................58 Summary..........................................................................................................................62 3 SUMMARY AND FUTURE DIRECTIONS.......................................................................105 Summary...............................................................................................................................105 Novel COX Sequences..................................................................................................105 Phylogenetic Analyses...................................................................................................106 Future Directions..................................................................................................................106 Completion of Novel Sequences...................................................................................107 Additional Phylogenetic Analysis.................................................................................107 Additional Sequences....................................................................................................108 Additional Taxa.............................................................................................................109 Function of Cyclooxygenases........................................................................................110 LIST OF REFERENCES.............................................................................................................117 BIOGRAPHICAL SKETCH.......................................................................................................130 7
PAGE 8
LIST OF TABLES Table page 2-1. Degenerate primers used in PCR.......................................................................................64 2-2. Amphioxus COX specific primers used in PCR................................................................65 2-3. Hagfish COX specific primers used in PCR......................................................................66 2-4. Lamprey COX specific primers used in PCR....................................................................67 2-5. Killifish COX specific primers used in PCR.....................................................................68 2-6. Sculpin COX specific primers used in PCR......................................................................69 2-7. Sequences included in phylogenetic analyses....................................................................70 2-8. LRTs between amino acid based models of COX evolution.............................................72 2-9. LRTs between tree topologies with altered positions of hCOX and lCOX.......................73 8
PAGE 9
LIST OF FIGURES Figure page 1-1. Cascade of reactions depicting the co nversion of arachidonic acid to the primary prostaglandins via catalyzation with either cyclooxygenase (COX) form........................35 1-2. General phylogeny representing the different COX forms in the fishes...........................36 1-3. Accepted evolution of ancestral chordates (Pough, et. al 2005).......................................37 1-4. Evolution of COX forms in th e chordates, based on Jarving et al (2004)........................38 2-1. Amino acid alignment of the forms of COX sequenced from the gills of the euryhaline killifish (Fundulus heteroclitus ).......................................................................74 2-2. Amino acid alignment of the three forms of COX sequenced from the gills of the longhorn sculpin ( Myoxocephalus octodecemspinosus ): COX-1a, COX-1b, and COX-2................................................................................................................................76 2-3. Amino acid alignment of the forms of COX sequenced from the gills of the lamprey ( Petromyzon marinus ), hagfish ( Myxine glutinosa ), and amphioxus ( Branchiostoma lancelolatum ), along with sCOX from the dogfish ( Squalus acanthias Accession #AAL37727)......................................................................................................................78 2-4. Amino acid alignment of lCOX sequenced from the gills of the lamprey ( Petromyzon marinus ), with COX-1 and COX-2 forms fr om vertebrates (Accession # NP_032995, NP_035328, XP_425326, XP_422297, AAS21313, and NP_705942).............................80 2-5. Amino acid alignment of lCOX sequenced from the gills of the lamprey ( Petromyzon marinus ), with COX-1 and COX-2 forms fr om vertebrates (Accession # NP_032995, NP_035328, XP_425326, XP_422297, AAS21313, and NP_705942).............................82 2-6. Amino acid alignment of COXc a nd COXd sequences from the lancelet ( Branchiostoma lancelolatum ) with COXa and COXb sequences from the sea squirt ( Ciona intestinalis ).............................................................................................................85 2-7. Evolutionary history of COX sequen ces inferred using the minimum evolution method with a distance optimality criterion.......................................................................87 2-8. Evolutionary history of COX seque nces excluding hagfish (hCOX) and lamprey (lCOX) sequences inferred using the mi nimum evolution method with a distance optimality criterion........................................................................................................... ..89 2-9. Evolutionary history of COX se quences inferred using a maximum liklihood optimality criterion........................................................................................................... ..90 9
PAGE 10
10 2-10. Evolutionary history of COX sequen ces excluding sequences from the hagfish (hCOX) and lamprey (lCOX) inferred using a maximum liklihood optimality criterion..............................................................................................................................92 2-11. Alternate likelihood ratio test (aLRT) scores for branch support of COX sequences.......94 2-12. Alternate likelihood ratio te st (aLRT) scores for branch support of COX sequences excluding hagfish (hCOX) and lamprey (lCOX) sequences..............................................96 2-13. Evolutionary history of COX sequences inferred using a Bayesian optimality criterion..............................................................................................................................97 2-14. Evolutionary history of COX sequen ces excluding hagfish (hCOX) and lamprey (lCOX) sequences inferred using a Bayesian optimality criterion....................................99 2-15. The 21 alternate tree topologies generated to test the effects of relocating the hagfish (hCOX) and lamprey (lCOX) sequences on the likelihood score (Ln L )........................100 2-16. Unrooted phylogeny showing the most likely scenario for COX evolution in the chordates based on several other analyses.......................................................................104 3-1. Euryhaline killifish ( Fundulus heteroclitus ) COX-1b mRNA expression following procedures outlined in Choe et al (2006)........................................................................113 3-2. Longhorn sculpin ( Myoxocephalus octodecemspinosus ) COX-1a and COX-2 mRNA expression following procedures outlined in Choe et al (2006).....................................114 3-3. Atlantic hagfish ( Myxine glutinosa ) COX (hCOX) mRNA expression following procedures outlined in Choe et al (2006) .......................................................................115 3-4. Sea lamprey ( Petromyzon marinus ) COX (lCOX) mRNA ex pression in various tissues following procedures outlined in Choe et al (2006)...........................................116
PAGE 11
Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science EVOLUTION OF CYCLOOXYGENASE IN THE CHORDATES By Justin Chase Havird May 2008 Chair: David Evans Major: Zoology Cyclooxygenase (COX) is the enzyme found in animals responsible for converting arachidonic acid into prostaglandins. These pros taglandins can then perform multiple functions including regulation of inflammation respons es, changes in vascular tone, and ion transport/osmoregulation. Before this thesis only two main forms of cyclooxygenase (COX-1 and COX-2) and one ancestral form ( Ciona COX) were known from the chordates. I sequenced COX genes from several hitherto uninvestigated chordates in order to determine a more comprehensive scenario for the evolutionary history of cyclooxygenase in the chordates. After designing primers, partial COX sequences were obtained from the Sea Lamprey ( Petromyzon marinus ) and amphioxus (Branchiostoma lancelolatum ). Complete COX sequences were obtained from the Atlantic Hagfish ( Myxine glutinosa ), euryhaline killifish ( Fundulus heteroclitus), and longhorn sculpin ( Myoxocephalus octodecemspinosus). These novel sequences along with other known COX sequences were then subjected to a variety of phylogenetic analyses using standa rd techniques. The results of these analyses suggest a complex history for the evolution of cyclooxygenase in the chordates. There appears to have been at least 3 evol utionary independent origins of COX in the chordates. The forms from the sea squirt ( Ciona intestinalis ) (COXa and COXb) represent the 11
PAGE 12
12 most ancestral forms, the forms from amphioxus represent another independent origin (COXc and COXd), and the vertebrate forms (COX-1 and COX-2) represent yet another origin, with the hagfish and lamprey as likely basal members. Fu rthermore, the teleosts appear to have duplicate COX-1 (COX-1a and COX-1b) or COX-2 (COX-2a and COX-2b) forms. These results agree with previous studies and reveal novel informa tion about COX forms in the chordates. More intense sampling of other chorda tes could reveal other novel orig ins of COX or contribute to a more robust understanding of COX evolution. Finally, the function of most non-mammalian COX forms remains to be investigated.
PAGE 13
CHAPTER 1 GENERAL INTRODUCTION Introduction Cyclooxygenase (COX) is the enzyme responsible for the rate limiting step that converts arachidonic acid to Prostaglandin G2 and Prostaglandin H2. Arachidonic acid is an essential fatty acid required by most mammals and can be obtaine d by the conversion of linoleic acid found in plants (occurring in herbivores). Some animals cannot convert linol eic acid into arachidonic acid and therefore must obtain it from other animals (obligate carnivores). COX oxidizes arachidonic acid to make the hydroperoxy endoperoxide Prostaglandin G2, which is then reduced to form the hydroxy endoperoxide Prostaglandin H2. Prostaglandin H2 can then be convert ed to the primary prostanoids (Pro staglandin I2, Prostaglandin D2, Prostaglandin E2, Prostaglandin F2 and Thromboxane A2) by a variety of enzymatic and non-en zymatic pathways. Therefore, the expression of COX can directly in fluence the amount of prostagla ndins that are synthesized due to the arachidonic ac id cascade (Vane et al ., 1998, Fig. 1-1). Prostaglandins are found in vertebrates, some invertebrates, and po ssibly even in plants (Rowley et al ., 2005; Ryan, 2000). In mammals, they are expressed in virtua lly all tissues and have a myriad of physiological functions, but they mainly regulate vascular tone in smooth muscle by acting as vasodilators and vasoconstr ictors. In addition, prostaglandins are key regulators of such diverse processes as salt and water homeostasis in the mammalian kidney (Harris and Breyer, 2001), protecti on of stomach lining (Kargman et al ., 1996), platelet aggregation in blood (F rster and Parratt, 1996), birth (Challis et al. 2002), and central nervous system function (Hopkins, 2007). They also may play prominent roles in pathological states such as fever, inflammation, and cancer (Vane et al ., 1998). 13
PAGE 14
Prostaglandins were first extracted from se men, prostate (hence the name prostaglandin), and seminal vesicles by Goldblatt and von Euler in the 1930s (von Euler, 1936) and were shown to affect blood pressure. The reac tion that catalyzes Prostaglandin G2 synthesis was first characterized in the 1970s (Ham berg and Samuelsson, 1973; Hamberg et al ., 1974), but it wasnt until 1988 that cyclooxygenase (COX) was first described as the enzyme responsible for endoperoxide synthesis by three separa te groups (DeWitt and Smith, 1988; Merlie et al ., 1988; Yokoyama et al ., 1988), all of which described a COX from sheep seminal vesicles which was later named COX-1. However, before COX-1 wa s isolated there was early evidence that multiple COX forms existed, based on prostaglandin activity in the rat brain and spleen (Flower and Vane, 1972). In the early 1990s, a sec ond COX form was cloned from mouse and chicken fibroblast cell cultures a nd was named COX-2 (Kujubu et al ., 1991; Xie et al ., 1991; O'Banion et al ., 1992). This discovery would later prove to be of major importance in the field of prostaglandin biology due to the development of dr ugs that could selectivel y inhibit the different forms of cyclooxygenase. Both forms catalyze the reactions that form prostaglandins from arachidonic acid, but early studies suggested that COX-1 was consti tutively expressed in many tissues, while COX-2 was an inducible isoform that was expressed in response to p hysiological stimuli (Funk, 2001). Because basal expression of COX-2 was low in some cell types and COX-2 was found to be expressed much more so than COX-1 in response to mitogens and cytokines, it was assumed that COX-1 was constitutively expressed at basal le vels and served a housekeeping function and that COX-2 was inducibly expressed to serve in infla mmatory functions (Kujubu et al ., 1991; Xie et al ., 1991; O'Banion et al ., 1992). However, this was later sh own to be an oversimplification (Funk, 2001). Notably, it was shown that COX-2 is constitutively expressed in the mammalian 14
PAGE 15
kidney, where it regulates blood flow and ion tran sport in the Loop of He nle (Harris and Breyer, 2001) and plays a role in cell su rvival in medullary interstiti al cells during dehydration (Yang et al ., 2002; Yang, 2003). Despite their potential functional differences in different tissues, COX-1 and COX-2 are biochemically very similar. Both are inhi bited by nonsteroid an ti-inflammatory drugs (NSAIDs), such as aspirin, due to competition with arachidonic acid for the active sites of the enzymes or irreversible acetyla tion (Vane, 1971). Both proteins in humans are ~600 amino acids and share about 63% sequence similarity with each other. However, the proteins originate from different genes on different chromosomes and en code different mRNA transcripts. COX-1 comes from a much larger gene (22 kilobases) than C OX-2 (8 kilobases). The amino acid residues that form the binding and catalytic sites (the functi onally important sites) as well as the adjacent amino acids are all identical in human COX-1 and COX-2, except for two substitutions. At positions 434 and 523, valine is substituted for isoleucine in COX-2 when compared to COX-1. These substitutions have been implicated as the most likely reason behind the biochemical differences between COX-1 and CO X-2. COX-1 appears to be more selective in terms of the fatty acids it will accept as substrates than COX-2 (Otto and Smith, 1995). COX-2 selective inhibitors have offered an insight into the bi ochemistry behind the different forms. These inhibitors can differentiate between COX1 and COX-2 with over 1000 fold specificity (Griswold and Adams, 1996). The rationale behind these selective inhibitors is that the smaller size of valine on position 523 in COX-2 allows th e inhibitor to bind to a side pocket in COX-2, but the larger isoleucine on position 523 of COX-1 appears to prev ent this binding. When this position is altered in both forms to the alternat e amino acid, the selectivity for inhibitors is reversed (Wong et al ., 1997; Gierse et al ., 1996; Guo et al ., 1996), suggesting th at this residue 15
PAGE 16
plays a critical role in differing biochemical activ ity. COX-2 selective inhi bitors were thought to be a major medical breakthrough due to the ability to stop disease induced inflammation without affecting normal prostaglandin levels produced by COX-1 and therefore causing minimal side effects. However, due to the constitutive expres sion of COX-2 in certain tissues (especially the kidney), COX-2 selective inhibitors may not be as affective as once thought. It is important to remember that COX-1 and COX-2 are very structur ally similar and that their main difference is due to expression. As outlined above, COX form and function have been extensively studied extensively in mammals, especially humans. However, non-ma mmalian COX forms and their functions have not been examined in great detail in other ch ordates. The goal of this research was to characterize COX evolution in the chordates, with specific attention on the more evolutionarily ancestral chordates. Describing ancestral COX form s is the first step in elucidating the evolution of COX function, which may lead to a better understanding of overall COX function, and even the development of new drugs. In this chapter, I give a brief overvie w of COX evolution and function in the chordates, concluding with a br ief summary of the research described in the following chapters. Cyclooxygenase Evolution Like mammals, most vertebrates appear to have two homologous copies of cyclooxygenase named COX-1 and COX-2. However, other C OX forms exist throughout the chordates, including variants of COX-1 and CO X-2 as well as forms representing independent origins of COX that are not closely related to COX-1 or COX-2. There are also COX forms in corals and other invertebrates (Jarving et al ., 2004), but the focus of this section is to review the forms found in the chordates. 16
PAGE 17
Mammals COX-1 and COX-2 sequences have been either cloned or derived from genomic sequences for most of the classic mammalian model species, includ ing mouse (Accession# NP_032995 and NP_035328), rat (Accession# NP _058739 and NP_058928), dog (Accession# NP_001003023 and NP_001003354), cow (Accession# XP_869575 and NP_776870), sheep (Accession# NP_001009476 and NP_001009432), a nd rabbit (Accession# NP_001076150 and NP_001075857). COX sequences also exist for rhesus monkey, guinea pig, pig, horse, and mink, but these are either partial sequences or are not available for both forms. A third COX variant has been proposed to occur in mammals (Flower and Vane, 1972) based on inhibitory studies (see below). This form is known as COX-3 or COX-1b because of its close similarity to COX-1. Recently, a potential COX-3 (Accession# AY547265) was cloned from mouse (Kis et al ., 2006) and rat (Accession# AY523672, Snipes et al ., 2005). However, these forms likely do not participate in prostaglandin synthesis (Kis et al ., 2006) and are not a separate COX form. Furthermore, a similar study using an artificially created human COX-1b found no inhibitory similarity to th e proposed COX-3 enzyme (Censarek et al ., 2006). Therefore, while a third, genuinely different COX variant may exist in mammals, its sequence remains elusive. Analysis of the recently completed gray short-tailed opossum genome yielded the predicted COX-1 and COX-2 forms (A ccession# XP_001370514 and XP_001375945), as well as two additional COX-1 forms, named COX-1b and COX-1b2 (Accession# XP_001370542 and XP_001370595). These forms are complete and di ffer in amino acid composition and length. These are the first COX-1 variants in a mammal that do not represent modified versions of normal COX-1 and although is has not been sugges ted, it is possible that one of these forms 17
PAGE 18
represents the elusive COX-3 va riant. Functional studies to determine how these forms differ from normal COX have not been published to date. Fishes Cyclooxygenase has been sequenced in several fishes, including the rainbow trout ( Oncorhynchus mykiss) (Zou et al ., 1999), brook trout ( Salvelinus fontinalis ) (Roberts et al ., 2000), Atlantic croaker ( Micropogonias undulates ), and zebrafish ( Danio rerio) (Grosser et al ., 2002). These fishes, along with genomic sequence data from the stickleback ( Gasterosteus aculeatus), green spotted puffer ( Tetraodon nigroviridis ), pufferfish ( Takifugu rubripes), and Japanese medaka ( Oryzias latipes ) suggest that fishes posses both COX-1 and COX-2 forms (Jarving et al ., 2004). Furthermore, COX-2 has recently been sequenced from European sea bass ( Dicentrarchus labrax) (Buonocore et al ., 2005) and euryhali ne killifish ( Fundulus heteroclitus ) (Choe et al ., 2006), supporting this prediction. A cyclooxygenase has also been sequen ced from the spiny dogfish shark ( Squalus acanthias ) (Yang et al ., 2002). However, this form was not designated as COX-1 or COX-2 because of its near equal identity to both forms (it is only slightly more similar to COX-1). This form was named sCOX by the authors and may repr esent an evolutionarily distinct form of cyclooxygenase found in the elasmobranchs. Ishikawa and colleagues have recently fu rther characterized COX evolution in the teleosts by demonstrating that both the zebrafish (Ishikawa et al ., 2007) and rainbow trout (Ishikawa and Herschman, 2007) have two functional COX-2 genes (named COX-2a and COX2b). In addition, by analyzing sequence databases, they concluded that not all teleosts have two COX-2 forms, but that the stickleback, green sp otted puffer, pufferfish, and Japanese medaka have two COX-1 forms instead (nam ed COX-1a and COX-1b) (Ishikawa et al ., 2007). They concluded that a genome duplication before the te leosts, and a subsequent loss of either one 18
PAGE 19
COX-1 or COX-2 form characterizes the spec ies mentioned (Ishikawa and Herschman 2007) (Fig. 1-2). This is supported by the apparent ge nome duplication events in teleosts (Crollius and Wessenbach, 2005). Other Vertebrates Although functional studies on prostagla ndin roles and cyclooxygenase are not uncommon in other vertebrates, cyclooxygenase sequences from non-mammalian and nonteleost vertebrates are lacking. Cyclooxygenase forms have been predicted from the chicken genome ( Gallus gallus ), suggesting that birds have both COX-1 and COX-2 forms (Urick and Johnson, 2006). Both forms have also been found in the African clawed frog ( Xenopus laevis ) and COX-2 has been found in the western clawed frog ( Xenopus tropicalis ), suggesting that both forms occur in amphibians (Klein et al ., 2002). Functional studies sugg est COX exists in reptiles as well (Seebacher and Franklin, 2003) and the recently completed Anolis carolinensis genome contains both COX forms. Taken together, the presence of COX-1 a nd COX-2 forms in the mammals, teleosts, birds, and amphibians suggest that both forms orig inated in the early vertebrates, possibly from the duplication of an ancestral COX gene. The presence of sCOX in the dogfish indicates that COX forms in the early vertebrate s may not have diverged (in sequence or possibly in function) to the extent that allows them to be labeled as COX-1 or COX-2, although it is predicted that two COX forms exist in all vertebra tes, since sCOX groups with COX1 and is not ancestral to the COX-1/COX-2 split (Fig. 1-2, unpublished phyl ogeny generated using the online program PHYML) (Guindon and Gascuel, 2003; Guindon et al ., 2005). Other Chordates Although the subphylum Vertebrata includes the majority of species in the phylum Chordata, two other subphyla are included in the chordates: the Urocho rdata (sea squirts or 19
PAGE 20
tunicates) and the Cephalochordata (amphioxus or lancelets). These ot her subphyla are basal to the vertebrates and represent evolutionarily dist inct monophyletic groups (Fig. 1-3). There is some debate as to whether the urochordates or cephalochordates represent the base of the chordate phylogeny, although urochordates have been generally been considered basal (Pough et al ., 2005). Interestingly, the genomes of sea squirts ( Ciona intestinalis and Ciona savignyi ) predict COX genes that are not homol ogous to COX-1 and COX-2. These forms (named COXa and COXb) are basal to vertebrate COX forms and indi cate an evolutionarily distinct branch in COX evolution (Jarving et al 2004) (Fig. 1-4). This separate bran ch is situated between the COX of corals and the vertebrate COX fo rms (Fig. 1-4). This suggests that the COX-1 and COX-2 forms originated sometime after the origin of the sea squirts but before the origin of the elasmobranchs (around 550-600 mya). Cyclooxygenase Function in Mammals Cyclooxygenase was first described from mamm als, and due to its medical importance, its function has been intensel y examined in mammals, specifically humans. Since COX stimulates the production of pros taglandins, COX function is intrin sically linked to the roles that prostaglandins play in everythi ng from water balance to reproduc tion. Here, I review the major functions of COX-1 and COX-2 in various ma mmalian physiological systems (modeled after Vane et al ., 1998), but it should be noted that ma mmalian COX studies are ongoing and that COX likely plays many roles not described here. Gastrointestinal Tract In the stomach, prostaglandins play a cy toprotective role, guarding the stomach lining against digestive acids and enzymes. The mech anism behind this protection is an increased mucosal blood flow caused by the vasodilating properties of prostagl andins (specifically 20
PAGE 21
Prostaglandin I2). It has generally been accepted that these prostaglandins are synthesized by COX-1, although COX-2 is present in small amounts (Kargman et al ., 1996). This is supported by studies showing that irradi ated mice have further redu ced numbers of crypt cells (cytoprotective mucosal cells) when indome thacin (a non-specific COX inhibitor) was administered, but not when a COX-2 sele ctive inhibitor was administered (Cohn et al ., 1997). COX-2 expression in the esopha gus of patients with gastro esophageal reflux disease was upregulated in response to acid e xposure, suggesting a protective function in the distal esophagus (Lurje et al ., 2007). This may indicate th at throughout the gastrointe stinal (GI) track, COX-2 is inducibly expressed in response to inflammation responses. Furthermore, throughout the GI track, inflam matory mediators such as COX may also serve to repair damaged mucosa (Martin and Wallace, 2006). While it seems that COX-1 is responsible for baseline prostagl andin expression in the GI tr act, COX-2 and not COX-1 seems to be more involved in response to GI tract cancers (Ristimki et al ., 1997). However, recent studies on patients with gastric cancer from Dalian, China show ed that COX-2 expression was infrequent in gastric cancers due to hypermethylation and that a COX-2 focused strategy for treatment would likely be ineffective (Huang et al ., 2006). The Kidney Prostaglandins can modulate blood flow in the kidney, as in other organs (Vane and Botting, 1994). In addition, COX-2 has been shown to play a major role in ion regulation and water balance in the mammalian kidney (Harris et al ., 1994). It has also been shown that the COX-2 selective inhibitor SC-58236 and seve ral non-selective COXinhibiting NSAIDs (including sulindac, ibuprofen, a nd indomethacin) cause medullary interstitial cell (MIC) death in the kidney, but the COX-1 selective inhib itor SC-58560 was 100-fold less potent for inducing MIC death (Hao et al ., 1999). COX-2 knockout mice also develop terminal nephropathy 21
PAGE 22
(abnormal kidney function) which results in death by the third month (Morham et al ., 1995). These results suggest that COX-2, and not COX1, plays a critical role in kidney function via MIC cell survival. This is contrary to the or iginal description of COX-1 as the constitutively expressed form responsible for maintaining hom eostasis and COX-2 as the inducible form responsible for inflammatory responses. In the kidney, COX-2 is present in a subset of tubular epithelial cells located in the cortex and outer medulla (Vio et al ., 1997). These cells ar e part of the medullary thick ascending limb (MTAL), the part of the nephron that regulates extracellular fluid vol ume by establishing the osmotic gradient in the medulla that concentrates urine (Hebert et al ., 1981; Hebert and Andrioli, 1984). In vitro MTAL preparations have demonstrat ed high levels of Prostaglandin E2 (PGE2) (Ferreri et al ., 1984; Hirano et al ., 1986; Lear et al ., 1990) which has been shown to inhibit basolateral Na+/K+-ATPase (NKA) and apical Na+/K+-2Clcotransporter (NKCC) in the MTAL (Kaji et al ., 1996; Stokes and Kokko, 1976; Wald et al ., 1990). These proteins are critical for NaCl reabsorption in the MTAL. Furthermore, the COX-2 selective inhibitor NS-398 has been shown to prevent tumor necrosis factor (TNF) mediated PGE2 production in cultured MTAL cells (Ferreri et al ., 1999). TNF is a cytokine expressed in the MTAL that may be an important mediator of ion transport by inhibiting ion uptake via a prostanoid dependant mechanism (Escalante et al ., 1994; van-Lanschot et al ., 1991). Taken as a whole these results show that COX-2 may influence ion uptake via TNF mediated prostaglandi n production in the MTAL. Therefore, COX-2 is involved in cell survival and ion transport in the mammalian kidney. This finding is critical for treatment prot ocols using COX inhibitors. NSAIDs are the most common drugs used to treat pain and inflammation with over 30 million uses per day worldwide (Singh and Triadafilopoulos, 1999) Although the negative side effects 22
PAGE 23
(hypertension, renal ischemia, a nd GI toxicity) associated with chronic NSAID use have been known for some time, the discovery of COX-2 se lective inhibitors sugge sted a treatment for inflammation without causing disruption of nor mal COX function (thought to be mediated by COX-1). The role of COX-2 in the kidney has reduced the supposed power of these new drugs and negative side effects associated with C OX-2 selective inhibitors are now well documented (see Harris and Breyer, 2006 for a review). Furthermore, in 2005 the Food and Drug Administration (FDA) required additional warning la beling of celecoxib to highlight side effects and required Pfizer to remove valdecoxib from the market, both of which are selective COX-2 inhibitors. Platelets Interestingly, COX-1 seems to be the onl y form expressed in mammalian platelets. COX-2 was not detected in human platelets using western blotting techniques (Reiter et al ., 2001) or in canine platelets using northe rn blotting techniques (Kay-Mugford et al ., 2000). In the platelet, COX-1 is involved in platelet aggregation (clott ing) through the production of Thromboxane A2 (TXA2). When inhibited with aspiri n, COX-1 mediated production of TXA2 is irreversibly inhibited, causing redu ced platelet aggregation (blood th inning). This effect occurs about an hour after oral ingestion of even low doses of aspirin and can inhibit platelet function for several days after a dose (Vane et al., 1998; Vane, 1971). Inhibiti on of COX-1 by resveratrol and other components of red wine is responsible for the anti-plate let activity associated with the cardiovascular benefits of red wine consumption (Szewczuk and Penning, 2004). Reproduction Cyclooxygenase and the prostaglandins it produc es play critical roles in almost every aspect of reproduction in mammals, although its ro le in pregnancy has been studied the most 23
PAGE 24
intensely. Both isoforms are likely involved in reproductive processes, with COX-1 expression constitutive and COX-2 expres sion inducible (Chakraborty et al ., 1996). In a recent study, COX-2 was shown to contri bute to fertility in male mice (Balaji et al ., 2007). COX-2 expression was intense in the epithe lial cells of mice vas deferens and the COX-2 specific inhibitor nimesulide was shown to decrea se sperm motility six hours after administration (Balaji et al ., 2007). Interestingly, although prostagl andin levels initially dropped after nimesulide administration, they tended to ri se after sustained C OX-2 inhibition (Balaji et al ., 2007). This suggests that COX-1 expression may recover COX func tion or at least contributes heavily to prostaglandin production in the vas defere ns. A similar situation exists with respect to human female fertility, in which COX-1 expres sion can recover function in cases where COX-2 is lost or inhibited (Wang et al ., 2004). It also seems that COX-2 is critical for follicle development and ovulation in women and that COX inhibitors can have se vere effects on female fertility (Sirois et al ., 2004). Prostaglandins are also important during pr egnancy and labor. Generally, COX-1 is expressed more than COX-2 during normal stages of pregnancy, and COX-1 may play an active role in maintaining a healthy pregnancy (Trautman et al ., 1996). COX-1 is expressed at high levels is the fetal brain, kidne ys, heart, and lungs (Gibb and Sun, 1996). However, both COX-1 and COX-2 are expressed in the uterine epitheliu m during early pregnancy and may play roles in implantation and placental formation (Chakraborty et al ., 1996). This was recently confirmed for COX-2 in rats (Diao et al ., 2007). COX-2 may play a domi nant role in inducing labor. During labor, contractions are stimulated by prostaglandins a nd therefore any effect causing increased COX activity will induce labor. Up-regul ation of COX-2 during intrauterine infection may increase prostaglandins to the poin t of causing premature labor (Spaziani et al ., 1996). 24
PAGE 25
Furthermore, a similar up-regulation in sheep may reduce progesterone levels to the point that pregnancy cannot be maintained (McLaren et al ., 1996). However, it was recently shown that during the course of pregnancy there is a gr adual change in COX expression, with COX-1 dominating during early pregnancy and COX-2 dominating in later stages (Johnson et al ., 2006). It was concluded that while COX-2 is expressed at high levels in late st ages of pregnancy, it is not up-regulated suddenly to induce labor (Johnson et al ., 2006). COX inhibitors can halt premature labor by inhibiting pr ostaglandin production (Sawdy et al ., 1997). However, as in treatment strategies that use COX inhibitors for other problems, there are negative side effects associated with the kidneys: re nal clearance and glomerular filtration were reduced during the first weeks of life in babies that were expos ed to COX inhibitors to stop premature birth (Allegaert et al ., 2006). As in other cases, COX-2 selectiv e inhibitors were thought to eliminate these negative side effects but have not proven more effective than non-specific NSAIDs (Olson 2005). Nervous System COX-1 and COX-2 are both expressed in the mammalian nervous system and may participate in nerve transmission, sensory processi ng, thermoregulation, and pain effects. As in other systems, the use of COX inhibitors to tr eat neural diseases (such as Alzheimers) is prevalent, but is accompanied by undesired side effects that are not prevented by using COX-2 selective inhibitors. The brain is one of the few organs that c onstitutively expresses COX-2 and contains high amounts of COX-2 in the cortex, hippocampus hypothalamus, and spinal cord (Breder et al ., 1995), whereas COX-1 is abundantly expressed in the forebrain, where it may be involved in sensory information processing (Yamagata et al ., 1993). COX-2 is expressed in both neuronal 25
PAGE 26
and non-neuronal cells in the brain and is up-regu lated in response to ab normal nerve activity, suggesting a role in nerve transmission (Breder et al ., 1995; Breder and Saper, 1996). COX-2 in the spinal cord has been shown to modulate pain responses (nociception) in both humans and rats due to peri pheral pain stimulation (Martin et al ., 2007; Gardiner et al ., 1997; Willingale et al ., 1997). These results are expected si nce prostaglandins are generated at the ends of sensory neurons during in flammatory pain (Ferreira 1972; Woolf et al ., 1997). However, the up-regulation of COX-2 over COX-1 in the spinal cord during inflammatory pain suggests a role during pain processing in the centr al as well as the peripheral nervous system. This up-regulation has been detected via increases in COX-2 mRNA (Gardiner et al ., 1997; Beiche et al ., 1996) in the spinal cord of rats during inflammatory stimuli. Furthermore, COX-2 selective inhibition also affected the withdrawal reflex of ra ts when stimulated using a noninflammatory, electrical stimulus (Willingale et al ., 1997). This supports the general role of COX-2 in nociception in the spinal cord. COX-2 also plays a thermoregulatory role in the brain by increasing prostaglandin production in the hypothalamus during periods of fever. Injection of bacterial lippopolysaccarides (LPS) into mammals causes a monophasic fever characterized by a single rise in core body temperature. It has long been known that PGE2 levels increase in the hypothalamus during fever and likely mediat e the response (Ivanov and Romanovsky, 2004). Early studies suggested that an up-regulation of COX-2 was the cause of increased PGE2 levels during fever (Cao et al ., 1997). This initial hypothesis has been confirmed, recently in COX-2 deficient mice which do not respond to LPS injec tion, in contrast to CO X-1 deficient mice or control mice (Steiner et al ., 2005). 26
PAGE 27
Cyclooxygenase Function in Non-Mammals Although COX has been extensively studied in mammals due to its role in inflammation and the correlating drug mark et surrounding COX inhibition, studies in non-mammals are lacking. Unfortunately, it seems that alt hough non-mammals posses the mammalian COX-1 and COX-2 forms, these forms may be substantially different (both in structure and function) from their mammalian counterparts (Grosser et al ., 2002) and may also represent independent evolutionary events, such as duplications of an isoform (teleosts) or independent origins (sea squirts). In this section, I review the major studies i nvestigating COX function in nonmammalian chordates. Fishes Cyclooxygenase has likely been studied in the fishes more than other non-mammals due to their status as model organisms. As noted above, COX forms have been found in several fishes and along with genomic sequence data there is a general consensus that fishes possess both COX-1 and COX-2 forms that are homologous to those found in mammals. However, there is recent evidence that teleosts also possess an additional form of COX-1 or COX-2 (Ishikawa et al ., 2007; Ishikawa and Herschman, 2007) and some functional data also exist for these forms. Osmoregulation In fishes, the gills are the dominant site of acid/base regulation, nitrogenous waste secretion, gas exchange, a nd ion transport (Evans et al ., 2005). Several studies have investigated the osmoregulatory role of COX in th e gills of the euryhaline killifish, Fundulus heteroclitus This teleost can withstand instant salinity transfer between full strength seawater and fresh water without any apparent physiol ogical stress (Wood and Laurent, 2003). One study showed that short circuit current across th e opercular epithelium (a ti ssue with known ion transport capabilities) was reduced by using a non-specific COX inhibitor (Evans et al ., 2004). This 27
PAGE 28
suggests that COX-2 plays a role in ion transport in fishes that ma y be similar to its function in the kidneys of mammals. Further studies suppor ted this initial hypothesis by showing that COX2 is expressed most abundantly in the gill, opercu lar epithelium, and kidney of the killifish (Choe et al ., 2006), tissues that are known to be involved in ion transport in fishes. This is supported in the zebrafish, which shows the highest le vels of COX-2 in the gills (Grosser et al ., 2002). Furthermore, COX-2 was localized in mitochondrionrich cells in the gills of killifish, the main sites of ion uptake and secr etion in the gills (Choe et al ., 2006). Finally, COX-2 expression was shown to significantly increas e following either hypotonic or hype rtonic salinity transfers, suggesting COX-2 may play a role in maintaining cell homeostasis or promoting cell survival during periods of osmotic shock (Choe et al ., 2006). Taken together these results strongly suggest that COX-2 in the gills of teleosts acts in the same way as COX-2 in the kidneys of mammals, including regulating ion transpor t and promoting cell survival. Cyclooxygenase has also been shown to play an osmoregulatory role in the rectal gland of the dogfish shark, Squalus acanthias. Sharks rely on the rectal gland rather than the gills for regulation of salt secretion and a COX form (sCOX) has been cloned from the rectal gland (Yang et al ., 2002). This form was expressed most abundan tly in the rectal glan d of the shark, where PGE2 production was also high (Yang et al ., 2002). Finally, using a COX-2 specific inhibitor, vasoactive intestinal peptide mediated chloride se cretion decreased in the rectal gland, but then recovered following removal of the inhibitor (Yang et al ., 2002). Even though sCOX is slightly more similar to COX-1 of mammals than COX-2, th is result can be explained by the presence of valine at position 523 instead of isoleucine (and t hus conferring COX-2 inhibitory properties). This result suggests that sCOX plays a role in ion transport in the rectal glands of sharks that may be similar to the gills of teleosts or the kidneys of mammals. Th e osmoregulatory role of 28
PAGE 29
COX in fishes needs to be examined in greater de tail using other species and diverse techniques. This may be highly feasible because COX seque nces exist for a wide range of teleosts. Reproduction As in mammals, COX has been shown to play a role in reproduction in fishes. By using the non-specific COX inhibitor indomethacin, it was shown in the Atlantic croaker ( Micropogonias undulatus ) that COX pathways may play a ro le in the maturation of ovarian follicles and ovulation through prostaglandin forma tion, although other proteins may play a more dominant role (Patio et al ., 2003). Results from the European sea bass also indicate a similar role for prostaglandins in ovulation, with indom ethacin inhibiting follic le maturation (Sorbera et al ., 2001). However, it has been shown in the brook trout that indomet hacin does not block ovulation (Goetz et al ., 1989), although it does in other fish species (Goetz et al ., 1991). This apparent loss of function in the brook trout may be explained by changes in the levels of COX-1 and COX-2 during ovulation. It was shown that COX-1 levels remain ed constant and high during ovulation but that COX-2 le vels did not increase prior to ovulation as they do in mammals (Roberts et al ., 2000). This is supported by data from the zebrafish which show high levels of COX-1 but not COX-2 in the ovaries (Grosser et al ., 2002). This suggests that COX-2 function in the reproduction of the brook trout (and perhaps other fishes) is different than in other vertebrates. In the Japanese medaka, it was show n that low chronic levels (perhaps comparable to those found in waste water) (Metcalfe et al ., 2003) of the non-specific COX inhibitor ibuprofen causes altered reproduction by decr easing the number of spawning events but increasing the number of eggs per spawning event (Flippin et al ., 2007). This conforms to delayed pregnancies associated with NSAID use in mammals. These results indicate that COX likely plays a role in fish reproduc tion, although it may vary among species. 29
PAGE 30
Differential functions of novel forms Recently, the discovery of novel COX2 forms in the zebrafish (Ishikawa et al ., 2007) and rainbow trout (Ishikawa and Herschman, 200 7) and analyses of genomic sequence data indicates that all teleosts ma y posses at least three COX form s: either COX-1a, COX-1b, and COX-2 or COX-1, COX-2a, and COX-2b. Prelim inary data on the described COX-2 forms suggests that they are functionally different fr om each other. In the zebrafish, COX-2a and COX-2b are expressed at differe nt levels in the gill, kidney, and other tissues and respond differently to 12-O-tetradecanoylphorbol-13-ace tate (TPA, which is known to induce COX-2 expression in mouse) (Ishikawa et al ., 2007). A similar result was found in the rainbow trout, with differential expression of the two forms due to TPA (Ishikawa and Herschman, 2007). Interestingly, COX-2a and not COX-2b was upregulated in response to lippopolysaccarides (Ishikawa and Herschman, 2007). Clearly, the discovery of these new forms indicates a new chapter in studying COX function and further studies are needed to de termine the functional differences among the teleosts forms of COX. Other functions Cyclooxygenase has also been implicated in determining body plan development during embryonic stages in the zebrafish (Grosser et al ., 2002; Cha et al ., 2005; Cha et al., 2006a; Cha et al ., 2006b; Yeh and Wang, 2006). Briefly, COX may function to signal cell motility during gastrulation (Cha et al ., 2006b), and COX-1 function has been s hown to participate in vascular tube formation during development (Cha et al ., 2005). COX-2 expression has also been shown to increase in response to lippopolysaccarid es in the rainbow trout (Brubacher et al ., 2000; Holland et al ., 2002), suggesting an immunoregulatory role for COX-2 in fish as in mammals ( e.g ., fever induction). Clearly, the function of COX in fishes is just beginning to be explored, 30
PAGE 31
and further studies will likely indicate more sim ilarities and differences with COX functions in mammals. Amphibians As in other vertebrates, there is evid ence that COX is involved in ovulation and reproduction in amphibians. Indomethacin was shown to inhibit oocyte ovulation in the frog ( Rana temporaria ), while Prostaglandin F2 was shown to promote ovulation (Skoblina et al ., 1997). This suggests another medi ator may be influencing PGF2 production in the frog, but this contrast between seemingly relate d proteins has not been confirme d in other species. COX may also function in testosterone synthesis via PGE2 in the crested newt, Triturus carnifex (Gobbetti and Zerani, 2002). Ion and water transport have also been studied and non-specific COX inhibitors reduce short circuit current and voltage potential across the frog skin, bladder, and corneal epithelium (Shakhmatova et al ., 1997; Carrasquer and Li, 2002). As in reptiles, fever and thermoregulatory changes are behaviorally induced in amphibian s. Injection of lippopolysaccarides (LPS) normally causes a behavior ally induced fever in toads. However, when indomethacin is administered, the fever indu ced by LPS is completely blocked in the toad ( Bufo paracnemis) (Bicego et al ., 2002). This indicates that the COX pathway plays a role in the behaviorally induced fever caused by LPS and that this response has an ancient origin that may be conserved in all tetrapods. Reptiles Cyclooxygenase has not been st udied extensively in the lizar ds, crocodilians, snakes, and turtles although some data suggest prominen t roles in thermoregulation. Reptiles undergo behavioral thermoregulation to adjust their body temperatures, and this process is amplified by changes in heart rate. Specifically, an increase in heart rate (heart-rate hysteresis) accompanies an increase in temperature. Prostaglandins and cyclooxygenase were shown to play a role in this 31
PAGE 32
process in the lizard Pogona vitticeps Inhibition of COX-1 and COX2 resulted in no increased heart rate during heating (Seebacher and Fra nklin, 2003). Furthermore, administration of Prostaglandin F2 and Prostaglandin I2, but not Thromboxane B2, caused an increased heart rate (Seebacher and Franklin, 2003). However, this effect was not seen in crocodiles ( Crocodylus porosus ) (Seebacher and Frank lin, 2004) or the lizard Phrynocephalus przewalskii (Liu et al ., 2006), suggesting that thermore gulatory functions of COX via changes in heart rates may vary between closely related species. Interestingly, COX was also implicated in tail regeneration in th e house lizard. Twenty days after tail removal (the time period associated with tissue differentiation), an increase in cyclooxygenase activity occurred along with the appearance of endogenous Prostaglandin E2, which may signal a cascade resulting in tissue differentiation (Jayadeep et al ., 1995). Finally, indomethacin was used in the turtle bladder to sh ow that prostaglandins likely do not play a role in the inhibition of Na+ or H+ transport due to high intracellu lar calcium (Arruda, 1982). COX function in reptiles obviously needs further stud y as many functions associated with COX in mammals have yet to be investigated in re ptiles and there have been no COX targeted sequencing efforts. Birds Cyclooxygenase and prostaglandin function have been studied somewhat in model avian species, specifically chickens, wh ere there is support that COX is involved in similar processes as in mammals, ( e.g., reproduction). COX-2 ( but not COX-1) and prostaglandin activity was shown to increase with administration of transfor ming growth factor in granulosa cells of white leghorn hen follicles, suggesting that COX-2 a nd not COX-1 plays a role in granulosa cell proliferation during follicular development (Li et al ., 1996). It was also shown that indomethacin reduced the prolifer ation of granulosa cells (Jin et al ., 2006). When using COX-1 32
PAGE 33
and COX-2 specific inhibitors, it was found that both types reduced granulosa cell proliferation, but that COX-2 specific inhibitor showed a strong er effect, supporting orig inal studies that COX2 is the dominant form in granulosa cell proliferation (Jin et al ., 2007). However, COX-1 was found to be expressed highly in the brain and seminal vesicle of the chicken (Reed et al ., 1996). COX-1 (but not COX-2) is also up-regulated in ovarian cancers occurring in hens, suggesting a target for treatment (Urick and Johnson, 2006). C OX inhibitors, both specific and non-specific, were shown to decrease sperm motility in the domes tic turkey, demonstrating a further role for COX in bird reproduction (Kennedy et al ., 2003). As in mammals, COX-2 has been implicated with the detecti on and persistence of peripheral inflammatory pain in chickens. The number of C OX-2 containing neurons increased significantly in laminae under inflammatory c onditions, but not under control conditions, 12-24 hours after injection of Fr eunds adjunvant (Yamada et al ., 2006). This suggests that the numbers of COX-2 containing neurons are re lated to inflammatory pain as in mammals. It appears that COX may not play a thermoregulator y role in birds as it does in mammals because indomethacin did not cause any change in rectal temperat ure during heat stress in chickens (Furlan et al ., 1998). However, COX and its prostaglandins may play a role in memory retention in birds, because inhibitors produce amnesic effects (Hlsc her, 1995). Finally, COX-2 is expressed in the kidneys during development in chickens (Mathonnet et al ., 2001) and at high levels in the kidneys of adults (Reed et al ., 1996). Study Overview As outlined above, the family of cyclooxygenase genes has an interesting and functionally important history in the vertebrates. Drugs deri ved from cyclooxygenase function are among the most important and widely used tr eatment methods in medicine. However, COX function and evolution is not fully understood in the ancestral chor dates, specifically the teleosts, 33
PAGE 34
sharks, hagfish, lampreys, cephalochordates, a nd urochordates. Indee d, the origin of the mammalian forms of COX is unclear since the sea squirt possesses evolutionarily distinct COX forms. Resolving the evolution of COX in the chordates could prove invaluable in determining new targets for drug development (Searls, 2003). In Chapter 2, I describe the cloning and characterization of 9 new COX genes in the chordates. These include COX-1 and COX-2 form s from teleosts as well as novel forms from the hagfish, lamprey, and amphioxus. In phylogenet ic analyses, these forms indicate that several origins of COX occurred during chordate e volution, with the mammalian COX-1 and COX-2 forms likely originating with the hagfish (craniates). Data from the teleosts supports the findings of Ishikawa and colleges that teleosts in the Acanthopterygii posses two COX-1 forms and one COX-2 form, whereas earlier teleosts such as the zebrafish and rainbow trout posses two COX-2 forms and one COX-1 form. Furthermore, analyses of protein alignments from novel sequences indicate conserved and derived functional resi dues and areas between novel COX forms and the COX-1 and COX-2 forms of mammals. In Chapter 3, I summarize the findings of th is research and suggest future avenues of research based on the results presented here. I also present preliminary data investigating the possible function of cyclooxygenase in these ancestral chordates. 34
PAGE 35
Arachidonic Acid Prostaglandin G2 COX-1 or COX-2 Prostaglandin H2 Prostaglandin D2 Prostaglandin E2 Thromboxane A2 Thromboxane B2 Prostaglandin I2 6-keto-Prostaglandin F1 COX-1 or COX-2 Prostaglandin F2 Figure 1-1. Cascade of reactions depicting th e conversion of arachidonic acid to the primary prostaglandins via catalyzation with either cyclooxygenase (COX) form. Arachidonic acid is obtained from the conversion of linolet ic acid found in plants or directly from other animals. It is then oxidi zed by COX to produce Prostaglandin G2 which is subsequently reduced by COX to form Prostaglandin H2. Prostaglandin H2 can then be converted into the other prostaglandi ns via other enzymes (Vane et al., 1998). 35
PAGE 36
COX-2a Danio COX-2a Oncorhynchus COX-2b Danio COX-2b Oncorhynchus COX-2 Acanthopterygians COX-1a Acanthopterygians COX-1b Acanthopterygians COX-1 Danio COX-1 Oncorhynchus sCOXSqualus Figure 1-2 General phylogeny repr esenting the different COX forms in the fishes. Based on recent studies, it appears that teleosts have multiple forms of COX-1 or COX-2 (Ishikawa et al ., 2007; Ishikawa and Herschman, 2007). It is hypothesized that with the origin of the teleosts, COX-1 and C OX-2 underwent duplication, but that in one lineage (represented by the zebrafish and rainbow trout) one COX-1 form was lost and in another lineage (represented by the Acanthopterygians) one COX-2 form was lost. Also included is the form cloned from the spiny dogfish (Yang et al ., 2002) which loosely groups with COX-1 forms, although it shares near equal identity with COX-1 and COX-2. This phylogeny was simplified from an unpublished phylogeny generated using the online program P HYML (Guindon and Gascuel, 2003; Guindon et al ., 2005). Abbreviations: Danio = Danio rerio (zebrafish), Oncorhynchus = Oncorhynchus mykiss (rainbow trout), Squalus = Squalus acanthias (spiny dogfish). The Acanthopterygians clade is represented by the stickleback ( Gasterosteus aculeatus), green spotted puffer ( Tetraodon nigroviridis ), pufferfish ( Takifugu rubripes ), and Japanese medaka ( Oryzias latipes ). 36
PAGE 37
Figure 1-3 Accepted evolution of ancestral chordates (Pough, et. al 2005). Black lines show relationships only; they do not indicat e times of divergence nor the unrecorded presence of taxa in the fossil record. Ba rs shaded red indicate ranges of time when the taxon is known to be present. Th e subphylum Vertebrata includes the Myxinoidea (hagfishes), Petromyzontoidea (lampreys), Teleoste i (teleosts), and tetrapods. There is some controversy over wh ether the Urochordates (sea squirts) or the Cephalochordates (lancelets) represent the most basal chordate lineage. Also, it has historically been suggested that the hagfish and lamprey constitute a monophyletic group (the Agnathans or Cyclostomes). 37
PAGE 38
38 Invertebrate COX forms COXb Urochordates COXa Urochordates COX-2 Vertebrates COX-1 Vertebrates Figure 1-4 Evolution of COX forms in the chordates, based on Jarving et al (2004). The Urochordates (represented by Ciona intestinalis and Ciona savignyi ) have COX forms that represent an evolutionary dis tinct branch ancestral to the COX-1 and COX-2 forms of the vertebrates. In this tree, the vertebrates are represented by the teleosts, mammals, Xenopus, and Gallus
PAGE 39
CHAPTER 2 CYCLOOXYGENASES IN THE ANCEST RAL CHORDATES: SEQUENCE AND PHYLOGENETIC ANALYSES Introduction Cyclooxygenase (COX) is the enzyme that catalyzes the oxida tion and subsequent reduction of arachidonic acid to form Prostaglandin G2 and Prostaglandin H2 (PGH2). PGH2 can then undergo additional reactions to produce the primary prostagla ndins, which act as vascular tone regulators in the verteb rates. Prostaglandins and C OX participate in a variety of physiological functions in the vertebrates, incl uding inducing fever, maintaining pregnancy, and regulating ion transport in the kidneys (Steiner et al ., 2005; McLaren et al ., 1996; Harris et al ., 1994). These functions have been ex tensively studied in mammals (Vane et al ., 1998), but comparatively little functional data exist for ot her animals. However, COX has been sequenced in several evolutionarily more ancestral chordates, particularly the teleosts (Zou et al ., 1999; Roberts et al ., 2000; Choe et al ., 2006). Based on functional studies of COX in the teleosts, it seems that some functions are conserved (Choe et al ., 2006; Sorbera et al ., 2001; Brubacher et al ., 2000; Holland et al ., 2003) while others may be altered in some species (Goetz et al ., 1989) or novel (Cha et al ., 2006b). In mammals there are two main forms of cyclooxygenase. The first form was isolated from sheep seminal vesicles in 1988 and later named COX-1 (DeWitt and Smith, 1988; Merlie et al ., 1988; Yokoyama et al ., 1988). A second form was isol ated from mouse and chicken fibroblast cell cultures in the early 1990s and named COX-2 (Kujubu et al ., 1991; Xie et al ., 1991; O'Banion et al ., 1992). Originally, COX-1 was considered to be a constitutive form that maintained normal cell functions and COX-2 was c onsidered to be an inducible form that was up-regulated in inflammatory res ponses (Funk, 2001). However, stud ies have shown that this is an oversimplification and COX-2 is expressed constitutively in the brain (Breder et al ., 1995) 39
PAGE 40
and kidneys (Harris and Breyer, 2001) of mammals. This has led to the abandonment of COX-2 selective inhibitors ( e.g., celecoxib and valdecoxib) which were thought to treat inflammatory pain without the negative side effects associ ated with non-selective COX inhibition. Although structurally and biochemically similar, COX-1 a nd COX-2 vary in expression in the vertebrates and participate in different functions. For example, COX-2 but not COX-1 plays a role in granulosa cell proliferation in chicken follicles (Li et al ., 1996; Jin et al ., 2007). Although the amino acid sequences of COX-1 and COX-2 share about 63% similarity, the presence of valine in COX-2 at position 523 instead of isoleucine is thought to be re sponsible for their differences in substrate selectivity and sensitivity to specific inhibito rs (Otto and Smith, 1995). Inhibitory studies in mammals have suggested a COX-3 form also exists, but it has not been characterized (Kis et al ., 2006; Censarek et al ., 2006). Analyses of genomic sequence and targeted cloning efforts have demonstrated that, like mammals, other vertebrates have COX-1 and COX-2 forms (Jarving et al ., 2004). However, some variation exists in the COX-1 and COX-2 dichotomy, notably in the more evolutionarily ancestral chordates. Ishikawa and colleagues (Ishikawa et al ., 2007; Ishikawa and Herschman, 2007) have recently shown that teleosts posse s three forms of COX. The zebrafish ( Danio rerio) and the rainbow trout ( Onchorhynchus mykiss ) both posses two COX-2 forms (named COX-2a and COX-2b) and one COX-1 form whereas the stickleback (Gasterosteus aculeatus), green spotted puffer ( Tetraodon nigroviridis ), pufferfish ( Takifugu rubripes ), and Japanese medaka ( Oryzias latipes) posses two COX-1 forms (named COX-1a and COX-1b) and one COX-2 form. They also showed that COX-2a and COX-2b ar e functionally different in the zebrafish and rainbow trout. Also, a COX form (named sC OX) has been cloned from the spiny dogfish ( Squalus acanthias ) and groups with COX-1 in phylogene tic analyses, but shares strong 40
PAGE 41
sequence identity to COX-2 as well (Yang et al ., 2002). Furthermore, sea squirts (subphylum Urochordata) possess two forms of COX (named COXa and COXb) that do not correspond to the COX-1 or COX-2 of vertebrate s and represent an ancestral or igin of COX in the chordates (Jarving et al ., 2004). The current view of the evolution of cyclooxygenase in the chordates represents an interesting but incomplete account of this gene family, both from a phylogenetic and functional standpoint. Therefore, the first goal of this study was to pinpoint the orig in of COX-1 and COX2 in the chordates by searching for COX forms in the sea lamprey ( Petromyzon marinus), Atlantic hagfish ( Myxine glutinosa ), and amphioxus ( Branchiostoma lancelolatum ). A second goal was to confirm the findings of Ishikawa and colleagues by searching for the 3 forms of COX in the euryhaline killifish ( Fundulus heteroclitus ) and longhorn sculpin ( Myoxocephalus octodecemspinosus ). It was hypothesized that both te leosts would posses COX-1a, COX-1b, and COX-2 since both are in the Acanthopterygii clade of Teleostei and all other acanthopterygiians posses COX-1a, COX-1b, and COX-2. Here, I re port the cloning of 9 novel COX sequences from the species mentioned above. Sequence an d phylogenetic analyses s uggest that ancestral COX forms may have had similar functions as COX-1 and COX-2 in mammals, which likely originated with the origin of the vertebrates, although the hagf ish and lamprey may have COX-1 and COX-2 forms that re present novel lineages. Materials and Methods Animals and Holding Conditions All procedures were approved prior to be ginning the experiment by the University of Florida Institutional Animal Care and Use Committee (IACUC). Euryhaline killifish ( Fundulus heteroclitus ) were captured from Northeast Creek n ear the Mount Desert Island Biological Laboratory (MDIBL), Salisbury Cove Maine using minnow traps. Th ey were transported to the 41
PAGE 42
MDIBL where they were kept in fiberglass ta nks containing 100%, flowi ng seawater from the Gulf of Maine. The tanks were expose d to natural conditions. Longhorn sculpin ( Myoxocephalus octodecemspinosus ) were purchased from fishermen and were housed in a similar way at the MDIBL. Atlantic hagfish ( Myxine glutinosa ) were also purchased from fishermen and were housed at the MDIBL in 100% seawater and maintained on a 12 h light/dark cycle. Female, non-migratory lampreys ( Petromyzon marinus ) were a generous gift from the USGS Great Lakes Science Center at the Ha mmond Bay Biological Stat ion in Millersburg, Michigan and were dissected there. Lancelets ( Branchiostoma lancelolatum ) were purchased from Gulf Marine Specimens (Panacea, FL) and were kept in the bags they were shipped in for a few days before they were sacrificed. All housed animals were fed to satiation regularly. Some killifish were shipped to the Univers ity of Florida (Gainesv ille, FL) for further experiments. These killifish were housed as described previously (Choe et al ., 2006). Some hagfish were also transported to the Universi ty of Florida (Gainesville, FL) for further experiments. These hagfish were housed for a few days in a large Rubbermaid tank in 100% seawater at 4 C before they were killed. Tissue Collection After initial anaesthetiza tion with MS-222 (~600 mg L-1), killifish, sculpin, and lampreys were pithed and/or decapitated. The gill arches (1st and 2nd arches for lampreys, 2nd and 3rd arches for teleosts) were then removed using ster ile, RNAse free dissecting tools. Hagfish were decapitated and all gill baskets were then rem oved using sterile, RNAse free tools. Lancelets were cut in half with sterile, RNAse free tools. After removal, tissues were immediately placed in liquid nitrogen and stored at -80 C for 2 weeks to 4 months before further processing. 42
PAGE 43
Reverse Transcription, Primer Design, PCR, Cloning, and Sequencing Reverse transcription, PCR, cloning, and sequencing were performed as described previously (Choe et al ., 2006) with modifications. Total R NA was isolated from the gills of killifish, lampreys, sculpin, and hagfish as well as the anterior half of lancelets using TRI-reagent (Sigma, St. Louis, MO), and revers e transcribed with a SuperscriptTM II or SuperscriptTM III reverse transcriptase kit (Invi trogen, Carlsbad, CA) following th e manufacturers protocol and using oligo-dT as a primer. The resulting cDNA was stored at -20 C until used for PCR. Degenerate primers (those labe led as CH COX, Table 2-1) were first designed to amplify non-specific cyclooxygenases in the hagfish, lamp rey, and lancelet based on conserved amino acid sequences between COX-1 and COX-2 of Mus musculus, COX-2 of Fundulus heteroclitus sCOX of Squalus acanthias (GenBank accession numbers: NP_032995, NP_035328, and AAS21313), and predicted COXa and COXb of Ciona intestinalis based on genomic data. These primers were also later used to amplify COX-2 in sculpin. Another set of degenerate primers (those labeled as CH COX-1 in Table 2-1) were designed to amplify COX-1 forms specifically over COX-2 forms in teleosts ba sed on conserved amino acid sequences between COX-1 in Oncorhynchus mykiss Danio rerio, Salvelinus fontanalis (GenBank accession numbers: CAC10360, NP_705942, and AAF14529) and a predicted COX-1 of Tetraodon nigroviridis based on genomic data. These primers firs t amplified a COX-1a form in sculpin and a COX-1b form in killifish (see below). A further set of degenerate primers (those labeled as CH COX-1a in Table 2-1) were designed to amp lify COX-1a forms specifically over COX-1b in teleosts based on conserved amino acid seque nces between predicted COX-1a forms in Gasterosteus aculeatus Oryzias latipes and Tetraodon nigroviridis based on genomic data and a COX-1a found in sculpin earlier in this study (see below). A fi nal set of degenerate primers (those labeled as CH COX-1b in Table 2-1) were designed to amplify COX-1b forms specifically 43
PAGE 44
over COX-1b in teleosts based on conserved am ino acid sequences between predicted COX-1b forms in Gasterosteus aculeatus Oryzias latipes and Tetraodon nigroviridis based on genomic data and a COX-1b found in killif ish earlier in this study (see be low). All degenerate primers were designed using the online program Consensus-Degenerate Hybrid Oligonucleotide Primers (CODEHOP) (Rose et al ., 1998). Initial PCR reactions were performed on 1/20th of a reverse transcri ptase reaction with a TaKaRa Ex Taq Hot Start DNA Polymerase Kit (Takara Bio Inc., Japan) in a PCR Express thermocycler (ThermoHybaid, Franklin, MA) with the following parameters: initial denaturing at 95 C for 5 minutes, then 35 cycles of 30 seconds at 95 C, 30 seconds at 60 C, and 1.5 minutes at 72 C. A final elongation step of 7 minutes at 72 C was performed for each PCR and then the reaction was held at 4 C. Initial PCR products were visualized by ethidi um bromide staining on 1-2% agarose gels and then ligated into PCR4-TOPO vectors and transformed into TOP10 chemically competent cells using a TOPO TA Cloning Kit (Invitrogen) following the manufacturers protocol. Plasmids were then isolated using a High Pure Plasmid Isolation Kit (Roche, Germany) following the manufacturers protocol and plasmid DNA was sequenced in both directions at the Marine DNA Se quencing Center at the MDIBL. After initial fragments were sequenced, Pr imer Express software (Applied Biosystems, Foster City, CA) was used to design specific prim ers based on the initial fragments (Tables 2-2, 2-3, 2-4, 2-5, and 2-6). These specific primers, along with the original de generate primers, were used to extend COX sequences. PCR was comple ted as above with slight modifications to annealing temperature and elongation time for each reaction, as well as nested PCR reactions in some cases to increase specificity. Products were visualized, cloned, and sequenced as above. 44
PAGE 45
5 and 3 Rapid Amplific ation of cDNA Ends (RACE) After initial sequences were extended with sp ecific primers, the 5 and 3 ends were sequenced using 5 and 3 RACE. Following th e manufacturers protocols, a Generacer Kit (Invitrogen) was used to make 5 and 3 cDNA for hagfish, k illifish, and scul pin. Initial sequence data was used to make 5 and 3 RACE specific primers for each species (those labeled with 5 or 3 in Tables 2-2, 2-3, 2-4, 2-5, and 2-6). These primers, along with kit primers that anneal to the ends of the 5 and 3 cDNA, we re used in touchdown PCR reactions with the following parameters: 5 cycles of 30 seconds at 94 C and 1.5 minutes at 72 C; 5 cycles of 30 seconds at 94 C and 1.5 minutes at 70 C; and 25 cycles of 30 seconds at 94 C, 30 seconds at 60 C, and 1.5 minutes at 72 C. As above, a fi nal elongation step of 7 minutes at 72 C terminated the reaction, which was held at 4 C. For some reactions, annealing temperatures and elongation times were slightly altered and nest ed PCR reactions were performed for some reactions using specific primers (Tables 2-2, 23, 2-4, 2-5, and 2-6) to increase specificity. Products were visualized, cloned, and sequenced as above. Sequence Analysis Initial sequences from dege nerate primers, sequence exte nsions, and 5 and 3 RACE were assembled using GeneTools software (BioT ools Inc., Edmonton, Alberta) and the resulting sequences were searched for open reading frames. The predicted amino acids were aligned with other COX proteins using PepTools software (BioTools Inc.) for phylogenetic analysis. Multiple alignments were also generated using ClustalW to search for conserved protein domains across COX sequences. Alignments were manually adju sted and the N and C-terminal areas were realigned using Microsoft Word based on previous COX alignments (Ishikawa et al ., 2007; Yang et al ., 2002; Kulmacz et al ., 2003). For the novel sequences, COX forms were aligned with 45
PAGE 46
other relevant COX proteins and annotate d using functional amino acids and domains highlighted from Ishikawa et al (2007), Yang et al (2002), and Kulmacz et al (2003). Phylogenetic Analysis To investigate the evolutionary relationshi ps of the different forms of COX in the chordates, phylogenies were ge nerated using the novel amino acid sequences generated in this study and several other COX protein sequences from GenBank and genome databases. All novel protein sequences were included in the data set. COX sequences from teleosts were included in the analysis only if all 3 forms of COX were av ailable. These included the zebrafish, rainbow trout, green spotted puffer, stickleback, eur yhaline killifish, and longhorn sculpin. COX sequences from non-teleosts were included in the analys is if at least two forms of COX were available and both sequences were designated as Reference Sequences (RefSeqs) in GenBank. These included the chicken, frog, mouse, rat, human, rabbit, dog, cow, sheep, and opossum. COX sequences were also included if they represented an evolutionarily interesting group. These included the dogfish shark an d the sea squirt. In total, 50 protein sequences were included in the analyses (Table 2-7). Models of evolution to be used in phylogenetic analyses were evaluated to account for amino acid substitutions, among-site variation, and invariable sites. Using likelihood ratio tests, a model using the WAG rate matrix (W helan and Goldman, 2001) and the gamma ( ) distribution for among-site varia tion was chosen (Table 2-8). A distance phylogeny was first generated using Molecular Evolutionary Genetics Analysis 4 (MEGA) software (Tamura et al ., 2007) with a minimum evolution (ME) analysis and 1000 bootstrap replicates. In this analysis the parameter was fixed to the value estimated from maximum likelihood (ML) analysis ( = 0.737, see below). Additionally, a JTT rate matrix (Jones et al ., 1992) was used instead of a WAG matrix because a W AG matrix is not available 46
PAGE 47
for MEGA 4. A complete deletion option was also selected. After comparing results between ML, ME, and Bayesian Phylogenetics (BP), lampre y and hagfish sequences were identified as possible rouges and another distance analysis with identical parameters (but with = 0.742) was performed excluding these two sequences. A maximum likelihood phylogeny (ML) wa s generated using the program PhyML (Guindon and Gascuel, 2003). An initial phylogeny was generated using WAG analysis, 1000 bootstrap replicates, 8 substitution rate categor ies, and a gamma shape parameter with an estimated Again, the analysis was repeated with the rouge lamprey and hagfish sequences removed. To evaluate support for monophyletic groups, an approximate likelihood ratio test (aLRT, Anisimova and Gascuel, 2006) was pe rformed both including and excluding the lamprey and hagfish sequences. This test used the mi nimum of Chi-Squared and Shimodaira-Hasegawa (SH) support values. Finally, A Bayesian phylogenetics (BP) analys is was also performed using an initial run with 2 million generations and a final run with 10 million generations using the program MrBayes 3.1.2. Once again, analyses were perf ormed with and without hagfish and lamprey sequences. Each run consisted of on e cold and three heated chains ( T = 0.2). The WAG + model was used with an estimated After discarding the firs t 10% of each run as burnin, posterior probabilities were calculated based on trees sa mpled every 500 generations. Due to inconsistencies in tree topologie s between the three methods used above, alternative topologies were ge nerated in which the positions of the hagfish and lamprey sequences were forced to be included in diffe rent clades (Figure 2-15). These alternative topologies were generated and viewed using TreeView 1.6.6. Alternative topologies were 47
PAGE 48
evaluated using Ln L scores generated by maximum likelihoo d criterion with online execution in PhyML (Guindon et al ., 2005). Results Molecular Identification of Cyclooxygenases Branchiostoma Initial PCR reactions using the degenerate primers CH COX F3 and CH COX R2 (Table 2-1) amplified two 532 bp products from the lancelet ( Branchiostoma lancelolatum ). These two products (referred to hereafter as COXc and COXd ) were found to share the most identity with COXa and COXb from the tunicat e. COXc was 50.9% identical to COXa from the tunicate and COXd was 58.8% identical to COXb from the tuni cate. All attempts to extend these sequences using specific primers designed against these se quences (Table 2-2) we re unsuccessful. After searching for open reading frames two putative 177 amino acid prot eins were predicted from the lancelet. Myxine Initial PCR reactions using the degenerate primers CH COX F1, CH COX F2, CH COX R1, and CH COX R2 (Table 2-1) were used to amplify a single 1454 bp product from the hagfish ( Myxine glutinosa ). This product (hereafte r referred to as hCOX) was found to be 63.8% identical to COX-1 of the zebrafish ( Danio rerio). RACE primers (5 and 3, Table 2-3) were designed against this initial pr oduct and used with kit primers to completely sequence the 2416 bp product. Searching for open reading frames yi elded a putative 610 amino acid protein. Petromyzon Initial PCR reactions using degenerate primers CH COX F3 and CH COX R1 (Table 2-1) amplified a single 541 bp product from gill tissue of the lamprey ( Petromyzon marinus). This product (hereafter referred to as lCOX) was later extended using degenerate primers (Table 2-1) 48
PAGE 49
and specific primers designed agains t the initial 541 bp product (Table 2-4). Despite all attempts to amplify the complete lCOX product (including RACE) the sequence was only extended to 860 bp. A 286 amino acid open reading frame was pred icted from this product. Fundulus Because the COX-2 sequence fo r the euryhaline killifish ( Fundulus heteroclitus ) was previously completed (Choe et al ., 2006), the degenerate primer CH COX-1 F2 (Table 2-1) designed to amplify COX-1 forms over COX-2 form s was used with the degenerate primer CH COX R1 (Table 2-1) to initially amplify a non-COX-2, 1335 bp product (hereafter referred to as COX-1b Fundulus ) from killifish gill tissue. This product was then extended and completed using specific primers and RACE primers (Table 2-5). The completed 2226 bp sequence contains a 598 amino acid open reading frame that was nested inside the teleost COX-1b clade in all phylogenetic analyses (see be low), prompting the design of degenerate primers (Table 2-1) that were used to amplify COX-1a forms in teleosts over COX-1b forms. Using the degenerate primers CH COX-1a F1 and CH COX-1a R1 (Table 2-1), a 759 bp product was initially amplified in PCR reactions. Using RACE primer s designed to complete this sequence (Table 25), the 3 end of this product was completed, resulting in a 1655 bp product (hereafter referred to as COX-1a Fundulus) that contains a 452 amino acid reading fr ame. However, the 5 end of this product was not sequenced despite multiple attemp ts. This putative protein grouped with the COX-1a sequences of other teleos ts in all phylogenetic analyses (see below), leading to the conclusion that the COX-1a, COX-1b, and COX-2 sequences included in the analyses all represent different COX products expressed in the gill tissu e of the euryhaline killifish. Myoxocephalus Using the degenerate primers CH COX F1 and CH COX R2 (Table 2-1), a 977 bp product was initially amplified in PCR reactions using cDNA from the gills of the longhorn 49
PAGE 50
sculpin ( Myoxocephalus octodecemspinosus). This product (hereafter referred to as COX-2 Myoxocephalus) was then extended and completed usi ng RACE and specific primers (Table 26), resulting in a 2618 bp product that contains a 605 amino aci d open reading frame. This sequence grouped with COX-2 forms from other te leosts in all phylogenetic analyses, inciting the use of COX-1 degenerate primers (Table 2-1) CH COX-1 F2 and CH COX-1 R2 in PCR reactions to amplify a 1000 bp product (hereafter referred to as COX-1a Myoxocephalus ). This product was then extended and completed using RA CE primers (Table 2-6), resulting in a 2633 bp product encoding a 622 amino acid open reading frame. This product grouped with COX-1a forms from other teleosts in all phylogenetic analyses, prompting the design of degenerate primers used to amplify COX-1b forms in teleosts over COX-1a forms (Table 2-1). Using the degenerate primers CH COX-1b F1 and CH CO X-1b R0 (Table 2-1), an initial 895 bp product (hereafter referred to as COX-1b Myoxocephalus ) was amplified in PCR reactions. This product was extended and completed using specific RACE primers (Table 2-6), resulting in a 2332 bp product containing a 600 amino acid open readin g frame. This product grouped with other teleost COX-1b forms in all phylogenetic analyses, suggesting that the COX-2, COX-1a, and COX-1b forms sequenced represent 3 different C OX forms expressed in the gills of the longhorn sculpin. Sequence Analyses Teleosts COX-1a, COX-1b, and COX-2 Three different COX forms are now available from the teleosts studied here: the COX-2 sequenced from the euryhaline killifish (Choe et al ., 2006), the COX-2 sequenced here from the longhorn sculpin, and two the COX-1 forms sequenced here from both the killifish and sculpin, named as COX-1a and COX-1b (following Ishikawa and Herschman, 2007). COX-1a and COX1b predicted proteins from the killifish were more similar to Mus COX-1 than COX-2 (70% 50
PAGE 51
versus 64% for COX-1a and 68% versus 59% for COX-1b, respectively). Similarly, COX-2 from the sculpin was more similar to Mus COX-2 than Mus COX-1 (70% versus 60%) and COX-1a and COX-1b were more similar to Mus COX-1 than Mus COX-2 (53% versus 47% for COX-1a and 67% versus 57% for COX-1b, respectively). An alignment of COX forms from the killifis h (Figure 2-1) shows that all the important amino acid residues for cyclooxygenase function are conserved in all forms. However, because COX-1a is incomplete it is unknown whether the Nterminal characteristics are conserved in this form. These include an active site tyrosine (Tyr-385, using ovine COX-1 numbering), haembinding histidines (His-207 and His-338), and the aspirin acetyla tion site (Ser530) (Ishikawa et al ., 2007; Ishikawa and Herschman, 2007; Yang et al ., 2002; Kulmacz et al ., 2003). In addition the COX-2 form contains a C-terminal amino aci d insertion and the COX-1 forms contain an Nterminal amino acid insertion. This is consis tent with the COX-1 and COX-2 forms of other vertebrates (Herschman et al ., 2003; Kulmacz et al ., 2003), although killifish COX-2 seems to have a shorter than average in sertion. In mammals, the ami no acid residues thought to give COX-1 and COX-2 their differing affinities for su bstrates are positions 513 (His in COX-1 and Arg in Cox-2) and 523 (Ile in COX-1 and Val in COX-2) (Guo et al ., 1996). However, as previously noted in other teleosts (Grosser et al ., 2002; Choe et al ., 2006; Ishikawa et al ., 2007) these differences appear to be absent in teleosts, with Arg an d Val present in positions 513 and 523, respectively, in both teleost COX-1 and COX-2. This is also the case in the killifish with Arg in position 513 and Val in position 523 in all COX forms reported here. As in the killifish, alignment of the three COX forms sequenced from the sculpin (Figure 2-2) shows conserved amino acid re sidues characteristic of COX func tion in all three forms. The characteristic terminal amino acid insertions f ound in mammalian COX forms are also present in 51
PAGE 52
their respective sculpin counterpa rts. However, COX-1b from th e sculpin seems to have both N and C-terminal insertions. Disulfide bonds near th e n-terminus of sculpin COX-1a also seem to be altered due to the substitution of other amino acids for cysteine. It also appears that some functional amino acid residues near the C-terminus of sculpin COX-1a are not conserved with the other forms. These include the substitution of Leu for Arg at position 513, Leu for Val at position 523, and Phe for Ser at position 530. Lamprey, hagfish, and amphioxus COX forms The one incomplete putative COX protein obtained from the gills of the lamprey shares near equal identity (63%) with Mus COX-1 and COX-2. The gene, referred to here as lamprey COX (lCOX) (Yang et al ., 2002) shares the most identity (70 %) with a form from the lancelet (COXc), but shares notable identi ty (64%) with the form from the hagfish as well. Based on an alignment of COX forms sequenced from select non-mammalian and non-teleost chordates (Figure 2-3), the partial COX form sequenced fro m the gills of the lamprey has the predicted histidine site (His-207) critical for peroxidase activity as well as conserved haem-binding domains. However, the membrane binding regi ons near the Arg-277 loop that differ between COX-1 and COX-2 forms appear to be different from both COX-1 and COX-2 in the lamprey form. This is seen when the lamprey form is aligned with COX-1 and COX-2 forms from vertebrates (Figure 2-4). The putative COX protein from the gills of the hagfish shares slightly more identity with Mus COX-2 than Mus COX-1 (63% versus 60%). The gene, referred to here as hCOX, shares the most identity (> 60%) with vertebrate COX-2 forms, but also shares notable identity with lCOX (64%) and forms from amphioxus (> 60%). An alignment of hCOX with other nonmammalian, non-teleost COX proteins (Figure 2-3) shows that all important amino acid residues and binding sites critical to cycl ooxygenase activity are conserved in hCOX. An alignment of 52
PAGE 53
hCOX with COX-1 and COX-2 protei ns from various vertebrates (Figure 2-5) shows that hCOX has an insertion of amino acids near the C-terminal; characteristic of COX-2 proteins in vertebrates. However, there are several re gions where COX-1 differs from COX-2 but hCOX shows no identity to COX-1 or COX-2 amino acids (indicated as sites with a + over them in Figure 2-5). Two putative incomplete COX sequences were obtained from amphioxus. These sequences, referred to here as COXc and COXd, are most identical (81%) to each other, but also share notable identity with hCOX (66%), lC OX (67%), vertebrate COX-1 (> 60%), and vertebrate COX-2 (> 60%). Th ey are the least identical to Ciona COXa (48%) and COXb (55%). Based on an alignment of non-mammalia n, non-teleost COX proteins (Figure 2-3), it appears that amino acids critical to COX functi on are conserved in COXc and COXd. However, the first haem-binding domain seems to be very di fferent in the amphioxus forms. This is also the case for Ciona COX forms and COXa, COXb, COXc, and COXd all share notable identity in this region as shown by alignment (Figure 2-6). Phylogenetic analyses A phylogeny (Figure 2-7) generated under a distance optimality criterion with 1000 bootstrap replicates using MEGA4 shows predic ted COX clades. COX-1a and COX-1b forms from teleosts form monophyletic sister clades The novel COX-1 forms from the killifish and sculpin group together with othe r teleosts in these clades as expected based on their designation here. Eutherian COX-1 forms group together in a monophyletic clade sister to the teleost COX-1 clade. The COX-1 proteins from Gallus, Monodelphis, and Xenopus are located at the base of this monophyletic vertebrate COX-1 clade, with sCOX being the most basal. Vertebrate COX-2 forms group together in another monophyletic clade that is sister to the vertebrate COX-1 clade. There are two monophyletic clades within the COX-2 group: a teleos t clade and a tetrapod clade. 53
PAGE 54
The forms from lamprey and hagfish (lCOX and hC OX) are basal to the vertebrate COX clade, while the forms from amphioxus (COXc and C OXd) form a monophyletic group basal to these forms. The forms from Ciona (COXa and COXb) form another monophyletic group that is basal to all other COX forms. Boot strap values give robust support for vertebrate clades, the amphioxus clade, and the Ciona clade, but give weak support for placement of hCOX, lCOX, and sCOX. Based on these re sults and other generated phyl ogenies (see below), hCOX and lCOX were identified as possible rogue sequences due to their inconsistent positions depending on the type of analysis used and another distan ce analysis was complete d without including these sequences (Figure 2-8). The t opology of this phylogeny is very similar to the complete COX phylogeny. However, sCOX is basal to the vert ebrate COX clade and not included with the other vertebrate COX-1 forms. Additionally, bootstrap values for most clades are considerably more robust than when hagfish and la mprey sequences are included. To further test relationships between COX proteins, a phyloge ny (Figure 2-9) was generated using the maximum likelihood criter ion, 1000 bootstrap replicates, and the program PhyML (Guindon and Gascuel, 2003). The topology of this tree is very similar to the one generated using distance criteri on. Briefly, it contains well supported, monophyletic clades for COX-1 (teleost COX-1a, teleost COX-1b, tele ost COX-1, mammalian COX-1, and vertebrate COX-1 with sCOX as the most basal sequence) and COX-2 (teleost COX-2, mammalian COX-2, and vertebrate COX-2). The lamprey (lCOX) is also basal to the vertebrate COX clade, the amphioxus COXc and COXd form a basal monophyletic clade, and the most basal clade in the tree is a monophyletic COXa and C OXb from the sea squirt. However, in the ML analysis, the hagfish (hCOX) sequence groups at the base of the vertebrate COX-2 clade instead of being basal to the vertebrate COX clade. Again, th e analysis was repeated without hCOX and lCOX 54
PAGE 55
(Figure 2-10), causing bootstrap values to incr ease dramatically. Also, the estimated gamma shape parameter increased slightly from 0.737 to 0.742. To further evaluate support for COX clades, an aLRT analysis (Anisimova and Gascuel, 2006) was performed. This type of analysis prov ides values of support similar to Felsensteins bootstrap support values but is much faster because bootstrap sampling requires many (1000 in the previous analyses) runs wh ile aLRT is run only once. The fundamental difference between the two is that bootstrap support values are based on repeatability whereas the aLRT is a measure of the likelihood gain of including a branch versus collapsing it (Anisimova and Gascuel, 2006). The resulting aLRT statistics can be interpreted using parametric Chi-Squa red distributions or a non-parametric Shimodaira-Hasegawa-like (SH) pr ocedure (Anisimova and Gascuel, 2006). The former seems to be more liberal (giving values similar to Bayesian posterior probabilities) and the later more conservative (Anisimova and Gascuel, 2006). Here, the most conservative approach was used and values represent the minimum of Chi-Squared and Shimodaira-Hasegawa (SH) support values. When all sp ecies are included (Figure 2-11) clades of interest have robust support (> 0.90) except for the vertebrate COX-2 clade (contai ning hCOX, 0.683) and the clade containing lCOX (0.87). When hCOX and lCOX ar e excluded from the analysis (Figure 2-12), support values generally increase, with a notable incr ease in the vertebrate C OX-2 clade (0.883). A Basesian phylogenetics (BP) analysis wa s also performed (Figure 2-13) using the program MrBayes 3.1.2, with one initial 2 million generation run, and one final 10 million generation run. The topology of the consensus tree (with burnin = 10%) is similar to those described above, with well-supporte d aforementioned vertebrate COX clades. However, in BP analysis lCOX groups at the base of the amphioxus COXc/COXd clade instead of being the basal member of the vertebrate COX clade. This grouping is well-supported by posterior probability 55
PAGE 56
scores (0.93). As in ML analysis, hCOX groups w ith vertebrate COX-2 prot eins at the base of the clade. Also similar to previous analys es, when lCOX and hCOX are excluded (Figure 2-14) scores improve and all clades have robust support. Due to the contrasting topologies genera ted using distance, ML, and BP methods, 21 different tree topologies (Figure 2-15) were generated by using th e most likely tree from ML analysis (Figure 2-9) and reloca ting hCOX and/or lCOX sequences to other branches of the tree. Positions of all other sequences were not altered from the most likely tree. A Ln likelihood score (Ln L ) was generated for each topology using onlin e execution of the program PhyML (Guindon et al ., 2005). Likelihood scores were compared to th e most likely tree. These tests indicate that lCOX and hCOX can be moved some what freely around the base of the tree without resulting in a noticeably different Ln L than the most likely topology (Table 2-9). It appears that moving lCOX to the base of the COX-1 clade causes the smallest change in Ln L whereas moving hCOX or lCOX deeply into COX-1 or COX-2 clades causes the largest changes in Ln L Discussion Here, I present the first forms of the en zyme cyclooxygenase (COX) sequenced from the hagfish, lamprey, and amphioxus. I also show that the killifish and sculpin conform to the current hypothesis that teleos ts posses additional COX-1 and COX-2 forms. The phylogenetic relationships among the various COX forms are stil l not definite, although the results presented here allow some conclusions to be drawn and pred ictions to be made. Specifically, it appears that COX sequences in the chordates form at least 3 well supported phylogenetic groups: the urochordata, cephalochordata, a nd vertebrata. Hagfish and lamprey positions are variable depending upon the analysis used, but likelihood scores seems to favor placement within the vertebrata clade. 56
PAGE 57
Novel COX Forms Although the two COX sequences from amphioxus were not able to be extended, they aligned readily with other COX se quences. They show a first haem-binding domain that shares little conservation with vertebrate forms (20%), as in the urochordates. However, the second haem-binding domain is completely conserved in th ese forms, as is histidine-207 which is crucial for peroxidase activity. The arginine-277 loop region does no t resemble COX-1 or COX-2 forms. Due to the truncated nature of these proteins, it is difficult to determine if COX in the cephalochordates plays similar roles as in the vertebrates, but some functions are likely conserved. Following the nomenclature used by Jarving et al (2004) these forms are named COXc and COXd. The COX sequence found in th e gills of the hagfish was extended to completion and readily aligned wi th other COX forms. It ha s conserved amino acids, binding domains, and bonds critical for COX function. Ther efore, it likely plays a similar role in the hagfish. The hagfish COX has an N-terminus deletion and valine-523 char acteristic of COX-2 forms, but there are multiple sites where it shares equal or no homology with COX-1 and COX-2 of vertebrates. Therefore, following the nomenclature of Yang et al (2002) this form is named hCOX. Although incomplete, the one COX sequen ce cloned from the gills of the lamprey readily aligns with other COX sequences and contains amino acids and regions critical to COX function. However, like the hagfish sequence it shares no biased homology with COX-1 or COX-2 forms. This form is therefore na med lCOX. As predicted based on phylogenetic position, COX-1a, COX-1b, and COX-2 were found in th e gills of the euryhaline killifish and the longhorn sculpin. With the exception of COX-1a in the sculpin, all COX forms from these teleosts contain conserved regions critical to COX function. As has been reported in other teleosts (Ishikawa et al ., 2007), these forms all contain va line at position 523, unlike in mammals. COX-1a from the sculpin does not have several amino acids critical for COX 57
PAGE 58
function located near the C and N terminus of the sequence. These include disruption of disulfide bonds near the N terminus due to cy stine replacements, little conserved homology in the membrane binding domain, and alternate residues for arginie-513 (leucine), valine-523 (leucine), and serine-530 (phenylalanine). This is not the case in the other acanthopterygiians, where all COX forms have these conserved regions. This may indicate th at in the sculpin COX1a has partially lost function, and COX-1b or COX-2 may therefore have additional responsibilities. Phylogenetic Analyses As has been seen when tracing the evolutionary histories of other gene families in evolutionary deep time (Abbasi and Grzeschik, 2007; Goudet et al ., 2007), the pedigree of the COX family is littered with duplications, losse s, and uncertainty. Despite utilizing a multifaceted approach to depict COX evolution in the chordates, the results indicate that COX has a complicated and under-examined history of evoluti on in the chordates. This is not surprising, considering the complex functions COX has been s hown to have in the chordates. Because of these complex functions, duplications, partitioning of roles, differential expression, and multiple forms with multiple functions are to be expected. In all phylogenetic analyses, the more recent history of COX evolution is more clear and consistent than the relationships of the ancestral taxa. The teleosts consistently form monophyletic COX clades nested within the gna thostomes, as expected. The gnathostomes consistently form two COX sister clades in phylogenetic analyses: COX-1 and COX-2. This is expected based on previous phylogenetic analyses in the vertebrates (Gu, 2001; Gu, 2006). The COX-1 clade is consistently rooted with the sequence from the dogfish (sCOX), although support values are not robust fo r this relationship when hagf ish and lamprey sequences are included. In distance analysis excluding hagfish and lamprey sequences, sCOX groups outside 58
PAGE 59
the COX-1 and COX-2 clades, representing anot her COX gnathostome lineage. These groupings are as expected, based on Yang et al .s (2002) original descripti on of sCOX as neither a COX-1 nor COX-2. These are the first phylogenetic analyses of COX to include sCOX and it is concluded that sCOX is most likely a ba sal member of the COX-1 clade. In all analyses, the COX-1 clade contains robustly supported, m onophyletic teleost and eutherian mammal clades. In the teleost cl ade, COX-1b and COX-1a form well-supported, monophyletic clades. Although this is expected based on initial studies (Ishikawa et al ., 2007; Ishikawa and Herachmann, 2007), this is the fi rst study to use sophisticated tree-building techniques to show the monophyly and legitimacy of the COX-1 forms in teleosts. It is also shown in each analysis that the non-duplicate COX-1 forms from the zebrafish and trout are more related to COX-1b forms from the acanthopte rygiians, grouping with them at the base of a monophyletic COX-1/COX-1b clade. The mammalian COX-1 group is not as robustl y supported using all t ypes of analyses. The expected relationships (based on assumed ev olutionary relationships in the tetrapods) are shown and supported with some confidence in th e maximum likelihood anal ysis (ML) when the hagfish and lamprey sequences are excluded. In this anlaysis, the tetr apods form a monophyletic group with the frog and chicken at the base of the clade, followed by the opossum sequences, and then the eutherians. Even in this analys is, the frog and chicken sequences group together, contrary to what is biologically expected (the frog should be more basal, representing an evolutionary ancient non-amniote lineage, and the chicke n should be a sister to the mammals). However, this grouping is not well-supported in any analysis. When hagfish and lamprey are included in ML analysis, the chicken/frog clade occurs at the base of the mammalian/teleost clade, contrary to biological predictions. Howeve r, this relationship is poorly supported. These 59
PAGE 60
low support values are confirmed in the aLRT anal ysis. In Bayesian a nd distance analyses, the frog is basal to the chicken, but these sequences are again at the ba se of a biologically irrelevant teleost/mammal clade. However, unlike dist ance and ML methods, Bayesian posterior probabilities strongly support this relationship. Bayesian posterior probabilities traditionally overestimate support values for a variety of reasons (Cummings et al ., 2003: Lewis, pers. comm..) and can give incorrect clades strong poste rior probabilities. This may explain why this biologically unexpected relationship is strongly supported only in the Bayesian trees. Therefore, ML analyses (supported by bootstrap and aLRT sc ores) show the most probable scenario for COX-1 evolution in the gnathostomes: a basa l sCOX, followed by a teleost COX-1 clade (as described above) sister to a tetrapod COX-1 clade. The COX-2 groups within the gnathostomes mi rrors the same general trends as the COX1 clade. Teleosts consistently form a well-supported, monophyletic group. The COX-2a sequences from zebrafish and trout group together to form the most basal members of this clade. The COX-2b sequences group with the non-duplicate COX-2 sequences from the acanthopterygiians to form the other teleost COX-2 clade. Sister to the teleosts, the tetrapods form another COX-2 clade in the gnathostomes. In this tetrapod COX-2 clade, frog and chicken sequences form the most basal (although not well-s upported) clade, with the mammals as a sister clade. This biologically rele vant relationship between COX-2 sequences is predicted in all analyses. Basal to the COX-1 and COX-2 clades in the gnathostomes are two clades that are consistently and robustly supporte d in all phylogenetic analyses: a urochordate COX clade and a cephalochordate COX clade. The cephalochorda te clade consists of the two COX sequences obtained in this study (COXc and COXd) and th e urochordate clade consists of the two 60
PAGE 61
sequences from Ciona (COXa and COXb). The urochordate COX forms always forms the base of the entire tree and the cepha lochordate COX forms always group as a sister clade to the gnathostome COX clade. This study is the first to show this biologically relevant relationship between COX forms in these ancestral chordates. The hagfish (hCOX) and lamprey (lCOX) sequences have alternate phylogenetic positions depending on the technique used to reconstruct their evolutionary histories. Also, when these sequences are removed from analys is, bootstrap, aLRT, and posterior probability scores supporting COX clades increase. Because of their uncertain placement in COX evolution, these sequences are designated as rogues a nd analyses excluding them likely show a more accurate representation of COX evolution. In distance analysis, both hCOX and lCOX are placed outside the gnathostome COX clade and form two separate COX lineages between the cephalochordate and gnathostome clades. In ML analysis, lCOX remains basal but hCOX is placed at the base of the gnathostome COX-2 clade. In BP analysis, hCOX remains in the COX2 clade but lCOX is placed within the cephaloch ordate clade. Support values for hCOX and lCOX placement are weak in all but BP analyses Clearly, the placement of hCOX and lCOX in COX evolution is not clear and without further evidence, placing them with any degree of certainty becomes difficult. In an attempt to find the most likely placement of these sequences, 21 different phylogenies were generated based on the mo st likely tree with hCOX and lCOX placed in variable positions. The trees were then evaluated in a maximum likelihood framework. It is important to note that although Ln L scores were generate d for each topology, other biological factors may influence which topology or type of topology is actually the most likely. For example, placing hCOX in the COX-2 clade and lCOX at the base of the tree (technically the most likely tree) implies that another COX lin eage exists for all vertebrates (the lineage 61
PAGE 62
represented by lCOX) because the hagfish is biol ogically ancestral to the lamprey. This would mean that dozens of novel COX forms have yet to be discovered in well-examined groups such as the mammals and teleosts. With genomes of se veral of these groups available, it is unlikely that such a vast amount of sequence data have yet to be found. Therefore, the most likely scenarios for COX evolution would imply that ei ther hagfish is basal and lamprey is derived, both are basal, or both are derived. However, placing lCOX, hCOX, or sCOX in the COX-1 or COX-2 clade implies that another COX form has yet to be found in these species. It appears that moving lCOX into the base of the COX-1 clade causes less of a change in Ln L scores than moving hCOX outside the gnathosto me COX clade. However, it is apparent that either sequence can be moved somewhat freely throughout the tr ee without resulting in noticeably worse (~10 Ln L ) tree scores. This leads to th e conclusion that although hCOX and lCOX may represent independent, basal lineages of COX, they are most likely basal members of the COX-2 and COX-1 clades (respectively) in the gna thostomes. If this is the case then at least one other form of COX should exist in the hagf ish and lamprey, as well as in the dogfish shark. Summary Presented here are the first COX sequences from amphioxus, the hagfish, and the lamprey. All predicted COX sequences from the killifish and sculpin are also described. These sequences all likely represent functional forms of COX found in the chordates based on conserved amino acids and domains critical to COX function. For the first time, a sophisticated phylogenetic analysis of COX forms from the majo rity of representative chordate lineages was attempted. Although still not definitive, a gene ral hypothesis for the evol utionary history of COX in the chordates can be made (Figure 2-16). In this scenario, there are three main COX groups corresponding to the three subphyla found in the phylum Chordata: the urochordata, cephalochordata, and vertebrata. Each contains two main COX groups, with the well62
PAGE 63
documented COX-1 and COX-2 forms only found in the vertebrata clade. These forms contain monophyletic groups for the teleosts and the tetr apods, with multiple forms in the teleosts confirming previous hypotheses (Ishikawa et al ., 2007; Ishikawa and Herschman, 2007). Hagfish and lamprey placement is uncertain, but most likely these sequences represent basal members of the vertebrata clade. This does not support the timing of geno mic duplications in the chordates according to the most recent view of the 2R Hypothesis (Kasahara, 2007), indicating two rounds of genome duplication af ter the origin of the urochordate s but prior to th e radiation of the jawed vertebrates. However, several developm entally important proteins have been shown to not support this timing and indicate an earlier duplication (Hughes, 1999). Multiple COX-1 and COX-2 genes in the teleosts do support the hypot hesis that the teleosts are characterized by another round of genome duplication after the origin of the vertebrates (Li et al ., 2007). Clearly, this uncertain period representing some 600 milli on years of evolution is characterized by many losses and duplications and remains to be resolved definitively. 63
PAGE 64
Table 2-1. Degenerate primers used in PCR Name Orientation Nucleotide sequence (5' to 3') CH COX F1 Sense ATG GAC GAC TAC CAG TGY GAY TGY AC CH COX F2 Sense GAT GTT TGC ATT TTT CGC TCA RCA YTT YAC CH COX F2.25 Sense TGG CG T GGA CCT AGG TCA NRT NTA YGG CH COX F2.5 Sense CGA GCT GCG GTT CCA TAA ARA YGG NAA RYT CH COX F3 Sense CCA CTA TA T GGC TGC GGG ARC AYA AYM G CH COX F5 Sense CAT GTG GAA TT C CAT CAC CTG TAY CAY TGG CA CH COX R1 Antisense CCC C GA AGG TGG ATG GYT TCC AVY A CH COX R2 Antisense GCA TCA GCG GGT GCC ART GRT A CH COX R3 Antisense TCC TCT TTT AGT ATG TCG CAG ACT CKR TTR TGY TC CH COX-1 F4 Sense TGC CAG ACA GCA TCC ACA THG AYG GNG A CH COX-1 F5 Sense CAG CAG ACA ATG CGC AGG NCA RAT HGG CH COX-1 R2 Antisense TGG GTG TA T GTT GTG TCC TCC NCC DAT YTG CH COX-1a F1 Sense CCC CAC C AA CCT ACA ATA CCA ART AYG GNT A CH COX-1a F2 Sense AAA GTG CTG ACT GAR MGN TTY TT CH COX-1a F3 Sense GGA CTA ATC TGA TGT TCG CGT TYA TGG CNC A CH COX-1a F4 Sense AGC GGA AGG CGG CTT YAC NAA NGC CH COX-1a F5 Sense CAA ATA GTA AAT GGG GAR AYN TAY CC CH COX-1a R1 Antisense ATA GGG AAT GTC GTC NCC RTC DAT CH COX-1a R2 Antisense GGG GCA CGT TTT CCG GRT ANA YCA T CH COX-1a R3 Antisense ACC TCA GTG ACA GTG GGA GGR TAN RTY TC CH COX-1b F1 Sense CGG AT G GGT GTG GGC TTY ACN AAR GC CH COX-1b R0 Antisense GGA GA A GGG AGC ACC CAT YTC NAC CAT CH COX-1b R3 Antisense CCA GCG TCC GCG CDA TYT CYT C 64
PAGE 65
Table 2-2. Amphioxus COX specific primers used in PCR Name Orientation Nucleo tide sequence (5 to 3) COXa F1 Sense TAC TCC ACC GTG TGG CTG C COXa F2 Sense TGG GAC GAC GAG AGG CTC COXa F2.5 Sense ACC GGT GAG ACC ATC AAC ATC AT COXa F3 Sense CAC CTG AGC GGC TAC AAC TTT COXa F4 Sense GAC CTG TTC TGG GAC CCT GAG COXa F5 Sense CCA GTA CCA GAA CCG CAT CTT C COXa R1 Antisense CGT ACG GCG GGT AGA TCA TG COXa R2 Antisense GCT GCC GCT CCA CGG COXa R3 Antisense CCC GT A GAT ATG GCT CAG GTC A COXa 3 F1 Sense GAG GCT CTT CCA GAC AGC CA COXa 3 F2 Sense CAC CTG AGC GGC TAC AAC TTT COXa 5 R1 Antisense T GG CTG TCT GGA AGA GCC TC COXa 5 R2 Antisense TGG AGT AGA CGA ACA GGC CC COXa 5 R3 Antisense CCC CGT AGA TAT GGC TCA GGT COXb F0 Sense CAG ACA GCT AGA CTC ATT CTT ATC AGT GA COXb F1 Sense TAT CAA CAT CGT CAT TGG AGA GTA TG COXb F1.5 Sense GGC TGG CAA AAA CTT CCA ACT COXb F2 Sense TGC AGT ACC AGA ACA GCA TAT TTG COXb F3 Sense AGC ATA TTT GTG GAG TTT AAC CAC TTG COXb R1 Antisense CGT CCG GAG GGT AAA TCA TGT COXb R2 Antisense ACT TCA GTT TCC CGT CCT GGA COXb R3 Antisense CGT A AA CAT GGC TCA TGT CCA CT COXb 3 F1 Sense GTC CCC GAG AAG AAG CGA TT COXb 3 F2 Sense GAA CG G CTC TAC CAG ACA GCT AG COXb 3 F3 Sense TGT TCT GGG ACC CTG AGC TG COXb 5 R1 Antisense TGG TAG AGC CGT TCG TCG TC COXb 5 R2 Antisense GGA GGG TAA ATC ATG TGT ACG GA COXb 5 R3 Antisense ACA TGG CTC ATG TCC ACT GC 65
PAGE 66
Table 2-3. Hagfish COX speci fic primers used in PCR Name Orientation Nucleo tide sequence (5 to 3) hCOX 3 F1 Sense TTA ACC ACC TGT ACC ACT GGC AC hCOX 3 F2 Sense CTC AAC CAT GGA GTA CGA GGT CT hCOX 3 F3 Sense CAA TGC TAT GGA GTT CTA CCT GGG hCOX 3 F0.8 Sense CAG TAC AGC AAC CGC ATC TCA hCOX 3 F0.6 Sense AAT GGG ACG ACG AGA GAA TCT TT hCOX 3 F0.4 Sense GGT TTG CTG TTG GTC ACG AAG hCOX 3 F0.2 Sense AGT CCA CCC ACA TAC AAT GCC hCOX 3 F0.1 Sense CTC ACG CAC TTT GCT CCC TT hCOX 3 F0 Sense GAA CTG CAC CTA CCC CGA GAC hCOX mid F1 Sense GTT TCG AGT CCA CCC ACA TAC AA hCOX mid F2 Sense CAT ACT CCA ACC TCA GCT ACT TCA CT hCOX mid R1 Antisense CAC ATC GAG TAC GTC ACA CAC G hCOX mid R2 Antisense AAG GGT CCC GTA GAG CAT GA hCOX 5 R1 Antisense TCT CCG AAT AAA GCT CCC TGT C hCOX 5 R2 Antisense TGC TGC CAT TTC CGT TTC TC hCOX 5 R3 Antisense CAT GTT TGA TGG ATG CTG TTG C hCOX 5 R4 Antisense AGG CG A GTG AAG TAG CTG AGG TT hCOX 5 R5 Antisense CCT CCC AGG ACT TGT ACC GTA AT hCOX 5 R6 Antisense GAG GA A GTA ATG CAC GGT ATC TGG 66
PAGE 67
Table 2-4. Lamprey COX speci fic primers used in PCR Name Orientation Nucleo tide sequence (5 to 3) lCOX F1 Sense GAG GAA CGC TTG AGA GAC AAC A lCOX R1 Antisense TCT TG T TGA TGT TT C GAC CGC lCOX R2 Antisense CCA CCG CCA ATC TGG C lCOX R3 Antisense AAT CAG GAA GCT CTC AGG CAT C lCOX 3 F1 Sense GAG TTC CTC TTC AAC CCC GG lCOX 3 F2 Sense TCG AAA CAT CAA CAA GAA CCT CC lCOX 3 F3 Sense TTC ACC TCC TTC CTG GAG CTC lCOX 5 R1 Antisense TGC TGC ACG TAC TCC TCG ATC lCOX 5 R2 Antisense GCA GCT GGT GTT GTC TCT CAA 67
PAGE 68
Table 2-5. Killifish COX specific primers used in PCR Name Orientation Nucleo tide sequence (5 to 3) Fh COX-1b 3 F1 Sense CTT CAC CGA TAG CGA GGA GAT AG Fh COX-1b 3 F2 Sense AGC TCT ACG GTG ACA TCG ACA CT Fh COX-1b 3 F3 Sense GGG TAA CCC CAT ATG TTC TCC AC Fh COX-1b 5 R1 Antisense CCC ATG TGA GAG AGC CTT AGT GA Fh COX-1b 5 R2 Antisense C AG CCT CTC AAA CAA CAC CTG AG Fh COX-1b 5 R3 Antisense GT A GTA GGA TTC CCA GCT GAG GTA GT Fh COX-1a 3 F1 Sens e GGT GTA CCC CGA AGG TTT CC Fh COX-1a 3 F2 Sense GAG TAC GTG CAG CAC CTG AGC Fh COX-1a 3 F3 Sense AGT ACA CCA ATC GCA TCG CC Fh COX-1a 5 R1 Antisense GAT GAT AAG TCT GGC GGT CTG G Fh COX-1a 5 R2 Antisense GAT GGT GGC ATA CAC GGT GA Fh COX-1a 5 R3 Antisense TTT ATG AAG CCG AAG CTG GTG 68
PAGE 69
Table 2-6. Sculpin COX speci fic primers used in PCR Name Orientation Nucleo tide sequence (5 to 3) Mo COX-2 3 F1 Sense CCC CGG TCT GAT GAT GTA CG Mo COX-2 3 F2 Sense ACC GAG TGT GTG ACG TGT TGA Mo COX-2 3 F3 Sense TC G AGG ACT ACG TGC AGC AC Mo COX-2 3 F4 Sense CAG GTC TAA CGC AAT CTT TGG G Mo COX-2 3 F5 Sense CCT CAA GGG CTT AAT GGG AAA C Mo COX-2 3 F6 Sense AAC ATC GTC AAC ACC GCC TC Mo COX-2 mid F1 Sense AC A GTT CAT TCC GGA TCC ACA Mo COX-2 mid F2 Sense CA C CAG CCT GAT GTT TGC ATT Mo COX-2 mid R1 Antisense CGA AGC GGT GAG ATT CAG GA Mo COX-2 mid R2 Antisense CTC CAT CCA GGA CCT GAT ATT TAA G Mo COX-2 5 R1 Antisense TCC GGA ATG AAC TGT CTT CTC A Mo COX-2 5 R2 Antisense AGG TCG GAG GAC TAT CAA TCA AGT Mo COX-2 5 R3 Antisense GAA GGA GAT GGA GTT AAT GAT GTT CC Mo COX-1a 3 F1 Sense CA T CAC CTC CGG CTT CAT AAA Mo COX-1a 3 F2 Sense GAG GGA GCA TAA CAG ACT CTG TGA Mo COX-1a 3 F3 Sense CTG GAG TTC TGC CAC CTC TAC C Mo COX-1a 3 F4 Sense GC A GCC CTT CAA TGA ATA CAG G Mo COX-1a 3 F5 Sense CAT GCT GGA GAA GAC CCT TCC Mo COX-1a 3 F6 Sense GAG TAT GTT GGA GAT GGG TGC TC Mo COX-1a 5 R1 Antisense GCC GAA ATG TCT TCC TCC TAA A Mo COX-1a 5 R2 Antisense CAT CGG TAA AGG GCA GTC CTC Mo COX-1a 5 R3 Antisense TGA CCG TCA GCA CTA ATC TCA TG Mo COX-1b 3 F1 Sense TCA CAC ACT ACG GCA TCG AGA Mo COX-1b 3 F2 Sense TTG GCG GTG GCT TTA ACA TC Mo COX-1b 3 F3 Sens e CAG CCC TTC AAC GAG TAC AGG Mo COX-1b 5 R1 Antisense GAG CGA TAC GGT TCC CAT ACT G Mo COX-1b 5 R2 Antisense GTT GTG CTC CCT GAG CCA GA Mo COX-1b 5 R3 Antisense AAT AGC CAT CTG AGC CTC AGG A 69
PAGE 70
Table 2-7. Sequences included in phylogenetic analyses # Name Organism Size (AA) 1 COX1a Fundulus Fundulus heteroclitus (killifish) 452 2 COX1b Fundulus Fundulus heteroclitus (killifish) 598 3 COX2 Fundulus Fundulus heteroclitus (killifish) 610 4 COX1a Myoxocephalus Myoxocephalus octodecimspinosus (lh. Sculpin) 622 5 COX1b Myoxocephalus Myoxocephalus octodecimspinosus (lh. Sculpin) 600 6 COX2 Myoxocephalus Myoxocephalus octodecimspinosus (lh. Sculpin) 605 7 COXa Ciona Ciona intestinalis (sea squirt) 653 8 COXb Ciona Ciona intestinalis (sea squirt) 600 9 COXc Branchiostoma Branchiostoma lancelolatum (amphioxus) 177 10 COXd Branchiostoma Branchiostoma lancelolatum (amphioxus) 177 11 hCOX Myxine Myxine glutinosa (Atlantic Hagfish) 610 12 lCOX Petromyzon Petromyzon marinus (Sea Lamprey) 286 13 COX1a Oryzias Oryzias latipes (Japanese medaka) 605 14 COX1b Oryzias Oryzias latipes (Japanese medaka) 606 15 COX2 Oryzias Oryzias latipes (Japanese medaka) 609 16 COX1a Gasterosteus Gasterosteus aculeatus (stickleback) 598 17 COX1b Gasterosteus Gasterosteus aculeatus (stickleback) 598 18 COX2 Gasterosteus Gasterosteus aculeatus (stickleback) 620 19 COX1a Tetraodon Tetraodon nigroviridis (Green Spotted Puffer) 1023 20 COX1b Tetraodon Tetraodon nigroviridis (Green Spotted Puffer) 589 21 COX2 Tetraodon Tetraodon nigroviridis (Green Spotted Puffer) 456 22 COX1 Oncorhynchus Oncorhynchus mykiss (rainbow trout) 624 23 COX2a Oncorhynchus Oncorhynchus mykiss (rainbow trout) 607 24 COX2b Oncorhynchus Oncorhynchus mykiss (rainbow trout) 609 25 COX1 Danio Danio rerio (zebrafish) 597 26 COX2a Danio Danio rerio (zebrafish) 601 27 COX2b Danio Danio rerio (zebrafish) 606 28 COX1 Monodelphis Monodelphis domestica (gr. S.t. opossum) 625 29 COX1b Monodelphis Monodelphis domestica (gr. S.t. opossum) 627 30 COX1b2 Monodelphis Monodelphis domestica (gr. S.t. opossum) 729 31 COX2 Monodelphis Monodelphis domestica (gr. S.t. opossum) 608 32 COX1 Gallus Gallus gallus (rd. jungle fowl) 649 33 COX2 Gallus Gallus gallus (rd. jungle fowl) 603 34 COX1 Bos Bos taurus (cattle) 600 35 COX2 Bos Bos taurus (cattle) 604 36 COX1 Xenopus l Xenopus laevis (Afr. Claw. Frog) 587 37 COX2 Xenopus l Xenopus laevis (Afr. Claw. Frog) 604 38 COX1 Canis Canis lupus familiaris (dog) 633 39 COX2 Canis Canis lupus familiaris (dog) 604 40 COX1 Homo Homo sapiens (human) 599 41 COX2 Homo Homo sapiens (human) 604 42 COX1 Rattus Rattus norvegicus (norway rat) 602 43 COX2 Rattus Rattus norvegicus (norway rat) 604 44 COX1 Mus Mus musculus (house mouse) 602 70
PAGE 71
Table 2-7. Continued. # Name Organism Size (AA) 45 COX2 Mus Mus musculus (house mouse) 604 46 COX1 Oryctolagus Oryctolagus cuniculus (rabbit) 606 47 COX2 Oryctolagus Oryctolagus cuniculus (rabbit) 604 48 COX1 Ovis Ovis aries (sheep) 600 49 COX2 Ovis Ovis aries (sheep) 603 50 sCOX Squalus Squalus acanthias (spiny dogfish) 593 71
PAGE 72
Table 2-8. LRTs between amino acid based models of COX evolution Evolutionary Model* Ln liklihood (Ln L ) Score (Ln L1LnL2) d.f.@ P# WAG -26632 WAG + I -26315 317 1 < 0.01 WAG + -25294 1021 1 < 0.01 WAG + I + -25294 0 2 > 0.05 *Evolutionary models are arranged from the most simple to the most complex. Only models using the WAG matrix were evaluated. I repr esents an invariable sites parameter and represents an among site variation parameter @ d.f. indicates the difference in free paramete rs between the complex and simple models #P values represent the comparison of simple to complex models and were obtained using a chisquared distribution test 72
PAGE 73
Table 2-9. LRTs between tree topologies with altered positions of hCOX and lCOX Tree Topology Name* Ln liklihood (Ln L ) Score Estimated @ (Ln L1LnL2)# COX2 Hag (most likely) -25294.22 0.737 COX1Lamp/COX2Hag -25295.57 0.737 2.70 COX2 Agnatha -25299.09 0.737 9.74 Lamp/Amp -25299.27 0.737 10.1 COX2 Vert -25299.61 0.736 10.7 COX1 Hag -25300.92 0.735 13.4 COX1 Vert -25301.22 0.735 14.0 Ancestral Hag -25301.56 0.735 14.6 COX1 Agnatha -25301.78 0.735 15.1 COX1Hag/COX2Lamp -25302.19 0.735 15.9 Ancestral agnatha -25302.55 0.735 16.7 COX1 Lamp -25303.25 0.735 18.1 COX2 Lamp -25303.85 0.735 19.3 Agnatha/Amp -25308.82 0.735 29.2 Hag/Amp -25308.90 0.735 29.4 Hag Internal -25321.15 0.737 53.9 Lamp Internal -25322.77 0.736 57.1 Lamp Mammal -25341.41 0.732 94.4 Lamp Teleost -25344.68 0.735 101 Hag Teleost -25350.30 0.734 112 Hag mammal -25408.58 0.731 229 *Each tree has the same topology as the most likely tree (Figure 2-9) but with alternative placements of hagfish (hCOX) and lamprey (l COX) sequences as noted in Figure 2-15. @ represents the shape value under the model #Where Ln1 is the most likely tree score and Ln2 is the tree score of the topology in question 73
PAGE 74
COX-1b MRASVLGSVCALLVLLREPGCQGDEVTTSTVNPCCYLPCKHWSVCVRYGEDKYECDC 57 COX-2 MNRITFAVFLLALC--------FSFHKEVLGNACCSEPCQNRGVCTAMGSDSYECDC 49 *N68 COX-1b THTGYYGENCSIPELWTRVRQFLKPSPDVVHYILTHFHWLWDIINN-TFLRNVLMRL 113 COX-2 TRTGYRGQNCTTPEFLTWIKISLKPTPNTVHYILTHFKGFWNIINNISFLRDAIMKY 106 *R120 *N144 COX-1a NLSYYTRLLPPVPKDCPLPMGTKGKPVLP 29 COX-1b VITARSNLIPSPPTFNSKYNYLSWESYYNLSYYTRILPPVPEDCPTPLGVKGRNGLP 170 COX-2 VLTSRSHMIDSPPTYNADYGYKSWEAYSNLSYYTRALPPVPEDCPTPMGVVGKKELP 163 Q203/H207 COX-1a DIKEISERYFKRTEFRPDPQGTNLMFAFMAQHFTHQFFKTSHKVDAGFTKALGHGVD 86 COX-1b DPQVLFERLLKRRTFRPDPQGSNIMFAFFAQHFTHQFFKTYNRMGLGFTKALSHGVD 227 COX-2 DVKVLAEKLLVRRRFIPDPQGTSLMFAFFAQHFTHQFFKSDMKNGPAFTVAKGHGVD 220 *R277 COX-1a ASNIYGEELERQHQLRLHKDGKLKYQLINGEMYPPTVSEVPVHMVYPEGFPAEQRLA 143 COX-1b AGHIYGDSLERQHLLRLFRDGKLKYQLIDGEVYPPSVTDAPVRMSYPPGIPVEKQMA 284 COX-2 LGHIYGENLEKQHKLRLFKDGKLKYTMVNGEVYPPLVKDVGVEMHYPPHVPDSQRFA 277 COX-1a IGQEVFGLVPGLTVYATIWLREHNRVCDILKGEHPTWDDEQLFQTARLIIIGEIINI 200 COX-1b IGQEVFGLLPGLSLYATLWLREHNRVCDILKAEHPTWDDEQLFQTTRLIIIGETIRI 341 COX-2 VGHEAFGLVPGLLMYATIWLREHNRVCDVLKGVHPDWDDERLFQTTRLILIGETIKI 334 Y385/H388 COX-1a IIEEYVQHLSGYYLKLKYDPSLLFGVRFQYTNRIALEFCYLYHWHPLMPDSFLIDGD 257 COX-1b VIEEYVQHLSGYLLQLKFDPTLLFNSNFQYGNRIALEFSQLYHWHPLMPDSFHISGD 398 COX-2 VIEDYVQHLSGYHFKLKFDPELLFNQRFQYQNRIASEFNTLYHWHPLMPDSFHIEEK 391 *N410 COX-1a ELPYSQFLYNTSILMHYGVEKLVDAFSRQPAGQIGGGRNIHQAVLRVAEMVIRDSRA 314 COX-1b ELSYSQFLFNTSVLTHYGVEKLVDAFSRQAAGQIGGGHNINAVITKVIVGTIEESRQ 455 COX-2 DYSYKEFVFNTSVVTEHGINNLVDSFSKQIAGRVAGGRNVPGPIMYVAIKSIENSRK 448 *R513 COX-1a ARLQPFNQYRKRFNLKPYSSFYELTGDEEMARGLEELYGDIDALEFYPGLLLEKTRP 371 COX-1b LRIQPFNEYRKRFNLEPYTSFRDFTDSEEIASTLEELYGDIDTLEFYPGLLLEKTRP 512 COX-2 MRYQSLNAYRKRFSMKPYTSFEDLTGEKEMAAILEELYGDVDAVELYPGLLVEKPRT 505 *V523 *S530 COX-1a SSIFGESMVEMGAPFSLKGLLGNPINSPEYWKPSTFGGETGFNIIKTSTLKKLVCLN 428 COX-1b GAIFGESMVEMGAPFSLKGLLGNPICSPQYWKPSTFGGKVGFDIVNSASLKKLVCLN 569 COX-2 NAIFGETMVEMGAPFSLKGLMGNPICSPEYWKPSTFGGTAGFDIVNTASLQRLVCNN 562 Figure 2-1. Amino acid alignment of the form s of COX sequenced from the gills of the euryhaline killifish (Fundulus heteroclitus ). The COX-2 sequence is from GenBank (Accession #AAS21313) and the others were obta ined in this study. Amino acids that share at least 60% identity across sequences are highlighted. *I ndicates amino acids (numbered based on ovine COX-1) involved in COX function: active site of COX (tyrosine-385, histidine-388, and serine-530) substrate binding site (arginine-120), N glycosylation sites (asparagine-68, 144, and 410), sites for peroxidase activity (glutamine-203 and histidine-207), and the two sites which define conformational differences in channels between COX-1 and COX-2 in mammals (arginine-513 and valine-523) (Ishikawa et al ., 2007). The black boxes indi cate the two domains that define the haem-binding sites and the da shed box indicates the membrane binding domain (Kulmacz et al ., 2003; Ishikawa et al ., 2007). The solid lines represent disulfide bonds, with the beginning and end of the l ongest bond represented by arrows (Kulmacz et al ., 2003). The dashed line represents the arginine-277 loop region, which has been shown to differ between COX-1 and COX-2 forms in mammals (Yang et al ., 2003). The numbers to the right of the sequences indicate the amino acid number of each sequence. 74
PAGE 75
COX-1a TK-WCPYVDFHVPRNEEERKPS-----------------------TEL 452 COX-1b SR-TCPYVAFSVPAEEEPGRNDGKERS------------------SEL 598 COX-2 VRGPCPVASFYVPDVKETGSMTINSSTSHSRDSNINPTVILKERTSEL 610 Figure 2-1. Continued. 75
PAGE 76
COX-1a MKRANLSNYFLETNHCEQNISDNYHFITLITIL-SRES--KSLASFPSGLTFCCWGA 54 COX-1b MRSPVLGPVCALLLL-LREPAC------------------RGDEVTSDTVNPCCY-36 COX-2 MYRFTFA-VF---LLALGVLVC-------------------------EGGNPCC--25 *N68 COX-1a HVLLTQVLIPVVITLVRTQECVCDSVQIT-TNVTALALASMETTALSRSSGPEFVRS 110 COX-1b --FPCQHWGVCVRYGEDKYECDCTRTGYTGGNCTVPEFWSRVHQFL-KPS-PEVLHY 89 COX-2 -SEPCQNRGVCTALGTDNYECDCTRTGYHGHNCTTPEFLTWVKISL-KPS-PNTVHY 79 *R120 COX-1a RTRRSSSSSPTSIGCWHLVNN-SFLRGTVMRLVLTVRSDLIPSPPTYNTKYGYLNWE 166 COX-1b ILTHFN-------WLWDIINH-TFLRDVLMRMVLTVRSNLIPSPPTYNSKYDYLSWE 138 COX-2 ILTHFK-------GFWNIINSISFFRDAIMRYVLTSRSHLIDSPPTFNADYGYKSWE 129 *N144 COX-1a SYYNISYYTRLLPPVPEDCPLPMGTKGRPDLPDPKVLTERFFRRKTFRPDPQGANLM 223 COX-1b SYYNLSYYTRILPPVPKDCPTPLGVKGKAGLPDPELLVERLLKRRTFRPDPQGSNLM 195 COX-2 AYSNLSYYTRTLPPVPEDCPTPMGVVGKKELPDAKLLAEKLFMRRQFIPDPQGTSLM 186 *Q203/H207 COX-1a FAFMAQHFTHQFFKTDHELQGGFTKALGHGVDAGNIYGDNLAKQHHLRLHKDGKLKY 280 COX-1b FAFFAQHFTHQFFKTYNRMGVGFTKALAHGVDAGHVYGDNLQRQLKLRLHKDGKLKY 252 COX-2 FAFFAQHFTHQFFKSDMKKGPAFTLATGHGVDLNHVYGGSMERQHKLRLRQDGKLKY 243 *R277 COX-1a QIVNGETYPPTTSEAPVHMMYPEDVPPEKRLAIGQEVFGLLPGLTMYATIWLREHNR 337 COX-1b QLVDGQIYPPSVVDAPVKMSYPPGVPPEAQMAICQEVYGLLPGLGMFATLWLREHNR 309 COX-2 QVLDGEVYPPTVKEVGADMHYPPHVPESHRFAVGHEAFGLVPGLMMYATIWLREHNR 300 COX-1a LCDILKAEHPTWDDEQLFQTTRLIVIGEIINIIIEEYVQQLSGYQLKLKFDPTLLFN 394 COX-1b VCDILKAEHPTWDDEQLFQTARFVIIGETIKIVIEEYVQQLSGYLLQLKFDPALLFN 366 COX-2 VCDVLKEVHPDWDDERLFQTTRLILIGETIKIVIEDYVQHLSGYNFKLKFDPELLFN 357 *Y385/H388 *N410 COX-1a ERFQYSNRIALEFCHLYHWHSLMPDSFLIDGDDIPYSQFFYNTSILMHYGVEKLVDA 451 COX-1b SNFQYGNRIALEFSQLYHWHPLMPESFLINGDELPYKRFLFNNTVLTHYGIENLVTA 423 COX-2 QRFQYQNRIASEFNTLYHWHPLMPDSFHIEEKDYSYKEFVFNTSVVTEHGIGNLVES 414 Figure 2-2. Amino acid alignment of the three forms of COX sequenced from the gills of the longhorn sculpin ( Myoxocephalus octodecemspinosus ): COX-1a, COX-1b, and COX2. All sequences were obtained during this study. Amino acids that share at least 60% identity across sequences are highlighted. *Indicates amino acids (numbered based on ovine COX-1) involved in COX func tion: active site of COX (tyrosine-385, histidine-388, and serine-530), subs trate binding site (arginine-120), N -glycosylation sites (asparagine-68, 144, and 410), sites fo r peroxidase activity (glutamine-203 and histidine-207), and the two sites which defi ne conformational differences in channels between COX-1 and COX-2 in mammal s (arginine-513 and valine-523) (Yang et al ., 2002; Ishikawa et al ., 2007). The black boxes indicate the two domains that define the haem-binding sites and the dashed box indicates the membra ne binding domain (Kulmacz et al ., 2003; Ishikawa et al ., 2007). The solid lines represent disulfide bonds, with the beginning and end of the longest bond represented by arrows (Kulmacz et al ., 2003). The dashed line repres ents the arginine-277 loop region, which has been shown to differ between COX-1 and COX-2 forms in mammals (Yang et al ., 2003; Kulmacz et al ., 2003). The numbers to the right of the sequences indicate the amino acid number of each sequence. 76
PAGE 77
COX-1a FSHQPAGQIGGGHNSHAVVLKVAEMVIRESRETRVQPFNEYRKKFNLQPYTSFYDLT 508 COX-1b FSRQVAGQIGGGFNINAAVTKVSVLTIKESRKLRMQPFNEYRKRFNLKPYTSFREFT 480 COX-2 FTNQIAGRVAGGRNVPGPIMYVAIKSIENSRKMRYQSLNAYRKRFSMKPYSSFEDMT 471 *R513 *V523 *S530 COX-1a GDIEMAKGLEELYGDIDAVEFYPGLMLEKTLPTRIFGESMLEMGAPLFPERPVGKPH 565 COX-1b DNEEIARELEEFYGDVDALEFYPGLLLERTREGSIFGESMVEMGAPFSLKGLLGNPI 537 COX-2 GEKEMAAILEEFYGHVDAVELYPGLLVEKPRSNAIFGETMVEMGAPFSLKGLMGNPI 528 COX-1a LLPV-LEAQHLWGRDGFQYRNFHSEETGVPQHQVVSIRGL-PCSPKGRRNQTNKSI620 COX-1b CSPVYWKPSTFGGKVGF---DIVNSAT---LKKLVCLNTR-TCSYVAFRVPTEEQLK 587 COX-2 CSPEYWKPSTFGGSEGF---NIVNTAS---LQRLVCNNVQGPCPVASFSGPDVKDSG 579 COX-1a -----------------------YTL 622 COX-1b TGNDDSKTRT-------------DEL 600 COX-2 SMIINSSTSNSDINPTVILKERTTEL 605 Figure 2-2. Continued. 77
PAGE 78
sCOX MEAARIILLLLPLCFLKMADTTAINPCCYYPCQNKGICVNVGKEGYECDCTRT 53 hCOX MTSEVFVVLCVAVVFAGAA--AADDPCCGSPCENKGVCVSVGFEDYECDCTRT 51 *N68 sCOX GYYGVNCTFPFSWSRVHKFLKPSPSSMHHVLTHYKWLWYIINNISFFSDTLMR 106 hCOX GYYGSNCTYPETWTWIVNMLKPLPDTVHYFLTHFAPFWSLVNRAAFLRDRVMR 104 *R120 *N144 sCOX LVLTVRANIIPSPPTYNSDYTYVSWEGYSNISYLTRLLPPVPKDCPTPTGTQG 159 hCOX HVLMSRAHMVSSPPTYNADYGYKSWEAYSNLSYFTRLLPPVPQGCPTPMGRTG 157 *Q203/H207 sCOX YKKLPDSEQLAEEFLLRRKFIPDPQGSNLMFAFFAQHFTHQFFRTDLDRGPGF 212 lCOX HQFFNTDPVRGPAF 14 hCOX KKELPDAQKLAERFLLRRTFIPDPQGSNLLFAFFAQHFTHQFFKTDFKRGPGF 210 COXc HQFFKTDFKKGAGR 14 COXd HQFFKTDFKKGPGR 14 sCOX TKALGHGVDLTHIYGDSLERQHHLRLFKDGKLKYQVVNGEVFPPSVKEAPIQM 265 lCOX TRALGHGVDLNHIYGGTLERQHQLRLFKDGKLKFQMIDGEAYPPVVRDAPVHM 67 hCOX TKALGHGVDLSHIYGDTLDKQHKLRLHNNGKLKFQMIDGEVFPPLVSEAPVDM 263 COXc TW-GDHGVDLSHIYGETVERQHQLRSFTDGKLKFQRVEGEVYPPSLADAPVHM 66 COXd TW-SDHAVDMSHVYGETVERQRQLRSFQDGKLKYQLVDGEAFPPSLQDASVHM 66 *R277 sCOX KYPSTLPEEKRLAIGHDTFGLIPGLMMYATIWLREHNRVCDILKEEHPVWSDE 318 lCOX VYPEHVPASLRFAVGHEVYGLLPGLLVYATVWLREHNRVCDVLHARHPRWDDE 120 hCOX IYPPHVPEAARFAVGHEAFGLVPGLMLYGTLWLREHNRVCDVLDVDHPEWDDE 316 COXc IYPPYVPEGKRFAIGHEFFGLLPGLFVYSTVWLREHNRVCDVMKELHPDWDDE 119 COXd IYPPDVPEKKRFALGHEFFGLLPGLFVWATVWLREHNRVCDVMKDLHPDWDDE 119 sCOX QLFQTTRLILIGETIKIVIEDYVQHLSGYNFRMMFNPELLFTEHFQYSNRIAV 371 lCOX RLFQTARLILTGETMKIVIEEYVQHLSGYNFHLKFDPTLLFGVNFQYSNRMSL 173 hCOX RIFQTARLILIGETIKIVIEDYVQHLSGYHFKLKFDPKLLFGQPFQYSNRISV 369 COXc RLFQTARLILTGETINIIINEYVQHLSGYNFDLFWDPELLFSDQFQYQNRIFV 172 COXd RLYQTARLILISETINIVIGEYVQHLAGKNFQLFWDPELLFEEQXQYQNSIFV 172 Figure 2-3. Amino acid alignment of the forms of COX sequenced from the gills of the lamprey ( Petromyzon marinus ), hagfish ( Myxine glutinosa ), and amphioxus ( Branchiostoma lancelolatum ), along with sCOX from the dogfis h (Squalus acanthias, Accession #AAL37727). Amino acids that share at least 60% identity across sequences are highlighted. *Indicates amino acids (num bered based on ovine COX-1) involved in COX function: active site of COX (tyrosine-385, his tidine-388, and serine-530), substrate binding site (arginine-120), N -glycosylation sites (a sparagine-68, 144, and 410), sites for peroxidase activity (gluta mine-203 and histidine-207), and the two sites which define conformational differen ces in channels between COX-1 and COX2 in mammals (arginine-513 and valine-523) (Yang et al ., 2002; Ishikawa et al ., 2007). The black boxes indicate the two domains that define the haem-binding sites and the dashed box indicates the membrane binding domain (Kulmacz et al ., 2003; Ishikawa et al ., 2007). The solid lines represent disulfide bo nds, with the beginning and end of the longest bond re presented by arrows (Kulmacz et al ., 2003). The dashed line represents the arginine-277 loop region, which has been shown to differ between COX-1 and COX-2 forms in mammals (Yang et al ., 2003; Kulmacz et al ., 2003). The numbers to the right of the seque nces indicate the am ino acid number of each sequence. 78
PAGE 79
* *Y385/H388 *N410 sCOX EFDHLYHWHPLMPDSFIVKGQDFSYKDFLFNTDILLNLGVDALVESFSKQIAG 424 lCOX EFNHLYHWHPLMPDSLLIDGRNYSYDEFLFNPGLLADKKLMPLVRSFMRQRAG 226 hCOX EFNHLYHWHGLNPDAFRVGTQEYQYSQFLFNNTILLNHGVRGLLEAFNVQQAG 422 COXc EFNHL 177 COXd EFNHL 177 sCOX RIGGGRNIHQSLLHIAIATIEHGRLLRFQPYNEYRKKLGLTPYKSFQELTGER 477 lCOX TVSGGRNINKNLLHVATSIIEHGRTLRLQSLNQYRHRFNMRPFTSFLELTGDE 279 hCOX RIGGGQNIHGALLHVATASIKHGRKMRFQSLNQYRKQFGLQPYQSFEQLTGET 475 *V523 *S530 sCOX EVAARLEKLYGHIDAMEFYPALLLEAPNKNSIFGESMVEMGAPFSLKGLMGNP 530 lCOX AMAAEME 286 hCOX EMAADLAELYSDINAMEFYLGLMVEKPRQGALFGETMVEAGAPFSLKGLMGNA 528 sCOX ICSPDYWKPSTFGGKTGFDIVNTATFEKLICLNVK-KCPYVGFHVPY-----582 hCOX ICSPEYWKPSTFGGNRGFEIVNSASLRRLVCLNLQGPCPDVAFHVPRDNQQDV 581 sCOX NVDNDYEREKGKP------------STEL 593 hCOX VVNVTGSQGGSDGVTTTPHYVADQQSREL 610 Figure 2-3. Continued. 79
PAGE 80
*H207 COX1 Mus HQFFKTSGKMGPGFTKALGHGVDLGHIYGDNLERQYHLRLFKDGKLKYQVLDG 261 COX1 G.gal HQFFKTSGKMGRGFTKALGHGVDLGHLYGDNLQRQHQLRLFQDGKLKFQVVNG 304 COX1 Danio HQFFKTHNRVGLGFTKGLGHGVDAGHIYGDSLDRQLELRLHKDGKLKYQVLNG 259 lCOX HQFFNTDPVRGPAFTRALGHGVDLNHIYGGTLERQHQLRLFKDGKLKFQMIDG 53 COX2 Mus HQFFKTDHKRGPGFTRGLGHGVDLNHIYGETLDRQHKLRLFKDGKLKYQVIGG 245 COX2 G.gal HQFFKTDHKKGPGFTKAYGHGVDLNHIYGETLERQLKLRLRKDGKLKYQMIDG 245 COX2 F.het HQFFKSDMKNGPAFTVAKGHGVDLGHIYGENLEKQHKLRLFKDGKLKYTMVNG 250 *R277 COX1 Mus EVYPPSVEQASVLMRYPPGVPPERQMAVGQEVFGLLPGLMLFSTIWLREHNRV 314 COX1 G.gal EVYPPSVTEVPVHMVYPPAIPKEKQLAMGQEVFGLLPGLCMYATLWLREHNRV 357 COX1 Danio DIYPPTVLHAQVKMSYPPSVPPEQQLAIGQEVFGLLPGLGMYATLWLREHNRV 312 lCOX EAYPPVVRDAPVHMVYPEHVPASLRFAVGHEVYGLLPGLLVYATVWLREHNRV 106 COX2 Mus EVYPPTVKDTQVEMIYPPHIPENLQFAVGQEVFGLVPGLMMYATIWLREHNRV 298 COX2 G.gal EMYPPTVKDTQAEMIYPPHVPEHLQFSVGQEVFGLVPGLMMYATIWLREHNRV 298 COX2 F.het EVYPPLVKDVGVEMHYPPHVPDSQRFAVGHEAFGLVPGLLMYATIWLREHNRV 303 COX1 Mus CDLLKEEHPTWDDEQLFQTTRLILIGETIKIVIEEYVQHLSGYFLQLKFDPEL 367 COX1 G.gal CDILKQEHPTWGDEQLFQTARLILIGETIKIVIEDYVQHLSGYFLNLKFDPEL 410 COX1 Danio CEILKQEHPTWGDEQLFQTARLIIIGETIRIVIEEYVQHLSGYRLKLHFDPTL 365 lCOX CDVLHARHPRWDDERLFQTARLILTGETMKIVIEEYVQHLSGYNFHLKFDPTL 159 COX2 Mus CDILKQEHPEWGDEQLFQTSRLILIGETIKIVIEDYVQHLSGYHFKLKFDPEL 351 COX2 G.gal CDVLKQEHPEWDDEQLFQTTRLILIGETIKIVIEDYVQHLSGYHFKLKFDPEL 351 COX2 F.het CDVLKGVHPDWDDERLFQTTRLILIGETIKIVIEDYVQHLSGYHFKLKFDPEL 356 *Y385/H388 *N410 COX1 Mus LFRAQFQYRNRIAMEFNHLYHWHPLMPNSFQVGSQEYSYEQFLFNTSMLVDYG 420 COX1 G.gal LFEQQFQYRNRIAVEFNQLYHWHALMPDSFTIQGQEYSYEQFLYNTSMLMDYG 463 COX1 Danio LFNSQFQYQNRISVEFNQLYHWHPLMPDSFYIDGDHIQYSKFIFNTSILTHYG 418 lCOX LFGVNFQYSNRMSLEFNHLYHWHPLMPDSLLIDGRNYSYDEFLFNPGLLADKK 212 COX2 Mus LFNQQFQYQNRIASEFNTLYHWHPLLPDTFNIEDQEYSFKQFLYNNSILLEHG 404 COX2 G.gal LFNQRFQYQNRIAAEFNTLYHWHPLLPDTFQIHNQEYTFQQFLYNNSIMLEHG 404 COX2 F.het LFNQRFQYQNRIASEFNTLYHWHPLMPDSFHIEEKDYSYKEFVFNTSVVTEHG 409 COX1 Mus VEALVDAFSRQRAGRIGGGRNFDYHVLHVAVDVIKESREMRLQPFNEYRKRFG 473 COX1 G.gal VEALAESFSMQTAGRIGGGQNINANVLGVAVGVIEESRQLRLQPFNEYRKRFG 516 COX1 Danio LEKLVEAFSIQPAGQIGGGHNIHPVVSGVAERVIVESRELRLQPFNEYRKRFN 471 lCOX LMPLVRSFMRQRAGTVSGGRNINKNLLHVATSIIEHGRTLRLQSLNQYRHRFN 265 COX2 Mus LTQFVESFTRQIAGRVAGGRNVPIAVQAVAKASIDQSREMKYQSLNEYRKRFS 457 COX2 G.gal LSHMVKSFSKQSAGRVAGGKNVPAAVQKVAKASIDQSRQMRYQSLNEYRKRFM 457 COX2 F.het INNLVDSFSKQIAGRVAGGRNVPGPIMYVAIKSIENSRKMRYQSLNAYRKRFS 462 Figure 2-4. Amino acid alignment lCOX seque nced from the gills of the lamprey ( Petromyzon marinus ), with COX-1 and COX-2 forms fr om vertebrates (Accession # NP_032995, NP_035328, XP_425326, XP_422297, AAS21313, and NP_705942). Amino acids that share at least 60% identity across se quences are highlighte d. *Indicates amino acids (numbered based on ovine COX-1) involved in COX function: active site of COX (tyrosine-385 and histidine-388), N -glycosylation site (asparagine-410), and sites for peroxidase activity (glu tamine-203 and histidine-207) (Yang et al ., 2002; Ishikawa et al ., 2007). The black boxes indicate th e two domains that define the haem-binding sites (Ishikawa et al ., 2007). The dashed line represents the arginine277 loop region, which has been shown to differ between COX-1 and COX-2 forms in mammals (Yang et al ., 2003; Kulmacz et al ., 2003). The numbers to the right of the sequences indicate the amino acid number of each sequence. 80
PAGE 81
COX1 Mus LKPYTSFQELTGEKEMAAEL 493 COX1 G.gal LKPYTSFQELTGEEDKAAEL 536 COX1 Danio LKPYTSFAELTGEQEMSKEL 491 lCOX MRPFTSFLELTGDEAMAAEM 285 COX2 Mus LKPYTSFEELTGEKEMAAEL 477 COX2 G.gal LKPFKSFEELTGEKEMAAEL 477 COX2 F.het MKPYTSFEDLTGEKEMAAIL 482 Figure 2-4. Continued. 81
PAGE 82
++ + COX1 Mus MSRRSLSLWFPPLLLLLLPPTPSVLLADPGVPSPVNPCCYYPCQNQGVCVRFG 53 COX1 Danio MRECNFLLKWTVI--LLLSVSFCAGEESPTSSNTANPCCYYPCQNQGICVRYG 51 COX1 Ovis MSRQSISLRFPLL--LLLLSPSPVFSADPGAPAPVNPCCYYPCQHQGICVRFG 51 hCOX MTSEVFVVLCVAVVFA-GAAAAD------------DPCCGSPCENKGVCVSVG 40 COX2 Mus MLFRAVLLCAALGLSQAA-----------------NPCCSNPCQNRGECMSTG 36 COX2 F.het MNRITFAVFLLALCFSFHKEVLG------------NACCSEPCQNRGVCTAMG 41 COX2 Ovis MLARALLLCAAV-VCGAA-----------------NPCCSHPCQNRGVCMSVG 35 + +*N68 + + + + + ++ + COX1 Mus LDNYQCDCTRTGYSGPNCTIPEIWTWLRNSLRPSPSFTHFLLTHGYWLWEFVN 106 COX1 Danio LERYECDCTRTGYYGENCTIPELWTRVYRLLKPSPNVVHYILTHFDWLWDLIN 104 COX1 Ovis LDRYQCDCTRTGYSGPNCTIPEIWTWLRTTLRPSPSFIHFLLTHGRWLWDFVN 104 hCOX FEDYECDCTRTGYYGSNCTYPETWTWIVNMLKPLPDTVHYFLTHFAPFWSLVN 93 COX2 Mus FDQYKCDCTRTGFYGENCTTPEFLTRIKLLLKPTPNTVHYILTHFKGVWNIVN 89 COX2 F.het SDSYECDCTRTGYRGQNCTTPEFLTWIKISLKPTPNTVHYILTHFKGFWNIIN 94 COX2 Ovis FDQYKCDCTRTGFYGENCTTPEFLTRIKLLLKPTPDTVHYILTHFKGVWNIVN 88 + ++ + *R120+ *N144 + COX1 Mus -ATFIREVLMRLVLTVRSNLIPSPPTYNSAHDYISWESFSNVSYYTRILPSVP 158 COX1 Danio R-SFLRDWLMRKVLTVRANLIPSPPTYNSRYDYLNWEAYSNITYYTRILPPVP 156 COX1 Ovis -ATFIRDTLMRLVLTVRSNLIPSPPTYNIAHDYISWESFSNVSYYTRILPSVP 156 hCOX RAAFLRDRVMRHVLMSRAHMVSSPPTYNADYGYKSWEAYSNLSYFTRLLPPVP 146 COX2 Mus NIPFLRSLIMKYVLTSRSYLIDSPPTYNVHYGYKSWEAFSNLSYYTRALPPVA 142 COX2 F.het NISFLRDAIMKYVLTSRSHMIDSPPTYNADYGYKSWEAYSNLSYYTRALPPVP 147 COX2 Ovis KISFLRNMIMRYVLTSRSHLIESPPTYNVHYSYKSWEAFSNLSYYTRALPPVP 141 Figure 2-5. Amino acid alignment of lCOX se quenced from the gills of the lamprey ( Petromyzon marinus ), with COX-1 and COX-2 form s from vertebrates (Accession # NP_032995, NP_035328, XP_425326, XP_422297, AAS21313, and NP_705942). Amino acids that share at least 60% iden tity across sequences are highlighted. *Indicates amino acids (numbered based on ovine COX-1) involved in COX function: active site of COX (tyrosine-385, histidine-388, and serine-530), substrate binding site (arginine-120), N -glycosylation sites (asparagine-68, 144, and 410), sites for peroxidase activity (glutamine-203 a nd histidine-207), and the two sites which define conformational differences in channels between COX-1 and COX-2 in mammals (arginine-513 a nd valine-523) (Yang et al ., 2002; Ishikawa et al ., 2007). +Indicates amino acids unique to hCOX. The black boxes indicate the two domains that define the haem-binding sites and the dashed box indicates the membrane binding domain (Kulmacz et al ., 2003; Ishikawa et al ., 2007). The solid lines represent disulfide bonds, with the beginni ng and end of the longest bond represented by arrows (Kulmacz et al ., 2003). The dashed line repr esents the arginine-277 loop region, which has been shown to differ between COX-1 and COX-2 forms in mammals (Yang et al ., 2003; Kulmacz et al ., 2003). The numbers to the right of the sequences indicate the amino acid number of each sequence. 82
PAGE 83
+ + + + *Q203/H207 COX1 Mus KDCPTPMGTKGKKQLPDVQLLAQQLLLRREFIPAPQGTNILFAFFAQHFTHQF 211 COX1 Danio NDCPTPMGTKGKIKLPDPKLLVEKFMLRRNFRLDPQGTNLMFAFFAQHFTHQF 209 COX1 Ovis RDCPTPMDTKGKKQLPDAEFLSRRFLLRRKFIPDPQSTNLMFAFFAQHFTHQF 209 hCOX QGCPTPMGRTGKKELPDAQKLAERFLLRRTFIPDPQGSNLLFAFFAQHFTHQF 199 COX2 Mus DDCPTPMGVKGNKELPDSKEVLEKVLLRREFIPDPQGSNMMFAFFAQHFTHQF 195 COX2 F.het EDCPTPMGVVGKKELPDVKVLAEKLLVRRRFIPDPQGTSLMFAFFAQHFTHQF 200 COX2 Ovis DDCPTPMGVKGRKELPDSKEVVKKVLLRRKFIPDPQGTNLMFAFFAQHFTHQF 194 + COX1 Mus FKTSGKMGPGFTKALGHGVDLGHIYGDNLERQYHLRLFKDGKLKYQVLDGEVY 264 COX1 Danio FKTHNRVGLGFTKGLGHGVDAGHIYGDSLDRQLELRLHKDGKLKYQVLNGDIY 262 COX1 Ovis FKTSGKMGPGFTKALGHGVDLGHIYGDNLERQYQLRLFKDGKLKYQMLNGEVY 262 hCOX FKTDFKRGPGFTKALGHGVDLSHIYGDTLDKQHKLRLHNNGKLKFQMIDGEVF 252 COX2 Mus FKTDHKRGPGFTRGLGHGVDLNHIYGETLDRQHKLRLFKDGKLKYQVIGGEVY 248 COX2 F.het FKSDMKNGPAFTVAKGHGVDLGHIYGENLEKQHKLRLFKDGKLKYTMVNGEVY 253 COX2 Ovis FKTDIERGPAFTKGKNHGVDLSHVYGESLERQHNRRLFKDGKMKYQMINGEMY 247 + + *R277++ COX1 Mus PPSVEQASVLMRYPPGVPPERQMAVGQEVFGLLPGLMLFSTIWLREHNRVCDL 317 COX1 Danio PPTVLHAQVKMSYPPSVPPEQQLAIGQEVFGLLPGLGMYATLWLREHNRVCEI 315 COX1 Ovis PPSVEEAPVLMHYPRGIPPQSQMAVGQEVFGLLPGLMLYATIWLREHNRVCDL 315 hCOX PPLVSEAPVDMIYPPHVPEAARFAVGHEAFGLVPGLMLYGTLWLREHNRVCDV 305 COX2 Mus PPTVKDTQVEMIYPPHIPENLQFAVGQEVFGLVPGLMMYATIWLREHNRVCDI 301 COX2 F.het PPLVKDVGVEMHYPPHVPDSQRFAVGHEAFGLVPGLLMYATIWLREHNRVCDV 306 COX2 Ovis PPTVKDTQVEMIYPPHIPEHLKFAVGQEVFGLVPGLMMYATIWLREHNRVCDV 300 + COX1 Mus LKEEHPTWDDEQLFQTTRLILIGETIKIVIEEYVQHLSGYFLQLKFDPELLFR 370 COX1 Danio LKQEHPTWGDEQLFQTARLIIIGETIRIVIEEYVQHLSGYRLKLHFDPTLLFN 368 COX1 Ovis LKAEHPTWGDEQLFQTARLILIGETIKIVIEEYVQQLSGYFLQLKFDPELLFG 368 hCOX LDVDHPEWDDERIFQTARLILIGETIKIVIEDYVQHLSGYHFKLKFDPKLLFG 358 COX2 Mus LKQEHPEWGDEQLFQTSRLILIGETIKIVIEDYVQHLSGYHFKLKFDPELLFN 354 COX2 F.het LKGVHPDWDDERLFQTTRLILIGETIKIVIEDYVQHLSGYHFKLKFDPELLFN 359 COX2 Ovis LKQEHPEWGDEQLFQTSRLILIGETIKIVIEDYVQHLSGYHFKLKFDPELLFN 353 + *Y385/H388 + *N410 + ++ COX1 Mus AQFQYRNRIAMEFNHLYHWHPLMPNSFQVGSQEYSYEQFLFNTSMLVDYGVEA 423 COX1 Danio SQFQYQNRISVEFNQLYHWHPLMPDSFYIDGDHIQYSKFIFNTSILTHYGLEK 421 COX1 Ovis AQFQYRNRIAMEFNQLYHWHPLMPDSFRVGPQDYSYEQFLFNTSMLVDYGVEA 421 hCOX QPFQYSNRISVEFNHLYHWHGLNPDAFRVGTQEYQYSQFLFNNTILLNHGVRG 411 COX2 Mus QQFQYQNRIASEFNTLYHWHPLLPDTFNIEDQEYSFKQFLYNNSILLEHGLTQ 407 COX2 F.het QRFQYQNRIASEFNTLYHWHPLMPDSFHIEEKDYSYKEFVFNTSVVTEHGINN 412 COX2 Ovis QQFQYQNRIAAEFNTLYHWHPLLPDVFQIDGQEYNYQQFIYNNSVLLEHGVTQ 406 ++ + + + ++ + + COX1 Mus LVDAFSRQRAGRIGGGRNFDYHVLHVAVDVIKESREMRLQPFNEYRKRFGLKP 476 COX1 Danio LVEAFSIQPAGQIGGGHNIHPVVSGVAERVIVESRELRLQPFNEYRKRFNLKP 474 COX1 Ovis LVDAFSRQPAGRIGGGRNIDHHILHVAVDVIKESRVLRLQPFNEYRKRFGMKP 474 hCOX LLEAFNVQQAGRIGGGQNIHGALLHVATASIKHGRKMRFQSLNQYRKQFGLQP 464 COX2 Mus FVESFTRQIAGRVAGGRNVPIAVQAVAKASIDQSREMKYQSLNEYRKRFSLKP 460 COX2 F.het LVDSFSKQIAGRVAGGRNVPGPIMYVAIKSIENSRKMRYQSLNAYRKRFSMKP 465 COX2 Ovis FVESFTRQIAGRVAGRRNLPAAVEKVSKASLDQSREMKYQSFNEYRKRFLLKP 459 *V523 COX1 Mus YTSFQELTGEKEMAAELEELYGDIDALEFYPGLLLEKCQPNSIFGESMIEMGA 529 COX1 Danio YTSFAELTGEQEMSKELEELYGHIDAMEFYPALLLEKTRPGAVFGESMVEMGA 527 COX1 Ovis YTSFQELTGEKEMAAELEELYGDIDALEFYPGLLLEKCHPNSIFGESMIEMGA 527 hCOX YQSFEQLTGETEMAADLAELYSDINAMEFYLGLMVEKPRQGALFGETMVEAGA 517 COX2 Mus YTSFEELTGEKEMAAELKALYSDIDVMELYPALLVEKPRPDAIFGETMVELGA 513 COX2 F.het YTSFEDLTGEKEMAAILEELYGDVDAVELYPGLLVEKPRTNAIFGETMVEMGA 518 COX2 Ovis YESFEELTGEKEMAAELEALYGDIDAMELYPALLVEKPAPDAIFGETMVEAGA 512 Figure 2-5. Continued 83
PAGE 84
*S530 + + + + + COX1 Mus PFSLKGLLGNPICSPEYWKPSTFGGDVGFNLVNTASLKKLVCLNTK-TCPYVS 581 COX1 Danio PFSLKGLMGNPICSPDYWKPSTFGGKTGFDIVNSATLKKLVCLNTK-WCPYVS 579 COX1 Ovis PFSLKGLLGNPICSPEYWKASTFGGEVGFNLVKTATLKKLVCLNTK-TCPYVS 579 hCOX PFSLKGLMGNAICSPEYWKPSTFGGNRGFEIVNSASLRRLVCLNLQGPCPDVA 570 COX2 Mus PFSLKGLMGNPICSPQYWKPSTFGGEVGFKIINTASIQSLICNNVKG-CPFTS 565 COX2 F.het PFSLKGLMGNPICSPEYWKPSTFGGTAGFDIVNTASLQRLVCNNVRGPCPVAS 571 COX2 Ovis PFSLKGLMGNPICSPEYWKPSTFGGEVGFKIINTASIQSLICSNVKG-CPFTS 564 + + + ++ COX1 Mus FR-VPDYPGDDGSVLVRR-------------------STEL 602 COX1 Danio FHTPPSDYKPQRTS-----------------------HGEL 597 COX1 Ovis FH-VPDPRQEDRPGVERP-------------------PTEL 600 hCOX FH-VPRDNQQDVVVNVTGSQGGSDGVTTTPHYVADQQSREL 609 COX2 Mus FN-V-QDPQPTKTATINASASHSRLDDINPTVLIKRRSTEL 603 COX2 F.het FY-VP-DVKETGSMTINSSTSHSRDSNINPTVILKERTSEL 609 COX2 Ovis FS-V-QDAHLTKTVTINASSSHSGLDDINPTVLLKERSTEL 602 Figure 2-5. Continued 84
PAGE 85
*H207 COXa Ciona HMFFKTDPMKGMPYQWGDQLVDLSQIYGHGEKRQHELRSHVNGKLKVSLVDGH 279 COXb Ciona HQFFKTNTIKGMPFQWGEHSVDLSHVYGHTIQRQHELRSHIDGKLKVFETNGE 226 COXc HQFFKTDFKKGAGRTWGDHGVDLSHIYGETVERQHQLRSFTDGKLKFQRVEGE 53 COXd HQFFKTDFKKGPGRTWSDHAVDMSHVYGETVERQRQLRSFQDGKLKYQLVDGE 53 *R277 COXa Ciona EFPPLSNQTTANMSNINLLPQEYQFVFGHQGFSLMPTFLIWSTIWLREHNRIC 332 COXb Ciona VFPPLTESANVTMSGEKLM-RGRKFAIGHPGFGAFPSFFVIATLWLREHNRVC 279 COXc VYPPSLADAPVHMIYPPYVPEGKRFAIGHEFFGLLPGLFVYSTVWLREHNRVC 106 COXd AFPPSLQDASVHMIYPPDVPEKKRFALGHEFFGLLPGLFVWATVWLREHNRVC 106 COXa Ciona DLIKEENPAWDDERIFQTARLVLTGETIKVVIEDYVQHLSGFHYKLLYDPELV 385 COXb Ciona DILKDLHPDWDDERLFQTARLILTGETLKIIVEDYVQHVSGFHFQLSYDPEIL 332 COXc DVMKELHPDWDDERLFQTARLILTGETINIIINEYVQHLSGYNFDLFWDPELL 159 COXd DVMKDLHPDWDDERLYQTARLILISETINIVIGEYVQHLAGKNFQLFWDPELL 159 COXa Ciona QGGSHSFHNQIHVEFQL 402 COXb Ciona HKSTFSYNNQIHAEFHI 349 COXc FSDQFQYQNRIFVEFNH 176 COXd FEEQXQYQNSIFVEFNH 176 Figure 2-6. Amino acid alignment of COXc and COXd sequences from the lancelet ( Branchiostoma lancelolatum ) with COXa and COXb sequences from the sea squirt ( Ciona intestinalis ). Amino acids that share at least 60% identity across sequences are highlighted. *Indicates the amino acid (numbered based on ovine COX-1) involved in peroxidase ac tivity (histidine-207) (Yang et al ., 2002; Ishikawa et al ., 2007). The black boxes indicate the two domains that define the haem-binding sites (Ishikawa et al ., 2007). The dashed line repres ents the arginine-277 loop region, which has been shown to differ between COX-1 and COX-2 forms in mammals (Yang et al ., 2003; Kulmacz et al ., 2003). The numbers to the right of the sequences indicate the amino acid number of each sequence. 85
PAGE 86
86
PAGE 87
Figure 2-7. Evolutionary hist ory of COX sequences inferred using the minimum evolution method with a distance optimality criterion. The optimal tree with the sum of branch length = 6.63322594 is shown. The percentage of replicate trees in which the associated taxa clustered toge ther in the bootstrap test ( 1000 replicates) is shown next to the branches (Felsenstein, 1985). The tree is drawn to scale, with branch lengths in the same units as those of the evolutiona ry distances used to infer the phylogenetic tree. The evolutionary distances were co mputed using the JTT matrix-based method (Jones et al ., 1992) and are in the units of the number of amino acid substitutions per site (scale bar = 0.05 amino acid substitutions per site). The rate variation among sites was modeled with a gamma distribution (s hape parameter = 0.737). The ME tree was searched using the Close-Neighbor-Interchange (CNI) algorithm at a search level of 1. The Neighbor-Joining algorithm was used to generate the initial tree. All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option). There were a total of 147 pos itions in the final dataset. Phylogenetic analyses were conducted in MEGA4 (Tamura et al ., 2007). Sequences enclosed in a box are novel and reported here for the first time. 87
PAGE 88
COX1b Gasterosteus COX1b Myoxocephalus COX1b Oryzias COX1b Fundulus COX1b Tetraodon COX1 Oncorhynchus COX1 Danio COX1a Tetraodon COX1a Fundulus COX1a Oryzias COX1a Gasterosteus COX1a Myoxocephalus COX1 Mus COX1 Rattus COX1 Bos COX1 Ovis COX1 Oryctolagus COX1 Homo COX1 Canis COX1 Gallus COX1 Monodelphis COX1b Monodelphis COX1b2 Monodelphis COX1 Xenopus COX2 Gasterosteus COX2 Myoxocephalus COX2b Oncorhynchus COX2 Oryzias COX2 Fundulus COX2 Tetraodon COX2b Danio COX2a Danio COX2a Oncorhynchus COX2 Xenopus COX2 Bos COX2 Ovis COX2 Gallus COX2 Homo COX2 Mus COX2 Rattus COX2 Monodelphis COX2 Canis COX2 Oryctolagus sCOX Squalus COXc Branchiostoma COXd Branchiostoma COXa Ciona COXb Ciona 100 99 94 71 57 80 44 61 99 62 86 87 34 51 92 98 85 37 99 71 60 26 32 35 91 93 52 82 43 39 41 34 40 70 26 77 94 83 39 20 32 38 41 38 13 0.1 88
PAGE 89
Figure 2-8. Evolutionary hist ory of COX sequences excluding hagfish (hCOX) and lamprey (lCOX) sequences inferred using the mi nimum evolution method with a distance optimality criterion. The optimal tree with the sum of branch length = 6.18990857 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap te st (1000 replicates) is show n next to the branches (Felsenstein, 1985). The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the JTT matrix-based method (Jones et al ., 1992) and are in the units of the number of amino acid substitutions per site (scale bar = 0.1 amino acid substitutions per site ). The rate variation among sites was modeled with a gamma distribution (sha pe parameter = 0.742). The ME tree was searched using the Close-Neighbor-Interchange (CNI) algorithm at a search level of 1. The Neighbor-Joining algorithm was used to generate the initial tree. All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option). There were a total of 147 pos itions in the final dataset. Phylogenetic analyses were conducted in MEGA4 (Tamura et al ., 2007). 89
PAGE 90
Figure 2-9. Evolutionary hi story of COX sequences infe rred using a maximum liklihood optimality criterion. The most likely tree with an Ln L score = -25294.222410 is shown. The number of replicate trees in wh ich the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches (Felsenstein, 1985). The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the WAG matrix-b ased method (Whelan and Goldman, 2001) and are in the units of the number of ami no acid substitutions per site (scale bar = 0.1 amino acid substitutions per site). The rate variation among sites was modeled with a gamma distribution (shape parameter estimated = 0.737). Phylogenetic analyses were conducted using PhyML (Guindon and Gascue l, 2003). Sequences enclosed in a rectangle are novel and reported here for the first time. 90
PAGE 91
91
PAGE 92
Figure 2-10. Evolutionary history of COX se quences excluding sequences from the hagfish (hCOX) and lamprey (lCOX) inferred using a maximum liklihood optimality criterion. The most likely tree with an Ln L score = -24242.30059 is shown. The number of replicate trees in which the a ssociated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches (Felsenstein, 1985). The tree is drawn to scale, with branch lengths in the sa me units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the WAG matrix-b ased method (Whelan and Goldman, 2001) and are in the units of the number of ami no acid substitutions per site (scale bar = 0.1 amino acid substitutions per site). The rate variation among sites was modeled with a gamma distribution (shape parameter estimated = 0.742). Phylogenetic analyses were conducted using PhyML (Guindon and Ga scuel, 2003). Sequence names are abbreviated from Figure 2-9. 92
PAGE 93
93
PAGE 94
Figure 2-11. Alternate likelihood ra tio test (aLRT) scores for bran ch support of COX sequences. The most likely tree is shown and the topology, branch lengths, and model of evolution are the same as in Figure 2-9. Scores represent the gain in likelihood when collapsing a particular branch and are inde pendent of bootstrap sc ores (Felsenstein, 1985). Scores were interpreted using a c onservative estimate of the minimum of parametric chi-squared and non-parametric Shimodaira-Hasegawa-like (SH) support value interpretations (Anisimova and Gascue l, 2006). Analysis was performed using PhyML (Guindon and Gascuel, 2003). Support va lues of interest to the present study are enclosed in boxes. 94
PAGE 95
95
PAGE 96
Figure 2-12. Alternate likelihood ra tio test (aLRT) scores for br anch support of COX sequences excluding hagfish (hCOX) and lamprey (lCOX) sequences. The most likely tree is shown and the topology, branch lengths, and model of evolution are the same as in Figure 2-11. Scores represent the gain in likelihood when collapsing a particular branch and are independent of bootstrap sc ores (Felsenstein, 1985). Scores were interpreted using a conservative estimate of the minimum of parametric chi-squared and non-parametric Shimodaira-Hasegawa-l ike (SH) support valu e interpretations (Anisimova and Gascuel, 2006). Analys is was performed using PhyML (Guindon and Gascuel, 2003). Support values of intere st to the present study are enclosed in boxes. 96
PAGE 97
Figure 2-13. Evolutionary history of COX se quences inferred using a Bayesian optimality criterion. The consensus tree is shown. The posterior probabilities (10 million generations) are shown next to the branches. Trees were sampled every 500 generations with a burnin of 10% (401 trees). The tree is dr awn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the WAG matrixbased method (Whelan and Goldman, 2001) and are in the units of the number of amino acid substitutions per site (scale bar = 0.1 amino acid substitutions per site). The rate variation among sites was mode led with a gamma distribution (shape parameter estimated = 0.672275). Phylogene tic analyses were conducted using MrBayes 3.1.2. Sequences enclosed in a re ctangle are novel and reported here for the first time. 97
PAGE 98
98
PAGE 99
Figure 2-14. The evolutionary history of COX sequences excluding hagfish (hCOX) and lamprey (lCOX) sequences inferred using a Bayesian optimality criterion. The consensus tree is shown. The posterior probabilities (10 million generations) are shown next to the branches. Trees were sampled every 500 generations with a burnin of 10% (401 trees). The tree is drawn to scal e, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the WAG matrix-based method (Whelan and Goldman, 2001) and are in the units of the number of amino acid substitutions per site (scale bar = 0.1 amino acid substitutio ns per site). The rate variation among sites was modeled with a gamma distri bution (shape parameter estimated = 0.667978). Phylogenetic analyses were conducted using MrBayes 3.1.2. 99
PAGE 100
a) b) c) d) Figure 2-15. The 21 alternate tr ee topologies generated to test the effects of relocating the hagfish (hCOX) and lamprey (lCOX) sequences on the likelihood score (Ln L ). Each topology is simplified, but all relationships remain the same as in the most likely topology (Figure 2-9) with the exception of hCOX and lCOX. Each topology was generated using TreeView1.6.6 based on the most likely topology. Ln L was calculated for each topology online using PhyML (Guindon et al ., 2005), branch lengths and model parameters were estimate d. The topologies are named in Table 2-9 as follows (from most to least likely) : a) COX2 Hag (most likely) b) COX1 Lamp/COX2 Hag c) COX2 Agnatha d) Lamp/Amp e) COX2 Vert f) COX1 Hag g) COX1 Vert h) Ancestral Hag i) COX1 Agnatha j) COX1 Hag/COX2 Lamp k) Ancestral Agnatha l) COX1 Lamp m) COX2 Lamp n) Agntha/Amp o) Hag/Amp p) Hag Internal q) Lamp Internal r) Lamp Mammal s) Lamp Teleost t) Hag Teleost u) Hag Mammal. 100
PAGE 101
e) f) g) h) i) j) Figure 2-15. Continued. 101
PAGE 102
k) l) m) n) o) 2 p) Figure 2-15. Continued. 102
PAGE 103
q) r) s) t) u) Figure 2-15. Continued. 103
PAGE 104
104 Figure 2-16. Unrooted phylogeny showing the mo st likely scenario for COX evolution in the chordates based on several other analyses. Clades of interest are highlighted on the right and sequences that repr esent unclear placements are highlighted in red boxes.
PAGE 105
CHAPTER 3 SUMMARY AND FUTURE DIRECTIONS Summary This thesis is the first attempt to charact erize the evolution of the enzyme cyclooxygenase in the phylum chordata. Traditional molecular protocols such as clon ing and polymerase chain reactions were used to search for cyclooxygenase sequences in a wide ra nge of evolutionarily key chordates. Using these novel sequences an d pre-existing sequences from GenBank and genomic databases, phylogenetic analyses were pe rformed to reveal the identity of the novel sequences and create the most lik ely scenario for the evolutiona ry history of cyclooxygenase in the chordates. Novel COX Sequences Chapter 2 describes the process of obtai ning and characterizing novel cyclooxygenase (COX) sequences in amphioxus ( Branchiostoma lancelolatum ), the Atlantic hagfish (Myxine glutinosa ), the sea lamprey ( Petromyzon marinus ), the euryhaline killifish ( Fundulus heteroclitus ), and the longhorn sculpin ( Myoxocephalus octodecemspinosus ). Total RNA was isolated and reverse transcribed from gill tis sue (or whole animals in amphioxus) and the resulting cDNA was exposed to PCR in the presence of degenerate and/or sp ecific primers. The products were then visualized and cloned. The resulting plasmids were sequenced at the Marine DNA Sequencing Center at the MDIBL. Complete amino acid sequences of putative COX proteins from the hagfish, killifish, and scul pin and incomplete amino acid sequences from amphioxus and the lamprey were predicted based on the sequences found. These putative amino acid sequences were aligned w ith known COX protein sequen ces and conserved domains, regions, and residues critical to COX function were describe d. Based on the novel sequences identity to known COX forms and phylogenetic an alyses (see below), it was concluded that 105
PAGE 106
COX-1a, COX-1b, and COX-2 are found in the sculpin; COX-1a and COX-1b are found in the killifish; a new COX form named lCOX is found in the lamprey; a new COX form named hCOX is found in the hagfish; and two ancestra l forms named COXc and COXd are found in amphioxus. Phylogenetic Analyses Chapter 2 also describes the phylogenetic analyses used to reconstr uct the evolutionary history of cyclooxygenase in the chordates. The nine novel COX protein sequences described here, as well as 41 additional COX protein sequences from GenBank and genomic databases, were subjected to phylogenetic analyses using three optimality criteria: minimum evolution/distance, maximum likelihood, and Baye sian phylogenetics. In each analysis, two clades were well-supported in the gnathostomes: a COX-1 clade and a COX-2 clade. Each of these clades contained a well suppor ted teleost and tetrapod clade. In the teleost clades, clades for COX-1a, COX-1b, COX-2a, and COX-2b were well-supported, confirming the findings of Ishikawa et al (2007) and Ishikawa and Herschman (2007). The placement of the more evolutionarily ancestral chordates was ultim ately uncertain, although it appears that the urochordates, cephalochordates, and vertebrate s (containing COX-1 and COX-2) each represent an independent origin of COX proteins. Th e hagfish and lamprey sequences most likely represent basal members of the COX-2 and COX-1 clades of vertebrates, respectively; however, these forms may belong closer to the root of the tree. Future Directions Although here I report 9 novel cyclooxygena se (COX) sequences from a range of chordates and present a likely scenario for the ev olution of this gene family in the chordates, many questions remain regarding COX evolution a nd function in the chordates. These include determining the true history of C OX evolution in the earliest chor dates, pinpointing the origin of 106
PAGE 107
COX-1 and COX-2, and investigating the history of COX duplication and loss in the teleosts. Finally, the physiological function of COX in the earliest chordate s has only been preliminarily investigated. Completion of novel sequences Several of the novel sequences documented here could not be completed despite multiple efforts using many primer sets and PCR techniqu es. Notably, only 532 bp were obtained of the sequences from amphioxus (COXc and COXd). Co mpared to the complete open reading frame of COX transcripts in other chordates (rangi ng from about 1700-2500 bp), this indicates that about 75% of these sequences remain unknown. Completing the rest of these sequences or thoroughly searching the recently complete genome of Branchiostoma floridae may yield putative proteins that provide novel groupings in phylogenetic analyses. Based on the available sequence data, however, COXc and COXd are we ll-supported as being n ear the base of the chordate COX lineage, sister to the vertebra te COX-1 and COX-2 clade. The lCOX sequence obtained from the lamprey is also only about 40% complete. Completing the open-reading frame of this putative protein or searching genomic data bases could possibly give a more definite view of its identity as this sequ ence groups with different clad es depending on the phylogenetic analysis used. In actuality, th e short size of lCOX compared to the other COX sequences used in the phylogenetic analyses may have contributed to its tendency to a ppear in different clades in different analyses. Finally, the COX-1a sequen ce obtained from the sculpin may warrant further investigation and reaffirmation via additional sequencing since many residues critical to COX function are missing. Additional Phylogenetic Analysis Unfortunately, the evolutionary relationship of the COX proteins in the chordates was not able to be unequivocally resolved, despite the use of three different phylogenetic approaches. 107
PAGE 108
This may be due to the incompleteness of some sequences and a lack of sufficient sampling across taxa or may represent a hi story that is irresolvable und er the current circumstances. However, the phylogenetic techniques used here may also be improved upon in future analyses. Phylogenetically uninformative areas of the se quences (such as near the C and N terminus, which show little homology across sequences) could be removed. Here, protein sequences were used to gain insight into the e volution of COX functionality, but in sights could also be gained by performing similar analyses using DNA sequence data instead of using their inferred protein amino acid sequences. This would allow for gr eater complexity of models to estimate substitution rates and could take additional f actors into account such as codon bias. Using nucleotide versus amino acid would also enable additional search strategies and phylogenetic programs to be used. Choosing the optimum search strategy and computer program may be critical in finding the optimum evolutionary r econstruction. For example, maximum likelihood analyses PhyML (the program used in this study) may find suboptimal likelihood scores when the data-set contains little phylogenetic information (Morrison, 2007). However, programs such as RAxML (Stamatakis, 2006), GARLI (Brauer et al ., 2002), and PAUP* (Rogers and Swafford, 1998) may perform better under the same circum stances (Morrison, 2007). Also, evaluating alternative tree topolog ies using frameworks other th an maximum likelihood may yield alternative results. Additional Sequences Based on the most likely scenario of COX evol ution in the chordates, there are several COX sequences in the chordates examined in this study that remain to be characterized. If hCOX in actuality represents a basal form of CO X-2, then there should also be a COX-1 form in the hagfish. Similarly, if lCOX and sCOX repres ent basal forms of COX-1 in the lamprey and dogfish, then they should also contain versions of COX-2. These additional forms of COX may 108
PAGE 109
be harder to clone and sequence due to accumulated mutations over evolutionary time, which would render the sequences unidentifiable and possi bly non-functional. However, this is not the case in the other vertebrates, as COX-1 and COX2 seem to each have critical functionality. Using Southern blotting techniqu es (Southern, 1975) to identify COX DNA sequences in the DNA of hagfish, lamprey, and dogfish could immedi ately reveal the pres ence of two forms of COX in these animals. If two forms are present, it could be concluded that they likely represent COX-1 and COX-2 forms, and thus the origin of these forms would coincide with the vertebrate lineage (as predicted in the current phylogenetic an alysis). However, if only one form of COX was found the results would suggest a more recent origin of COX-1 and COX-2. This technique could be applied to a wide diversity of chordates to determine how many COX variants they contain. Additional Taxa In this study, a comparativ e evolutionary approach was taken by sampling a wide range of chordates, most of which had no previous record of COX gene s. However, there are still many evolutionarily interesting c hordates that have not been ex amined for COX genes. While COX genes are known from multiple mammals and teleosts, other chordate groups have only been minimally sampled. Most notably is the l ack of any reported COX genes from any reptile. COX has also only been documente d from a single avian species ( Gallus gallus), a single elasmobranch species ( Squalus acanthias ), and a single marsupial ( Monodelphis domestica ). These groups represent major lin eages within the vertebrates and may contain novel forms of COX. The results of this and other recent studies also warran t further investigation within certain chordate groups. For example, evolutio narily ancestral fishes such as the sturgeon, paddlefish, and bowfin could likely contain a single COX-1 and C OX-2 gene, while the teleosts studied here have multiple COX-1 or COX2 genes. Ishikawa and Herschman (2007) 109
PAGE 110
hypothesize that a genome duplication in the tele ost lineage and subsequent loss of either a COX-1 (in the zebrafish/trout lineage) or COX2 gene (in the acanthopterygian lineage) is responsible for this observation. If this is the case, then examining taxa near the base of the teleost lineage may reveal fish with all four variants of COX (COX-1a, COX-1b, COX-2a, and COX-2b). The COX variants from the dogfish shark, lamprey, and hagfish all group weakly with either COX-1 or COX-2. Obtaining C OX sequences from other species of hagfish, lampreys, and elasmobranchs may confirm this in itial relationship or reveal novel groupings. Finally, although support values for the urochorda ta and cephalochordata clades are strong regardless of the phylogenetic technique use d, only one species from each subphylum was included in phylogenetic analyses. Including more species could gi ve a more complete view of COX evolution in the evolutionarily ancestral ch ordates. COX sequences from these additional taxa may be readily available in the near fu ture due to the growing regularity of genome sequencing across many taxa. For example, the recently completed Anolis carolinensis genome contains COX-1 and COX-2 genes. However, due to the length of time required for whole genome projects, the limited number of genomes available, and the difficulty of searching recently completed, non-annotated genomes, studi es that specifically target sequencing COX genes in these taxa are needed. Function of Cyclooxygenases As reviewed in Chapter 1, cyclooxygenase play s several roles in chordates because of its role in prostaglandin production. These functions have been examined in some detail in the mammals and to a lesser degree in the fishes. Ho wever, no studies have examined the functions of the novel COX forms presented here or any no n-vertebrate COX forms in the chordates. Therefore, I have begun preliminary st udies investigating the function of these COX forms. 110
PAGE 111
One preliminary method used to infer possible functions for a gene is to determine where its mRNA transcript or protei n product is expressed in the organism. For example, Choe et al (2006) concluded that COX-2 may play a role in os moregulation in the killif ish partly due to its high level of expression in the gills and its pr esence within mitochondrio n-rich cells. Following similar procedures, I have found that COX-1b mRNA is also expressed highly in the gills of the killifish (Fig. 3-1a). Similar results were also found for COX-1a and COX-2 in the sculpin (Fig. 3-2a, Fig. 3-2b), hCOX in the hagfish (Fig. 3-3a ), and lCOX in the lamprey (Fig. 3-4). To investigate which cells express COX mRNA transcripts, in situ hybridization (ISH) was performed using probes designed against the novel COX sequences described here using previous protocols (Choe et al ., 2006). However, no staining wa s observed for COX transcripts in the gills of the killifish (Fig. 3-1b) or sculpin (Fig. 32b). This is likely due to errors in probe design or hybridization protocol, since COX mRNA was abundant in the gills of both the killifish and sculpin. In the hagfish, probes designed to show hCOX mRNA expression did produce staining patterns in a population of epith elial cells lining gill filaments and lamellae (Fig. 3-3b). This gill structure is similar to those of teleosts but may be functionally different (Evans et al ., 2005) and since ion transport mechanisms in hagfish may not be directly related to those in teleosts (Choe et al ., 1999), it is difficult to assign an osmoregulatory function of hCOX based on this staining pattern. Finally, since COX-2 mRNA expression was shown to increase following salinity transfers using qPCR (Choe et al ., 2006), a similar procedure was used to investigate COX expression after sa linity acclimation in the gills of the killifish (Fig. 3-1c) and sculpin (Fig. 3-2c). The only significant result from this e xperiment was that COX-1b was expressed more abundantly in killifish chronica lly acclimated to fresh water than seawater. However, the same general trend show n for COX-2 in the killifish (Choe et al ., 2006) was also 111
PAGE 112
shown for COX-1b in the killifish: an increase in expression 3 hours af ter salinity transfer (although P = 0.06 in this case) and a return to pr e-transfer expression le vels by 8 hours after transfer. Taken as a whole, these preliminary results suggest that the different COX forms may be playing various roles in osmoregulation in th e killifish, sculpin, lamprey, and hagfish. Clearly, COX function in the non-mammalian chor dates has just begun to be investigated and many questions still remain. Future studies may investigate COX roles in these chordates by examining protein expression patterns usi ng immunohistochemistry, blocking COX function via commercially available COX i nhibitors, or examining COX expression during reproductive events. Of particular interest are the differential roles of the va rious COX forms in the teleosts and the possible partitioning of functionality between COX-1a and COX-1b or COX-2a and COX-2b. 112
PAGE 113
GOB HS I K COX 1b A) B) C) 18S Figure 3-1. Euryhaline killifish ( Fundulus heteroclitus ) COX-1b mRNA expression following procedures outlined in Choe et al (2006). A) COX-1b mRNA expression in various tissues in the killifish. Expression was constant across tissues for the ribosomal protein 18S. COX-1b expression was most abundant in the gills (labeled as G), followed by the heart (H), kidney (K), ope rcular epithelium (O), stomach (S), intestine (I), and brain (B). B) In situ hybridization of COX-1b in the gills of the killifish. C) Expression of COX-1b mRNA in the gills of the killifish following salinity transfers. Expression is normalized to the ribosomal gene L8 and is shown relative to seawater (SW) (FW = fresh water). 113
PAGE 114
GO B H SI K GO B H SIK COX-2 18S COX-1a 18S COX-1a COX-2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 24 hrs 72 hrs TimeExpression 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 24 hrs 72 hrs TimeExpression Figure 3-2. Longhorn sculpin (Myoxocephalus octodecemspinosus ) COX-1a and COX-2 mRNA expression following procedures outlined in Choe et al (2006). A) COX-1a and COX-2 mRNA expression in various tissues in the sculpin. Expr ession was constant across tissues for the ribosomal protein 18S. COX-1a was most abundant in the opercular epithelium (labeled as O), followe d by the gills (G), ki dney (K), heart (H), brain (B), intestine (I), a nd stomach (S). COX-2 was most abundant in the gills, followed by the opercular epithelium, and kidney. B) In situ hybridization of COX1b and COX-2 in the gills of the sculpin. C) Expression of COX-1b and COX-2 mRNA in the gills of the sculpin after 24 and 72 hours following salinity transfers to 100% and 20% seawater. Expression is normalized to the ribosomal gene L8. Seawater 20% Seawater A) B) C) COX-1a COX-2 114
PAGE 115
GMBH S I K hCOX 18S A) B) Antisense Antisense Sense Sense Figure 3-3. Atlantic hagfish ( Myxine glutinosa ) COX (hCOX) mRNA expression following procedures outlined in Choe et al (2006). A) COX mRNA expression in various tissues in the hagfish. Expression was c onstant across tissues for the ribosomal protein 18S. COX (hCOX) expres sion was most abundant in the gills (labeled as G), followed by the heart (H), kidney (K), stomach (S), intestine (I), muscle (M), and brain (B). B) In situ hybridization of hCOX in the gills of the hagfish from two contrary gill cross-sectio ns. Arrows highlight staining, indicating hCOX mRNA expression. Staining for antisense and sense (control) probes is shown. 115
PAGE 116
116 GM BH S I K lCOX 18S Figure 3-4. Sea lamprey ( Petromyzon marinus ) COX (lCOX) mRNA ex pression in various tissues following procedures outlined in Choe et al. (2006). Expr ession was constant across tissues for the ribosomal protei n 18S. COX (lCOX) expression was most abundant in the heart (labeled as H), follo wed by the intestine (I), stomach (S), gills (G), kidney (K), muscle (M), and brain (B)
PAGE 117
LIST OF REFERENCES Abbasi, A. A. and Grzeschik, K. H. (2007). An insight into th e phylogenetic history of HOX linked gene families in vertebrates. BMC Evol. Biol. 7, 239. Allegaert, K., Rayyan, M., and Anderson, B.J. (2006). Impact of ibuprofen administration on renal drug clearance in the first weeks of life. Methods Find. Exp. Clin. Pharmacol. 28, 519-522. Anisimova, M. and Gascuel, O. (2006). Approximate likelihood-ra tio test for branches: A fast, accurate, and powerful alternative. Syst. Biol. 55, 539-552. Arruda, J. A. (1982). Interaction of calcium and cycl ooxygenase inhibitors on transport by the turtle and toad bladders. Arch. Int. Pharmacodyn. Ther. 257, 319-334. Balaji, T., Ramanathan, M., and Menon, V. P. (2007). Localization of cyclooxygenase-2 in mice vas deferens and its effects on fert ility upon suppression using nimesulide: a preferential cyclooxygenase-2 inhibitor. Toxicology. 234, 135-144. Beiche, F., Scheuerer, S., Brune, K., Ge isslinger, G., and Goppelt-Struebe, M. (1996). Upregulation of cyclooxygenase-2 mRNA in the rat spinal cord following peripheral inflammation. FEBS Lett. 390, 165-169. Bicego, K. C., Steiner, A. A., Ant unes-Rodrigues, J., and Branco, L. G. (2002). Indomethacin impairs LPS-induced behavioral fever in toads. J. Appl. Physiol. 93, 512-516. Buonocore, F., Forlenza, M., Randelli, E., Bene detti, S., Boss, P., Meloni, S., Secombes, C. J., Mazzini, M. and Scapigliati, G. (2005). Biological act ivity of sea bass ( Dicentrarchus labrax L.) recombinant interleukin-1beta. Mar. Biotechnol. (NY). 7, 609617. Brauer, M. J., Holder, M. T., Dries, L. A., Zwickl, D. A., Lewis, P. O., and Hillis, D. M. (2002). Genetic algorithms and parallel pr ocessing in maximum-likelihood phylogeny inference. Mol. Biol. Evol. 19, 1717-1726. Breder, C. D., Dewitt, D., and Kraig, R. P. (1995). Characterization of inducible cyclooxygenase in rat brain. J. Comp. Neurol. 355, 296-315. Breder, C. D. and Saper, C. B. (1996). Expression of inducible cyclooxygenase mRNA in the mouse brain after systemic administ ration of bacterial lipopolysaccharide. Brain Res. 713, 64-69. Brubacher, J. L ., Secombes, C. J ., Zou, J ., and Bols, N. C (2000). Constitutive and LPSinduced gene expression in a macrophage-like cell line from the rainbow trout ( Oncorhynchus mykiss). Dev. Comp. Immunol. 24, 565-574. 117
PAGE 118
Cao, C., Matsumura, K., Yamagata, K., and Watanabe, Y. (1996). Endothelial cells of the brain vasculature express cyclooxygenase-2 mR NA in response to systemic interleukin-1: a possible site of prostaglandin synthesis responsible for fever. Brain Res. 733, 263-272. Carrasquer, G and Li, M (2002). Effect of ibuprofen and ro fecoxib transport parameters in the frog corneal epithelium. J. Membr. Biol. 190 127-132. Censarek, P., Freidel, K., Hohlfeld T., Schrr, K. and Weber, A. A. (2006). Human cyclooxygenase-1b is not the el usive target of acetaminophen. Eur. J. Pharmacol. 551, 50-53. Cha, Y. I., Kim, S. H., SolnicaKrezel, L., and Dubois, R. N. (2005). Cyclooxygenase1 signaling is required for vascular tube formation during development. Dev. Biol. 282, 274-283. Cha, Y. I., Solnica-Krezel, L., and DuBois, R. N. (2006a). Fishing for prostanoids: deciphering the developmental f unctions of cyclooxygenase-derived prostaglandins. Dev. Biol. 289, 263-272. Cha, Y. I., Kim, S. H., Sepich, D., Buchanan, F. G., Solnica-Krezel, L., and DuBois, R. N. (2006b). Cyclooxygenase-1-derived PGE2 pr omotes cell motility via the G-proteincoupled EP4 receptor during vertebrate gastrulation. Genes Dev. 20, 77-86. Challis, J. R., Sloboda, D. M., Alfaidy, N., Lye, S. J., Gibb, W., Patel, F. A., Whittle, W. L. and Newnham, J. P. (2002). Prostaglandins and m echanisms of preterm birth. Reproduction. 124, 1-17. Chakraborty, I., Das, S. K.,Wang, J., and Dey, S. K. (1996). Developmental expression of the cyclo-oxygenase-1 and cyclooxygenase-2 genes in the peri -implantation mouse uterus and their differential regulation by th e blastocyst and ovarian steroids. J. Mol. Endocrinol. 16, 107 Choe, K. P., Edwards, S., Morrison-Shetlar A. I., Toop, T., and Claiborne, J. B. (1999). Immunolocalization of Na+/K(+)-ATPase in mitochondrion-rich cells of the atlantic hagfish ( Myxine glutinosa ) gill. Comp. Biochem. Physiol. 124, 161-168. Choe, K. P., Havird, J., Rose, R., Hyndman K., Piermarini, P. and Evans, D. H. (2006). COX2 in a euryhaline teleost, Fundulus heteroclitus : primary sequence, distribution, localization, and potential function in gills during salinity acclimation. J. Exp. Biol. 209 1696-1708. Cohn, S. M., Schloemann, S., Tessner, T., Seibert, K., and Stenson, W. F. (1997). Crypt stem cell survival in the mouse intestinal epitheliu m is regulated by prostaglandins synthesized through cyclooxygenase-1. J. Clin. Invest. 99, 1367-1379. 118
PAGE 119
Diao, H. L., Zhu, H., Ma, H., Tan, H. N., Cong, J., Su, R. W., and Yang, Z. M. (2007). Rat ovulation, implantation and decidualizati on are severely compromised by COX-2 inhibitors. Front. Biosci. 12, 3333-3342. Crollius, H. R. and Wessenbach, J. (2005). Fish genomics and biology. Genome Res. 15, 1675-1682. Cummings, M. P., Handley, S. A., Myers, D. S., Reed, D. L., Rokas, A., and Winka, K. (2003). Comparing bootstrap and posterior pr obability values in the four-taxon case. Syst. Biol. 52, 477-487. DeWitt, D. L. and Smith, W. L. (1988). Primary structure of pros taglandin G/H synthase from sheep vesicular gland determined from the complementary DNA sequence. Proc. Natl. Acad. Sci. USA. 85, 1412-1416. Escalante, B. A., Ferreri, N. R., Dunn, C. E., and McGiff, J. C. (1994). Cytokines affect ion transport in primary cultured thick ascending limb of Henle's loop cells. Am. J. Physiol. 266, C1568-1576. Evans, D. H., Rose, R. E., Roeser, J. M. and Stidham, J. D. (2004). NaCl transport across the opercular epithelium of Fundulus heteroclitus is inhibited by an endothelin to NO, superoxide, and prostanoid signaling axis. Am. J. Physiol. 286 R560-R568. Evans, D. H., Piermarini, P. M. and Choe, K. P. (2005). The multifunctional fish gill: dominant site of gas exchange, osmoregul ation, acid-base regulation, and excretion of nitrogenous waste. Physiol. Rev. 85, 97-177. Felsenstein, J. (1985). Confidence limits on phylogenies : An approach using the bootstrap. Evolution. 39 783-791. Ferreri, N. R., Schwartzman, M., Ibraham, N. G., Chander, P. N., and McGiff, J. C. (1984). Arachidonic acid metabolism in a cell suspen sion isolated from rabbit renal outer medulla. J. Pharmacol. Exp. Ther. 231, 441-448. Ferreri, N. R., An, S. J., and McGiff, J. C. (1999). Cyclooxygenase-2 expression and function in the medullary thick ascending limb. Am. J. Physiol. 277, F360-368. Ferreira, S. H. (1972). Prostaglandins, aspiri n-like drugs and hyperalgesia. Nature 240 200203. Flippin, J. L., Huggett, D., and Foran, C. M. (2007). Changes in the timing of reproduction following chronic exposure to ibuprofen in Japanese medaka, Oryzias latipes Aquat. Toxicol. 81, 73-78. 119
PAGE 120
Flower, R. J. and Vane, J. R. (1972). Inhibition of prostaglandin synthetase in brain explains the anti-pyretic activity of paracetamol (4-acetamidophenol). Nature 240, 410-411. F rster, W. and Parratt, J. R. (1996). The case for low-dose aspirin for the prevention of myocardial infarction: But how low is low? Cardiovasc. Drugs Ther. 10, 727-734. Funk, C. D. (2001). Prostaglandins and leukotrienes : advances in eicosanoid biology. Science 294, 1871-1875. Furlan, R. L., Macari, M., Malheiros, E. B., Secato, E. R., and Guerreiro, J. R. (1998). Effect of indomethacin on hyperthermia induced by heat stress in broiler chickens. Int. J. Biometeorol. 42, 73-76. Gardiner, N. J., Giblett, S., and Grubb, B. D. (1997). Cyclooxygenases in rat spinal cord: selective induction of COX2 during peripheral inflammation. Br. J. Pharmacol. 120, 71P. Gibb, W. and Sun, M. (1996). Localization of prostaglandi n H synthase type 2 protein and mRNA in term human fetal membranes and decidua. J. Endocrinol. 150, 497-503. Gierse, J. K., McDonald, J. J., Hauser, S. D., Rangwala, S. H., Koboldt, C. M. and Seibert, K. (1996). A single amino acid differen ce between cyclooxygenase-1 (COX-1) and -2 (COX-2) reverses the select ivity of COX-2 specific inhibitors. J. Biol. Chem. 271, 15810-15814. Gobbetti, A. and Zerani, M. (2002). Pituitary adenylate cyclase-activating polypeptide induces testicular testoste rone synthesis through PGE(2) mediation in crested newt, Triturus carnifex. J. Exp. Zool. 293, 73-80. Goetz, F. W., Duman, P., Ranjan, M. and Herman, C. A. (1989). Prostaglandin F and E synthesis by specific tissue components of the brook trout ( Salvelinus fontinalis ) ovary. J. Exp. Zool. 250, 196-205. Goetz, F. W., Hsu, S.-Y. and Selover, A. (1991). Stimulation of prostagl andin synthesis in fish follicles by a phorbol ester and calcium ionophore. J. Exp. Zool. 259, 355-364. Goudet, G., Mugnier, S., Callebaut, I., and Monget, P. (2007). Phylogenetic analysis and identification of pseudogenes reveal a progressive loss of zona pellucida genes during evolution of vertebrates. Biol. Reprod. Epub. Ahead of Print. Griswold, D. E. and Adams, J. L. (1996). Constitutive cyclooxygenase (COX-1) and inducible cyclooxygenase (COX-2): rationale for selective inhibition and progress to date. Med. Res. Rev. 16 181-206. 120
PAGE 121
Grosser, T., Yusuff, S., Cheskis, E., Pack, M. A. and FitzGerald, G. A. (2002). Developmental expression of func tional cyclooxygenases in zebrafish. Proc. Natl. Acad. Sci. USA 99, 8418-8423. Guindon, S. and Gascuel, O. (2003). A simple, fast, and accur ate algorithm to estimate large phylogenies by maximum likelihood. Syst. Biol. 52, 679-704. Guindon, S., Lethiec, F., Duroux, P. and Gascuel, O. (2005). PHYML Online--a web server for fast maximum likelihood-ba sed phylogenetic inference. Nucleic Acids Res. 33, W557-559. Gu, X. (2001). Maximum-Likelihood approach for gene family evolution under functional divergence. Mol. Biol. Evol. 18, 453-464. Gu, X. (2006). A simple statistical method for esti mating type-II (cluster specific) functional divergence of protein sequences. Mol. Biol. Evol. 23, 1937. Guo, Q., Wang, L., Ruan, K. and Kulmacz, R. J. (1996). Role of Val509 in time-dependent inhibition of human prostaglandin H synt hase-2 cyclooxygenase activity by isoformselective agents. J. Biol. Chem. 271, 19134-19139. Hao, C.-M., Komhoff, M., Guan, Y., Redha, R. and Breyer, M. D. (1999). Selective targeting of cyclooxygenase-2 reveals its ro le in renal medullary interstitial cell survival. Am. J. Physiol. 277, F352-F359. Harris, R. C. and Breyer, M. D. (2001). Physiological regulati on of cyclooxygenase-2 in the kidney. Am. J. Physiol. 281 F1-F11. Harris, R. C. and Breyer, M. D. (2006). Update on cycloo xygenase-2 inhibitors. Clin. J. Am. Soc. Nephrol. 1, 236-245. Harris, R. C., McKanna, J. A., Akai, Y., Jacobson, H. R., Dubois, R. N. and Breyer, M. D. (1994). Cyclooxygenase-2 is associated with the macula densa of rat kidney and increases with salt restriction. J. Clin. Invest. 94 2504-2510. Hamberg, M. and Samuelsson, B. (1973). Detection and isol ation of an endoperoxide intermediate in prostaglandin biosynthesis. Proc. Natl. Acad. Sci. USA. 70, 899-903. Hamberg, M., Svensson, J. and Samuelsson, B. (1974). Prostaglandin endoperoxides. A new concept concerning the mode of ac tion and release of prostaglandins. Proc. Natl. Acad. Sci. USA. 71, 3824-3828. Hebert, S. C., Culpepper, R. M., and Andreoli, T. E. (1981). NaCl transport in mouse medullary thick ascending limb. I. Functional nephron heterogeneity and ADHstimulated NaCl cotransport. Am. J. Physiol 241 F412F431. 121
PAGE 122
Hebert, S. C. and Andreoli, T. E. (1984). Control of NaCl tran sport in the thick ascending limb. Am. J. Physiol. 246 F745F756. Herschman, H. R., Talley, J. J., and DuBois, R. (2003). Cyclooxygenase 2 (COX-2) as a target for therapy and noninvasive imaging. Mol. Imaging Biol. 5, 286-303. Hirano, T., Yasukawa, K., Harada, H., Taga, T., Watanabe, Y., Matsuda, T., Kashiwamura, S., Nakajima, K., Koyam a, K., Iwamatsu, A., Tsunasawa, S., Sakiyama, F., Matsui, H., Takahara, H., Taniguchi, T., and Kishimoto, T. (1986). Complementary DNA for a novel human interleu kin (BSF-2) that i nduces B lymphocytes to produce immunoglobulin. Nature 324, 73-76. Holland, J. W., Pottinger, T. G. and Secombes, C. J. (2002). Recombinant interleukin-1 beta activates the hypothalamic-pituitaryinterrenal axis in rainbow trout, Oncorhynchus mykiss. J. Endocrinol. 175 261-267. Hlscher, C. (1995). Prostaglandins play a role in memory consolidation in the chick. Eur. J. Pharmacol. 294, 253-259. Hopkins, S. J. (2007). Central nervous system recogniti on of peripheral infl ammation: a neural, hormonal collaboration. Acta. Biomed. 78 Suppl. 1 231-247. Huang, L., Zhang, K. L., Li, H., Chen, X. Y., Kong, Q. Y., Sun, Y., Gao, X., Guan, H. W., and Liu, J. (2006). Infrequent COX-2 expression due to promoter hypermethylation in gastric cancers in Dalian, China. Hum. Pathol. 37, 1557-1567. Hughes, A. L. (1999). Phylogenies of developmentally important proteins do not support the hypothesis of two rounds of genome dup lication early in vertebrate history. J. Mol. Evol. 48, 565-576. Ishikawa, T. O., Griffin, K. J., Banerjee, U. and Herschman, H. R. (2007). The zebrafish genome contains two inducible, functional cyclooxygenase-2 genes. Biochem. Biophys. Res. Commun. 352, 181-187. Ishikawa, T. O. and Herschman, H. R. (2007). Two inducible, f unctional cyclooxygenase-2 genes are present in the rainbow trout genome. J. Cell. Biochem. 102, 1486-1492. Ivanov, A. I. and Romanovsky, A. A. (2004). Prostaglandin E2 as a mediator of fever: synthesis and catabolism. Front. Biosci. 9, 1977-1993. Jarving, R., Jarving, I., Kurg, R., Brash, A. R. and Samel, N. (2004). On the evolutionary origin of cyclooxygenase (COX) isozymes: ch aracterization of mari ne invertebrate COX genes points to independent duplication events in vertebrate and i nvertebrate lineages. J. Biol. Chem. 279, 13624-13633. 122
PAGE 123
Jayadeep, A., Reddanna, P., Sailesh, S., Das, U. N., Ramesh, G., Kumar, K. V., and Menon, V. P. (1995). Prostaglandin metabolism during growth and differentiation of the regenerating vertebrate appendage. Prostaglandins Leukot. Essent. Fatty Acids. 52, 235239. Jin, Y., Zhang, C., Lin, X., and Zeng, W. (2006). Prostaglandin involvement in folliclestimulating hormone-induced proliferat ion of granulosa cells from chicken prehierarchical follicles. Prostaglandins Other Lipid. Mediat. 81, 45-54. Jin, Y., Zhang, C., Zeng, W., Taya, K., and Tan, T. Q. (2007). Interactive actions of prostaglandin and epidermal growth factor to enhance proliferati on of granulosa cells from chicken prehierarchical follicles. Prostaglandins Ot her Lipid Mediat. 83, 285-294. Johnson, R. F., Mitchell, C. M., Gile s, W. B., Bisits, A., and Zakar, T. (2006). Mechanisms regulating prostaglandin H2 synthase-2 mRNA level in the amnion and chorion during pregnancy. J. Endocrinol. 188, 603-610. Jones, D. T., Taylor, W. R., and Thornton, J. M. (1994). A mutation data matrix for transmembrane proteins. FEBS Lett. 339, 269-275. Kaji, D. M., Chase, H. S. Jr, Eng, J. P., and Diaz, J. (1996). Prostaglandin E2 inhibits Na-K2Cl cotransport in medullary thick ascending limb cells. Am. J. Physiol 271 C354-C361. Kargman, S., Charleson, S., Cartwright, M., Frank, J., Riendeau, D., Mancini, J., Evans, J. and O'Neill, G. (1996). Characterization of Prostaglandin G/H Synthase 1 and 2 in rat, dog, monkey, and human gast rointestinal tracts. Gastroenterology 111, 445-454. Kasahara, M. (2007). The 2R hypothesis: an update. Curr. Opin. Immunol. 19, 547-552. Kay-Mugford, P. A., Benn, S. J., LaMarre, J., and Conlon, P. D. (2000). Cyclooxygenase expression in canine platelets and Madin-Darby canine kidney cells. Am. J. Vet. Res. 61, 1512-1516. Kennedy, J. H., Korn, N., and Thurston, R. J. (2003). Prostaglandin levels in seminal plasma and sperm extracts of the domestic turkey, and the effects of cyclooxygenase inhibitors on sperm mobility. Reprod. Biol. Endocrinol. 1, 74. Kis, B., Snipes, J. A., Gaspar, T., Lenzser, G., Tulbert, C. D. and Busija, D. W. (2006). Cloning of cyclooxygenase-1b (p utative COX-3) in mouse. Inflamm. Res. 55, 274-278. Klein, S. L., Strausberg, R. L., Wagner, L., Pontius, J., Clifton, S. W. and Richardson, P. (2002). Genetic and genomic tools for Xenopus research: The NIH Xenopus initiative. Dev. Dyn. 225, 384-391. Kulmacz, R. J., van der Donk, W. A., and Tsai, A. L. (2003). Comparison of the properties of prostaglandin H synthase-1 and -2. Prog. Lipid Res. 42, 377-404. 123
PAGE 124
Kujubu, D., Fletcher, B., Varnum, B., Lim, R. and Herschman, H. (1991). TIS10, a phorbol ester tumor promoter-inducible mRNA from Swiss 3T3 cells, encodes a novel prostaglandin synthase /cyclooxygenase homologue. J. Biol. Chem. 266, 12866-12872. Lear, S., Silva, P., Kelley, V. E., and Epstein, F. H. (1990). Prostaglandin E2 inhibits oxygen consumption in rabbit medullary thick ascending limb. Am. J. Physiol. 258 F1372-1378. Li, G., Zhang, Q. J., Ji, Z. L., and Wang, Y. Q. (2007). Origin and evolution of vertebrate ABCA genes: a story from amphioxus. Gene. 405, 88-95. Li, J., Simmons, D. L., and Tsang, B. K. (1996). Regulation of hen granulosa cell prostaglandin production by transforming gr owth factors during follicular development: involvement of cyclooxygenase II. Endocrinology. 137, 2522-2529. Liu, C., Li, R., Liu, Z., Yin, S., and Wang, Z. (2006). The role of prostaglandins and the hypothalamus in thermoregulation in the lizard, Phrynocephalus przewalskii (Agamidae). J. Comp. Physiol. [B]. 176 321-328. Lurje, G., Vallbohmer, D., Collet, P. H., Xi, H., Baldus, S. E., Brabender, J., Metzger, R., Heitmann, M., Neiss, S., Drebber, U., Holscher, A. H., and Schneider, P. M. (2007). COX-2 mRNA expression is significantly increased in acid-exposed compared to nonexposed squamous epithelium in gastroesophageal reflux disease. J. Gastrointest. Surg. 11, 1105-1111. Mathonnet, M., Lalloue, F., Danty, E., Comte, I., and Livre, C. A. (2001). Cyclooxygenase 2 tissue distribution and developmental pattern of expression in the chicken. Clin. Exp. Pharmacol. Physiol. 28, 425-432. Martin, G. R. and Wallace, J. L. (2006). Gastrointestinal infla mmation: a central component of mucosal defense and repair. Exp. Biol. Med. (Maywood). 231 130-137. Martin, F., Fletcher, D., Ch auvin, M., and Bouhassira, D. (2007). Constitutive cyclooxygenase-2 is involved in central nociceptive processes in humans. Anesthesiology. 106, 1013-1018. McLaren, W. J., Young, I. R.,Wong, M. H., and Rice, G. E. (1996). Expression of prostaglandin G/H synthase-1 and -2 in ovine amnion and placenta following glucocorticoidinduced labour onset. J. Endocrinol. 151, 125. Merlie, J. P., Fagan, D., Mudd, J. and Needleman, P. (1988). Isolation and characterization of the complementary DNA for sheep seminal vesicle prostaglandin e ndoperoxide synthase (cyclooxygenase). J. Biol. Chem. 263, 3550. 124
PAGE 125
Metcalfe, C. D., Koenig, B. G., Bennie, D. T ., Servos, M., Ternes, T. A., and Hirsch, R. (2003). Occurrence of neutral and acidic dr ugs in the effluents of Canadian sewage treatment plants. Environ. Toxicol. Chem. 22, 2872-2880. Morham, S. G., Langenbach, R., Loftin, C. D., Tiano, H. F., and Vouloumanos, N. (1995). Prostaglandin synthase 2 gene disruption cau ses severe renal pathology in the mouse. Cell 83, 473-482. Morrison, D. A. (2007). Increasing the efficiency of s earches for the maximum likelihood tree in a phylogenetic analysis of up to 150 nucleotide sequences. Syst. Biol. 56, 988-1010. OBanion, M., Winn, V. and Young, D. (1992). cDNA cloning and functional activity of a glucocorticoid-regulated in flammatory cyclooxygenase. Proc. Natl. Acad. Sci. USA 89, 4888-4892. Olson, D. M. (2005). The promise of prostaglandins: have they fulfilled their potential as therapeutic targets for the delay of preterm birth? J. Soc. Gynecol. Investig. 12, 466-478. Otto, J. C. and Smith, W. L. (1995). Prostaglandin endoperoxide synthases-1 and -2. J. Lipid Mediat. Cell Signal. 12, 139. Patio, R., Yoshizaki, G., Bolamba, D., and Thomas P. (2003). Role of arachidonic acid and protein kinase C during maturation-induci ng hormone-dependent meiotic resumption and ovulation in ovarian follic les of Atlantic croaker. Biol. Reprod. 68, 516-523. Pough, F. H., Janis, C. M. Heiser, J. B. (2005). Vertebrate life: Seventh edition. pp. 16-48. Prentice Hall. Reed, D. W., Bradshaw, W. S., Xie, W., and Simmons, D. L. (1996). In vivo and in vitro expression of a non-mammalian cyclooxygenase-1. Prostaglandins. 52, 269-284. Reiter, R., Resch, U., and Sinzinger, H. (2001). Do human plat elets express COX-2? Prostaglandins Leukot. Essent. Fatty Acids. 64, 299-305. Ristimki, A., Honkanen, N., Jnkl H., Sipponen, P., and Hrknen, M. (1997). Expression of cyclooxygenase-2 in human gastriccarcinoma. Cancer Res. 57, 1276-1280. Roberts, S. B., Langenau, D. M. and Goetz, F. W. (2000). Cloning and characterization of prostaglandin endoperoxide synthase-1 and -2 from the brook trout ovary. Mol. Cell. Endocrinol. 160, 89-97. Rogers, J. S., and Swofford, D. L. (1998). A fast method fo r approximating maximum likelihoods of phylogenetic tree s from nucleotide sequences. Syst. Biol. 47 77-89. 125
PAGE 126
Rose, T. M., Schultz, E. R., Henikoff, J. G., Pietrokovski, S., McCallum, C. M., and Henikoff, S. (1998). Consensus-degenerate hybrid oligonucleotide primers for amplification of distantly related sequences. Nucleic Acids Res. 26, 1628-1635. Rowley, A. F., Vogan, C. L., T aylor, G. W. and Clare, A. S. (2005). Prostaglandins in noninsectan invertebrates: recent insights and unsolved problems. J. Exp. Biol. 208, 3-14. Ryan, C. A. (2000). The systemin signaling pathway: differential activation of plant defensive genes. Biochim. Biophys. Acta. 1477, 112-121. Sawdy, R., Slater, D., Fisk, N., Edmonds, D. K., and Bennett, P. (1997). Use of a cyclooxygenase type-2-selective nonsteroidal an ti-inflammatory agent to prevent preterm delivery. Lancet 350, 265. Searls, D. B. (2003). Pharmacophylogenomics: genes, evolution and drug targets. Nat. Rev. Drug. Discov. 2, 613-623. Seebacher, F. and Franklin, C. E. (2003). Prostaglandins are important in thermoregulation of a reptile (Pogona vitticeps ). J. Exp. Biol. 206, 1143-1151. Seebacher, F. and Franklin, C. E. (2004). Integration of autonomic and local mechanisms in regulating cardiovascular responses to heating and cooling in a reptile ( Crocodylus porosus ). J. Comp. Physiol. [B]. 174, 577-585. Shakhmatova, E. I., Kuznetsova, A. A., Prutskova, N. P., Bogolepova, A. E., and Natochin, I. V. (1997). The effect of cyclooxygenase inhib itors on ion and water transport in the human kidney and frog skin and bladder. Ross. Fiziol. Zh. Im. I. M. Sechenova. 83, 6875. Singh, G. and Triadafilopoulos, G. (1999). Epidemiology of NSAID induced gastrointestinal complications. J. Rheumatol. 26, 18-24. Sirois, J., Sayasith, K., Brown, K. A., Stock, A. E., Bouchard, N., and Dor, M. (2004). Cyclooxygenase-2 and its role in ovulation: a 2004 account. Hum. Reprod. Update. 10, 373-385. Skoblina, M. N., Kondrat'eva, O. T., Ni kiforova, G. P., and Huhtaniemi, I. (1997). The role of eicosanoids and progesterone in the ovulation of the oocytes in the common frog. Ontogenez. 28, 211-216. Snipes, J. A., Kis, B., Shelness, G. S., Hewett, J. A. and Busija, D. W. (2005). Cloning and characterization of cyclooxygenase-1 b (putative cyclooxygenase-3) in rat. J. Pharmacol. Exp. Ther. 313, 668-676. 126
PAGE 127
Sorbera, L. A., Asturiano, J. F., Carrillo, M., and Zanuy, S. (2001). Effects of polyunsaturated fatty acids and prostaglandins on oocyte matu ration in a marine teleost, the European sea bass ( Dicentrarchus labrax ). Biol. Reprod. 64, 382-389. Southern, E. M. (1975). Detection of specific sequen ces among DNA fragments separated by gel electrophoresis. J Mol Biol. 98, 503-517. Spaziani, E. P., Lantz, M. E., Benoit, R. R., and OBrien, W. F. (1996). The induction of cyclooxygenase-2 (COX-2) in intact human amnion tissue by interleukin-4. Prostaglandins. 51, 215-223 Stamatakis, A. (2006). RAxML-VI-HPC: Maximum likelihood-based phylogenetic analyses with thousands of ta xa and mixed models. Bioinformatics. 22 2688-2690. Steiner, A. A., Rudaya, A. Y., Robbins, J. R., Dragic, A. S., Langenbach, R., and Romanovsky, A. A. (2005). Expanding the febrigenic ro le of cyclooxygenase-2 to the previously overlooked responses. Am. J. Physiol. Regul. Integr. Comp. Physiol. 289, R1253-1257. Stokes, J. B. and Kokko, J. P. (1977). Inhibition of sodium transport by prostaglandin E2 across the isolated, perfused rabbit collecting tubule. J. Clin. Invest. 59, 1099-1104. Szewczuk, L. M. and Penning, T. M. (2004). Mechanism-based inactivation of COX-1 by red wine m-hydroquinones: a struct ure-activity relationship study. J. Nat. Prod. 67, 17771782. Tamura, K., Dudley, J., Nei, M., and Kumar, S. (2007). MEGA4: Molecular Evolutionary Genetics Analysis (ME GA) software version 4.0. Mol. Biol. Evol. 24, 1596-1599. Trautman, M. S., Edwin, S. S., Collmer, D., Dudley, D. J., Simmons, D., and Mitchell, M. D. (1996). Prostaglandin H synthase-2 in human gestational tissues: regulation in amnion. Placenta 17, 239-245. Urick, M. E. and Johnson, P. A. (2006). Cyclooxygenase 1 and 2 mRNA and protein expression in the Gallus domesticus model of ovarian cancer. Gynecol. Oncol. 103, 673678. Vane, J. R. (1971). Inhibition of prostaglandin synt hesis as a mechanism of action for the aspirin-like drugs. Nature 231, 232-235. Vane, J. R. and Botting, R. M. (1994). Biological properties of cyclooxygenase products. In Lipid Mediators, ed. F Cunningham, pp. 61. London: Academic. Vane, J. R., Bakhle, Y. S. and Botting, R. M. (1998). Cyclooxygenases 1 and 2. Annu. Rev. Pharmacol. Toxicol. 38, 97-120. 127
PAGE 128
Vio, C. P., Cespedes, C., Gallardo, P., and Masferrer, J. L. (1997). Renal identification of cyclooxygenase-2 in a subset of thick ascending limb cells. Hypertension. 30, 687-692. van-Lanschot, J. J., Mealy, K., Jacobs, D. O., Evans, D. A., and Wilmore, D. W. (1991). Splenectomy attenuates the inappropriate diur esis associated with tumor necrosis factor administration. Surg. Gynecol. Obstet 172, 293-297. von Euler, U. S. (1936). On the specific vaso-dilating a nd plain muscle stimulating substances from accessory genital glands in man and certain animals (prostaglandin and vesiglandin). J. Physiol. 88, 213-234. Wald, H., Scherzer, P., Rubinger, D., and Popovtzer, M. M. (1990). Effect of indomethacin in vivo and PGE2 in vitro on mTAL Na-K-A TPase of the rat kidney. Pflgers Arch. 415 648. Wang, H., Ma, W. G., Tejada, L., Zhang, H., Morrow, J. D., Das, S. K., and Dey, S. K. (2004). Rescue of female infertility from the loss of cyclooxygenase-2 by compensatory up-regulation of cyclooxygenase-1 is a function of genetic makeup. J. Biol. Chem. 279, 10649-1058. Whelan, S. and Goldman, N. (2001). A general empirical mode l of protein evolution derived from multiple protein families using a maximum-likelihood approach. Mol. Biol. Evol. 18, 691-699. Willingale, H. L., Gardiner, N. J., McLym ont, N., Giblett, S., and Grubb, B. D. (1997). Prostanoids synthesized by cyclooxygenase isoforms in rat spinal cord and their contribution to the development of neuronal hyperexcitability. Br. J. Pharmacol. 122, 1593-1604. Woolf, C. J., Allchorne, A., Safieh-Garabedian, B., and Poole, S. (1997). Cytokines, nerve growth factor and inflammatory hyperalgesia: the contribution of tum our necrosis factor. Br. J. Pharmacol. 121, 417-424 Wood, C. M. and Laurent, P. (2003). Na+ versus Cltransport in the intact killifish after rapid salinity transfer. Biochim. Biophys. Acta. 1618, 106-119. Wong, E., Bayly, C., Waterman, H. L., Riendeau, D. and Mancini, J. A. (1997). Conversion of prostaglandin G/H synthase-1 into an en zyme sensitive to PGHS-2-selective inhibitors by a double His513 to Arg and Ile523 to Val mutation. J. Biol. Chem. 272, 9280-9286. Xie, W., Chipman, J., Robertson, D., Erikson, R. and Simmons, D. (1991). Expression of a mitogen-responsive gene encoding prostagl andin synthase is regulated by mRNA splicing. Proc. Natl. Acad. Sci. USA 88, 2692-2696. 128
PAGE 129
129 Yamada, S., Kawate, T., Sakamoto, H., Aoki, K., Hamada, Y., and Atsumi, S. (2006). Cyclo-oxygenase-2-immunoreactive neurons in the lumbar dorsal horn in a chicken acute inflammation model. Anat. Sci. Int. 81 164-172. Yamagata, K., Andreassson, K. I., Kaufman, W. E., Barnes, C. A., and Worley, P. F. (1993). Expression of a mitogen-inducible cyclooxygenase in brain neurons; regulation by synaptic activity and glucocorticoids. Neuron 11, 371-386. Yang, T. (2003). Regulation of cyclooxygenase-2 in renal medulla. Acta Physiol. Scand. 177, 417-421. Yang, T., Forrest, S. J., Stine, N., Endo, Y., Pa sumarthy, A., Castrop, H., Aller, S., Forrest, J. N., Jr, Schnermann, J. and Briggs, J. (2002). Cyclooxygenase cloning in dogfish shark, Squalus acanthias and its role in rect al gland Cl secretion. Am. J. Physiol. 283, R631-R637. Yeh, H. C. and Wang, L. H. (2006). Profiling of prostanoids in zebrafish embryonic development. Prostaglandins Leukot. Essent. Fatty Acids. 75 397-402. Yokoyama, C., Takai, T. and Tanabe, T. (1988). Primary structure of sheep prostaglandin endoperoxide synthase deduced from cDNA sequence. FEBS Lett. 231, 347-351. Zou, J., Neumann, N. F., Holland, J. W., Belosevic, M., Cunningham, C., Secombes, C. J. and Rowley, A. F. (1999). Fish macrophages expre ss a cyclo-oxygenase-2 homologue after activation. Biochem. J. 340, 153-159.
PAGE 130
BIOGRAPHICAL SKETCH Justin Chase Havird was born in 1984, in Valdosta, Georgia to Dawn and Kurt Havird. Justin has one younger brother, Jos hua. The Havirds soon moved to Lake City, Florida, where Justin spent the majority of his childhood and adolescence. Justin attended Columbia High School in Lake C ity and graduated with top honors in 2002. Justin then attended the Univer sity of Florida in Gainesville where he majored in zoology after an initial interest in aerospace engineering. While at UF, Justin was heavily involved in the universitys band program, pl aying trumpet in the marching, basketball, volleyball, concert, and symphonic bands for se veral years. As an undergraduate, Justin worked in the laboratory of Dr. David H. Ev ans for a number of years, investigating ion transporters in the gills of fishes. His se nior thesis was titled Neuronal Nitric Oxide Synthase in the Gill of the Euryhaline Killifish, Fundulus heteroclitus. Justin graduated summa cum laude from UF in 2006 with a B.S. in zoology. In 2006, Justin entered graduate school at the University of Florida in the Department of Zoology. Continuing to work w ith Dr. Evans, Justin spent three summers at the Mount Desert Island Biological Laboratory in Salis bury Cove, Maine where he collaborated with other researchers inve stigating fish gill physiology. During his graduate studies, Justin has earned several awards and grants, including a GIAR from Sigma Xi. Justin has also presented his resear ch at scientific meeti ngs and conferences. While earning his M.S., Justin also designed experiments and instructed students in a senior level animal physiology course for 2 years. Justin has also collaborated with other scientists on a wide ra nge of marine life science projec ts, most notably a revision of the cobitid genus Lepidocephalichthys in SE Asia with Dr. Larry Page. After earning his 130
PAGE 131
131 M.S., Justin plans to continue his academi c career by completing a Ph.D. in a related marine science field.
and specific primers designed against the initial 541 bp product (Table 2-4). Despite all attempts
to amplify the complete 1COX product (including RACE) the sequence was only extended to 860
bp. A 286 amino acid open reading frame was predicted from this product.
Fundulus
Because the COX-2 sequence for the euryhaline killifish (Fundulus heteroclitus) was
previously completed (Choe et al., 2006), the degenerate primer CH COX-1 F2 (Table 2-1)
designed to amplify COX-1 forms over COX-2 forms was used with the degenerate primer CH
COX R1 (Table 2-1) to initially amplify a non-COX-2, 1335 bp product (hereafter referred to as
COX-lb Fundulus) from killifish gill tissue. This product was then extended and completed
using specific primers and RACE primers (Table 2-5). The completed 2226 bp sequence
contains a 598 amino acid open reading frame that was nested inside the teleost COX-lb clade in
all phylogenetic analyses (see below), prompting the design of degenerate primers (Table 2-1)
that were used to amplify COX-la forms in teleosts over COX-lb forms. Using the degenerate
primers CH COX-la Fl and CH COX-la R1 (Table 2-1), a 759 bp product was initially
amplified in PCR reactions. Using RACE primers designed to complete this sequence (Table 2-
5), the 3' end of this product was completed, resulting in a 1655 bp product (hereafter referred to
as COX-la Fundulus) that contains a 452 amino acid reading frame. However, the 5' end of this
product was not sequenced despite multiple attempts. This putative protein grouped with the
COX-la sequences of other teleosts in all phylogenetic analyses (see below), leading to the
conclusion that the COX-la, COX-lb, and COX-2 sequences included in the analyses all
represent different COX products expressed in the gill tissue of the euryhaline killifish.
Myoxocephalus
Using the degenerate primers CH COX Fl and CH COX R2 (Table 2-1), a 977 bp
product was initially amplified in PCR reactions using cDNA from the gills of the longhorn
COXb Clona
COXa Clona
Canis
COX2 Xenopus
COX2a Oncorhynchus
COX2aDanio
COX2 Tetraodon
sCOX Squalus
CCX1 Monodelphis
995 60 COXlb2 Monrdelphis
100 COX b Monodelphis
978 r-l I Rattus
10Q X1 Oryatolagus
1 66*i1X1 Homo
34 -83Eril Canis
ICOXla Fundulus7
COXb a Tetraodon
ICOX II Ca taratmsus
I COXI a1 :aaphals
COXI aryziaa
614 COXIa Fundulus
0-7
SCOXI Cncorhynchus
CCXI Xenopus
L .57 COXI Gallus
4 CXd BranchiOs tora-4
0.1
Figure 2-9. Evolutionary history of COX sequences inferred using a maximum likelihood
optimality criterion. The most likely tree with an Ln L score = -25294.222410 is
shown. The number of replicate trees in which the associated taxa clustered together
in the bootstrap test (1000 replicates) is shown next to the branches (Felsenstein,
1985). The tree is drawn to scale, with branch lengths in the same units as those of the
evolutionary distances used to infer the phylogenetic tree. The evolutionary distances
were computed using the WAG matrix-based method (Whelan and Goldman, 2001)
and are in the units of the number of amino acid substitutions per site (scale bar = 0.1
amino acid substitutions per site). The rate variation among sites was modeled with a
gamma distribution (shape parameter estimated = 0.737). Phylogenetic analyses were
conducted using PhyML (Guindon and Gascuel, 2003). Sequences enclosed in a
rectangle are novel and reported here for the first time.
hypothesize that a genome duplication in the teleost lineage and subsequent loss of either a
COX-1 (in the zebrafish/trout lineage) or COX-2 gene (in the acanthopterygian lineage) is
responsible for this observation. If this is the case, then examining taxa near the base of the
teleost lineage may reveal fish with all four variants of COX (COX-la, COX-lb, COX-2a, and
COX-2b). The COX variants from the dogfish shark, lamprey, and hagfish all group weakly
with either COX-1 or COX-2. Obtaining COX sequences from other species of hagfish,
lampreys, and elasmobranchs may confirm this initial relationship or reveal novel groupings.
Finally, although support values for the urochordata and cephalochordata clades are strong
regardless of the phylogenetic technique used, only one species from each subphylum was
included in phylogenetic analyses. Including more species could give a more complete view of
COX evolution in the evolutionarily ancestral chordates. COX sequences from these additional
taxa may be readily available in the near future due to the growing regularity of genome
sequencing across many taxa. For example, the recently completed Anolis carolinensis genome
contains COX-1 and COX-2 genes. However, due to the length of time required for whole
genome projects, the limited number of genomes available, and the difficulty of searching
recently completed, non-annotated genomes, studies that specifically target sequencing COX
genes in these taxa are needed.
Function of Cyclooxygenases
As reviewed in Chapter 1, cyclooxygenase plays several roles in chordates because of its
role in prostaglandin production. These functions have been examined in some detail in the
mammals and to a lesser degree in the fishes. However, no studies have examined the functions
of the novel COX forms presented here or any non-vertebrate COX forms in the chordates.
Therefore, I have begun preliminary studies investigating the function of these COX forms.
specific primers used in PCR
Name
hCOX 3' Fl
hCOX 3' F2
hCOX 3' F3
hCOX 3' F0.8
hCOX 3' F0.6
hCOX 3' F0.4
hCOX 3' F0.2
hCOX 3' F0.1
hCOX 3' FO
hCOX mid Fl
hCOX mid F2
hCOX mid R1
hCOX mid R2
hCOX 5'R1
hCOX 5' R2
hCOX 5' R3
hCOX 5' R4
hCOX 5' R5
hCOX 5'R6
Orientation
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Antisense
Antisense
Antisense
Antisense
Antisense
Nucleotide sequence (5' to 3')
TTA ACC ACC TGT ACC ACT GGC AC
CTC AAC CAT GGA GTA CGA GGT CT
CAA TGC TAT GGA GTT CTA CCT GGG
CAG TAC AGC AAC CGC ATC TCA
AAT GGG ACG ACG AGA GAA TCT TT
GGT TTG CTG TTG GTC ACG AAG
AGT CCA CCC ACA TAC AAT GCC
CTC ACG CAC TTT GCT CCC TT
GAA CTG CAC CTA CCC CGA GAC
GTT TCG AGT CCA CCC ACA TAC AA
CAT ACT CCA ACC TCA GCT ACT TCA CT
CAC ATC GAG TAC GTC ACA CAC G
AAG GGT CCC GTA GAG CAT GA
TCT CCG AAT AAA GCT CCC TGT C
TGC TGC CAT TTC CGT TTC TC
CAT GTT TGA TGG ATG CTG TTG C
AGG CGA GTG AAG TAG CTG AGG TT
CCT CCC AGG ACT TGT ACC GTA AT
GAG GAA GTA ATG CAC GGT ATC TGG
Table 2-3. Hagfish COX
(abnormal kidney function) which results in death by the third month (Morham et al., 1995).
These results suggest that COX-2, and not COX-1, plays a critical role in kidney function via
MIC cell survival. This is contrary to the original description of COX-1 as the constitutively
expressed form responsible for maintaining homeostasis and COX-2 as the inducible form
responsible for inflammatory responses.
In the kidney, COX-2 is present in a subset of tubular epithelial cells located in the cortex
and outer medulla (Vio et al., 1997). These cells are part of the medullary thick ascending limb
(MTAL), the part of the nephron that regulates extracellular fluid volume by establishing the
osmotic gradient in the medulla that concentrates urine (Hebert et al., 1981; Hebert and Andrioli,
1984). In vitro MTAL preparations have demonstrated high levels of Prostaglandin E2 (PGE2)
(Ferreri et al., 1984; Hirano et al., 1986; Lear et al., 1990) which has been shown to inhibit
basolateral Na+/K -ATPase (NKA) and apical Na+/K+-2C1 cotransporter (NKCC) in the MTAL
(Kaji et al., 1996; Stokes and Kokko, 1976; Wald et al., 1990). These proteins are critical for
NaCl reabsorption in the MTAL. Furthermore, the COX-2 selective inhibitor NS-398 has been
shown to prevent tumor necrosis factor (TNF) mediated PGE2 production in cultured MTAL
cells (Ferreri et al., 1999). TNF is a cytokine expressed in the MTAL that may be an important
mediator of ion transport by inhibiting ion uptake via a prostanoid dependant mechanism
(Escalante et al., 1994; van-Lanschot et al., 1991). Taken as a whole these results show that
COX-2 may influence ion uptake via TNF mediated prostaglandin production in the MTAL.
Therefore, COX-2 is involved in cell survival and ion transport in the mammalian kidney.
This finding is critical for treatment protocols using COX inhibitors. NSAIDs are the
most common drugs used to treat pain and inflammation with over 30 million uses per day
worldwide (Singh and Triadafilopoulos, 1999). Although the negative side effects
COX-1 Vertebrates
COX-2 Vertebrates
COXb Urochordates
COXa Urochordates
Invertebrate COX forms
Figure 1-4 Evolution of COX forms in the chordates, based on Jarving et al. (2004). The
Urochordates (represented by Ciona intestinalis and Ciona savignyi) have COX
forms that represent an evolutionary distinct branch ancestral to the COX-1 and
COX-2 forms of the vertebrates. In this tree, the vertebrates are represented by the
teleosts, mammals, Xenopus, and Gallus
CHAPTER 3
SUMMARY AND FUTURE DIRECTIONS
Summary
This thesis is the first attempt to characterize the evolution of the enzyme cyclooxygenase
in the phylum chordata. Traditional molecular protocols such as cloning and polymerase chain
reactions were used to search for cyclooxygenase sequences in a wide range of evolutionarily
key chordates. Using these novel sequences and pre-existing sequences from GenBank and
genomic databases, phylogenetic analyses were performed to reveal the identity of the novel
sequences and create the most likely scenario for the evolutionary history of cyclooxygenase in
the chordates.
Novel COX Sequences
Chapter 2 describes the process of obtaining and characterizing novel cyclooxygenase
(COX) sequences in amphioxus (Branchiostoma lancelolatum), the Atlantic hagfish (Myxine
glutinosa), the sea lamprey (Petromyzon marinus), the euryhaline killifish (Fundulus
heteroclitus), and the longhorn sculpin (Myoxocephalus octodecemspinosus). Total RNA was
isolated and reverse transcribed from gill tissue (or whole animals in amphioxus) and the
resulting cDNA was exposed to PCR in the presence of degenerate and/or specific primers. The
products were then visualized and cloned. The resulting plasmids were sequenced at the Marine
DNA Sequencing Center at the MDIBL. Complete amino acid sequences of putative COX
proteins from the hagfish, killifish, and sculpin and incomplete amino acid sequences from
amphioxus and the lamprey were predicted based on the sequences found. These putative amino
acid sequences were aligned with known COX protein sequences and conserved domains,
regions, and residues critical to COX function were described. Based on the novel sequences'
identity to known COX forms and phylogenetic analyses (see below), it was concluded that
Differential functions of novel forms
Recently, the discovery of novel COX-2 forms in the zebrafish (Ishikawa et al., 2007)
and rainbow trout (Ishikawa and Herschman, 2007) and analyses of genomic sequence data
indicates that all teleosts may posses at least three COX forms: either COX-la, COX-lb, and
COX-2 or COX-1, COX-2a, and COX-2b. Preliminary data on the described COX-2 forms
suggests that they are functionally different from each other. In the zebrafish, COX-2a and
COX-2b are expressed at different levels in the gill, kidney, and other tissues and respond
differently to 12-O-tetradecanoylphorbol-13-acetate (TPA, which is known to induce COX-2
expression in mouse) (Ishikawa et al., 2007). A similar result was found in the rainbow trout,
with differential expression of the two forms due to TPA (Ishikawa and Herschman, 2007).
Interestingly, COX-2a and not COX-2b was up-regulated in response to lippopolysaccarides
(Ishikawa and Herschman, 2007). Clearly, the discovery of these new forms indicates a new
chapter in studying COX function and further studies are needed to determine the functional
differences among the teleosts' forms of COX.
Other functions
Cyclooxygenase has also been implicated in determining body plan development during
embryonic stages in the zebrafish (Grosser et al., 2002; Cha et al., 2005; Cha et al., 2006a; Cha
et al., 2006b; Yeh and Wang, 2006). Briefly, COX may function to signal cell motility during
gastrulation (Cha et al., 2006b), and COX-1 function has been shown to participate in vascular
tube formation during development (Cha et al., 2005). COX-2 expression has also been shown
to increase in response to lippopolysaccarides in the rainbow trout (Brubacher et al., 2000;
Holland et al., 2002), suggesting an immunoregulatory role for COX-2 in fish as in mammals
(e.g., fever induction). Clearly, the function of COX in fishes is just beginning to be explored,
Reverse Transcription, Primer Design, PCR, Cloning, and Sequencing
Reverse transcription, PCR, cloning, and sequencing were performed as described
previously (Choe et al., 2006) with modifications. Total RNA was isolated from the gills of
killifish, lampreys, sculpin, and hagfish as well as the anterior half of lancelets using TRI-reagent
(Sigma, St. Louis, MO), and reverse transcribed with a SuperscriptTM II or SuperscriptTM III
reverse transcriptase kit (Invitrogen, Carlsbad, CA) following the manufacturer's protocol and
using oligo-dT as a primer. The resulting cDNA was stored at -20 OC until used for PCR.
Degenerate primers (those labeled as CH COX, Table 2-1) were first designed to amplify
non-specific cyclooxygenases in the hagfish, lamprey, and lancelet based on conserved amino
acid sequences between COX-1 and COX-2 of Mus musculus, COX-2 of Fundulus heteroclitus,
sCOX of Squalus acanthias (GenBank accession numbers: NP_032995, NP_035328, and
AAS21313), and predicted COXa and COXb of Ciona intestinalis based on genomic data.
These primers were also later used to amplify COX-2 in sculpin. Another set of degenerate
primers (those labeled as CH COX-1 in Table 2-1) were designed to amplify COX-1 forms
specifically over COX-2 forms in teleosts based on conserved amino acid sequences between
COX-1 in Oncorhynchus mykiss, Danio rerio, Salvelinusfontanalis (GenBank accession
numbers: CAC10360, NP_705942, and AAF14529), and a predicted COX-1 of Tetraodon
nigroviridis based on genomic data. These primers first amplified a COX-la form in sculpin and
a COX-lb form in killifish (see below). A further set of degenerate primers (those labeled as CH
COX-la in Table 2-1) were designed to amplify COX-la forms specifically over COX-lb in
teleosts based on conserved amino acid sequences between predicted COX-la forms in
Gasterosteus aculeatus, Oryzias latipes, and Tetraodon nigroviridis based on genomic data and a
COX-la found in sculpin earlier in this study (see below). A final set of degenerate primers
(those labeled as CH COX-lb in Table 2-1) were designed to amplify COX-lb forms specifically
Yamada, S., Kawate, T., Sakamoto, H., Aoki, K., Hamada, Y., and Atsumi, S. (2006).
Cyclo-oxygenase-2-immunoreactive neurons in the lumbar dorsal horn in a chicken acute
inflammation model. Anat. Sci. Int. 81, 164-172.
Yamagata, K., Andreassson, K. I., Kaufman, W. E., Barnes, C. A., and Worley, P. F.
(1993). Expression of a mitogen-inducible cyclooxygenase in brain neurons; regulation
by synaptic activity and glucocorticoids. Neuron 11, 371-386.
Yang, T. (2003). Regulation of cyclooxygenase-2 in renal medulla. Acta Physiol. Scand. 177,
417-421.
Yang, T., Forrest, S. J., Stine, N., Endo, Y., Pasumarthy, A., Castrop, H., Aller, S., Forrest,
J. N., Jr, Schnermann, J. and Briggs, J. (2002). Cyclooxygenase cloning in dogfish
shark, Squalus acanthias, and its role in rectal gland Cl secretion. Am. J. Physiol. 283,
R631-R637.
Yeh, H. C. and Wang, L. H. (2006). Profiling of prostanoids in zebrafish embryonic
development. Prostaglandins Leukot. Essent. Fatty Acids. 75, 397-402.
Yokoyama, C., Takai, T. and Tanabe, T. (1988). Primary structure of sheep prostaglandin
endoperoxide synthase deduced from cDNA sequence. FEBS Lett. 231, 347-351.
Zou, J., Neumann, N. F., Holland, J. W., Belosevic, M., Cunningham, C., Secombes, C. J.
and Rowley, A. F. (1999). Fish macrophages express a cyclo-oxygenase-2 homologue
after activation. Biochem. J 340, 153-159.
Arachidonic Acid
COX-1
or
COX-2
0
0
Prostaglandin 12
HO C ,-OOH
OH OH
6-keto-Prostaglandin F1
OH
o OH
Prostaglandin D2
ODH
Prostaglandin G2
COX-1
or
COX-2
: %-Q--^G^^OOH
0 .... V
OH
Prostaglandin H2
H iH
-- oOH
OH
Thromboxane A2
HO C -H
OH
Thromboxane B2
v\ COOH
OH OH
Prostaglandin E2
Prostaglandin F2a
Figure 1-1. Cascade of reactions depicting the conversion of arachidonic acid to the primary
prostaglandins via catalyzation with either cyclooxygenase (COX) form. Arachidonic
acid is obtained from the conversion of linoletic acid found in plants or directly from
other animals. It is then oxidized by COX to produce Prostaglandin G2 which is
subsequently reduced by COX to form Prostaglandin H2. Prostaglandin H2 can then
be converted into the other prostaglandins via other enzymes (Vane et al., 1998).
Sequence Analysis............................... ... ...... .... ............ ......45
Phylogenetic Analysis .................................... ..... .......... .............. .. 46
R esu lts .. ............ .... .... ....... .....................................................4 8
M olecular Identification of Cyclooxygenases.............................................................. 48
B ra n ch io sto m a ................................................................................................... 4 8
M y x in e ..............................................................................4 8
P e tro m y z o n ......................................................................................................... 4 8
F u n d u lu s ............................................................................................................. 4 9
Myoxocephalus ............................................. 49
Sequence Analyses ................... ......................... 50
Teleosts COX-la, COX-lb, and COX-2 .................................... .................... 50
Lamprey, hagfish, and amphioxus COX forms............................................... 52
P hylogenetic analy ses ............................................................53
D iscu ssio n ................... ...................5...................6..........
N o v el C O X F o rm s ..................................................................................................... 5 7
P hy log en etic A n aly ses............................................................................................... 5 8
S u m m ary ............................................................................... 6 2
3 SUMMARY AND FUTURE DIRECTIONS ................................................105
S u m m ary ................... ...................1...................0.........5
N ovel COX Sequences ............................................................................ 105
P hy log en etic A n aly ses............................................................................................. 10 6
F u tu re D ire ctio n s ............................................................................................................ 10 6
Completion of Novel Sequences ......................................... ...............107
Additional Phylogenetic Analysis ................................. ........................... ...107
Additional Sequences ......................................... ................... ...........108
A d d itio n al T ax a ....................................................................................................... 10 9
Function of C yclooxygenases ........................................................................ 110
L IST O F R E F E R E N C E S ....................................................................................................... 117
BIOGRAPHICAL SKETCH .........................................................................130
7
intensely. Both isoforms are likely involved in reproductive processes, with COX-1 expression
constitutive and COX-2 expression inducible (Chakraborty et al., 1996).
In a recent study, COX-2 was shown to contribute to fertility in male mice (Balaji et al.,
2007). COX-2 expression was intense in the epithelial cells of mice vas deferens and the COX-2
specific inhibitor nimesulide was shown to decrease sperm motility six hours after administration
(Balaji et al., 2007). Interestingly, although prostaglandin levels initially dropped after
nimesulide administration, they tended to rise after sustained COX-2 inhibition (Balaji et al.,
2007). This suggests that COX-1 expression may recover COX function or at least contributes
heavily to prostaglandin production in the vas deferens. A similar situation exists with respect to
human female fertility, in which COX-1 expression can recover function in cases where COX-2
is lost or inhibited (Wang et al., 2004). It also seems that COX-2 is critical for follicle
development and ovulation in women and that COX inhibitors can have severe effects on female
fertility (Sirois et al., 2004).
Prostaglandins are also important during pregnancy and labor. Generally, COX-1 is
expressed more than COX-2 during normal stages of pregnancy, and COX-1 may play an active
role in maintaining a healthy pregnancy (Trautman et al., 1996). COX-1 is expressed at high
levels is the fetal brain, kidneys, heart, and lungs (Gibb and Sun, 1996). However, both COX-1
and COX-2 are expressed in the uterine epithelium during early pregnancy and may play roles in
implantation and placental formation (Chakraborty et al., 1996). This was recently confirmed
for COX-2 in rats (Diao et al., 2007). COX-2 may play a dominant role in inducing labor.
During labor, contractions are stimulated by prostaglandins and therefore any effect causing
increased COX activity will induce labor. Up-regulation of COX-2 during intrauterine infection
may increase prostaglandins to the point of causing premature labor (Spaziani et al., 1996).
sCOX MEAARIILLLLPLCFLKMADTTAINPCCYYPCQNKGICVNVGKEGYECDCTRT 53
hCOX MTSEVFVVLCVAVVFAGAA--AADDPCCGSPCENKGVCVSVGFEDYECDCTRT 51
t *N68 .................................................................................
sCOX GYYGVNCTFPFSWSRVHKFLKPSPSSMHHVLTHYKWLWYIINNISFFSDTLMR 106
hCOX GYYGSNCTYPTWTWIVNMLKPLPDTVHYFLTHFAPFWSLVNRAAFLRDRVMR 104
*R120 N1 -""
sCOX L.LTVRANIIPSPPTYNSDYTYVSWEGYSNISYLTRLLPPVPKDCPTPTGTQG 159
hCOX -.LMSRAHMVSSPPTYNADYGYKSWEAYSNLSYFTRLLPPVPQGCPTPMGRTG 157
*Q203/H207
sCOX YKKLPDSEQLAEEFLLRRKFIPDPQGSNLMFAFFAQHFTHQFFRTDLDRGPGF 212
1COX HQFFNTDPVRGPAF 14
hCOX KKELPDAQKLAERFLLRRTFIPDPQGSNLLFAFFAQHFTHQFFKTDFKRGPGF 210
COXc HQFFKTDFKKGAGR 14
COXd HQFFKTDFKKGPGR 14
sCOX TCALGH3VDLTHIYGDSLERQHHLRLFKDGKLKYQVVNGEVFPPSVKEAPIQM 265
ICOX TPALGH[VDLNHIYGGTLERQHQLRLFKDGKLKFQMIDGEAYPPVVRDAPVHM 67
hCOX TCALGH3VDLSHIYGDTLDKQHKLRLHNNGKLKFQMIDGEVFPPLVSEAPVDM 263
COXc TY-GDH3VDLSHIYGETVERQHQLRSFTDGKLKFQRVEGEVYPPSLADAPVHM 66
COXd TV-SDHVDMSHVYGETVERQRQLRSFQDGKLKYQLVDGEAFPPSLQDASVHM 66
--*R277----
sCOX KYPSTLPEEKRLAIGHDTFGLIPGLMMYATI LREHNRVCDILKEEHPVWSDE 318
1COX VYPEHVPASLRFAVGHEVYGLLPGLLVYATVILREHNRVCDJLHARHPRWDDE 120
hCOX IYPPHVPEAARFAVGHEAFGLVPGLMLYGTL'LREHNRVCDJLDVDHPEWDDE 316
COXc IYPPYVPEGKRFAIGHEFFGLLPGLFVYSTVILREHNRVCDJMKELHPDWDDE 119
COXd IYPPDVPEKKRFALGHEFFGLLPGLFVWATV LREHNRVCDJMKDLHPDWDDE 119
sCOX QLFQTTRLILIGETIKIVIEDYVQHLSGYNFRMMFNPELLFTEHFQYSNRIAV 371
1COX RLFQTARLILTGETMKIVIEEYVQHLSGYNFHLKFDPTLLFGVNFQYSNRMSL 173
hCOX RIFQTARLILIGETIKIVIEDYVQHLSGYHFKLKFDPKLLFGQPFQYSNRISV 369
COXc RLFQTARLILTGETINIIINEYVQHLSGYNFDLFWDPELLFSDQFQYQNRIFV 172
COXd RLYQTARLILISETINIVIGEYVQHLAGKNFQLFWDPELLFEEQXQYQNSIFV 172
Figure 2-3. Amino acid alignment of the forms of COX sequenced from the gills of the lamprey
(Petromyzon marinus), hagfish (Myxine glutinosa), and amphioxus (Branchiostoma
lancelolatum), along with sCOX from the dogfish (Squalus acanthias, Accession
#AAL37727). Amino acids that share at least 60% identity across sequences are
highlighted. *Indicates amino acids (numbered based on ovine COX-1) involved in
COX function: active site of COX (tyrosine-385, histidine-388, and serine-530),
substrate binding site (arginine-120), N-glycosylation sites (asparagine-68, 144, and
410), sites for peroxidase activity (glutamine-203 and histidine-207), and the two
sites which define conformational differences in channels between COX-1 and COX-
2 in mammals (arginine-513 and valine-523) (Yang et al., 2002; Ishikawa et al.,
2007). The black boxes indicate the two domains that define the haem-binding sites
and the dashed box indicates the membrane binding domain (Kulmacz et al., 2003;
Ishikawa et al., 2007). The solid lines represent disulfide bonds, with the beginning
and end of the longest bond represented by arrows (Kulmacz et al., 2003). The
dashed line represents the arginine-277 loop region, which has been shown to differ
between COX-1 and COX-2 forms in mammals (Yang et al., 2003; Kulmacz et al.,
2003). The numbers to the right of the sequences indicate the amino acid number of
each sequence.
COX1 Mus
COX1 Danio
COX1 Ovis
hCOX
COX2 Mus
COX2 F.het
COX2 Ovis
COX1 Mus
COX1 Danio
COX1 Ovis
hCOX
COX2 Mus
COX2 F.het
COX2 Ovis
*S530 + + + + +
PFSLKGLLGNPICSPEYWKPSTFGGDVGFNLVNTASLKKLVCLNTK-TCPYVS
PFSLKGLMGNPICSPDYWKPSTFGGKTGFDIVNSATLKKLVCLNTK-WCPYVS
PFSLKGLLGNPICSPEYWKASTFGGEVGFNLVKTATLKKLVCLNTK-TCPYVS
PFSLKGLMGNAICSPEYWKPSTFGGNRGFEIVNSASLRRLVCLNLQGPCPDVA
PFSLKGLMGNPICSPQYWKPSTFGGEVGFKIINTASIQSLICNNVKG-CPFTS
PFSLKGLMGNPICSPEYWKPSTFGGTAGFDIVNTASLQRLVCNNVRGPCPVAS
PFSLKGLMGNPICSPEYWKPSTFGGEVGFKIINTASIQSLICSNVKG-CPFTS
+ + + ++
FR-VPDYPGDDGSVLVRR-------------------STEL
FHTPPSDYKPQRTS-----------------------HGEL
FH-VPDPRQEDRPGVERP-------------------PTEL
FH-VPRDNQQDVVVNVTGSQGGSDGVTTTPHYVADQQSREL
FN-V-QDPQPTKTATINASASHSRLDDINPTVLIKRRSTEL
FY-VP-DVKETGSMTINSSTSHSRDSNINPTVILKERTSEL
FS-V-QDAHLTKTVTINASSSHSGLDDINPTVLLKERSTEL
Figure 2-5. Continued
COX-lb MRASVLGSVCALLVLLREPGCQGDEVTTSTVNPCCYLPCKHWSVCVRYGEDKYECDC 57
COX-2 MNRITFAVFLLALC--------FSFHKEVLGNACCSEPCQNRGVCTAMGSDSYECDC 49
-N 68 .........................................................................................
COX-lb THTGYYGENCSIPELWTRVRQFLKPSPDVVHYILTHFHWLWDIINN-TFLRNVLMRL 113
COX-2 TRTGYRGQNCTTPEFLTWIKISLKPTPNTVHYILTHFKGFWNIINNISFLRDAIMKY 106
*R120 *N144 -
COX-la I" NLSYYTRLLPPVPKDCPLPMGTKGKPVLP 29
COX-lb TARSNLIPSPPTFNSKYNYLSWESYYNLSYYTRILPPVPEDCPTPLGVKGRNGLP 170
COX-2 LLTSRSHMIDSPPTYNADYGYKSWEAYSNLSYYTRALPPVPEDCPTPMGVVGKKELP 163
Q203/H207
COX-la DIKEISERYFKRTEFRPDPQGTNLMFAFMAQHFTHQFFKTSHKVDAGFT LGHVD 86
COX-lb DPQVLFERLLKRRTFRPDPQGSNIMFAFFAQHFTHQFFKTYNRMGLGFT[ALSH[VD 227
COX-2 DVKVLAEKLLVRRRFIPDPQGTSLMFAFFAQHFTHQFFKSDMKNGPAFTAKGHVD 220
------- -----R277----
COX-la ASNIYGEELERQHQLRLHKDGKLKYQLINGEMYPPTVSEVPVHMVYPEGFPAEQRLA 143
COX-lb AGHIYGDSLERQHLLRLFRDGKLKYQLIDGEVYPPSVTDAPVRMSYPPGIPVEKQMA 284
COX-2 LGHIYGENLEKQHKLRLFKDGKLKYTMVNGEVYPPLVKDVGVEMHYPPHVPDSQRFA 277
COX-la IGQEVFGLVPGLTVYATIFLREHNRVCDELKGEHPTWDDEQLFQTARLIIIGEIINI 200
COX-lb IGQEVFGLLPGLSLYATLLREHNRVCD LKAEHPTWDDEQLFQTTRLIIIGETIRI 341
COX-2 VGHEAFGLVPGLLMYATIILREHNRVCD LKGVHPDWDDERLFQTTRLILIGETIKI 334
Y385/H388
COX-la IIEEYVQHLSGYYLKLKYDPSLLFGVRFQYTNRIALEFCYLYHWHPLMPDSFLIDGD 257
COX-lb VIEEYVQHLSGYLLQLKFDPTLLFNSNFQYGNRIALEFSQLYHWHPLMPDSFHISGD 398
COX-2 VIEDYVQHLSGYHFKLKFDPELLFNQRFQYQNRIASEFNTLYHWHPLMPDSFHIEEK 391
*N410
COX-la ELPYSQFLYNTSILMHYGVEKLVDAFSRQPAGQIGGGRNIHQAVLRVAEMVIRDSRA 314
COX-lb ELSYSQFLFNTSVLTHYGVEKLVDAFSRQAAGQIGGGHNINAVITKVIVGTIEESRQ 455
COX-2 DYSYKEFVFNTSVVTEHGINNLVDSFSKQIAGRVAGGRNVPGPIMYVAIKSIENSRK 448
*R513
COX-la ARLQPFNQYRKRFNLKPYSSFYELTGDEEMARGLEELYGDIDALEFYPGLLLEKTRP 371
COX-lb LRIQPFNEYRKRFNLEPYTSFRDFTDSEEIASTLEELYGDIDTLEFYPGLLLEKTRP 512
COX-2 MRYQSLNAYRKRFSMKPYTSFEDLTGEKEMAAILEELYGDVDAVELYPGLLVEKPRT 505
*V523 *S530
COX-la SSIFGESMVEMGAPFSLKGLLGNPINSPEYWKPSTFGGETGFNIIKTSTLKKLVCLN 428
COX-lb GAIFGESMVEMGAPFSLKGLLGNPICSPQYWKPSTFGGKVGFDIVNSASLKKLVCLN 569
COX-2 NAIFGETMVEMGAPFSLKGLMGNPICSPEYWKPSTFGGTAGFDIVNTASLQRLVCNN 562
Figure 2-1. Amino acid alignment of the forms of COX sequenced from the gills of the
euryhaline killifish (Fundulus heteroclitus). The COX-2 sequence is from GenBank
(Accession #AAS21313) and the others were obtained in this study. Amino acids that
share at least 60% identity across sequences are highlighted. *Indicates amino acids
(numbered based on ovine COX-1) involved in COX function: active site of COX
(tyrosine-385, histidine-388, and serine-530), substrate binding site (arginine-120), N-
glycosylation sites (asparagine-68, 144, and 410), sites for peroxidase activity
(glutamine-203 and histidine-207), and the two sites which define conformational
differences in channels between COX-1 and COX-2 in mammals (arginine-513 and
valine-523) (Ishikawa et al., 2007). The black boxes indicate the two domains that
define the haem-binding sites and the dashed box indicates the membrane binding
domain (Kulmacz et al., 2003; Ishikawa et al., 2007). The solid lines represent
disulfide bonds, with the beginning and end of the longest bond represented by
arrows (Kulmacz et al., 2003). The dashed line represents the arginine-277 loop
region, which has been shown to differ between COX-1 and COX-2 forms in
mammals (Yang et al., 2003). The numbers to the right of the sequences indicate the
amino acid number of each sequence.
represented by 1COX) because the hagfish is biologically ancestral to the lamprey. This would
mean that dozens of novel COX forms have yet to be discovered in well-examined groups such
as the mammals and teleosts. With genomes of several of these groups available, it is unlikely
that such a vast amount of sequence data have yet to be found. Therefore, the most likely
scenarios for COX evolution would imply that either hagfish is basal and lamprey is derived,
both are basal, or both are derived. However, placing 1COX, hCOX, or sCOX in the COX-1 or
COX-2 clade implies that another COX form has yet to be found in these species.
It appears that moving 1COX into the base of the COX-1 clade causes less of a change in
Ln L scores than moving hCOX outside the gnathostome COX clade. However, it is apparent
that either sequence can be moved somewhat freely throughout the tree without resulting in
noticeably worse (-10 Ln L) tree scores. This leads to the conclusion that although hCOX and
1COX may represent independent, basal lineages of COX, they are most likely basal members of
the COX-2 and COX-1 clades (respectively) in the gnathostomes. If this is the case then at least
one other form of COX should exist in the hagfish and lamprey, as well as in the dogfish shark.
Summary
Presented here are the first COX sequences from amphioxus, the hagfish, and the
lamprey. All predicted COX sequences from the killifish and sculpin are also described. These
sequences all likely represent functional forms of COX found in the chordates based on
conserved amino acids and domains critical to COX function. For the first time, a sophisticated
phylogenetic analysis of COX forms from the majority of representative chordate lineages was
attempted. Although still not definitive, a general hypothesis for the evolutionary history of
COX in the chordates can be made (Figure 2-16). In this scenario, there are three main COX
groups corresponding to the three subphyla found in the phylum Chordata: the urochordata,
cephalochordata, and vertebrata. Each contains two main COX groups, with the well-
documented COX-1 and COX-2 forms only found in the vertebrata clade. These forms contain
monophyletic groups for the teleosts and the tetrapods, with multiple forms in the teleosts
confirming previous hypotheses (Ishikawa et al., 2007; Ishikawa and Herschman, 2007).
Hagfish and lamprey placement is uncertain, but most likely these sequences represent basal
members of the vertebrata clade. This does not support the timing of genomic duplications in the
chordates according to the most recent view of the 2R Hypothesis (Kasahara, 2007), indicating
two rounds of genome duplication after the origin of the urochordates but prior to the radiation of
the jawed vertebrates. However, several developmentally important proteins have been shown to
not support this timing and indicate an earlier duplication (Hughes, 1999). Multiple COX-1 and
COX-2 genes in the teleosts do support the hypothesis that the teleosts are characterized by
another round of genome duplication after the origin of the vertebrates (Li et al., 2007). Clearly,
this uncertain period representing some 600 million years of evolution is characterized by many
losses and duplications and remains to be resolved definitively.
Figure 2-8. Evolutionary history of COX sequences excluding hagfish (hCOX) and lamprey
(1COX) sequences inferred using the minimum evolution method with a distance
optimality criterion. The optimal tree with the sum of branch length = 6.18990857 is
shown. The percentage of replicate trees in which the associated taxa clustered
together in the bootstrap test (1000 replicates) is shown next to the branches
(Felsenstein, 1985). The tree is drawn to scale, with branch lengths in the same units
as those of the evolutionary distances used to infer the phylogenetic tree. The
evolutionary distances were computed using the JTT matrix-based method (Jones et
al., 1992) and are in the units of the number of amino acid substitutions per site (scale
bar = 0.1 amino acid substitutions per site). The rate variation among sites was
modeled with a gamma distribution (shape parameter = 0.742). The ME tree was
searched using the Close-Neighbor-Interchange (CNI) algorithm at a search level of
1. The Neighbor-Joining algorithm was used to generate the initial tree. All positions
containing gaps and missing data were eliminated from the dataset (Complete
deletion option). There were a total of 147 positions in the final dataset. Phylogenetic
analyses were conducted in MEGA4 (Tamura et al., 2007).
COX-la, COX-lb, and COX-2 are found in the sculpin; COX-la and COX-lb are found in the
killifish; a new COX form named 1COX is found in the lamprey; a new COX form named hCOX
is found in the hagfish; and two ancestral forms named COXc and COXd are found in
amphioxus.
Phylogenetic Analyses
Chapter 2 also describes the phylogenetic analyses used to reconstruct the evolutionary
history of cyclooxygenase in the chordates. The nine novel COX protein sequences described
here, as well as 41 additional COX protein sequences from GenBank and genomic databases,
were subjected to phylogenetic analyses using three optimality criteria: minimum
evolution/distance, maximum likelihood, and Bayesian phylogenetics. In each analysis, two
clades were well-supported in the gnathostomes: a COX-1 clade and a COX-2 clade. Each of
these clades contained a well supported teleost and tetrapod clade. In the teleost clades, clades
for COX-la, COX-lb, COX-2a, and COX-2b were well-supported, confirming the findings of
Ishikawa et al. (2007) and Ishikawa and Herschman (2007). The placement of the more
evolutionarily ancestral chordates was ultimately uncertain, although it appears that the
urochordates, cephalochordates, and vertebrates (containing COX-1 and COX-2) each represent
an independent origin of COX proteins. The hagfish and lamprey sequences most likely
represent basal members of the COX-2 and COX-1 clades of vertebrates, respectively; however,
these forms may belong closer to the root of the tree.
Future Directions
Although here I report 9 novel cyclooxygenase (COX) sequences from a range of
chordates and present a likely scenario for the evolution of this gene family in the chordates,
many questions remain regarding COX evolution and function in the chordates. These include
determining the true history of COX evolution in the earliest chordates, pinpointing the origin of
CHAPTER 2
CYCLOOXYGENASES IN THE ANCESTRAL CHORDATES: SEQUENCE AND
PHYLOGENETIC ANALYSES
Introduction
Cyclooxygenase (COX) is the enzyme that catalyzes the oxidation and subsequent
reduction of arachidonic acid to form Prostaglandin G2 and Prostaglandin H2 (PGH2). PGH2 can
then undergo additional reactions to produce the primary prostaglandins, which act as vascular
tone regulators in the vertebrates. Prostaglandins and COX participate in a variety of
physiological functions in the vertebrates, including inducing fever, maintaining pregnancy, and
regulating ion transport in the kidneys (Steiner et al., 2005; McLaren et al., 1996; Harris et al.,
1994). These functions have been extensively studied in mammals (Vane et al., 1998), but
comparatively little functional data exist for other animals. However, COX has been sequenced
in several evolutionarily more ancestral chordates, particularly the teleosts (Zou et al., 1999;
Roberts et al., 2000; Choe et al., 2006). Based on functional studies of COX in the teleosts, it
seems that some functions are conserved (Choe et al., 2006; Sorbera et al., 2001; Brubacher et
al., 2000; Holland et al., 2003) while others may be altered in some species (Goetz et al., 1989)
or novel (Cha et al., 2006b).
In mammals there are two main forms of cyclooxygenase. The first form was isolated
from sheep seminal vesicles in 1988 and later named COX-1 (DeWitt and Smith, 1988; Merlie et
al., 1988; Yokoyama et al., 1988). A second form was isolated from mouse and chicken
fibroblast cell cultures in the early 1990s and named COX-2 (Kujubu et al., 1991; Xie etal.,
1991; O'Banion et al., 1992). Originally, COX-1 was considered to be a constitutive form that
maintained normal cell functions and COX-2 was considered to be an inducible form that was
up-regulated in inflammatory responses (Funk, 2001). However, studies have shown that this is
an oversimplification and COX-2 is expressed constitutively in the brain (Breder et al., 1995)
Cyclooxygenase Function in Non-Mammals
Although COX has been extensively studied in mammals due to its role in inflammation
and the correlating drug market surrounding COX inhibition, studies in non-mammals are
lacking. Unfortunately, it seems that although non-mammals posses the mammalian COX-1 and
COX-2 forms, these forms may be substantially different (both in structure and function) from
their mammalian counterparts (Grosser et al., 2002) and may also represent independent
evolutionary events, such as duplications of an isoform (teleosts) or independent origins (sea
squirts). In this section, I review the major studies investigating COX function in non-
mammalian chordates.
Fishes
Cyclooxygenase has likely been studied in the fishes more than other non-mammals due
to their status as model organisms. As noted above, COX forms have been found in several
fishes and along with genomic sequence data there is a general consensus that fishes possess
both COX-1 and COX-2 forms that are homologous to those found in mammals. However, there
is recent evidence that teleosts also possess an additional form of COX-1 or COX-2 (Ishikawa et
al., 2007; Ishikawa and Herschman, 2007) and some functional data also exist for these forms.
Osmoregulation
In fishes, the gills are the dominant site of acid/base regulation, nitrogenous waste
secretion, gas exchange, and ion transport (Evans et al., 2005). Several studies have investigated
the osmoregulatory role of COX in the gills of the euryhaline killifish, Fundulus heteroclitus.
This teleost can withstand instant salinity transfer between full strength seawater and fresh water
without any apparent physiological stress (Wood and Laurent, 2003). One study showed that
short circuit current across the opercular epithelium (a tissue with known ion transport
capabilities) was reduced by using a non-specific COX inhibitor (Evans et al., 2004). This
The forms from lamprey and hagfish (1COX and hCOX) are basal to the vertebrate COX clade,
while the forms from amphioxus (COXc and COXd) form a monophyletic group basal to these
forms. The forms from Ciona (COXa and COXb) form another monophyletic group that is basal
to all other COX forms. Bootstrap values give robust support for vertebrate clades, the
amphioxus clade, and the Ciona clade, but give weak support for placement of hCOX, 1COX,
and sCOX. Based on these results and other generated phylogenies (see below), hCOX and
1COX were identified as possible rogue sequences due to their inconsistent positions depending
on the type of analysis used and another distance analysis was completed without including these
sequences (Figure 2-8). The topology of this phylogeny is very similar to the complete COX
phylogeny. However, sCOX is basal to the vertebrate COX clade and not included with the
other vertebrate COX-1 forms. Additionally, bootstrap values for most clades are considerably
more robust than when hagfish and lamprey sequences are included.
To further test relationships between COX proteins, a phylogeny (Figure 2-9) was
generated using the maximum likelihood criterion, 1000 bootstrap replicates, and the program
PhyML (Guindon and Gascuel, 2003). The topology of this tree is very similar to the one
generated using distance criterion. Briefly, it contains well supported, monophyletic clades for
COX-1 teleostt COX-la, teleost COX-lb, teleost COX-1, mammalian COX-1, and vertebrate
COX-1 with sCOX as the most basal sequence) and COX-2 teleostt COX-2, mammalian COX-2,
and vertebrate COX-2). The lamprey (1COX) is also basal to the vertebrate COX clade, the
amphioxus COXc and COXd form a basal monophyletic clade, and the most basal clade in the
tree is a monophyletic COXa and COXb from the sea squirt. However, in the ML analysis, the
hagfish (hCOX) sequence groups at the base of the vertebrate COX-2 clade instead of being
basal to the vertebrate COX clade. Again, the analysis was repeated without hCOX and 1COX
COXa Ciona
COXb Ciona
sCOX Squalus
COX1 Xenopus
1.) COX1 Gallus
COX1 Oryctolagus
1.00 98COX1 Homo
OOCOX1 Canis
1 9 OX1 Bos
1.00
SCOX1 Ovis
0.92 COX1 Mus
1. 1.00
COX1 Rattus
COX1 Monodelphis
S 1.00 d1 COX b Monodelphis
COXlb2 Monodelphis
0.97 COX1 Oncorhynchus
COX1 Danio
0 6 COX1b Tetraodon
COX1b Fundulus
9 COXlb Oryzias
0.99
1.00 r .0COX b Gasterosteus
1.00 1.00
COX1b Myoxocephalus
COX1a Fundulus
0.98
COXla Oryzias
1.i0 COX1a Myoxocephalus
COX1a Gasterosteus
C0.88
COXI a Tetraodon
S COX2 Monodelphis
COX2 Oryctolagus
8a0X2 Canis
75COX2Homo
1 00OX2 Bos
1 'XK
I "TTM 1.00
L COX2 Rattus
COX2 Gallus
0.96
COX2Xenopus
10 COX2b Danio
COX2b Oncorhynchus
1.00 COX2 Tetraodon
COX2 Fundulus
S 1.0 COX2 Oryzias
COX2 Gasterosteus
1.00
0COX2 Myoxocephalus
COX2a Danio
CO1.00
COX2a Oncorhynchus
COXc Branchiostorna
1.0 Branchistma
COXd Branchiostoma
- 1.00
I 1.0(
0.1
other relevant COX proteins and annotated using functional amino acids and domains
highlighted from Ishikawa et al. (2007), Yang et al. (2002), and Kulmacz et al. (2003).
Phylogenetic Analysis
To investigate the evolutionary relationships of the different forms of COX in the
chordates, phylogenies were generated using the novel amino acid sequences generated in this
study and several other COX protein sequences from GenBank and genome databases. All novel
protein sequences were included in the data set. COX sequences from teleosts were included in
the analysis only if all 3 forms of COX were available. These included the zebrafish, rainbow
trout, green spotted puffer, stickleback, euryhaline killifish, and longhorn sculpin. COX
sequences from non-teleosts were included in the analysis if at least two forms of COX were
available and both sequences were designated as Reference Sequences (RefSeqs) in GenBank.
These included the chicken, frog, mouse, rat, human, rabbit, dog, cow, sheep, and opossum.
COX sequences were also included if they represented an evolutionarily interesting group.
These included the dogfish shark and the sea squirt. In total, 50 protein sequences were included
in the analyses (Table 2-7).
Models of evolution to be used in phylogenetic analyses were evaluated to account for
amino acid substitutions, among-site variation, and invariable sites. Using likelihood ratio tests,
a model using the WAG rate matrix (Whelan and Goldman, 2001) and the gamma (F)
distribution for among-site variation was chosen (Table 2-8).
A distance phylogeny was first generated using Molecular Evolutionary Genetics
Analysis 4 (MEGA) software (Tamura et al., 2007) with a minimum evolution (ME) analysis
and 1000 bootstrap replicates. In this analysis the a parameter was fixed to the value estimated
from maximum likelihood (ML) analysis (a = 0.737, see below). Additionally, a JTT rate matrix
(Jones et al., 1992) was used instead of a WAG matrix because a WAG matrix is not available
Prostaglandin 12). It has generally been accepted that these prostaglandins are synthesized by
COX-1, although COX-2 is present in small amounts (Kargman et al., 1996). This is supported
by studies showing that irradiated mice have further reduced numbers of crypt cells
(cytoprotective mucosal cells) when indomethacin (a non-specific COX inhibitor) was
administered, but not when a COX-2 selective inhibitor was administered (Cohn et al., 1997).
COX-2 expression in the esophagus of patients with gastroesophageal reflux disease was
upregulated in response to acid exposure, suggesting a protective function in the distal esophagus
(Lurje et al., 2007). This may indicate that throughout the gastrointestinal (GI) track, COX-2 is
inducibly expressed in response to inflammation responses.
Furthermore, throughout the GI track, inflammatory mediators such as COX may also
serve to repair damaged mucosa (Martin and Wallace, 2006). While it seems that COX-1 is
responsible for baseline prostaglandin expression in the GI tract, COX-2 and not COX-1 seems
to be more involved in response to GI tract cancers (Ristimaki et al., 1997). However, recent
studies on patients with gastric cancer from Dalian, China showed that COX-2 expression was
infrequent in gastric cancers due to hypermethylation and that a COX-2 focused strategy for
treatment would likely be ineffective (Huang et al., 2006).
The Kidney
Prostaglandins can modulate blood flow in the kidney, as in other organs (Vane and
Botting, 1994). In addition, COX-2 has been shown to play a major role in ion regulation and
water balance in the mammalian kidney (Harris et al., 1994). It has also been shown that the
COX-2 selective inhibitor SC-58236 and several non-selective COX-inhibiting NSAIDs
(including sulindac, ibuprofen, and indomethacin) cause medullary interstitial cell (MIC) death
in the kidney, but the COX-1 selective inhibitor SC-58560 was 100-fold less potent for inducing
MIC death (Hao et al., 1999). COX-2 knockout mice also develop terminal nephropathy
COX Petromyzon
COX Myxine
COX2 Monodelphis
,&W canis
- COX2 Oryctolagus
0 Rattus
_QR6g2 Mus
000 COX2 Homo
SCOY2 Bos
COX2 Gallus
)0000
- COX2 Xenopus
Danio
b Oncorhynchus
! Oryzias
Monodelphis
COX1a Myoxoi
1 a Oryzias
I a Fundulus
I ... COXlaTetraodon
0.999 COX1 b Tetraodon
1 b Myoxocephalus
C- OX1 b Gasterosteus
0.939000
.74 0 0.59901 b Fundulus
COX1b Oryzias
COX1 Danio
0.776000
COX1 Oncorhynchus
COXd Branchiostoma
-- C c Branchiostoma
0.1
COXb Ciona
COXa Ciona
COM1 Bos
Mammals
COX-1 and COX-2 sequences have been either cloned or derived from genomic
sequences for most of the classic mammalian model species, including mouse (Accession#
NP_032995 and NP_035328), rat (Accession# NP_058739 and NP_058928), dog (Accession#
NP_001003023 and NP_001003354), cow (Accession# XP_869575 and NP_776870), sheep
(Accession# NP_001009476 and NP_001009432), and rabbit (Accession# NP_001076150 and
NP_001075857). COX sequences also exist for rhesus monkey, guinea pig, pig, horse, and
mink, but these are either partial sequences or are not available for both forms.
A third COX variant has been proposed to occur in mammals (Flower and Vane, 1972)
based on inhibitory studies (see below). This form is known as COX-3 or COX-lb because of its
close similarity to COX-1. Recently, a potential COX-3 (Accession# AY547265) was cloned
from mouse (Kis et al., 2006) and rat (Accession# AY523672, Snipes et al., 2005). However,
these forms likely do not participate in prostaglandin synthesis (Kis et al., 2006) and are not a
separate COX form. Furthermore, a similar study using an artificially created human COX-lb
found no inhibitory similarity to the proposed COX-3 enzyme (Censarek et al., 2006).
Therefore, while a third, genuinely different COX variant may exist in mammals, its sequence
remains elusive.
Analysis of the recently completed gray short-tailed opossum genome yielded the
predicted COX-1 and COX-2 forms (Accession# XP_001370514 and XP_001375945), as well
as two additional COX-1 forms, named COX-lb and COX-lb2 (Accession# XP_001370542 and
XP_001370595). These forms are complete and differ in amino acid composition and length.
These are the first COX-1 variants in a mammal that do not represent modified versions of
normal COX-1 and although is has not been suggested, it is possible that one of these forms
A)
hCOX
18S-
.\lllisejise ''
S.-.. .o -..., .
,".'*''\,i ^
I .w ^'^ <^ ;
Sense
Sense
Figure 3-3. Atlantic hagfish (Myxine glutinosa) COX (hCOX) mRNA expression following
procedures outlined in Choe et al. (2006). A) COX mRNA expression in various
tissues in the hagfish. Expression was constant across tissues for the ribosomal
protein 18S. COX (hCOX) expression was most abundant in the gills (labeled as G),
followed by the heart (H), kidney (K), stomach (S), intestine (I), muscle (M), and
brain (B). B) In situ hybridization of hCOX in the gills of the hagfish from two
contrary gill cross-sections. Arrows highlight staining, indicating hCOX mRNA
expression. Staining for antisense and sense (control) probes is shown.
G M B H S I K
i i i i
GOBH S IK
iUUU11
A)
COX-la
18S
B)
40
35
30
25
20
15
05
00
24 hrs
72 hrs
GO BH SI K
[ [ [
iOX-2
18S-
COX-la 16 COX-2
o Seaw ater
14
20% Seaw ater
S10
10
W 06
04
02
00
24 hrs
72 hrs
Time
Time
Figure 3-2. Longhorn sculpin (Myoxocephalus octodecemspinosus) COX-la and COX-2 mRNA
expression following procedures outlined in Choe et al. (2006). A) COX-la and
COX-2 mRNA expression in various tissues in the sculpin. Expression was constant
across tissues for the ribosomal protein 18S. COX-la was most abundant in the
opercular epithelium (labeled as O), followed by the gills (G), kidney (K), heart (H),
brain (B), intestine (I), and stomach (S). COX-2 was most abundant in the gills,
followed by the opercular epithelium, and kidney. B) In situ hybridization of COX-
lb and COX-2 in the gills of the sculpin. C) Expression of COX-lb and COX-2
mRNA in the gills of the sculpin after 24 and 72 hours following salinity transfers to
100% and 20% seawater. Expression is normalized to the ribosomal gene L8.
This may be due to the incompleteness of some sequences and a lack of sufficient sampling
across taxa or may represent a history that is irresolvable under the current circumstances.
However, the phylogenetic techniques used here may also be improved upon in future analyses.
Phylogenetically uninformative areas of the sequences (such as near the C and N terminus,
which show little homology across sequences) could be removed. Here, protein sequences were
used to gain insight into the evolution of COX functionality, but insights could also be gained by
performing similar analyses using DNA sequence data instead of using their inferred protein
amino acid sequences. This would allow for greater complexity of models to estimate
substitution rates and could take additional factors into account such as codon bias. Using
nucleotide versus amino acid would also enable additional search strategies and phylogenetic
programs to be used. Choosing the optimum search strategy and computer program may be
critical in finding the optimum evolutionary reconstruction. For example, maximum likelihood
analyses PhyML (the program used in this study) may find suboptimal likelihood scores when
the data-set contains little phylogenetic information (Morrison, 2007). However, programs such
as RAxML (Stamatakis, 2006), GARLI (Brauer et al., 2002), and PAUP* (Rogers and Swafford,
1998) may perform better under the same circumstances (Morrison, 2007). Also, evaluating
alternative tree topologies using frameworks other than maximum likelihood may yield
alternative results.
Additional Sequences
Based on the most likely scenario of COX evolution in the chordates, there are several
COX sequences in the chordates examined in this study that remain to be characterized. If
hCOX in actuality represents a basal form of COX-2, then there should also be a COX-1 form in
the hagfish. Similarly, if ICOX and sCOX represent basal forms of COX-1 in the lamprey and
dogfish, then they should also contain versions of COX-2. These additional forms of COX may
Novel COX Forms
Although the two COX sequences from amphioxus were not able to be extended, they
aligned readily with other COX sequences. They show a first haem-binding domain that shares
little conservation with vertebrate forms (20%), as in the urochordates. However, the second
haem-binding domain is completely conserved in these forms, as is histidine-207 which is crucial
for peroxidase activity. The arginine-277 loop region does not resemble COX-1 or COX-2
forms. Due to the truncated nature of these proteins, it is difficult to determine if COX in the
cephalochordates plays similar roles as in the vertebrates, but some functions are likely
conserved. Following the nomenclature used by Jarving et al. (2004) these forms are named
COXc and COXd. The COX sequence found in the gills of the hagfish was extended to
completion and readily aligned with other COX forms. It has conserved amino acids, binding
domains, and bonds critical for COX function. Therefore, it likely plays a similar role in the
hagfish. The hagfish COX has an N-terminus deletion and valine-523 characteristic of COX-2
forms, but there are multiple sites where it shares equal or no homology with COX-1 and COX-2
of vertebrates. Therefore, following the nomenclature of Yang et al. (2002) this form is named
hCOX. Although incomplete, the one COX sequence cloned from the gills of the lamprey
readily aligns with other COX sequences and contains amino acids and regions critical to COX
function. However, like the hagfish sequence it shares no biased homology with COX-1 or
COX-2 forms. This form is therefore named 1COX. As predicted based on phylogenetic
position, COX-la, COX-lb, and COX-2 were found in the gills of the euryhaline killifish and the
longhorn sculpin. With the exception of COX-la in the sculpin, all COX forms from these
teleosts contain conserved regions critical to COX function. As has been reported in other
teleosts (Ishikawa et al., 2007), these forms all contain valine at position 523, unlike in
mammals. COX-la from the sculpin does not have several amino acids critical for COX
94 COX1 b Gasterosteus
44 I= COX1b Myoxocephalus
61 COXlb Oryzias
51 COX1b Fundulus
37 COX1b Tetraodon
1 COX1 Oncorhynchus
60 COX1 Danio
I- COX1a Tetraodon
71 COX1a Fundulus
92 COX1a Oryzias
26 80 -[ COX1a Gasterosteus
57 COX1a Myoxocephalus
99 COX1 Mus
COX1 Rattus
32 8 86 COX1 Bos
98
COX1 Ovis
87 COX1 Oryctolagus
35 34 COX1 Homo
62 COX1 Canis
COX1 Gallus
91 COX1 Monodelphis
99 COX1b Monodelphis
85 COXlb2 Monodelphis
COX1 Xenopus
82 COX2 Gasterosteus
43'1 COX2 Myoxocephalus
52 3 COX2b Oncorhynchus
41 COX2 Oryzias
40 COX2 Fundulus
34= COX2 Tetraodon
0 -- COX2b Danio
COX2a Danio
26 CO)X2a Oncorhynchus
77 COX2 Xenopus
S94 COX2 Bos
0 COX2 Ovis
83 CO)X2 Gallus
3 COX2 Homo
383 COX2 Mus
COX2 Rattus
38 COX2 Monodelphis
13 COX2 Canis
41 COX2 Oryctolagus
sCOX Squalus
COXc Branchiostoma
99 COXd Branchiostoma
COXa Ciona
100 COXb Ciona
01
the COX-1 and COX-2 clades, representing another COX gnathostome lineage. These groupings
are as expected, based on Yang et al.'s (2002) original description of sCOX as neither a COX-1
nor COX-2. These are the first phylogenetic analyses of COX to include sCOX and it is
concluded that sCOX is most likely a basal member of the COX-1 clade.
In all analyses, the COX-1 clade contains robustly supported, monophyletic teleost and
eutherian mammal clades. In the teleost clade, COX-lb and COX-la form well-supported,
monophyletic clades. Although this is expected based on initial studies (Ishikawa et al., 2007;
Ishikawa and Herachmann, 2007), this is the first study to use sophisticated tree-building
techniques to show the monophyly and legitimacy of the COX-1 forms in teleosts. It is also
shown in each analysis that the non-duplicate COX-1 forms from the zebrafish and trout are
more related to COX-lb forms from the acanthopterygiians, grouping with them at the base of a
monophyletic COX-1/COX-lb clade.
The mammalian COX-1 group is not as robustly supported using all types of analyses.
The expected relationships (based on assumed evolutionary relationships in the tetrapods) are
shown and supported with some confidence in the maximum likelihood analysis (ML) when the
hagfish and lamprey sequences are excluded. In this anlaysis, the tetrapods form a monophyletic
group with the frog and chicken at the base of the clade, followed by the opossum sequences,
and then the eutherians. Even in this analysis, the frog and chicken sequences group together,
contrary to what is biologically expected (the frog should be more basal, representing an
evolutionary ancient non-amniote lineage, and the chicken should be a sister to the mammals).
However, this grouping is not well-supported in any analysis. When hagfish and lamprey are
included in ML analysis, the chicken/frog clade occurs at the base of the mammalian/teleost
clade, contrary to biological predictions. However, this relationship is poorly supported. These
Table 2-8. LRTs between amino acid based models of COX evolution
Evolutionary Model*
WAG
WAG + I
WAG + F
WAG + I + F
Ln likelihood (Ln L) Score (Ln Li-LnL2) 2 A d.f.@
-26632
-26315 317 1
-25294 1021 1
-25294 0 2
*Evolutionary models are arranged from the most simple to the most complex. Only models
using the WAG matrix were evaluated. I represents an invariable sites parameter and F
represents an among site variation parameter
A d.f. indicates the difference in free parameters between the complex and simple models
#P values represent the comparison of simple to complex models and were obtained using a chi-
squared distribution test
< 0.01
< 0.01
> 0.05
Table 2-2. Amphioxus COX specific primers used in PCR
Name
COXa Fl
COXaF2
COXa F2.5
COXa F3
COXaF4
COXaF5
COXa R1
COXa R2
COXa R3
COXa 3' Fl
COXa 3' F2
COXa 5' R1
COXa 5' R2
COXa 5' R3
COXbFO
COXb Fl
COXb F1.5
COXbF2
COXb F3
COXb R1
COXb R2
COXb R3
COXb 3' Fl
COXb 3' F2
COXb 3' F3
COXb 5' R1
COXb 5' R2
COXb 5' R3
Orientation
Sense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Nucleotide sequence (5' to 3')
TAC TCC ACC GTG TGG CTG C
TGG GAC GAC GAG AGG CTC
ACC GGT GAG ACC ATC AAC ATC AT
CAC CTG AGC GGC TAC AAC TTT
GAC CTG TTC TGG GAC CCT GAG
CCA GTA CCA GAA CCG CAT CTT C
CGT ACG GCG GGT AGA TCA TG
GCT GCC GCT CCA CGG
CCC GTA GAT ATG GCT CAG GTC A
GAG GCT CTT CCA GAC AGC CA
CAC CTG AGC GGC TAC AAC TTT
TGG CTG TCT GGA AGA GCC TC
TGG AGT AGA CGA ACA GGC CC
CCC CGT AGA TAT GGC TCA GGT
CAG ACA GCT AGA CTC ATT CTT ATC AGT GA
TAT CAA CAT CGT CAT TGG AGA GTA TG
GGC TGG CAA AAA CTT CCA ACT
TGC AGT ACC AGA ACA GCA TAT TTG
AGC ATA TTT GTG GAG TTT AAC CAC TTG
CGT CCG GAG GGT AAA TCA TGT
ACT TCA GTT TCC CGT CCT GGA
CGT AAA CAT GGC TCA TGT CCA CT
GTC CCC GAG AAG AAG CGA TT
GAA CGG CTC TAC CAG ACA GCT AG
TGT TCT GGG ACC CTG AGC TG
TGG TAG AGC CGT TCG TCG TC
GGA GGG TAA ATC ATG TGT ACG GA
ACA TGG CTC ATG TCC ACT GC
MDIBL where they were kept in fiberglass tanks containing 100%, flowing seawater from the
Gulf of Maine. The tanks were exposed to natural conditions. Longhorn sculpin
(Myoxocephalus octodecemspinosus) were purchased from fishermen and were housed in a
similar way at the MDIBL. Atlantic hagfish (Myxine glutinosa) were also purchased from
fishermen and were housed at the MDIBL in 100% seawater and maintained on a 12 h light/dark
cycle. Female, non-migratory lampreys (Petromyzon marinus) were a generous gift from the
USGS Great Lakes Science Center at the Hammond Bay Biological Station in Millersburg,
Michigan and were dissected there. Lancelets (Branchiostoma lancelolatum) were purchased
from Gulf Marine Specimens (Panacea, FL) and were kept in the bags they were shipped in for a
few days before they were sacrificed. All housed animals were fed to satiation regularly.
Some killifish were shipped to the University of Florida (Gainesville, FL) for further
experiments. These killifish were housed as described previously (Choe et al., 2006). Some
hagfish were also transported to the University of Florida (Gainesville, FL) for further
experiments. These hagfish were housed for a few days in a large Rubbermaid tank in 100%
seawater at 4 C before they were killed.
Tissue Collection
After initial anaesthetization with MS-222 (-600 mg L-1), killifish, sculpin, and lampreys
were pithed and/or decapitated. The gill arches (1st and 2nd arches for lampreys, 2nd and 3rd
arches for teleosts) were then removed using sterile, RNAse free dissecting tools. Hagfish were
decapitated and all gill baskets were then removed using sterile, RNAse free tools. Lancelets
were cut in half with sterile, RNAse free tools. After removal, tissues were immediately placed
in liquid nitrogen and stored at -80 OC for 2 weeks to 4 months before further processing.
(Figure 2-10), causing bootstrap values to increase dramatically. Also, the estimated gamma
shape parameter increased slightly from 0.737 to 0.742.
To further evaluate support for COX clades, an aLRT analysis (Anisimova and Gascuel,
2006) was performed. This type of analysis provides values of support similar to Felsenstein's
bootstrap support values but is much faster because bootstrap sampling requires many (1000 in
the previous analyses) runs while aLRT is run only once. The fundamental difference between
the two is that bootstrap support values are based on repeatability whereas the aLRT is a measure
of the likelihood gain of including a branch versus collapsing it (Anisimova and Gascuel, 2006).
The resulting aLRT statistics can be interpreted using parametric Chi-Squared distributions or a
non-parametric Shimodaira-Hasegawa-like (SH) procedure (Anisimova and Gascuel, 2006). The
former seems to be more liberal (giving values similar to Bayesian posterior probabilities) and
the later more conservative (Anisimova and Gascuel, 2006). Here, the most conservative
approach was used and values represent the minimum of Chi-Squared and Shimodaira-Hasegawa
(SH) support values. When all species are included (Figure 2-11) clades of interest have robust
support (> 0.90) except for the vertebrate COX-2 clade (containing hCOX, 0.683) and the clade
containing 1COX (0.87). When hCOX and 1COX are excluded from the analysis (Figure 2-12),
support values generally increase, with a notable increase in the vertebrate COX-2 clade (0.883).
A Basesian phylogenetics (BP) analysis was also performed (Figure 2-13) using the
program MrBayes 3.1.2, with one initial 2 million generation run, and one final 10 million
generation run. The topology of the consensus tree (with burnin = 10%) is similar to those
described above, with well-supported aforementioned vertebrate COX clades. However, in BP
analysis 1COX groups at the base of the amphioxus COXc/COXd clade instead of being the basal
member of the vertebrate COX clade. This grouping is well-supported by posterior probability
represents the elusive COX-3 variant. Functional studies to determine how these forms differ
from normal COX have not been published to date.
Fishes
Cyclooxygenase has been sequenced in several fishes, including the rainbow trout
(Oncorhynchus mykiss) (Zou et al., 1999), brook trout (Salvelinusfontinalis) (Roberts et al.,
2000), Atlantic croaker (Micropogonias undulates), and zebrafish (Danio rerio) (Grosser et al.,
2002). These fishes, along with genomic sequence data from the stickleback (Gasterosteus
aculeatus), green spotted puffer (Tetraodon nigroviridis), pufferfish (Takifugu rubripes), and
Japanese medaka (Oryzias latipes) suggest that fishes posses both COX-1 and COX-2 forms
(Jarving et al., 2004). Furthermore, COX-2 has recently been sequenced from European sea bass
(Dicentrarchus labrax) (Buonocore et al., 2005) and euryhaline killifish (Fundulus heteroclitus)
(Choe et al., 2006), supporting this prediction.
A cyclooxygenase has also been sequenced from the spiny dogfish shark (Squalus
acanthias) (Yang et al., 2002). However, this form was not designated as COX-1 or COX-2
because of its near equal identity to both forms (it is only slightly more similar to COX-1). This
form was named sCOX by the authors and may represent an evolutionarily distinct form of
cyclooxygenase found in the elasmobranchs.
Ishikawa and colleagues have recently further characterized COX evolution in the
teleosts by demonstrating that both the zebrafish (Ishikawa et al., 2007) and rainbow trout
(Ishikawa and Herschman, 2007) have two functional COX-2 genes (named COX-2a and COX-
2b). In addition, by analyzing sequence databases, they concluded that not all teleosts have two
COX-2 forms, but that the stickleback, green spotted puffer, pufferfish, and Japanese medaka
have two COX-1 forms instead (named COX-la and COX-lb) (Ishikawa et al., 2007). They
concluded that a genome duplication before the teleosts, and a subsequent loss of either one
plays a critical role in differing biochemical activity. COX-2 selective inhibitors were thought to
be a major medical breakthrough due to the ability to stop disease induced inflammation without
affecting normal prostaglandin levels produced by COX-1 and therefore causing minimal side
effects. However, due to the constitutive expression of COX-2 in certain tissues (especially the
kidney), COX-2 selective inhibitors may not be as affective as once thought. It is important to
remember that COX-1 and COX-2 are very structurally similar and that their main difference is
due to expression.
As outlined above, COX form and function have been extensively studied extensively in
mammals, especially humans. However, non-mammalian COX forms and their functions have
not been examined in great detail in other chordates. The goal of this research was to
characterize COX evolution in the chordates, with specific attention on the more evolutionarily
ancestral chordates. Describing ancestral COX forms is the first step in elucidating the evolution
of COX function, which may lead to a better understanding of overall COX function, and even
the development of new drugs. In this chapter, I give a brief overview of COX evolution and
function in the chordates, concluding with a brief summary of the research described in the
following chapters.
Cyclooxygenase Evolution
Like mammals, most vertebrates appear to have two homologous copies of
cyclooxygenase named COX-1 and COX-2. However, other COX forms exist throughout the
chordates, including variants of COX-1 and COX-2 as well as forms representing independent
origins of COX that are not closely related to COX-1 or COX-2. There are also COX forms in
corals and other invertebrates (Jarving et al., 2004), but the focus of this section is to review the
forms found in the chordates.
One preliminary method used to infer possible functions for a gene is to determine where
its mRNA transcript or protein product is expressed in the organism. For example, Choe et al.
(2006) concluded that COX-2 may play a role in osmoregulation in the killifish partly due to its
high level of expression in the gills and its presence within mitochondrion-rich cells. Following
similar procedures, I have found that COX-lb mRNA is also expressed highly in the gills of the
killifish (Fig. 3-la). Similar results were also found for COX-la and COX-2 in the sculpin (Fig.
3-2a, Fig. 3-2b), hCOX in the hagfish (Fig. 3-3a), and 1COX in the lamprey (Fig. 3-4). To
investigate which cells express COX mRNA transcripts, in situ hybridization (ISH) was
performed using probes designed against the novel COX sequences described here using
previous protocols (Choe et al., 2006). However, no staining was observed for COX transcripts
in the gills of the killifish (Fig. 3-1b) or sculpin (Fig. 3-2b). This is likely due to errors in probe
design or hybridization protocol, since COX mRNA was abundant in the gills of both the
killifish and sculpin. In the hagfish, probes designed to show hCOX mRNA expression did
produce staining patterns in a population of epithelial cells lining gill filaments and lamellae
(Fig. 3-3b). This gill structure is similar to those ofteleosts but may be functionally different
(Evans et al., 2005) and since ion transport mechanisms in hagfish may not be directly related to
those in teleosts (Choe et al., 1999), it is difficult to assign an osmoregulatory function of hCOX
based on this staining pattern. Finally, since COX-2 mRNA expression was shown to increase
following salinity transfers using qPCR (Choe et al., 2006), a similar procedure was used to
investigate COX expression after salinity acclimation in the gills of the killifish (Fig. 3-1c) and
sculpin (Fig. 3-2c). The only significant result from this experiment was that COX-lb was
expressed more abundantly in killifish chronically acclimated to fresh water than seawater.
However, the same general trend shown for COX-2 in the killifish (Choe et al., 2006) was also
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science
EVOLUTION OF CYCLOOXYGENASE IN THE CHORDATES
By
Justin Chase Havird
May 2008
Chair: David Evans
Major: Zoology
Cyclooxygenase (COX) is the enzyme found in animals responsible for converting
arachidonic acid into prostaglandins. These prostaglandins can then perform multiple functions
including regulation of inflammation responses, changes in vascular tone, and ion
transport/osmoregulation. Before this thesis, only two main forms of cyclooxygenase (COX-1
and COX-2) and one ancestral form (Ciona COX) were known from the chordates. I sequenced
COX genes from several hitherto uninvestigated chordates in order to determine a more
comprehensive scenario for the evolutionary history of cyclooxygenase in the chordates.
After designing primers, partial COX sequences were obtained from the Sea Lamprey
(Petromyzon marinus) and amphioxus (Branchiostoma lancelolatum). Complete COX
sequences were obtained from the Atlantic Hagfish (Myxine glutinosa), euryhaline killifish
(Fundulus heteroclitus), and longhorn sculpin (Myoxocephalus octodecemspinosus). These novel
sequences along with other known COX sequences were then subjected to a variety of
phylogenetic analyses using standard techniques. The results of these analyses suggest a
complex history for the evolution of cyclooxygenase in the chordates.
There appears to have been at least 3 evolutionary independent origins of COX in the
chordates. The forms from the sea squirt (Ciona intestinalis) (COXa and COXb) represent the
a) b) .
c) d)
Figure 2-15. The 21 alternate tree topologies generated to test the effects of relocating the
hagfish (hCOX) and lamprey (ICOX) sequences on the likelihood score (Ln L). Each
topology is simplified, but all relationships remain the same as in the most likely
topology (Figure 2-9) with the exception of hCOX and ICOX. Each topology was
generated using TreeViewl.6.6 based on the most likely topology. Ln L was
calculated for each topology online using PhyML (Guindon et al., 2005), branch
lengths and model parameters were estimated. The topologies are named in Table 2-9
as follows (from most to least likely): a) COX2 Hag (most likely) b) COX1
Lamp/COX2 Hag c) COX2 Agnatha d) Lamp/Amp e) COX2 Vert f) COX1 Hag g)
COX1 Vert h) Ancestral Hag i) COX1 Agnathaj) COX1 Hag/COX2 Lamp k)
Ancestral Agnatha 1) COX1 Lamp m) COX2 Lamp n) Agntha/Amp o) Hag/Amp p)
Hag Internal q) Lamp Internal r) Lamp Mammal s) Lamp Teleost t) Hag Teleost u)
Hag Mammal.
100
Metcalfe, C. D., Koenig, B. G., Bennie, D. T., Servos, M., Ternes, T. A., and Hirsch, R.
(2003). Occurrence of neutral and acidic drugs in the effluents of Canadian sewage
treatment plants. Environ. Toxicol. Chem. 22, 2872-2880.
Morham, S. G., Langenbach, R., Loftin, C. D., Tiano, H. F., and Vouloumanos, N. (1995).
Prostaglandin synthase 2 gene disruption causes severe renal pathology in the mouse.
Cell 83, 473-482.
Morrison, D. A. (2007). Increasing the efficiency of searches for the maximum likelihood tree
in a phylogenetic analysis of up to 150 nucleotide sequences. Syst. Biol. 56, 988-1010.
O'Banion, M., Winn, V. and Young, D. (1992). cDNA cloning and functional activity of a
glucocorticoid-regulated inflammatory cyclooxygenase. Proc. Natl. Acad. Sci. USA 89,
4888-4892.
Olson, D. M. (2005). The promise of prostaglandins: have they fulfilled their potential as
therapeutic targets for the delay of preterm birth? J. Soc. Gynecol. Investig. 12, 466-478.
Otto, J. C. and Smith, W. L. (1995). Prostaglandin endoperoxide synthases-1 and -2. J. Lipid
Mediat. Cell Signal. 12, 139-156.
Patiiio, R., Yoshizaki, G., Bolamba, D., and Thomas P. (2003). Role of arachidonic acid and
protein kinase C during maturation-inducing hormone-dependent meiotic resumption and
ovulation in ovarian follicles of Atlantic croaker. Biol. Reprod. 68, 516-523.
Pough, F. H., Janis, C. M. Heiser, J. B. (2005). Vertebrate life: Seventh edition. pp. 16-48.
Prentice Hall.
Reed, D. W., Bradshaw, W. S., Xie, W., and Simmons, D. L. (1996). In vivo and in vitro
expression of a non-mammalian cyclooxygenase-1. Prostaglandins. 52, 269-284.
Reiter, R., Resch, U., and Sinzinger, H. (2001). Do human platelets express COX-2?
Prostaglandins Leukot. Essent. Fatty Acids. 64, 299-305.
Ristimiki, A., Honkanen, N., Jinkilii, H., Sipponen, P., and Hark6nen, M. (1997).
Expression of cyclooxygenase-2 in human gastriccarcinoma. Cancer Res. 57, 1276-1280.
Roberts, S. B., Langenau, D. M. and Goetz, F. W. (2000). Cloning and characterization of
prostaglandin endoperoxide synthase-1 and -2 from the brook trout ovary. Mol. Cell.
Endocrinol. 160, 89-97.
Rogers, J. S., and Swofford, D. L. (1998). A fast method for approximating maximum
likelihood of phylogenetic trees from nucleotide sequences. Syst. Biol. 47, 77-89.
sharks, hagfish, lampreys, cephalochordates, and urochordates. Indeed, the origin of the
mammalian forms of COX is unclear since the sea squirt possesses evolutionarily distinct COX
forms. Resolving the evolution of COX in the chordates could prove invaluable in determining
new targets for drug development (Searls, 2003).
In Chapter 2, I describe the cloning and characterization of 9 new COX genes in the
chordates. These include COX-1 and COX-2 forms from teleosts as well as novel forms from
the hagfish, lamprey, and amphioxus. In phylogenetic analyses, these forms indicate that several
origins of COX occurred during chordate evolution, with the mammalian COX-1 and COX-2
forms likely originating with the hagfish craniatess). Data from the teleosts supports the findings
of Ishikawa and colleges that teleosts in the Acanthopterygii posses two COX-1 forms and one
COX-2 form, whereas earlier teleosts such as the zebrafish and rainbow trout posses two COX-2
forms and one COX-1 form. Furthermore, analyses of protein alignments from novel sequences
indicate conserved and derived functional residues and areas between novel COX forms and the
COX-1 and COX-2 forms of mammals.
In Chapter 3, I summarize the findings of this research and suggest future avenues of
research based on the results presented here. I also present preliminary data investigating the
possible function of cyclooxygenase in these ancestral chordates.
Table 2-9. LRTs between tree topologies with altered positions of hCOX and 1COX
Tree Topology Name* Ln likelihood (Ln L) Score Estimated ag (Ln L1-LnL2)-2T
COX2 Hag (most likely) -25294.22 0.737
COX1Lamp/COX2Hag -25295.57 0.737 2.70
COX2 Agnatha -25299.09 0.737 9.74
Lamp/Amp -25299.27 0.737 10.1
COX2 Vert -25299.61 0.736 10.7
COX1 Hag -25300.92 0.735 13.4
COX1 Vert -25301.22 0.735 14.0
Ancestral Hag -25301.56 0.735 14.6
COX1 Agnatha -25301.78 0.735 15.1
COX1Hag/COX2Lamp -25302.19 0.735 15.9
Ancestral agnatha -25302.55 0.735 16.7
COX1 Lamp -25303.25 0.735 18.1
COX2 Lamp -25303.85 0.735 19.3
Agnatha/Amp -25308.82 0.735 29.2
Hag/Amp -25308.90 0.735 29.4
Hag Internal -25321.15 0.737 53.9
Lamp Internal -25322.77 0.736 57.1
Lamp Mammal -25341.41 0.732 94.4
Lamp Teleost -25344.68 0.735 101
Hag Teleost -25350.30 0.734 112
Hag mammal -25408.58 0.731 229
*Each tree has the same topology as the most likely tree (Figure 2-9) but with alternative
placements of hagfish (hCOX) and lamprey (1COX) sequences as noted in Figure 2-15.
@a represents the shape value under the F model
Where Lnl is the most likely tree score and Ln2 is the tree score of the topology in question
COX-2a Danio
I COX-2a Oncorhynchus
COX-2b Danio
COX-2b Oncorhynchus
COX-2 Acanthopterygians
sCOX Squalus
COX-la Acanthopterygians
COX-1 b Acanthopterygians
COX-1 Danio
COX-1 Oncorhynchus
Figure 1-2 General phylogeny representing the different COX forms in the fishes. Based on
recent studies, it appears that teleosts have multiple forms of COX-1 or COX-2
(Ishikawa et al., 2007; Ishikawa and Herschman, 2007). It is hypothesized that with
the origin of the teleosts, COX-1 and COX-2 underwent duplication, but that in one
lineage (represented by the zebrafish and rainbow trout) one COX-1 form was lost
and in another lineage (represented by the Acanthopterygians) one COX-2 form was
lost. Also included is the form cloned from the spiny dogfish (Yang et al., 2002)
which loosely groups with COX-1 forms, although it shares near equal identity with
COX-1 and COX-2. This phylogeny was simplified from an unpublished phylogeny
generated using the online program PHYML (Guindon and Gascuel, 2003; Guindon
et al., 2005). Abbreviations: Danio = Danio rerio (zebrafish), Oncorhynchus =
Oncorhynchus mykiss (rainbow trout), Squalus = Squalus acanthias (spiny dogfish).
The Acanthopterygians clade is represented by the stickleback (Gasterosteus
aculeatus), green spotted puffer (Tetraodon nigroviridis), pufferfish (Takifugu
rubripes), and Japanese medaka (Oryzias latipes).
M.S., Justin plans to continue his academic career by completing a Ph.D. in a related
marine science field.
sculpin (Myoxocephalus octodecemspinosus). This product (hereafter referred to as COX-2
Myoxocephalus) was then extended and completed using RACE and specific primers (Table 2-
6), resulting in a 2618 bp product that contains a 605 amino acid open reading frame. This
sequence grouped with COX-2 forms from other teleosts in all phylogenetic analyses, inciting
the use of COX-1 degenerate primers (Table 2-1) CH COX-1 F2 and CH COX-1 R2 in PCR
reactions to amplify a 1000 bp product (hereafter referred to as COX- la Myoxocephalus). This
product was then extended and completed using RACE primers (Table 2-6), resulting in a 2633
bp product encoding a 622 amino acid open reading frame. This product grouped with COX-la
forms from other teleosts in all phylogenetic analyses, prompting the design of degenerate
primers used to amplify COX-lb forms in teleosts over COX-la forms (Table 2-1). Using the
degenerate primers CH COX-lb Fl and CH COX-lb RO (Table 2-1), an initial 895 bp product
(hereafter referred to as COX-lb Myoxocephalus) was amplified in PCR reactions. This product
was extended and completed using specific RACE primers (Table 2-6), resulting in a 2332 bp
product containing a 600 amino acid open reading frame. This product grouped with other
teleost COX-lb forms in all phylogenetic analyses, suggesting that the COX-2, COX-la, and
COX-lb forms sequenced represent 3 different COX forms expressed in the gills of the longhorn
sculpin.
Sequence Analyses
Teleosts COX-la, COX-lb, and COX-2
Three different COX forms are now available from the teleosts studied here: the COX-2
sequenced from the euryhaline killifish (Choe et al., 2006), the COX-2 sequenced here from the
longhorn sculpin, and two the COX-1 forms sequenced here from both the killifish and sculpin,
named as COX-la and COX-lb (following Ishikawa and Herschman, 2007). COX-la and COX-
lb predicted proteins from the killifish were more similar to Mus COX-1 than COX-2 (70%
94 COX1b Gasterosteus
44 0ICOX1b Myoxocephalus
60 COX1b Oryzias
50 ICOX1b Fundulus
3 COX1 b Tetraodon
72 0COX1 Oncorhynchus
54 COX1 Danio
SCOX1 a Tetraodon
70 COX1a Fundulus
92 0COX1a Oryzias
22 79 COX1a Gasterosteus
55 0cOX1a Myoxocephalus
Sr COX1 Mus
COX1 Rattus
30 COX1 Oryctolagus
2 COX1 Homo
87 COX1 Canis
45 0COX1 Bos
86 COX1 Ovis
COX1 Gallus
81 0 COX1 Monodelphis
3 COX1b Monodelphis
56 84 COX1 b2 Monodelphis
COX1 Xenopus
sCOX Squalus
83 COX2 Gasterosteus
44 ICO2 Myoxocephalus
538 COX2b Oncorhynchus
53
40 COX2 Oryzlas
3 COX2 Fundulus
34 COX2 Tetraodon
74 COX2b Danio
COX2a Danio
27 COX2a Oncorhynchus
77 -COX2 Xenopus
94 COX2 Bos
26
23 COX2 Ovis
80 COX2 Gallus
4 COX2 Homo
434 COX2 Canis
57 COX2 Oryctolagus
38 COX2 Monodelphis
2 COX2 Mus
32- COX2 Rattus
1COX Petromyzon I
|hCOX Myxine
[COXc Branchiostoma
99 0-COXd Branchiostoma
COXa Clona
100 COXb Ciona
0.05
Kujubu, D., Fletcher, B., Varnum, B., Lim, R. and Herschman, H. (1991). TIS10, a phorbol
ester tumor promoter-inducible mRNA from Swiss 3T3 cells, encodes a novel
prostaglandin synthase/cyclooxygenase homologue. J. Biol. Chem. 266, 12866-12872.
Lear, S., Silva, P., Kelley, V. E., and Epstein, F. H. (1990). Prostaglandin E2 inhibits oxygen
consumption in rabbit medullary thick ascending limb. Am. J. Physiol. 258, F1372-1378.
Li, G., Zhang, Q. J., Ji, Z. L., and Wang, Y. Q. (2007). Origin and evolution of vertebrate
ABCA genes: a story from amphioxus. Gene. 405, 88-95.
Li, J., Simmons, D. L., and Tsang, B. K. (1996). Regulation of hen granulosa cell
prostaglandin production by transforming growth factors during follicular development:
involvement of cyclooxygenase II. Endocrinology. 137, 2522-2529.
Liu, C., Li, R., Liu, Z., Yin, S., and Wang, Z. (2006). The role of prostaglandins and the
hypothalamus in thermoregulation in the lizard, Phrynocephalus przewalskii (Agamidae).
J. Comp. Physiol. [B]. 176, 321-328.
Lurje, G., Vallbohmer, D., Collet, P. H., Xi, H., Baldus, S. E., Brabender, J., Metzger, R.,
Heitmann, M., Neiss, S., Drebber, U., Holscher, A. H., and Schneider, P. M. (2007).
COX-2 mRNA expression is significantly increased in acid-exposed compared to
nonexposed squamous epithelium in gastroesophageal reflux disease. J. Gastrointest.
Surg. 11, 1105-1111.
Mathonnet, M., Lalloue, F., Danty, E., Comte, I., and Lievre, C. A. (2001). Cyclo-
oxygenase 2 tissue distribution and developmental pattern of expression in the
chicken. Clin. Exp. Pharmacol. Physiol. 28, 425-432.
Martin, G. R. and Wallace, J. L. (2006). Gastrointestinal inflammation: a central component of
mucosal defense and repair. Exp. Biol. Med. (Maywood). 231, 130-137.
Martin, F., Fletcher, D., Chauvin, M., and Bouhassira, D. (2007). Constitutive
cyclooxygenase-2 is involved in central nociceptive processes in humans.
Anesthesiology. 106, 1013-1018.
McLaren, W. J., Young, I. R.,Wong, M. H., and Rice, G. E. (1996). Expression of
prostaglandin G/H synthase-1 and -2 in ovine amnion and placenta following
glucocorticoidinduced labour onset. J. Endocrinol. 151, 125-135.
Merlie, J. P., Fagan, D., Mudd, J. and Needleman, P. (1988). Isolation and characterization of
the complementary DNA for sheep seminal vesicle prostaglandin endoperoxide synthase
(cyclooxygenase). J. Biol. Chem. 263, 3550-3553.
Table 2-6. Sculpin COX specific primers used in PCR
Name
Mo COX-2 3' Fl
Mo COX-2 3' F2
Mo COX-2 3' F3
Mo COX-2 3' F4
Mo COX-2 3' F5
Mo COX-2 3' F6
Mo COX-2 mid Fl
Mo COX-2 mid F2
Mo COX-2 mid R1
Mo COX-2 mid R2
Mo COX-2 5' R1
Mo COX-2 5' R2
Mo COX-2 5' R3
Mo COX-la 3' Fl
Mo COX-la 3' F2
Mo COX-la 3' F3
Mo COX-la 3' F4
Mo COX-la 3' F5
Mo COX-la 3' F6
Mo COX-la 5' R1
Mo COX-la 5' R2
Mo COX-la 5' R3
Mo COX-lb 3' Fl
Mo COX-lb 3' F2
Mo COX-lb 3' F3
Mo COX-lb 5' R1
Mo COX-lb 5' R2
Mo COX-lb 5' R3
Orientation
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Antisense
Antisense
Sense
Sense
Sense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Sense
Sense
Sense
Antisense
Antisense
Antisense
Nucleotide sequence (5' to 3')
CCC CGG TCT GAT GAT GTA CG
ACC GAG TGT GTG ACG TGT TGA
TCG AGG ACT ACG TGC AGC AC
CAG GTC TAA CGC AAT CTT TGG G
CCT CAA GGG CTT AAT GGG AAA C
AAC ATC GTC AAC ACC GCC TC
ACA GTT CAT TCC GGA TCC ACA
CAC CAG CCT GAT GTT TGC ATT
CGA AGC GGT GAG ATT CAG GA
CTC CAT CCA GGA CCT GAT ATT TAA G
TCC GGA ATG AAC TGT CTT CTC A
AGG TCG GAG GAC TAT CAA TCA AGT
GAA GGA GAT GGA GTT AAT GAT GTT CC
CAT CAC CTC CGG CTT CAT AAA
GAG GGA GCA TAA CAG ACT CTG TGA
CTG GAG TTC TGC CAC CTC TAC C
GCA GCC CTT CAA TGA ATA CAG G
CAT GCT GGA GAA GAC CCT TCC
GAG TAT GTT GGA GAT GGG TGC TC
GCC GAA ATG TCT TCC TCC TAA A
CAT CGG TAA AGG GCA GTC CTC
TGA CCG TCA GCA CTA ATC TCA TG
TCA CAC ACT ACG GCA TCG AGA
TTG GCG GTG GCT TTA ACA TC
CAG CCC TTC AAC GAG TAC AGG
GAG CGA TAC GGT TCC CAT ACT G
GTT GTG CTC CCT GAG CCA GA
AAT AGC CAT CTG AGC CTC AGG A
Grosser, T., Yusuff, S., Cheskis, E., Pack, M. A. and FitzGerald, G. A. (2002).
Developmental expression of functional cyclooxygenases in zebrafish. Proc. Natl.
Acad. Sci. USA 99, 8418-8423.
Guindon, S. and Gascuel, 0. (2003). A simple, fast, and accurate algorithm to estimate large
phylogenies by maximum likelihood. Syst. Biol. 52, 679-704.
Guindon, S., Lethiec, F., Duroux, P. and Gascuel, 0. (2005). PHYML Online--a web server
for fast maximum likelihood-based phylogenetic inference. Nucleic Acids Res. 33,
W557-559.
Gu, X. (2001). Maximum-Likelihood approach for gene family evolution under functional
divergence. Mol. Biol. Evol. 18, 453-464.
Gu, X. (2006). A simple statistical method for estimating type-II (cluster specific) functional
divergence of protein sequences. Mol. Biol. Evol. 23, 1937-1945.
Guo, Q., Wang, L., Ruan, K and Kulmacz, R. J. (1996). Role of Val509 in time-dependent
inhibition of human prostaglandin H synthase-2 cyclooxygenase activity by isoform-
selective agents. J. Biol. Chem. 271, 19134-19139.
Hao, C.-M., Komhoff, M., Guan, Y., Redha, R. and Breyer, M. D. (1999). Selective
targeting of cyclooxygenase-2 reveals its role in renal medullary interstitial cell
survival. Am. J. Physiol. 277, F352-F359.
Harris, R. C. and Breyer, M. D. (2001). Physiological regulation of cyclooxygenase-2 in the
kidney. Am. J. Physiol. 281, F1-F11.
Harris, R. C. and Breyer, M. D. (2006). Update on cyclooxygenase-2 inhibitors. Clin. J. Am.
Soc. Nephrol. 1, 236-245.
Harris, R. C., McKanna, J. A., Akai, Y., Jacobson, H. R., Dubois, R. N. and Breyer, M. D.
(1994). Cyclooxygenase-2 is associated with the macula densa of rat kidney and
increases with salt restriction. J. Clin. Invest. 94, 2504-2510.
Hamberg, M. and Samuelsson, B. (1973). Detection and isolation of an endoperoxide
intermediate in prostaglandin biosynthesis. Proc. Natl. Acad. Sci. USA. 70, 899-903.
Hamberg, M., Svensson, J. and Samuelsson, B. (1974). Prostaglandin endoperoxides. A new
concept concerning the mode of action and release of prostaglandins. Proc. Natl. Acad.
Sci. USA. 71, 3824-3828.
Hebert, S. C., Culpepper, R. M., and Andreoli, T. E. (1981). NaCl transport in mouse
medullary thick ascending limb. I. Functional nephron heterogeneity and ADH-
stimulated NaCl cotransport. Am. J. Physiol. 241, F412-F431.
Rose, T. M., Schultz, E. R., Henikoff, J. G., Pietrokovski, S., McCallum, C. M., and
Henikoff, S. (1998). Consensus-degenerate hybrid oligonucleotide primers for
amplification of distantly related sequences. Nucleic Acids Res. 26, 1628-1635.
Rowley, A. F., Vogan, C. L., Taylor, G. W. and Clare, A. S. (2005). Prostaglandins in non-
insectan invertebrates: recent insights and unsolved problems. J. Exp. Biol. 208, 3-14.
Ryan, C. A. (2000). The systemin signaling pathway: differential activation of plant defensive
genes. Biochim. Biophys. Acta. 1477, 112-121.
Sawdy, R., Slater, D., Fisk, N., Edmonds, D. K., and Bennett, P. (1997). Use of a cyclo-
oxygenase type-2-selective nonsteroidal anti-inflammatory agent to prevent preterm
delivery. Lancet 350, 265-266.
Searls, D. B. (2003). Pharmacophylogenomics: genes, evolution and drug targets. Nat. Rev.
Drug. Discov. 2, 613-623.
Seebacher, F. and Franklin, C. E. (2003). Prostaglandins are important in thermoregulation of
a reptile (Pogona vitticeps). J. Exp. Biol. 206, 1143-1151.
Seebacher, F. and Franklin, C. E. (2004). Integration of autonomic and local mechanisms in
regulating cardiovascular responses to heating and cooling in a reptile (Crocodylus
porosus). J. Comp. Physiol. [B]. 174, 577-585.
Shakhmatova, E. I., Kuznetsova, A. A., Prutskova, N. P., Bogolepova, A. E., and Natochin,
I. V. (1997). The effect of cyclooxygenase inhibitors on ion and water transport in the
human kidney and frog skin and bladder. Ross. Fiziol. Zh. Im. I. M. Sechenova. 83, 68-
75.
Singh, G. and Triadafilopoulos, G. (1999). Epidemiology ofNSAID induced gastrointestinal
complications. J. Rheumatol. 26, 18-24.
Sirois, J., Sayasith, K., Brown, K. A., Stock, A. E., Bouchard, N., and Dor6, M. (2004).
Cyclooxygenase-2 and its role in ovulation: a 2004 account. Hum. Reprod. Update. 10,
373-385.
Skoblina, M. N., Kondrat'eva, O. T., Nikiforova, G. P., and Huhtaniemi, I. (1997). The role
of eicosanoids and progesterone in the ovulation of the oocytes in the common frog.
Ontogenez. 28, 211-216.
Snipes, J. A., Kis, B., Shelness, G. S., Hewett, J. A. and Busija, D. W. (2005). Cloning and
characterization of cyclooxygenase-ib (putative cyclooxygenase-3) in rat. J. Pharmacol.
Exp. Ther. 313, 668-676.
most ancestral forms, the forms from amphioxus represent another independent origin (COXc
and COXd), and the vertebrate forms (COX-1 and COX-2) represent yet another origin, with the
hagfish and lamprey as likely basal members. Furthermore, the teleosts appear to have duplicate
COX-1 (COX-la and COX-lb) or COX-2 (COX-2a and COX-2b) forms. These results agree
with previous studies and reveal novel information about COX forms in the chordates. More
intense sampling of other chordates could reveal other novel origins of COX or contribute to a
more robust understanding of COX evolution. Finally, the function of most non-mammalian
COX forms remains to be investigated.
suggests that COX-2 plays a role in ion transport in fishes that may be similar to its function in
the kidneys of mammals. Further studies supported this initial hypothesis by showing that COX-
2 is expressed most abundantly in the gill, opercular epithelium, and kidney of the killifish (Choe
et al., 2006), tissues that are known to be involved in ion transport in fishes. This is supported in
the zebrafish, which shows the highest levels of COX-2 in the gills (Grosser et al., 2002).
Furthermore, COX-2 was localized in mitochondrion-rich cells in the gills of killifish, the main
sites of ion uptake and secretion in the gills (Choe et al., 2006). Finally, COX-2 expression was
shown to significantly increase following either hypotonic or hypertonic salinity transfers,
suggesting COX-2 may play a role in maintaining cell homeostasis or promoting cell survival
during periods of osmotic shock (Choe et al., 2006). Taken together, these results strongly
suggest that COX-2 in the gills of teleosts acts in the same way as COX-2 in the kidneys of
mammals, including regulating ion transport and promoting cell survival.
Cyclooxygenase has also been shown to play an osmoregulatory role in the rectal gland
of the dogfish shark, Squalus acanthias. Sharks rely on the rectal gland rather than the gills for
regulation of salt secretion and a COX form (sCOX) has been cloned from the rectal gland (Yang
et al., 2002). This form was expressed most abundantly in the rectal gland of the shark, where
PGE2 production was also high (Yang et al., 2002). Finally, using a COX-2 specific inhibitor,
vasoactive intestinal peptide mediated chloride secretion decreased in the rectal gland, but then
recovered following removal of the inhibitor (Yang et al., 2002). Even though sCOX is slightly
more similar to COX-1 of mammals than COX-2, this result can be explained by the presence of
valine at position 523 instead of isoleucine (and thus conferring COX-2 inhibitory properties).
This result suggests that sCOX plays a role in ion transport in the rectal glands of sharks that
may be similar to the gills of teleosts or the kidneys of mammals. The osmoregulatory role of
COXb Ciona
SCOXa Ciona
COX2 Monod
CQX2 Canis
X2b Onco
COX2 Tetra
COX1a Myox
COXa Tetr
COX b Oryz
COX b Mono
O"uWX1 Canis
0.O99' 51)
L- CI1 BDos
COXd Branc
Branc
0.1
Cao, C., Matsumura, K., Yamagata, K., and Watanabe, Y. (1996). Endothelial cells of the
brain vasculature express cyclooxygenase-2 mRNA in response to systemic interleukin-1:
a possible site of prostaglandin synthesis responsible for fever. Brain Res. 733, 263-272.
Carrasquer, G. and Li, M. (2002). Effect of ibuprofen and rofecoxib transport parameters in
the frog corneal epithelium. J. Membr. Biol. 190, 127-132.
Censarek, P., Freidel, K., Hohlfeld, T., Schr6r, K. and Weber, A. A. (2006). Human
cyclooxygenase-lb is not the elusive target of acetaminophen. Eur. J. Pharmacol.
551, 50-53.
Cha, Y. I., Kim, S. H., Solnica-Krezel, L., and Dubois, R. N. (2005). Cyclooxygenase- 1
signaling is required for vascular tube formation during development. Dev. Biol. 282,
274-283.
Cha, Y. I., Solnica-Krezel, L., and DuBois, R. N. (2006a). Fishing for prostanoids:
deciphering the developmental functions of cyclooxygenase-derived
prostaglandins. Dev. Biol. 289, 263-272.
Cha, Y. I., Kim, S. H., Sepich, D., Buchanan, F. G., Solnica-Krezel, L., and DuBois, R. N.
(2006b). Cyclooxygenase-1-derived PGE2 promotes cell motility via the G-protein-
coupled EP4 receptor during vertebrate gastrulation. Genes Dev. 20, 77-86.
Challis, J. R., Sloboda, D. M., Alfaidy, N., Lye, S. J., Gibb, W., Patel, F. A., Whittle, W. L.
and Newnham, J. P. (2002). Prostaglandins and mechanisms of preterm birth.
Reproduction. 124, 1-17.
Chakraborty, I., Das, S. K.,Wang, J., and Dey, S. K. (1996). Developmental expression of the
cyclo-oxygenase-1 and cyclo-oxygenase-2 genes in the peri-implantation mouse uterus
and their differential regulation by the blastocyst and ovarian steroids. J. Mol.
Endocrinol. 16, 107-122
Choe, K. P., Edwards, S., Morrison-Shetlar, A. I., Toop, T., and Claiborne, J. B. (1999).
Immunolocalization of Na+/K(+)-ATPase in mitochondrion-rich cells of the atlantic
hagfish (Myxine glutinosa) gill. Comp. Biochem. Physiol. 124, 161-168.
Choe, K. P., Havird, J., Rose, R., Hyndman, K, Piermarini, P. and Evans, D. H. (2006).
COX2 in a euryhaline teleost, Fundulus heteroclitus: primary sequence, distribution,
localization, and potential function in gills during salinity acclimation. J. Exp. Biol. 209,
1696-1708.
Cohn, S. M., Schloemann, S., Tessner, T., Seibert, K., and Stenson, W. F. (1997). Crypt stem
cell survival in the mouse intestinal epithelium is regulated by prostaglandins synthesized
through cyclooxygenase-1. J. Clin. Invest. 99, 1367-1379.
LIST OF REFERENCES
Abbasi, A. A. and Grzeschik, K. H. (2007). An insight into the phylogenetic history of HOX
linked gene families in vertebrates. BMC Evol. Biol. 7, 239.
Allegaert, K., Rayyan, M., and Anderson, B.J. (2006). Impact of ibuprofen administration on
renal drug clearance in the first weeks of life. Methods Find. Exp. Clin. Pharmacol. 28,
519-522.
Anisimova, M. and Gascuel, 0. (2006). Approximate likelihood-ratio test for branches: A fast,
accurate, and powerful alternative. Syst. Biol. 55, 539-552.
Arruda, J. A. (1982). Interaction of calcium and cyclooxygenase inhibitors on transport by the
turtle and toad bladders. Arch. Int. Pharmacodyn. Ther. 257, 319-334.
Balaji, T., Ramanathan, M., and Menon, V. P. (2007). Localization of cyclooxygenase-2 in
mice vas deferens and its effects on fertility upon suppression using nimesulide: a
preferential cyclooxygenase-2 inhibitor. Toxicology. 234, 135-144.
Beiche, F., Scheuerer, S., Brune, K., Geisslinger, G., and Goppelt-Struebe, M. (1996). Up-
regulation of cyclooxygenase-2 mRNA in the rat spinal cord following peripheral
inflammation. FEBS Lett. 390, 165-169.
Bicego, K. C., Steiner, A. A., Antunes-Rodrigues, J., and Branco, L. G. (2002).
Indomethacin impairs LPS-induced behavioral fever in toads. J. Appl. Physiol. 93,
512-516.
Buonocore, F., Forlenza, M., Randelli, E., Benedetti, S., Bossfi, P., Meloni, S., Secombes, C.
J., Mazzini, M. and Scapigliati, G. (2005). Biological activity of sea bass
(Dicentrarchus labrax L.) recombinant interleukin-lbeta. Mar. Biotechnol. (NY). 7, 609-
617.
Brauer, M. J., Holder, M. T., Dries, L. A., Zwickl, D. A., Lewis, P. 0., and Hillis, D. M.
(2002). Genetic algorithms and parallel processing in maximum-likelihood phylogeny
inference. Mol. Biol. Evol. 19, 1717-1726.
Breder, C. D., Dewitt, D., and Kraig, R. P. (1995). Characterization of inducible
cyclooxygenase in rat brain. J. Comp. Neurol. 355, 296-315.
Breder, C. D. and Saper, C. B. (1996). Expression of inducible cyclooxygenase mRNA in the
mouse brain after systemic administration of bacterial lipopolysaccharide. Brain Res.
713, 64-69.
Brubacher, J. L., Secombes, C. J., Zou, J., and Bols, N. C. (2000). Constitutive and LPS-
induced gene expression in a macrophage-like cell line from the rainbow trout
(Oncorhynchus mykiss). Dev. Comp. Immunol. 24, 565-574.
scores (0.93). As in ML analysis, hCOX groups with vertebrate COX-2 proteins at the base of
the clade. Also similar to previous analyses, when 1COX and hCOX are excluded (Figure 2-14)
scores improve and all clades have robust support.
Due to the contrasting topologies generated using distance, ML, and BP methods, 21
different tree topologies (Figure 2-15) were generated by using the most likely tree from ML
analysis (Figure 2-9) and relocating hCOX and/or 1COX sequences to other branches of the tree.
Positions of all other sequences were not altered from the most likely tree. A Ln likelihood score
(Ln L) was generated for each topology using online execution of the program PhyML (Guindon
et al., 2005). Likelihood scores were compared to the most likely tree. These tests indicate that
1COX and hCOX can be moved somewhat freely around the base of the tree without resulting in
a noticeably different Ln L than the most likely topology (Table 2-9). It appears that moving
1COX to the base of the COX-1 clade causes the smallest change in Ln L, whereas moving
hCOX or 1COX deeply into COX-1 or COX-2 clades causes the largest changes in Ln L.
Discussion
Here, I present the first forms of the enzyme cyclooxygenase (COX) sequenced from the
hagfish, lamprey, and amphioxus. I also show that the killifish and sculpin conform to the
current hypothesis that teleosts posses additional COX-1 and COX-2 forms. The phylogenetic
relationships among the various COX forms are still not definite, although the results presented
here allow some conclusions to be drawn and predictions to be made. Specifically, it appears
that COX sequences in the chordates form at least 3 well supported phylogenetic groups: the
urochordata, cephalochordata, and vertebrata. Hagfish and lamprey positions are variable
depending upon the analysis used, but likelihood scores seems to favor placement within the
vertebrata clade.
shown for COX-lb in the killifish: an increase in expression 3 hours after salinity transfer
(although P = 0.06 in this case) and a return to pre-transfer expression levels by 8 hours after
transfer. Taken as a whole, these preliminary results suggest that the different COX forms may
be playing various roles in osmoregulation in the killifish, sculpin, lamprey, and hagfish.
Clearly, COX function in the non-mammalian chordates has just begun to be investigated
and many questions still remain. Future studies may investigate COX roles in these chordates by
examining protein expression patterns using immunohistochemistry, blocking COX function via
commercially available COX inhibitors, or examining COX expression during reproductive
events. Of particular interest are the differential roles of the various COX forms in the teleosts
and the possible partitioning of functionality between COX-la and COX-lb or COX-2a and
COX-2b.
*Y385/H388 *N410
sCOX EFDHLYHWHPLMPDSFIVKGQDFSYKDFLFNTDILLNLGVDALVESFSKQIAG 424
1COX EFNHLYHWHPLMPDSLLIDGRNYSYDEFLFNPGLLADKKLMPLVRSFMRQRAG 226
hCOX EFNHLYHWHGLNPDAFRVGTQEYQYSQFLFNNTILLNHGVRGLLEAFNVQQAG 422
COXc EFNHL 177
COXd EFNHL 177
sCOX RIGGGRNIHQSLLHIAIATIEHGRLLRFQPYNEYRKKLGLTPYKSFQELTGER 477
1COX TVSGGRNINKNLLHVATSIIEHGRTLRLQSLNQYRHRFNMRPFTSFLELTGDE 279
hCOX RIGGGQNIHGALLHVATASIKHGRKMRFQSLNQYRKQFGLQPYQSFEQLTGET 475
*V523 *S530
sCOX EVAARLEKLYGHIDAMEFYPALLLEAPNKNSIFGESMVEMGAPFSLKGLMGNP 530
1COX AMAAEME 286
hCOX EMAADLAELYSDINAMEFYLGLMVEKPRQGALFGETMVEAGAPFSLKGLMGNA 528
sCOX ICSPDYWKPSTFGGKTGFDIVNTATFEKLICLNVK-KCPYVGFHVPY------ 582
hCOX ICSPEYWKPSTFGGNRGFEIVNSASLRRLVCLNLQGPCPDVAFHVPRDNQQDV 581
sCOX NVDNDYEREKGKP----------- STEL 593
hCOX VVNVTGSQGGSDGVTTTPHYVADQQSREL 610
Figure 2-3. Continued.
tunicates) and the Cephalochordata amphioxuss or lancelets). These other subphyla are basal to
the vertebrates and represent evolutionarily distinct monophyletic groups (Fig. 1-3). There is
some debate as to whether the urochordates or cephalochordates represent the base of the
chordate phylogeny, although urochordates have been generally been considered basal (Pough et
al., 2005).
Interestingly, the genomes of sea squirts (Ciona intestinalis and Ciona savignyi) predict
COX genes that are not homologous to COX-1 and COX-2. These forms (named COXa and
COXb) are basal to vertebrate COX forms and indicate an evolutionarily distinct branch in COX
evolution (Jarving et al. 2004) (Fig. 1-4). This separate branch is situated between the COX of
corals and the vertebrate COX forms (Fig. 1-4). This suggests that the COX-1 and COX-2 forms
originated sometime after the origin of the sea squirts but before the origin of the elasmobranchs
(around 550-600 mya).
Cyclooxygenase Function in Mammals
Cyclooxygenase was first described from mammals, and due to its medical importance,
its function has been intensely examined in mammals, specifically humans. Since COX
stimulates the production of prostaglandins, COX function is intrinsically linked to the roles that
prostaglandins play in everything from water balance to reproduction. Here, I review the major
functions of COX-1 and COX-2 in various mammalian physiological systems (modeled after
Vane et al., 1998), but it should be noted that mammalian COX studies are ongoing and that
COX likely plays many roles not described here.
Gastrointestinal Tract
In the stomach, prostaglandins play a cytoprotective role, guarding the stomach lining
against digestive acids and enzymes. The mechanism behind this protection is an increased
mucosal blood flow caused by the vasodilating properties of prostaglandins (specifically
and non-neuronal cells in the brain and is up-regulated in response to abnormal nerve activity,
suggesting a role in nerve transmission (Breder et al., 1995; Breder and Saper, 1996).
COX-2 in the spinal cord has been shown to modulate pain responses (nociception) in
both humans and rats due to peripheral pain stimulation (Martin et al., 2007; Gardiner et al.,
1997; Willingale et al., 1997). These results are expected since prostaglandins are generated at
the ends of sensory neurons during inflammatory pain (Ferreira 1972; Woolf et al., 1997).
However, the up-regulation of COX-2 over COX-1 in the spinal cord during inflammatory pain
suggests a role during pain processing in the central as well as the peripheral nervous system.
This up-regulation has been detected via increases in COX-2 mRNA (Gardiner et al., 1997;
Beiche et al., 1996) in the spinal cord of rats during inflammatory stimuli. Furthermore, COX-2
selective inhibition also affected the withdrawal reflex of rats when stimulated using a non-
inflammatory, electrical stimulus (Willingale et al., 1997). This supports the general role of
COX-2 in nociception in the spinal cord.
COX-2 also plays a thermoregulatory role in the brain by increasing prostaglandin
production in the hypothalamus during periods of fever. Injection of bacterial
lippopolysaccarides (LPS) into mammals causes a monophasic fever characterized by a single
rise in core body temperature. It has long been known that PGE2 levels increase in the
hypothalamus during fever and likely mediate the response (Ivanov and Romanovsky, 2004).
Early studies suggested that an up-regulation of COX-2 was the cause of increased PGE2 levels
during fever (Cao et al., 1997). This initial hypothesis has been confirmed, recently in COX-2
deficient mice which do not respond to LPS injection, in contrast to COX-1 deficient mice or
control mice (Steiner et al., 2005).
and COX-2 specific inhibitors, it was found that both types reduced granulosa cell proliferation,
but that COX-2 specific inhibitor showed a stronger effect, supporting original studies that COX-
2 is the dominant form in granulosa cell proliferation (Jin et al., 2007). However, COX-1 was
found to be expressed highly in the brain and seminal vesicle of the chicken (Reed et al., 1996).
COX-1 (but not COX-2) is also up-regulated in ovarian cancers occurring in hens, suggesting a
target for treatment (Urick and Johnson, 2006). COX inhibitors, both specific and non-specific,
were shown to decrease sperm motility in the domestic turkey, demonstrating a further role for
COX in bird reproduction (Kennedy et al., 2003).
As in mammals, COX-2 has been implicated with the detection and persistence of
peripheral inflammatory pain in chickens. The number of COX-2 containing neurons increased
significantly in laminae under inflammatory conditions, but not under control conditions, 12-24
hours after injection of Freund's adjunvant (Yamada et al., 2006). This suggests that the numbers
of COX-2 containing neurons are related to inflammatory pain as in mammals. It appears that
COX may not play a thermoregulatory role in birds as it does in mammals because indomethacin
did not cause any change in rectal temperature during heat stress in chickens (Furlan et al.,
1998). However, COX and its prostaglandins may play a role in memory retention in birds,
because inhibitors produce amnesic effects (Holscher, 1995). Finally, COX-2 is expressed in the
kidneys during development in chickens (Mathonnet et al., 2001) and at high levels in the
kidneys of adults (Reed et al., 1996).
Study Overview
As outlined above, the family of cyclooxygenase genes has an interesting and
functionally important history in the vertebrates. Drugs derived from cyclooxygenase function
are among the most important and widely used treatment methods in medicine. However, COX
function and evolution is not fully understood in the ancestral chordates, specifically the teleosts,
Figure 2-14. The evolutionary history of COX sequences excluding hagfish (hCOX) and
lamprey (1COX) sequences inferred using a Bayesian optimality criterion. The
consensus tree is shown. The posterior probabilities (10 million generations) are
shown next to the branches. Trees were sampled every 500 generations with a bumin
of 10% (401 trees). The tree is drawn to scale, with branch lengths in the same units
as those of the evolutionary distances used to infer the phylogenetic tree. The
evolutionary distances were computed using the WAG matrix-based method (Whelan
and Goldman, 2001) and are in the units of the number of amino acid substitutions
per site (scale bar = 0.1 amino acid substitutions per site). The rate variation among
sites was modeled with a gamma distribution (shape parameter estimated =
0.667978). Phylogenetic analyses were conducted using MrBayes 3.1.2.
Table 2-7. Sequences included in phylogenetic analyses
Name
COX1a Fundulus
COXlb Fundulus
COX2 Fundulus
COX1 a Myoxocephalus
COXlb Myoxocephalus
COX2 Myoxocephalus
COXa Ciona
COXb Ciona
COXc Branchiostoma
COXd Branchiostoma
hCOX Myxine
1COX Petromyzon
COXla Oryzias
COXlb Oryzias
COX2 Oryzias
COXIa Gasterosteus
COXlb Gasterosteus
COX2 Gasterosteus
COX1a Tetraodon
COXlb Tetraodon
COX2 Tetraodon
COX1 Oncorhynchus
COX2a Oncorhynchus
COX2b Oncorhynchus
COX1 Danio
COX2a Danio
COX2b Danio
COX1 Monodelphis
COXlb Monodelphis
COXlb2 Monodelphis
COX2 Monodelphis
COX1 Gallus
COX2 Gallus
COX1 Bos
COX2 Bos
COX1 Xenopus 1
COX2 Xenopus 1
COX1 Canis
COX2 Canis
COX1 Homo
COX2 Homo
COX1 Rattus
COX2 Rattus
COX1 Mus
Organism
Fundulus heteroclitus killifishh)
Fundulus heteroclitus killifishh)
Fundulus heteroclitus killifishh)
Myoxocephalus octodecimspinosus (lh. Sculpin)
Myoxocephalus octodecimspinosus (lh. Sculpin)
Myoxocephalus octodecimspinosus (lh. Sculpin)
Ciona intestinalis (sea squirt)
Ciona intestinalis (sea squirt)
Branchiostoma lancelolatum amphioxuss)
Branchiostoma lancelolatum amphioxuss)
Myxine glutinosa (Atlantic Hagfish)
Petromyzon marinus (Sea Lamprey)
Oryzias latipes (Japanese medaka)
Oryzias latipes (Japanese medaka)
Oryzias latipes (Japanese medaka)
Gasterosteus aculeatus (stickleback)
Gasterosteus aculeatus (stickleback)
Gasterosteus aculeatus (stickleback)
Tetraodon nigroviridis (Green Spotted Puffer)
Tetraodon nigroviridis (Green Spotted Puffer)
Tetraodon nigroviridis (Green Spotted Puffer)
Oncorhynchus mykiss (rainbow trout)
Oncorhynchus mykiss (rainbow trout)
Oncorhynchus mykiss (rainbow trout)
Danio rerio (zebrafish)
Danio rerio (zebrafish)
Danio rerio (zebrafish)
Monodelphis domestic (gr. S.t. opossum)
Monodelphis domestic (gr. S.t. opossum)
Monodelphis domestic (gr. S.t. opossum)
Monodelphis domestic (gr. S.t. opossum)
Gallus gallus (rd. jungle fowl)
Gallus gallus (rd. jungle fowl)
Bos taurus (cattle)
Bos taurus (cattle)
Xenopus laevis (Afr. Claw. Frog)
Xenopus laevis (Afr. Claw. Frog)
Canis lupusfamiliaris (dog)
Canis lupusfamiliaris (dog)
Homo sapiens (human)
Homo sapiens (human)
Rattus norvegicus (norway rat)
Rattus norvegicus (norway rat)
Mus musculus (house mouse)
Size (AA)
452
598
610
622
600
605
653
600
177
177
610
286
605
606
609
598
598
620
1023
589
456
624
607
609
597
601
606
625
627
729
608
649
603
600
604
587
604
633
604
599
604
602
604
602
versus 64% for COX-la and 68% versus 59% for COX-lb, respectively). Similarly, COX-2
from the sculpin was more similar to Mus COX-2 than Mus COX-1 (70% versus 60%) and
COX-la and COX-lb were more similar toMus COX-1 than Mus COX-2 (53% versus 47% for
COX-la and 67% versus 57% for COX-lb, respectively).
An alignment of COX forms from the killifish (Figure 2-1) shows that all the important
amino acid residues for cyclooxygenase function are conserved in all forms. However, because
COX-la is incomplete it is unknown whether the N-terminal characteristics are conserved in this
form. These include an active site tyrosine (Tyr-385, using ovine COX-1 numbering), haem-
binding histidines (His-207 and His-338), and the aspirin acetylation site (Ser-530) (Ishikawa et
al., 2007; Ishikawa and Herschman, 2007; Yang et al., 2002; Kulmacz et al., 2003). In addition
the COX-2 form contains a C-terminal amino acid insertion and the COX-1 forms contain an N-
terminal amino acid insertion. This is consistent with the COX-1 and COX-2 forms of other
vertebrates (Herschman et al., 2003; Kulmacz et al., 2003), although killifish COX-2 seems to
have a shorter than average insertion. In mammals, the amino acid residues thought to give
COX-1 and COX-2 their differing affinities for substrates are positions 513 (His in COX-1 and
Arg in Cox-2) and 523 (Ile in COX-1 and Val in COX-2) (Guo et al., 1996). However, as
previously noted in other teleosts (Grosser et al., 2002; Choe et al., 2006; Ishikawa et al., 2007)
these differences appear to be absent in teleosts, with Arg and Val present in positions 513 and
523, respectively, in both teleost COX-1 and COX-2. This is also the case in the killifish with
Arg in position 513 and Val in position 523 in all COX forms reported here.
As in the killifish, alignment of the three COX forms sequenced from the sculpin (Figure
2-2) shows conserved amino acid residues characteristic of COX function in all three forms. The
characteristic terminal amino acid insertions found in mammalian COX forms are also present in
COX-1 and COX-2, and investigating the history of COX duplication and loss in the teleosts.
Finally, the physiological function of COX in the earliest chordates has only been preliminarily
investigated.
Completion of novel sequences
Several of the novel sequences documented here could not be completed despite multiple
efforts using many primer sets and PCR techniques. Notably, only 532 bp were obtained of the
sequences from amphioxus (COXc and COXd). Compared to the complete open reading frame
of COX transcripts in other chordates (ranging from about 1700-2500 bp), this indicates that
about 75% of these sequences remain unknown. Completing the rest of these sequences or
thoroughly searching the recently complete genome of Branchiostomafloridae may yield
putative proteins that provide novel groupings in phylogenetic analyses. Based on the available
sequence data, however, COXc and COXd are well-supported as being near the base of the
chordate COX lineage, sister to the vertebrate COX-1 and COX-2 clade. The 1COX sequence
obtained from the lamprey is also only about 40% complete. Completing the open-reading frame
of this putative protein or searching genomic databases could possibly give a more definite view
of its identity as this sequence groups with different clades depending on the phylogenetic
analysis used. In actuality, the short size of 1COX compared to the other COX sequences used in
the phylogenetic analyses may have contributed to its tendency to appear in different clades in
different analyses. Finally, the COX-la sequence obtained from the sculpin may warrant further
investigation and reaffirmation via additional sequencing since many residues critical to COX
function are missing.
Additional Phylogenetic Analysis
Unfortunately, the evolutionary relationship of the COX proteins in the chordates was not
able to be unequivocally resolved, despite the use of three different phylogenetic approaches.
k) 1)
--- ______________ ^ ^-- --- -- -'llli
m) n)
o) p)
Figure 2-15. Continued
Figure 2-15. Continued.
Table 2-4. Lamprey COX specific primers used in PCR
Name
1COX Fl
1COX R1
1COX R2
1COX R3
1COX 3' Fl
1COX 3' F2
1COX 3' F3
1COX 5' R1
1COX 5' R2
Orientation
Sense
Antisense
Antisense
Antisense
Sense
Sense
Sense
Antisense
Antisense
Nucleotide sequence (5' to 3')
GAG GAA CGC TTG AGA GAC AAC A
TCT TGT TGA TGT TTC GAC CGC
CCA CCG CCA ATC TGG C
AAT CAG GAA GCT CTC AGG CAT C
GAG TTC CTC TTC AAC CCC GG
TCG AAA CAT CAA CAA GAA CCT CC
TTC ACC TCC TTC CTG GAG CTC
TGC TGC ACG TAC TCC TCG ATC
GCA GCT GGT GTT GTC TCT CAA
and kidneys (Harris and Breyer, 2001) of mammals. This has led to the abandonment of COX-2
selective inhibitors (e.g., celecoxib and valdecoxib) which were thought to treat inflammatory
pain without the negative side effects associated with non-selective COX inhibition. Although
structurally and biochemically similar, COX-1 and COX-2 vary in expression in the vertebrates
and participate in different functions. For example, COX-2 but not COX-1 plays a role in
granulosa cell proliferation in chicken follicles (Li et al., 1996; Jin et al., 2007). Although the
amino acid sequences of COX-1 and COX-2 share about 63% similarity, the presence of valine
in COX-2 at position 523 instead of isoleucine is thought to be responsible for their differences
in substrate selectivity and sensitivity to specific inhibitors (Otto and Smith, 1995). Inhibitory
studies in mammals have suggested a COX-3 form also exists, but it has not been characterized
(Kis et al., 2006; Censarek et al., 2006).
Analyses of genomic sequence and targeted cloning efforts have demonstrated that, like
mammals, other vertebrates have COX-1 and COX-2 forms (Jarving et al., 2004). However,
some variation exists in the COX-1 and COX-2 dichotomy, notably in the more evolutionarily
ancestral chordates. Ishikawa and colleagues (Ishikawa et al., 2007; Ishikawa and Herschman,
2007) have recently shown that teleosts posses three forms of COX. The zebrafish (Danio rerio)
and the rainbow trout (Onchorhynchus mykiss) both posses two COX-2 forms (named COX-2a
and COX-2b) and one COX-1 form, whereas the stickleback (Gasterosteus aculeatus), green
spotted puffer (Tetraodon nigroviridis), pufferfish (Takifugu rubripes), and Japanese medaka
(Oryzias latipes) posses two COX-1 forms (named COX-la and COX-lb) and one COX-2 form.
They also showed that COX-2a and COX-2b are functionally different in the zebrafish and
rainbow trout. Also, a COX form (named sCOX) has been cloned from the spiny dogfish
(Squalus acanthias) and groups with COX-1 in phylogenetic analyses, but shares strong
function located near the C and N terminus of the sequence. These include disruption of
disulfide bonds near the N terminus due to cystine replacements, little conserved homology in
the membrane binding domain, and alternate residues for arginie-513 leucinee), valine-523
leucinee), and serine-530 phenylalaninee). This is not the case in the other acanthopterygiians,
where all COX forms have these conserved regions. This may indicate that in the sculpin COX-
la has partially lost function, and COX-lb or COX-2 may therefore have additional
responsibilities.
Phylogenetic Analyses
As has been seen when tracing the evolutionary histories of other gene families in
evolutionary deep time (Abbasi and Grzeschik, 2007; Goudet et al., 2007), the pedigree of the
COX family is littered with duplications, losses, and uncertainty. Despite utilizing a multi-
faceted approach to depict COX evolution in the chordates, the results indicate that COX has a
complicated and under-examined history of evolution in the chordates. This is not surprising,
considering the complex functions COX has been shown to have in the chordates. Because of
these complex functions, duplications, partitioning of roles, differential expression, and multiple
forms with multiple functions are to be expected.
In all phylogenetic analyses, the more recent history of COX evolution is more clear and
consistent than the relationships of the "ancestral" taxa. The teleosts consistently form
monophyletic COX clades nested within the gnathostomes, as expected. The gnathostomes
consistently form two COX sister clades in phylogenetic analyses: COX-1 and COX-2. This is
expected based on previous phylogenetic analyses in the vertebrates (Gu, 2001; Gu, 2006). The
COX-1 clade is consistently rooted with the sequence from the dogfish (sCOX), although
support values are not robust for this relationship when hagfish and lamprey sequences are
included. In distance analysis excluding hagfish and lamprey sequences, sCOX groups outside
To the teachers who inspired, directed, and challenged me. My high school marine biology
teacher, Larry Joye, first introduced me to the fascinating world of life in the water and since
then my endeavors have never strayed far from it. Dr. David Evans enthusiastically welcomed
me into his lab as an undergraduate, although I had no former research experience. What started
as a part-time project soon evolved into an academic career as I was given the opportunity to
conduct original research. During this time, Dr. Keith Choe was an invaluable asset and
inspiration. Growing from this original partnership, I continued working with David on my
master's research and he has always supported my academic and research endeavors; I cannot
envision a better advisor.
COX in fishes needs to be examined in greater detail using other species and diverse techniques.
This may be highly feasible because COX sequences exist for a wide range of teleosts.
Reproduction
As in mammals, COX has been shown to play a role in reproduction in fishes. By using
the non-specific COX inhibitor indomethacin, it was shown in the Atlantic croaker
(Micropogonias undulatus) that COX pathways may play a role in the maturation of ovarian
follicles and ovulation through prostaglandin formation, although other proteins may play a more
dominant role (Patifio et al., 2003). Results from the European sea bass also indicate a similar
role for prostaglandins in ovulation, with indomethacin inhibiting follicle maturation (Sorbera et
al., 2001). However, it has been shown in the brook trout that indomethacin does not block
ovulation (Goetz et al., 1989), although it does in other fish species (Goetz et al., 1991). This
apparent loss of function in the brook trout may be explained by changes in the levels of COX-1
and COX-2 during ovulation. It was shown that COX-1 levels remained constant and high
during ovulation but that COX-2 levels did not increase prior to ovulation as they do in mammals
(Roberts et al., 2000). This is supported by data from the zebrafish which show high levels of
COX-1 but not COX-2 in the ovaries (Grosser et al., 2002). This suggests that COX-2 function
in the reproduction of the brook trout (and perhaps other fishes) is different than in other
vertebrates. In the Japanese medaka, it was shown that low chronic levels (perhaps comparable
to those found in waste water) (Metcalfe et al., 2003) of the non-specific COX inhibitor
ibuprofen causes altered reproduction by decreasing the number of spawning events but
increasing the number of eggs per spawning event (Flippin et al., 2007). This conforms to
delayed pregnancies associated with NSAID use in mammals. These results indicate that COX
likely plays a role in fish reproduction, although it may vary among species.
Jayadeep, A., Reddanna, P., Sailesh, S., Das, U. N., Ramesh, G., Kumar, K. V., and
Menon, V. P. (1995). Prostaglandin metabolism during growth and differentiation of the
regenerating vertebrate appendage. Prostaglandins Leukot. Essent. Fatty Acids. 52, 235-
239.
Jin, Y., Zhang, C., Lin, X., and Zeng, W. (2006). Prostaglandin involvement in follicle-
stimulating hormone-induced proliferation of granulosa cells from chicken
prehierarchical follicles. Prostaglandins Other Lipid. Mediat. 81, 45-54.
Jin, Y., Zhang, C., Zeng, W., Taya, K., and Tan, T. Q. (2007). Interactive actions of
prostaglandin and epidermal growth factor to enhance proliferation of granulosa cells
from chicken prehierarchical follicles. Prostaglandins Other LipidMediat. 83, 285-294.
Johnson, R. F., Mitchell, C. M., Giles, W. B., Bisits, A., and Zakar, T. (2006). Mechanisms
regulating prostaglandin H2 synthase-2 mRNA level in the amnion and chorion during
pregnancy. J. Endocrinol. 188, 603-610.
Jones, D. T., Taylor, W. R., and Thornton, J. M. (1994). A mutation data matrix for
transmembrane proteins. FEBS Lett. 339, 269-275.
Kaji, D. M., Chase, H. S. Jr, Eng, J. P., and Diaz, J. (1996). Prostaglandin E2 inhibits Na-K-
2C1 cotransport in medullary thick ascending limb cells. Am. J. Physiol. 271, C354-C361.
Kargman, S., Charleson, S., Cartwright, M., Frank, J., Riendeau, D., Mancini, J., Evans, J.
and O'Neill, G. (1996). Characterization of Prostaglandin G/H Synthase 1 and 2 in rat,
dog, monkey, and human gastrointestinal tracts. Gastroenterology 111, 445-454.
Kasahara, M. (2007). The 2R hypothesis: an update. Curr. Opin. Immunol. 19, 547-552.
Kay-Mugford, P. A., Benn, S. J., LaMarre, J., and Conlon, P. D. (2000). Cyclooxygenase
expression in canine platelets and Madin-Darby canine kidney cells. Am. J. Vet. Res. 61,
1512-1516.
Kennedy, J. H., Korn, N., and Thurston, R. J. (2003). Prostaglandin levels in seminal plasma
and sperm extracts of the domestic turkey, and the effects of cyclooxygenase inhibitors
on sperm mobility. Reprod. Biol. Endocrinol. 1, 74.
Kis, B., Snipes, J. A., Gaspar, T., Lenzser, G., Tulbert, C. D. and Busija, D. W. (2006).
Cloning of cyclooxygenase-lb (putative COX-3) in mouse. Inflamm. Res. 55, 274-278.
Klein, S. L., Strausberg, R. L., Wagner, L., Pontius, J., Clifton, S. W. and Richardson, P.
(2002). Genetic and genomic tools for Xenopus research: The NIH Xenopus initiative.
Dev. Dyn. 225, 384-391.
Kulmacz, R. J., van der Donk, W. A., and Tsai, A. L. (2003). Comparison of the properties of
prostaglandin H synthase-1 and -2. Prog. LipidRes. 42, 377-404.
Hebert, S. C. and Andreoli, T. E. (1984). Control of NaCl transport in the thick ascending
limb. Am. J Physiol. 246, F745-F756.
Herschman, H. R., Talley, J. J., and DuBois, R. (2003). Cyclooxygenase 2 (COX-2) as
a target for therapy and noninvasive imaging. Mol. Imaging Biol. 5, 286-303.
Hirano, T., Yasukawa, K., Harada, H., Taga, T., Watanabe, Y., Matsuda, T.,
Kashiwamura, S., Nakajima, K., Koyama, K., Iwamatsu, A., Tsunasawa, S.,
Sakiyama, F., Matsui, H., Takahara, H., Taniguchi, T., and Kishimoto, T. (1986).
Complementary DNA for a novel human interleukin (BSF-2) that induces B lymphocytes
to produce immunoglobulin. Nature 324, 73-76.
Holland, J. W., Pottinger, T. G. and Secombes, C. J. (2002). Recombinant interleukin-1 beta
activates the hypothalamic-pituitary-interrenal axis in rainbow trout, Oncorhynchus
mykiss. J. Endocrinol. 175, 261-267.
Hilscher, C. (1995). Prostaglandins play a role in memory consolidation in the chick. Eur. J
Pharmacol. 294, 253-259.
Hopkins, S. J. (2007). Central nervous system recognition of peripheral inflammation: a neural,
hormonal collaboration. Acta. Biomed. 78 Suppl. 1, 231-247.
Huang, L., Zhang, K. L., Li, H., Chen, X. Y., Kong, Q. Y., Sun, Y., Gao, X., Guan, H. W.,
and Liu, J. (2006). Infrequent COX-2 expression due to promoter hypermethylation in
gastric cancers in Dalian, China. Hum. Pathol. 37, 1557-1567.
Hughes, A. L. (1999). Phylogenies of developmentally important proteins do not support the
hypothesis of two rounds of genome duplication early in vertebrate history. J. Mol. Evol.
48, 565-576.
Ishikawa, T. O., Griffin, K. J., Banerjee, U. and Herschman, H. R. (2007). The zebrafish
genome contains two inducible, functional cyclooxygenase-2 genes. Biochem. Biophys.
Res. Commun. 352, 181-187.
Ishikawa, T. 0. and Herschman, H. R. (2007). Two inducible, functional cyclooxygenase-2
genes are present in the rainbow trout genome. J. Cell. Biochem. 102, 1486-1492.
Ivanov, A. I. and Romanovsky, A. A. (2004). Prostaglandin E2 as a mediator of fever:
synthesis and catabolism. Front. Biosci. 9, 1977-1993.
Jarving, R., Jarving, I., Kurg, R., Brash, A. R. and Samel, N. (2004). On the evolutionary
origin of cyclooxygenase (COX) isozymes: characterization of marine invertebrate COX
genes points to independent duplication events in vertebrate and invertebrate lineages. J.
Biol. Chem. 279, 13624-13633.
COX-1 or COX-2 form characterizes the species mentioned (Ishikawa and Herschman 2007)
(Fig. 1-2). This is supported by the apparent genome duplication events in teleosts (Crollius and
Wessenbach, 2005).
Other Vertebrates
Although functional studies on prostaglandin roles and cyclooxygenase are not
uncommon in other vertebrates, cyclooxygenase sequences from non-mammalian and non-
teleost vertebrates are lacking. Cyclooxygenase forms have been predicted from the chicken
genome (Gallus gallus), suggesting that birds have both COX-1 and COX-2 forms (Urick and
Johnson, 2006). Both forms have also been found in the African clawed frog (Xenopus laevis)
and COX-2 has been found in the western clawed frog (Xenopus tropicalis), suggesting that both
forms occur in amphibians (Klein et al., 2002). Functional studies suggest COX exists in reptiles
as well (Seebacher and Franklin, 2003) and the recently completed Anolis carolinensis genome
contains both COX forms.
Taken together, the presence of COX-1 and COX-2 forms in the mammals, teleosts,
birds, and amphibians suggest that both forms originated in the early vertebrates, possibly from
the duplication of an ancestral COX gene. The presence of sCOX in the dogfish indicates that
COX forms in the early vertebrates may not have diverged (in sequence or possibly in function)
to the extent that allows them to be labeled as COX-1 or COX-2, although it is predicted that two
COX forms exist in all vertebrates, since sCOX groups with COX-1 and is not ancestral to the
COX-1/COX-2 split (Fig. 1-2, unpublished phylogeny generated using the online program
PHYML) (Guindon and Gascuel, 2003; Guindon et al., 2005).
Other Chordates
Although the subphylum Vertebrata includes the majority of species in the phylum
Chordata, two other subphyla are included in the chordates: the Urochordata (sea squirts or
A)
COX-lb
18S
GOB HS K
1111111
Tkm(h)
Figure 3-1. Euryhaline killifish (Fundulus heteroclitus) COX-lb mRNA expression following
procedures outlined in Choe et al. (2006). A) COX-lb mRNA expression in various
tissues in the killifish. Expression was constant across tissues for the ribosomal
protein 18S. COX-lb expression was most abundant in the gills (labeled as G),
followed by the heart (H), kidney (K), opercular epithelium (0), stomach (S),
intestine (I), and brain (B). B) In situ hybridization of COX-lb in the gills of the
killifish. C) Expression of COX-lb mRNA in the gills of the killifish following
salinity transfers. Expression is normalized to the ribosomal gene L8 and is shown
relative to seawater (SW) (FW = fresh water).
COX1 Mus LKPYTSFQELTGEKEMAAEL 493
COX1 G.gal LKPYTSFQELTGEEDKAAEL 536
COX1 Danio LKPYTSFAELTGEQEMSKEL 491
1COX MRPFTSFLELTGDEAMAAEM 285
COX2 Mus LKPYTSFEELTGEKEMAAEL 477
COX2 G.gal LKPFKSFEELTGEKEMAAEL 477
COX2 F.het MKPYTSFEDLTGEKEMAAIL 482
Figure 2-4. Continued.
|
