Molecular analysis of F1F0 adenosine triphosphate synthase and renal H+, K+ -adenosine triphosphatases


Material Information

Molecular analysis of F1F0 adenosine triphosphate synthase and renal H+, K+ -adenosine triphosphatases
Alternate title:
Molecular analysis of F1F0 adenosine triphosphate synthase and renal hydrogen, potassium-adenosine triphosphatases
Physical Description:
xiii, 195 leaves : ill. ; 29 cm.
Otto, Tamara Caviston, 1971-
Publication Date:


Subjects / Keywords:
Proton-Translocating ATPases   ( mesh )
Adenosinetriphosphatase   ( mesh )
Amino Acid Substitution   ( mesh )
Uncoupling Agents   ( mesh )
Kidney -- enzymology   ( mesh )
Department of Biochemistry and Molecular Biology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Medicine -- Department of Biochemistry and Molecular Biology -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 2001.
Bibliography: leaves 171-194.
Statement of Responsibility:
by Tamara Caviston Otto.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002684551
oclc - 53948324
notis - ANF1821
System ID:

Full Text







This dissertation is dedicated to my family and to my husband for their

continuous love and support.


I would like to thank my advisor, Dr. Brian Cain. Dr. Cain has been a continuous

source of encouragement and support throughout my career as a graduate student. He has

provided me with the training and experience necessary for me to become a successful

scientist. I would also like to express my gratitude to the other members of my

committee--Dr. Michael Kilberg, Dr. Mary Jo Koroly, Dr. Daniel Purich, and Dr. Charles

Wingo--for their help and support during my time spent at the University of Florida. I

want to thank the members of my lab--Grady Campbell, James Gardner, Paul Sorgen,

Debbie Zies, Tammy Bohannon, Michelle Gumz, and Andy Hardy--as well as Sue Lee

Kang and Chien Chen of the Yang lab for the stimulating discussions we have shared and

for providing a great place to work.

I want to especially thank my family for their love and support throughout these

difficult years. They worked long hours to provide me with an excellent education,

which has allowed me to achieve my goals. I thank them dearly.

Finally, I wish to thank my dear husband, Scott. Scott has been my best friend

and my greatest supporter during the demands of graduate school. He has always been

there giving me the strength and courage to continue. I could not have accomplished this

great achievement without his support.


ACKNOWLEDGMENTS................................................................... iii

LIST OF TABLES.......................................................................... vi

LIST OF FIGURES......................................................................... vii

ABBREVIATIONS......................................................................... x

ABSTRACT................................................................................. xii


1 BACKGROUND AND SIGNIFICANCE............................................ 1

Ion Motive ATPases..................................................................... 1
E. coli FiFo ATP Synthase............................................................. 6
unc Operon Genetics.................................................................... 8
FiF0 ATP Synthase Mechanisms....................................................... 10
b Subunit................................................................................ 20
H+,K+-ATPase.............................................................................. 27
Gastric H+,K+-ATPase.......................................................................... 33
Renal H+,K+-ATPase.................................................................... 43
H+ Versus Na,K -ATPase........................................................................ 54

2 EXPERIMENTAL PROCEDURES................................................... 56

Recombinant DNA Techniques........................................................ 56
Plasmid Constructions................................................................... 56
FiFo ATP Synthase Methods........................................................... 73
3' R A C E ...................................... ...... ............... .......... .... ......... 77
5' R A C E ...................................... ............ ...... .......................... 78
Tissue Culture............................................................................ 78
H+,K+-ATPase Biochemical Assays................................................... 80
Northern Analysis........................................................................ 83
Immunoblot Analysis................................................................... 83
Immunofluorescence Microscopy..................................................... 85

FROM ESCHERICHIA COLI............................................................ 88

Introduction ......................... ...... ............... ............................... 88
Results ...................................... ..... ......... ..................... .......... 90
Discussion ............................. ........... ............ ........................... 105

4 AMPLIFICATION OF RABBIT NaKp3i cDNA..................................... 109

Introduction................. .. ............... ............... .... 109
Results....................................................................................... 111
Discussion ........................... ... ....... ......... ...... ............. ......... ..... 118

5 EXPRESSION OF RABBIT RENAL H+,K+-ATPases............................. 126

Introduction............................................................................... 126
R esults......................................... ........ .............. ...... ............... 128
Discussion...................................................... ......................... 151

6 CONCLUSIONS AND FUTURE STUDIES........................................ 158

E. coli F Fo ATP Synthase...............................................................158
Renal H+,K+-ATPases................................................................... 164

LIST OF REFERENCES................................................................... 171

BIOGRAPHICAL SKETCH.............................................................. 195


Table 2-1. Primers for the PCR mutagenesis of bArg-36 in the uncF(b) gene ....... 57

Table 2-2. RT-PCR and RACE primers................................................. 61

Table 2-3. Mutagenesis primers.......................................................... 65

Table 3-1. E. coli strains and plasmids.................................................. 95

Table 3-2. In vivo and in vitro activities of bArg-36 mutants........................... 96

Table 3-3. Alterations in F, activity associated with bArg-36 substitutions.......... 104

Table 4-1. Percent amino acid similarity and identity of rabbit NaKp I with
other known NaKP subunit isoforms................................................. 122


Figure 1-1. Subunit organization of E. coli FiF0 ATP synthase...................... 7

Figure 1-2. Transmembrane organization of H+,K-ATPase......................... 29

Figure 1-3. Enzymatic mechanism of H+,K+-ATPase................................. 32

Figure 1-4. Variation in the N-termini of the HKoC2 subunits from rabbit,
hum an, and rat........................................................................... 49

Figure 1-5. Distance analysis of HKa and NaKa subunits........................... 50

Figure 2-1. Rabbit HKca2a cDNA construction in pREP4............................ 60

Figure 2-2. Rabbit HKat2c cDNA construction in pREP4............................ 62

Figure 2-3. Rabbit HKp cDNA construction in pREP8 ............................... 64

Figure 2-4. Rabbit NaKPi cDNA construction in pREP8............................ 66

Figure 2-5. Rabbit HKa2c cDNA construction in pBudCE4......................... 69

Figure 2-6. Rabbit HKa2a cDNA construction in pBudCE4......................... 70

Figure 2-7. Rabbit NaK[iI cDNA construction in pBudCE4......................... 71

Figure 2-8. Plasmids pTLC47 and pTLC37 expression constructs.................. 72

Figure 3-1. Amino acid sequence alignment of bacterial b and b' subunits........ 91

Figure 3-2. Oligonucleotides for PCR mutagenesis of the uncF(b) gene........... 92

Figure 3-3. MunI restriction endonuclease analysis of uncF(b) gene mutations
in pKAM14................ .. ..... .......... ............... 94

Figure 3-4. Complementation of uncF(b) gene mutants in KM2(Ab)................ 97

Figure 3-5. Immunoblot analysis of uncF(b) gene mutants........................... 99

Figure 3-6. ATP-driven energization of membrane vesicles prepared
from uncF(b) gene mutants ............................................................. 101

Figure 3-7. Proton permeabililty of stripped membrane vesicles prepared
from uncF(b) gene mutants............................................................. 103

Figure 4-1. Partial nucleotide sequence alignment of known NaKP3 subunits...... 112

Figure 4-2. Agarose gel analysis of rabbit NaKp31 cDNA PCR amplifications...... 114

Figure 4-3. Nucleotide sequence alignment of rabbit NaK1 i cDNA................. 115

Figure 4-4. Agarose gel analysis of rabbit NaK3 i 5' RACE reactions............. 117

Figure 4-5. GenBank accession record for rabbit NaK3 i cDNA sequence ...........119

Figure 4-6. Western analysis of HEK293c18 cells expressing NaKpi from
an episom al vector....................................................................... 120

Figure 4-7. Kyte-Doolitle hydropathy plot of rabbit NaKpI3 amino acid sequence.. 123

Figure 4-8. Amino acid sequence analysis of rabbit NaKp3i protein................ 124

Figure 5-1. Episomal expression constructs............................................ 130

Figure 5-2. Northern analysis of HEK293cl8 cells expressing HKa2a, HKa2c,
and NaKpi1 from episomal vectors ......................................................131

Figure 5-3. Western analysis of HEK293cl8 cells expressing NaKpi from
an episom al vector....................................................................... 133

Figure 5-4. Control plasmid pBudCE4/lacZ/CAT expression vector............... 135

Figure 5-5. Western analysis of HEK293 cells stably expressing control
plasmid pBudCE4/lacZ/CAT.......................................................... 137

Figure 5-6. Stable expression constructs................................................ 138

Figure 5-7. Western analysis of COS-1 cells transiently expressing HKa2a-V5,
HKa2c-V5, and NaKpi proteins....................................................... 142

Figure 5-8. Differential expression of rabbit NaKp31 protein in stable HEK293
clones .................................... ........... ......................... ..... ...... 143

Figure 5-9. Northern analysis of HEK293 cells stably expressing HKct2a
or HKa2c transcripts..................................................................... 145

Figure 5-10. Agarose gel analysis of HKo2 RT-PCR reactions from stable
H EK 293 cells...............................................................................146

Figure 5-11. Localization of HKX2a-V5 and HKcE2c-V5 fusion proteins in
C O S-1 cells............................. ............ ...... ........ ............. .. 149

Figure 5-12. Western analysis of plasma membranes from COS-1 transient
transfections.............................. ..... ......................................... 150

Figure 5-13. Western analysis of COS-1 cells transiently expressing rabbit
NaKP33-myc fusion protein.............................................................. 152


ACMA, 9-amino-6-chloro-2-methoxyacridine

ADP, adenosine-5'-diphosphate

Ap, ampicillin

ATP, adenosine-5'-triphosphate -

ATPase, adenosine triphosphatase

bArg-36-ne, substitution of isoleucine for arginine at amino acid 36 in the b subunit

bp, base pair

CAT, chloramphenicol acetyltransferase

CCCP, carbonyl cyanide-m-chlorophenylhydrazone

CCD, cortical collecting duct

CD, collecting duct

Cm, chloramphenicol

CMV, cytomegalovirus

DCCD, dicyclohexylcarbodiimide

D-MEM, Dulbecco's modified Eagle medium

DNA, deoxyribonucleic acid

EBNA-1, Epstein Barr virus nuclear antigen-1

EBV, Epstein Barr virus

E. coli, Escherichia coli

FBS, fetal bovine serum

GAP3DH, glyceraldehyde-3-phosphate dehydrogenase

IMCD, inner medullary collecting duct

IPTG, isopropyl- 1 -thio-p-D-galactoside

kbp, kilobase pair

kD, kilodalton

NADH, p-nicotinamide adenine dinucleotide, reduced form

N. crassa, Neurospora crassa

NMR, nuclear magnetic resonance

OMCD, outer medullary collecting duct

PCR, polymerase chain reaction

P. modestum, Propionigenium modestum

PMSF, phenylmethylsulfonyl fluoride

RT-PCR, reverse-transcriptase PCR

S. cerevisiae, Saccharomyces cerevisiae

SDS, sodium dodecyl sulfate

SR, sarcoplasmic reticulum

UTR, untranslated region

v/v, volume/volume

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Tamara Caviston Otto

May 2001

Chairman: Brian D. Cain
Major Department: Biochemistry and Molecular Biology

The FiFo ATP synthase is responsible for the production of ATP in almost every

cell. The enzyme utilizes the electrochemical gradient of protons to drive ATP synthesis.

It is composed of two stalks that attach the FI sector to the Fo sector. The peripheral stalk

is composed of a dimer of b subunits. A specific b subunit arginine, bArg-36 in

Escherichia coli, displays evolutionary conservation among bacterial F iF ATP

synthases. Site-directed mutagenesis was used to generate a collection of mutations

affecting bArg-36. The phenotype differed depending upon the substitution, and the bArg-

36--Glu and bArg-36-+Ile substitutions virtually abolished enzyme function. Although total

amounts of the enzyme present in membranes prepared from mutant strains were

reduced, the primary effect of the bArg-36 substitutions was on activities of the intact

enzyme complexes. The most interesting outcome was that the bArg-36-Glu substitution

results in the uncoupling of a functional F0 from FI ATP hydrolysis activity. This is the

first uncoupling mutant identified in the b subunit.

The renal H+,K+-ATPases are membrane proteins located on apical membranes of

specialized epithelial cells in the collecting duct that are largely responsible for potassium

conservation in the kidney. The enzymes utilize ATP hydrolysis to transport protons

outward in an electroneutral exchange for the inward transport of potassium. There are

three H+,K+-ATPases in the kidney. Each H+,K+-ATPase is composed of an a subunit

and a P subunit. Two of the rabbit renal HKa subunits, HKa2a and HKO2c, are very

closely related and are the products of one gene. The HKa2c subunit has a sixty-one

amino acid extension at its amino-terminus. To study differences between HKa2a and

HKa2c, a mammalian transient expression system was established. The rabbit NaKPi

subunit was amplified from rabbit medulla using 5' and 3' rapid amplification of cDNA

ends to use as the P subunit in the expression system. Later, the NaK133 subunit was also

studied. Both HKo2 subunits were readily identified by Western analysis. The proteins

were also seen by immunocytochemistry. However, there was no apparent association

with either p subunit. This is in agreement with recent observations reported by other



Ion Motive ATPases

Ion motive ATPases are cellular enzymes that utilize the free energy stored in

adenosine-5'-triphosphate (ATP) to generate electrochemical ion gradients. These

gradients maintain intracellular ionic or pH balance and stimulate an assortment of

processes including absorption, secretion, and transmembrane signaling. Some ATPases

are responsible for coupling energy to catalyze processes such as muscle relaxation and

receptor recycling. There are three distinct classes of ion motive ATPases. These classes

are designated F-type, V-type, and P-type ATPases (177). The work described in this

dissertation represents studies conducted on an F-type ATPase and a P-type ATPase.

This section will provide a brief overview of the general characteristics of the families of

ion motive ATPases. The rest of the chapter will cover in detail the FiFo ATP synthase

and the H+,K+-ATPase.

Properties of F-type ATPases

F-type ATPases are a family of very large multimeric membrane proteins that

contain eight or more different subunits with a complex stoichiometry. The enzymes are

responsible for synthesizing the majority of ATP in a cell and are more correctly referred

to as FIFo ATP synthases. FiFo ATP synthases are located in the inner mitochondrial

membrane, the chloroplast thylakoid membrane, and the bacterial cytoplasmic

membrane. Although they vary in oligomeric complexity (125, 251), each of the FiFo


ATP synthases from mitochondria, chloroplasts, and bacteria has homologous subunits

and share a common molecular architecture. The FiFo ATP synthases are composed of

two sectors termed Fi and Fo. They are routinely separated in vitro to facilitate the study

of the enzyme. F, is a globular head attached to Fo by an asymmetric central stalk and a

peripheral stalk (1, 256). The soluble Fi sector houses the catalytic subunits for ATP

synthesis, and the membrane-bound Fo sector is responsible for proton translocation.

The chemiosmotic hypothesis, first postulated by Peter Mitchell in the early

1960's, describes the general mechanism of ATP production by the F F0 ATP synthase

(158). The hypothesis states that an electrochemical gradient of protons (ApH) is

generated during photosynthesis or oxidative phosphorylation. The gradient leads to the

energetically favorable movement of protons through FiFo ATP synthase to form ATP.

We now know that energy captured from the passage of protons down the

electrochemical gradient through the membrane-bound Fo sector of the enzyme causes

the rotation of the central stalk within the Fi sector. Rotation of the central stalk drives

the three nucleotide-binding sites in FI through the conformational states to produce ATP

(26, 169).

In mitochondria and in chloroplasts, the physiological role of the FIFo ATP

synthase is to produce ATP. The enzymes from facultative bacteria, however, have two

physiological roles that vary depending on the environmental conditions. In an aerobic

environment where AutH is generated through oxidative phosphorylation, the enzyme

utilizes the gradient and synthesizes ATP. Under anaerobic conditions where AgH is

low, the physiological role of the bacterial FiFo ATP synthase is to conduct ATP-driven

proton translocation. The enzyme functions in reverse and works as an ATPase,

hydrolyzing cytoplasmic ATP to translocate protons across the membrane generating the

proton motive force (AtH+ ). The AipH is used to power flagellar rotation and proton

symporters for nutrient uptake of sugars or amino acids. The ATPase function is

inhibited in the mitochondrial and chloroplast enzymes to avoid wasteful ATP hydrolysis

in vivo. However, purified enzymes from these sources can act as ATPases in vitro.

Properties of V-type ATPases

V-type ATPases are very large multimeric membrane protein complexes that have

thirteen or more subunits with a complex stoichiometry. The enzymes catalyze ATP-

dependent proton translocation in eukaryotic cells and are often called H+-ATPases (70,

223). H-ATPases acidify intracellular compartments, including clathrin-coated vesicles,

endosomes, and lysosomes. Acidification of intracellular compartments is responsible

for a variety of cellular processes, such as receptor-mediated endocytosis, membrane

trafficking, and molecular processing. The enzymes have also been identified in the

plasma membranes of specific cells where they are responsible for activities such as urine

acidification (128), bone resorption (23), and cytoplasmic pH maintenance (77).

The H -ATPase has a similar structural organization to the FiFo ATP synthase,

and both enzymes share segments of conserved primary sequence homology in specific

subunits. Like the F-type enzymes, the H -ATPase is composed of two sectors termed Vi

and V0. The soluble V1 sector is responsible for ATP hydrolysis, and the membrane-

bound Vo sector conducts proton translocation. Vi and Vo are connected by both a

central stalk and a peripheral stalk with additional projections at the base of the central

stalk (24, 59). Like the FIF0 ATP synthase, the two sectors of the H+-ATPase can be

separated in vitro. However, unlike the FIF0 ATP synthase, the detached Vi sector has

very little ATPase activity, and the isolated Vo sector does not seem to form an open

proton pore (86, 267, 270). Although the method by which the H+-ATPase conducts

ATP-dependent proton translocation is not known, it is believed that the mechanism is

similar to the rotation that occurs in the FIFo ATP synthase.

Kane (118) demonstrated that fully assembled yeast H-ATPase complexes

disassembled into inactive Vi and Vo sectors in vivo in response to changes in carbon

source. This disassembly was completely reversible. The process may be a common

feature of all H+-ATPases (86, 227). Indeed, many cells contain pools of free VI and Vo

sectors along with fully assembled H+-ATPase complexes (56, 167, 178, 227, 239).

Disassembly of the enzyme probably acts to conserve cytosolic ATP during glucose

deprivation. Therefore, the disassembly and reassembly of the enzyme is likely to

function as a general regulatory mechanism for the H+-ATPase.

Properties of P-type ATPases

P-type ATPases are a family of comparatively simple membrane proteins with

one to three subunits. These enzymes utilize the energy of ATP hydrolysis to maintain

specific transmembrane ion gradients (161, 199). Members of the P-type ATPase family

are grouped into two classes, Type I and Type II. Type I P-type ATPases transport

transition-metal ions, and Type II P-type ATPases transport alkali cations and other small

cations. Type II P-type ATPases are further divided into Type IIA and Type IIB

depending on the number of subunits present in the enzyme. In eukaryotic cells,

members of the P-type ATPase family include the Ca2+-ATPases involved in cellular

regulation of Ca2+ and found in plasma membranes and intracellular membrane systems

like the sarcoplasmic and endoplasmic reticulums; the Na+,K -ATPases involved in

cellular volume regulation, development of membrane potential, and Na+ transport in

kidney, intestine, and other epithelial tissues; H+,K+-ATPases involved in the

acidification of the stomach and K+ reabsorption in the distal colon and renal collecting

ducts; and H+-ATPases in yeast and plant cells responsible for proton-dependent nutrient

uptake. The prokaryotic P-type ATPases include the Kdp-ATPases involved in the high

affinity K+-uptake in E. coli (8); the Mg2+-ATPases responsible for outward transport of
Mg2+ in Salmonella typhimurium (233); and the Cu+-ATPases involved in uptake of

heavy metals in Enterococcus hirae (216).

Each member of the P-type ATPase family contains the catalytic a subunit, which

is responsible for ATP hydrolysis. In addition, Type IIB P-type ATPases also have a

glycosylated P subunit, which is responsible for trafficking the a subunit to the

membrane and for modulating enzymatic activity. The only members of the Type IIB

family that have been identified to date are the Na+,K+-ATPase and the H+,K+-ATPase.

In certain tissues, however, the Na+,K+-ATPase contains a third membrane protein

referred to as the y subunit. The y subunit seems to modulate the affinity of the enzyme

for ATP, Na+, and K+ (11, 235, 236). Although each member of the P-type ATPase

family is similar in size, there must be important differences that reflect their specific ion

motive properties. The specificity of the translocated ions most likely is characteristic of

the non-conserved amino acids of the ion binding domains or channels of these ATPases.

Unlike the F-type and V-type ATPases, enzymes of the P-type form a high-energy

aspartyl-phosphoryl-enzyme intermediate during ATP hydrolysis. However, the

mechanism by which they utilize ATP hydrolysis and phosphorylation to translocate ions

across the membrane is not clearly understood. The most popular model for the catalytic

mechanism of these enzymes is the Ei-E2 model (51, 108, 117, 206). This mechanism

assumes two major conformational states, El and E2, which are distinguished by their

ability to be phosphorylated by ATP and Pi, respectively. Phosphorylation forces the

enzyme into the E2 state and following dephosphorylation, the enzyme returns to the E,

state. Phosphorylation and dephosphorylation cause changes to occur within the enzyme,

and the bound ion is translocated across the membrane and released.

This dissertation describes the molecular analysis of two ion motive ATPases: the

FiFo ATP synthase from E. coli and the rabbit renal H+,K+-ATPase. This chapter will

cover both enzymes in detail.

E. coli F F0 ATP Synthase

The E. coli FiFo ATP synthase is the simplest and most widely studied FiFo ATP

synthase. It has the stoichiometric subunit composition of ot3P38& for the Fi sector (253)

and ab2clo-12 for the Fo sector (97) (Figure 1-1). The FI sector is connected to the Fo

sector by the central stalk and the peripheral stalk (1, 256). The ye subunits make up the

central stalk, and the b subunit dimer forms the peripheral or second stalk. The enzyme

can function either as a proton-driven ATP synthase or an ATP-driven proton pump in

vivo or in vitro. Mutants with a defective FIFo ATP synthase cannot support growth on

nonfermentable substrates such as succinate because of a lack of oxidative

phosphorylation, but are still able to grow aerobically on glycolytic substrates.

The FIFo ATP synthase from bacteria is an advantageous system to study for the

following reasons: (1) the genes can be over-expressed from plasmid vectors; (2)

molecular techniques can be utilized to change the genes on the chromosome as well as

genes cloned into plasmids; (3) E. coli can grow by glycolysis permitting study of strains

Figure 1-1. Subunit organization of E. coli F,Fo ATP synthase.

defective in FiFO ATP synthase; (4) the enzyme from E. coli is an appropriate model for

studying diseases in the human mitochondrial FIF0 ATP synthase.

The work described in Chapter 3 addresses the biochemical characterization of

the conserved residue bArg-36 in the b subunit of the E. coli enzyme. This section will

discuss what is currently known about the structure and rotational catalysis of the F Fo

ATP synthase and will detail the role of the b subunit as the stator of the FiFo ATP


unc Operon Genetics

Many E. coli mutants which are defective in FIFo ATP synthase have been

isolated. All mutants mapped in eight complementation groups of the unc operon located

at about 83 min on the genetic map (58, 76). The order of the structural genes was shown

to be uncB(a), uncE(c), uncF(b), uncH(8), uncA(a), uncG(y), uncD(P3), and uncC(c) (90).

The gene order of the unc operon was confirmed by determining its complete sequence

(253). The FiFo ATP synthase genes were preceded by a ninth gene, uncI, whose role

remains a mystery. The uncI gene encoded a basic and hydrophobic protein (253).

Using antibodies, the uncI gene product has been shown to be located on the cytoplasmic

membrane as well as in preparations ofFo or F1F0 (203). However, since the gene was

deleted without affecting assembly or activity of the FiFo ATP synthase (82), the protein

is not believed to be a subunit of the functional enzyme.

The functional promoter for the genes of the unc operon is upstream of uncI, two

weak promoters are present in the unci reading frame, and the transcriptional terminator

is immediately downstream of uncC(s). A single polycistronic mRNA transcript was

observed (84, 113, 121, 170). Although there is only one copy of each gene in the

transcript, some subunits, such as c, are found in multiple copies in the intact complex,

whereas other subunits are present in single copies. How is the expression of the genes

regulated so that the final stoichiometries are attained? The regulation is believed to be

under translational control. Brusilow et al. (28) identified loop and stem structures

immediately preceding initiation codons for the uncF(b), uncH(8) and uncG(y) genes. In

each case, these genes follow a more highly expressed gene in the unc operon. The RNA

secondary structures probably function to lower the expression of the genes immediately


Ikemura (106, 107) analyzed the order of preference among identical codons and

named the most preferred codon the "optimal codon." His theory was that E. coli genes

encoding for multi-copy proteins used optimal codons selectively whereas other genes

used optimal and nonoptimal codons to almost equivalent levels. Kanazawa et al. (116)

utilized this calculation for the genes of the FIFo subunits and found that the frequency of

usage of the optimal codon was indeed higher in the multi-copy subunit than in the single

copy subunit, suggesting that the number of subunits is decided at least to some degree by

the frequency of codon usage in each gene.

However, regulation of translation is more likely related to RNA secondary

structure and ribosome binding efficiency. Schaefer et al. (201) used primer extension

inhibition or "toeprinting" to analyze ribosome-binding activities of the translation

initiation sites of the unc operon genes. The toeprint for uncE(c) was at least twice as

strong as those from the other genes. Since there is a direct correlation between the

intensity of a toeprint band in vitro and the strength of the ribosome-binding site in vivo,

the uncE(c) gene has the most efficient ribosome-binding site in the unc operon. This

observation is in agreement with the fact that the c subunit is the most highly expressed

subunit in the FIFo ATP synthase.

Recently, Brusilow's group identified an intragenic ribosome binding site within

the uncB(a) reading frame that preceded a five-codon reading frame that was shifted one

base relative to the uncB(a) reading frame (145). When this binding site was destroyed

by mutagenesis, there was a four- to fivefold increase in expression of an uncB(a)/lacZ

fusion gene, suggesting that expression of uncB(a) was lowered by translational

frameshifting at this site within the uncB(a) reading frame.

FjF0 ATP Synthase Mechanisms

The FiFo ATP synthase utilizes the energy stored in AiHf+ to synthesize ATP by a

rotary catalytic mechanism. A variety of experimental evidence indicated that proton

translocation drives the rotation of the c subunit ring that causes rotation of the y subunit

within the ao3/03 hexamer. Rotation leads to catalysis of the three catalytic sites in the 0

subunits because of the asymmetric y subunit. Proton translocation and rotational

catalysis will be discussed separately in this section, although both are required for ATP


Proton Translocation

Proton translocation occurs through the Fo sector, which in the E. coli enzyme

consists of ab2clo.12 (Figure 1-1). Although the organization of the subunits in Fo remains

to be elucidated, low resolution electron microscopic and atomic force microscopic

images proposed a ring of c subunits with the a and b subunits lying at the periphery of

the ring (22, 212, 230). Chemical cross-linking studies have supported this interpretation

(69, 111, 149). The crystal structure of the FiFo ATP synthase from Saccharomyces

cerevisiae (S. cerevisiae) demonstrated that there is indeed a c oligomeric ring, and each

c subunit forms an a-helical hairpin (224). The a-helical hairpin structure correlated

well with earlier structural data from nuclear magnetic resonance (NMR) spectroscopy of

a monomeric c subunit (85). Neither the a nor the b subunits were resolved in the crystal

structure. The a subunit has five transmembrane a-helices (243, 249), and one of these

helices has been shown to interact with an a-helix of the c subunit (111). Chemical

cross-linking and direct binding assays showed that the two b subunits interact to form a

dimer (61, 150, 218) which extends as paired a-helices through the length of the enzyme

with interactions inside the lipid bilayer (55).

All three subunits (a, b, c) were required for proton translocation (53). Insertion

of the b subunit was necessary for proper assembly of the Fo complex and for insertion of

the a subunit in the membrane (96, 222, 229). The hydrophilic region of the b subunit

was not involved in ion translocation because trypsin digestion experiments that removed

the soluble portion of the subunit did not affect proton movement through Fo (98, 179,

180). The b subunit does not appear to play a role in the actual translocation of protons

across the membrane, but instead may store elastic energy during translocation. This

property will be discussed in further detail later in the chapter.

Both the a and c subunits contain functional groups essential for the transport of

protons across the membrane, but the c subunit houses the site which undergoes

protonation-deprotonation during proton translocation. The conserved residue, CAsp-61 (E.

coli nomenclature), is buried in the membrane in the center of the second transmembrane

helix (85, 224). This amino acid has long been believed to be the site of proton binding

in Fo. Fo-mediated proton translocation and ATP-driven proton translocation can be

blocked by reaction of dicyclohexylcarbodiimide (DCCD) with the carboxyl group of

CAsp-61. Modification of one c subunit is enough to completely inhibit the ATPase activity

of the enzyme (97). Mutations affecting CAsp.61 severely limit HI translocation. Even the

very conservative CAsp-61-*+Gu mutation had a marked effect on the enzyme, reducing

activity to 50% of wild type (155), while the CAsp-61-4Asn substitution virtually inactivated

the enzyme (102).

Kluge and Dimroth offered the most convincing evidence for the role of CAsp-61 in

proton translocation from their studies of the bacterium Propionigenium modestum (P.

modestum). The FIF0 ATP synthase from P. modestum synthesizes ATP by utilizing a

Na+-gradient. At low Na+ concentrations, the enzyme switched from Na+ to H transport,

suggesting a common mechanism for the translocation of both ions. The homologous

DCCD-sensitive site in the c subunit of P. modestum is CGlu-65. The extent of DCCD

sensitivity was strongly influenced by the pH of the incubation medium. At pH 5-7,

where the reaction rate of DCCD is at its maximum, Na+ weakly protected the site. In

contrast, at pH 8-9, where the reaction rate is markedly reduced, Na efficiently protected

the enzyme. The authors concluded that the carboxyl group of CGou-65 was the site of ion

binding and that Na+ only binds to the deprotonated active site residue (129, 130).

Zhang and Fillingame (272) used sequence comparisons between E. coli and P.

modestum to construct multiple mutations in the E. coli c subunit near CAsp-61 to generate

a Na+-translocating F0. One combination of four mutations, CValAsp-61AIalle-+AlaGlu-61SerThr,

did produce an F0 that, although unable to bind Na+, did bind Li\ in a way that inhibited

proton translocation. The dependence of protons on Li+ inhibition suggested competition

for the same binding site. The characteristics of the mutant provided the most direct

evidence that CAsp-61 is the proton binding site in the E. coli subunit.

The a subunit is also believed to play an essential role in proton translocation.

Cain and Simoni (30) identified the first two single amino acid substitutions in the a

subunit, aHis-245-~Tyr and aser-206-Leu, that impaired Fo-mediated proton translocation.

Neither substitution altered FI binding. The site occupied by aHis-245 was at a highly

conserved position, while aSer-206 was adjacent to a conserved Leu residue. It seemed

likely that the a subunit contained at least part of the Fo channel, and many mutations

have been constructed to map out the amino acids that contribute to proton translocation.

The only amino acid in the a subunit absolutely required for FIFo ATP synthase function

was aArg-210. This amino acid is in the fourth transmembrane helix and in close contact

with the expected position of CAsp-61 (111, 243). The site occupied by aArg-210 could not be

substituted by other amino acids without loss of activity (31, 64, 138). Both ATP-driven

proton translocation and passive proton translocation were affected. However, the

assembly of the enzyme was essentially normal since Fi retained ATPase activity in the

absence of Fo. This observation was corroborated by purifying stable intact enzyme

complexes with the aArg-210-lle mutation (79).

Mendel-Hartvig and Capaldi (152) demonstrated that during ATP hydrolysis of

wild type F1iF ATP synthase, the s subunit cross-linked to the P subunit and was trypsin

insensitive. When the enzyme was substrate inhibited, the c subunit had little cross-

linking to the P subunit and was trypsin sensitive. These observations suggested that the

& subunit changed its position during catalysis, which has been shown to be the result of

rotation of the s subunit within the 0303 hexamer (122). Examining trypsin sensitivity

and chemical cross-linking of the s subunit in purified FIFo enzyme complexes with Arg-

210-+iie substitutions revealed a similar observation (79). Therefore, this mutation is an

example of a single amino acid substitution that uncouples a functional Fi-ATPase

capable of rotation from a nonfunctional F0 within an intact FiFo complex.

Current models propose that the a subunit contains two half-channels that offer

access to the site of protonation of CAsp-61 from the periplasmic space and deprotonation

of CAsp-61 to the cytoplasm. It is believed that the aArg-210 residue facilitates a pKa shift of

the CAsp-61 carboxylate to a low pKa form. Protonation of this form is driven by the high

local concentration of protons. Once the site is protonated, a pKa-dependent

conformational change occurs that causes the proton to be released from the other side of

the membrane (188).

The exact number of c subunits in an intact Fo complex remains controversial.

Radiolabeling experiments originally suggested a possible range of nine to twelve c

subunits per F0 (74, 248). Jones and Fillingame (114) conducted cross-linking

experiments with c subunit dimers and trimers and demonstrated that an Fo with twelve c

subunits was functional. This result showed that the stoichiometry could be twelve.

However, Schemidt et al. (202) established that the stoichiometry of the c subunits varied

with growth conditions. Translational rates of uncE(c) were higher in cells grown on

glucose but not on succinate. Perhaps the most compelling evidence for the number of c

subunits came from the crystal structure of the FiFo ATP synthase from S. cerevisiae.

The crystal structure indicated ten c subunits (224). It is possible that some of the c

subunits were lost during the crystallization process. However, it is more likely that the

number of c subunits in an intact enzyme complex is not fixed and certainly not fixed at

twelve. The stoichiometry of c subunits is important because it dictates the number of

protons translocated per ATP synthesized. It is estimated that at least three and maybe

four protons are translocated per one ATP molecule synthesized (68, 246). Since there

are three catalytic sites in Fi, the most pleasing number of c subunits would be a multiple

of three, either nine or twelve. This would correspond to a 120 degree rotation of the y

subunit within Fi (see below).

The ring of c subunits has been shown to rotate (195). The authors immobilized

E. coli F1F0 ATP synthase on a Ni2+ coverslip through a His-tag linked to the N-terminus

of each a subunit. A fluorescently labeled actin filament was attached to the c subunit by

a cysteine residue. Once MgATP was added, the actin filament rotated, indicating that

the c subunit rotated during ATP hydrolysis. It is hypothesized that rotation occurs in the

opposite direction during ATP synthesis. Based on experimental evidence, the current

model for proton translocation is as follows. The CAsp-61 residue is in close proximity to

aArg-210. A proton enters a half-channel in the a subunit and protonates CAsp-61. Once CAsp-

61 becomes protonated, there is a loss of attraction between CAsp-6i and aArg-210. This loss

of attraction causes the c subunit ring to rotate one step, making another CAsp-61 site

available for protonation. One proton remains bound to the CAsp-6i residue during rotation

until a conformational change causes the proton to be released to the other side through a

second half-channel in the a subunit. The rotation of the c subunit ring is believed to

drive the rotation of the y subunit within the a3/P33 hexamer. How rotation is coupled to

catalysis will be discussed below.

Rotational Catalysis

Catalysis occurs in the Fi sector, which in the E. coli enzyme contains oaP3y8s.

The catalysis ofF Fo ATP synthase relies on a strong cooperativity at the three catalytic

sites. Substrate binding at one site occurs simultaneously with product release at the next

site. The cooperativity can be explained by the binding change mechanism which states

that the structures of the three catalytic sites are always different but cycle through

"open," "loose," and "tight" states (27). Experimental evidence has demonstrated that

the FiFo ATP synthase is a rotary motor enzyme. Proton translocation through F0 drives

the rotation of the internal y subunit, producing conformational changes in each catalytic

3 subunit. The conformational changes lead to cooperativity.

The structure of the Fi sector has been well documented (Figure 1-1). The high-

resolution structural analysis of the bovine mitochondrial Fi demonstrated that the a and

P subunits alternate to form a hexamer (1). The hexamer was the shape of a compacted

sphere 80 A high and 100 A across. The 0x33 hexamer was arranged around the coiled-

coil structure of the y subunit, often referred to as the central stalk. Reconstitution

experiments have demonstrated that the a3P33y complex is the minimal complex necessary

for ATP hydrolysis activity (168). The y subunit was able to cross-link to the c subunit

ring, indicating that the y subunit extends the full length of the stalk (255). The e subunit

was also shown to be present in the central stalk as a two-domain structure (241, 258).

Interactions between the C-terminal domain of E with either the a or the 0 subunits have

been identified (3, 4, 257), as well as interactions between the N-terminal domain of

with the y subunit (232, 255) and the c subunit ring (273). Oddly, the yeast crystal

structure did not show any apparent interaction between the E subunit and the Xa3P3

hexamer (224). Therefore, the crystal may not entirely reflect an active state for the


Wilkens and Capaldi (256) used electron microscopy to analyze detergent

solubilized E. coli FiFO ATP synthase. The negatively stained images showed the

presence of two narrow stalks 40-45 A in length that joined Fi and F0. One stalk was the

central stalk composed of the y and s subunits. The peripheral or second stalk extended

down the side of the enzyme from F1 to Fo and was made up of the 8 subunit and b

subunit dimer. NMR of the 8 subunit showed a largely globular a-helical protein (260).

Cross-linking studies have determined that 8 is attached near the top of Fi through

interactions with an a subunit (139, 173). Cross-linking studies have established that the

C-terminal part of the b subunit dimer interacts with the C-terminal region of the 8

subunit and with both the a and p subunits (62, 148, 149, 191, 192, 198, 256). The

second stalk is thought to act as the stator against which the y, s, and the ring of c

subunits rotate relative to the stationary 043 3 hexamer of Fi.

There are six possible nucleotide-binding sites in the F1 moiety: three non-

catalytic sites mainly on the a subunits and three catalytic sites mainly on the p subunits.

The bovine mitochondrial F, was crystallized in the presence of various nonhydrolyzable

analogs and showed that the six nucleotide-binding sites are at the interfaces between the

a and p subunits (1). It was clear from the structure that the catalytic sites in the three p

subunits exist in differing conformations due to their positions relative to the y subunit.

The structure supported Boyer's hypothesis of the binding change mechanism. The

catalytic sites in the p subunits interconvert among three sites: "open," "loose," and

"tight." The open site has a very low affinity for ligands and is catalytically inactive, the

loose site loosely binds ligands and is catalytically inactive, while the tight site tightly

binds ligands and is catalytically active. Proton translocation leads to conformational

changes in each of the P subunits to convert a tight site with bound ATP to an open site,

thereby releasing the nucleotide. Concurrently, a loose site with loosely bound ADP and

phosphate is transformed to a tight site where ATP forms. Fresh substrate binds to the

open site, converting it to a loose site, and the cycle repeats. The steps that require

energy are substrate binding and product release. The conformational changes that occur

in each of the 0 subunit are explained by the rotation of the asymmetric y subunit.

The rotary motion of the y subunit has been proven experimentally for the isolated

F, portion acting as an ATPase. Duncan et al. (60) made a PAsp380-'Cys mutant and

demonstrated that the Cys residue formed a disulfide bond with YCys87. If the bond was

reduced, the enzyme allowed to hydrolyze ATP and then put back into oxidizing

conditions for bond reformation, the disulfide bond was formed randomly among all

three P subunits. If the enzyme was not put under catalyzing conditions, the bond usually

formed with the same 0 subunit. The most logical explanation was that the y subunit

rotated among the 3 subunits during ATP hydrolysis. Switching of the disulfide bond

formation among P subunits could be inhibited by blocking proton transport or oxidative

phosphorylation (274, 275). These results established that rotation was linked to


Sabbert et al. (193) used another approach to detect rotation. A phosphorescent

eosin molecule was attached to the C-terminus of the y subunit and after photobleaching,

the anisotropic decay was followed by polarized absorption recovery. The experiment

did not establish rotation, but after analyzing the decay, the results suggested that the

assumed rotation was "stepped." The data best fit a model with the y subunit acting as a

unidirectional three-step rotor.

The most convincing evidence for the rotation of the y subunit was conducted by

Noji et al. (172). The authors added a His-tag to the N-terminus of the P subunit and

affixed the Fi complex to a slide coated with Ni2+. They also attached a fluorescently

labeled actin filament to the y subunit. Rotation of the y subunit could be observed under

a fluorescent microscope only with ATP hydrolysis. These experiments proved that only

the catalytic domain of the enzyme (43p3y) was necessary to cause rotation. However,

the same results were achieved when the e subunit was fluorescently labeled in an aA33y3

Fi complex (122). Rotation was always counterclockwise, as observed from the

membranous side of the hexamer. In addition, at nanomolar ATP concentrations, the

rotary motion was interrupted. Discrete 120 degree steps were detected with intermittent

backward steps. This stepped motion confirmed the model of Sabbert et al. (193). At

low ATP concentrations, binding of ATP to the catalytic site would be expected to be a

rate-limiting step. The binding of ATP facilitated the release of tightly bound ADP,

suggesting that ATP binding at one catalytic site promoted release at another (2). The

stepped motion at low ATP concentrations suggested that at least two of the three sites

had to be filled by nucleotide in order for a 120 degree rotation to occur. This

observation implies that rotation is coupled to cooperative catalysis involving more than

one catalytic site.

The data suggest a model of energy coupling within the F IFo complex in which

rotation of the ring of c subunits occurs in one direction for ATP hydrolysis and in the

other direction for ATP synthesis. Protons are consecutively translocated for each c

subunit across the membrane through associations with the a subunit. Because the c

subunit is in direct contact with the y and e subunits (255, 273), the rotation of the c

subunits is driven in one direction or drives in the other direction the rotation of the y and

e subunits among catalytic sites.

b Subunit

The Fi and Fo sectors of the enzyme are attached by two stalks. The central stalk

is composed of the y and 6 subunits and rotates within the a3P3 hexamer to drive

conformational changes in the three P3 subunits during catalysis. The b subunit dimer

forms the peripheral or second stalk and participates in interactions with 8, a, and P

subunits in Fi (Figure 1-1). The second stalk is believed to act as a stator, holding the

aC3p3 hexamer against rotating with ys so that the conformational changes can occur.

In E. coli, the b subunit contains 156 amino acids. There is an N-terminal

transmembrane sequence followed by a polar, highly charged region that is predicted to

be mostly a-helical. Analysis of the secondary structure of the hydrophilic region of the

b subunit proposed a random coil between two long a-helices (205, 252). Although these

characteristics were apparent in b subunit sequences from the mitochondrial and

chloroplast enzymes, the b subunit has the least primary sequence conservation among all

subunits of the FIF0 ATP synthase.

Multi-dimensional NMR spectroscopy was used to determine the structure of the

synthetic peptide bl-33, which contains the transmembrane region of the E. coli b subunit

(55). The structure revealed an a-helix between residues bAsn-4 and bMet-22, which was

broken by a rigid bend around bLys-23 to bTrp-26. The a-helix continued at residue bpro-27 at

a 20 degree angle offset. The a-helix stretching from bAsn-4 to bTrp-26 is likely to span the

lipid bilayer to anchor the b subunit in the membrane. Single Cys residues were

substituted for residues 2-21 of the native b subunit, and dimer formation between b

subunits after cross-linking was tested. High yields of b subunit dimers were found at

positions 2, 6, and 10, and less intense cross-links were observed at positions 3, 4, 8, 9,

11, 13, 14, 17, and 18. The results demonstrated that the transmembrane domains of the

two b subunits are capable of close contact with each other. The fact that the less intense

cross-links stretch along the a-helices may indicate that the helices are mobile within the

membrane. Using cross-linking distance constraints and the NMR data, the authors

proposed a model for possible transmembrane b subunit interactions. In the model, the

transmembrane a-helices of each b subunit are at a 23 degree angle to each other. The 20

degree angle offset after the bend may counteract for the 23 degree tilt of the helices so

that the following helical segments at an angle more perpendicular to the membrane.

Attempts to determine the high-resolution structure of the cytoplasmic region of

the b subunit have been unsuccessful. Most of the structural information available on

hydrophilic portion of the subunit has been obtained through circular dichroism

spectroscopy, analytical centrifugation, and chemical cross-linking experiments. Dunn

(61) expressed and purified the cytoplasmic portion of the E. coli b subunit (bVai25-Leu156)

and termed the protein bsol. The polypeptide displayed characteristics indicative of a

highly extended protein during sedimentation equilibrium, and circular dichroism

spectroscopy demonstrated that bsol was highly a-helical. Analysis of the molecular

weight of the bso01 polypeptide by sedimentation equilibrium resulted in a value close to

the calculated dimer value. A variety of experiments have demonstrated that bso, can bind

to the Fi sector (61, 148, 198, 219), but in an intact enzyme complex, dimer formation is

necessary for interaction with F, (218).

Dimerization interactions between b subunits were analyzed by N-terminal

deletions and site-directed mutagenesis using the bsol polypeptide (150). When cysteine

residues were substituted at positions 124, 128, 132, and 139, disulfide formation

occurred with the same residue on the other subunit, implying a parallel a-helical

interaction in this area. Adding bsolSer60-+Cys with bsotLeu65-sCys and bsolAla61-*Cys with

bsolLeu65-*cys resulted in a mixed disulfide formation between pairs of the polypeptides,

suggesting a non-parallel a-helical interaction in this region. N-terminal deletion

analysis demonstrated that residues 53-66 but not 24-52 are essential for dimerization.

The exact boundaries of the dimerization domain are unknown, but probably reside

within residues 53-122 (150, 190). The results propose a model in which the two b

subunits interact in more than one area, but probably not simply as a symmetrical parallel

alignment of a-helices.

I observed that FiFo ATP synthase complexes with a two amino acid deletion in

the hydrophilic region of the b subunit were able to support growth on minimal succinate

that was indistinguishable from the wild type enzyme. This result was unexpected. In

fact, the only defect of the deletion was reduced assembled complexes. To analyze the

effects of larger deletions on the enzyme, our laboratory constructed deletion mutations

within the cytoplasmic region of the b subunit (219). The deletion mutations were

designed to represent deletions of turns in an a-helix. A deletion of two amino acids

approximated a half-turn of an a-helix, while a deletion of seven amino acids was

roughly two turns of an a-helix. Surprisingly, deletions of up to seven amino acids

resulted in fully functional FIFo ATP synthases, and a deletion of eleven amino acids still

produced considerable activity. Removing twelve amino acids, however, destroyed

activity. In each case, the major effects of the mutations were on assembly and stability

of the enzyme complex. Somehow the b subunit was able to compensate for its

shortening of up to 16 A, as would be the case in the deletion of eleven amino acids. The

b subunit was believed to be a rigid rod-like structure with specific points of contact with

FI (150, 173). However, the results of the b subunit deletions suggested that the b

subunit has some inherent flexibility and is not necessarily a rigid structure.

To further examine the flexibility of the b subunit, our laboratory also constructed

insertion mutations in the same area of the subunit (217). In an attempt to retain an a-

helical conformation and dimerization between b subunits, the amino acid sequence

inserted was a duplication of b subunit sequence. Insertion of seven amino acids resulted

in a virtually normal phenotype, and insertions of up to fourteen amino acids retained

significant levels of ATP synthase activity. Any loss of function corresponded to a defect

in assembly of the enzyme complex. In fact, the assembly of b subunits with an eighteen

amino acid insertion was so harsh that no b subunit could be found in membrane

fractions. Insertion of fourteen amino acids would lengthen the stalk by about 20 A.

These data taken with the comparable results from the deletion analysis strongly suggest

that the b subunit has significant flexibility to allow such large changes in length. The

second stalk observed in the electron micrograph of Wilkens and Capaldi (256) was bent

approximately 30 degrees, offering independent evidence for the inherent flexibility of

the b subunit dimer.

Mutagenesis of the hydrophilic region of the b subunit has demonstrated that the b

subunit plays an important role in the proper assembly of the enzyme. Mutating the

conserved residue bAla-79 prevented the proper assembly of the enzyme due to a failure of

b subunit dimerization (147, 218). Early termination of the b subunit resulted in a

nonfunctional Fo complex even when only bGlu-155 and bLu-i156 were missing at the C-

terminus (182, 229). This originally led the field to believe that the b subunit bent back

toward the membrane. However, in light of the current structural information based on

electron microscopy, cross-linking, and nuclear magnetic resonance spectroscopy, the C-

terminus of the b subunit is thought to extend towards Fi (191, 218, 256). Mutagenesis

affecting bAia-i28 abolished dimerization of the b subunits (104).

The residues near the C-terminus of the b subunit are necessary for the proper

assembly of FiFo ATP synthase in vivo (229) and for interaction with F1 in vitro (148).

Cross-linking experiments have demonstrated that the 8 subunit binds near the top of the

ac333 hexamer (139, 173). The 8 subunit was essential for binding Fi to Fo (198).

Although it was believed that the interaction between the b subunit and F, was through its

association with the 8 subunit, cross-linking analysis of the E. coli enzyme failed to

produce a b-8 interaction (10). In the chloroplast enzyme, however, a zero-length cross-

linking agent did reveal a subunit 1-5 (E. coli b-8) product, suggesting that the subunits

are adjacent to each other (18). The b-6 interaction in the E. coli enzyme has since been

demonstrated with the yeast two-hybrid system, NMR spectroscopy, and protease

protection experiments (192, 198). Sedimentation equilibrium analysis of purified bsol

and 8 subunits showed the complex had a stoichiometry of b28, and sedimentation

velocity studies revealed the complex was organized in an end-to-end instead of a side-

by-side relationship (62). In order for the b subunit to reach up the side of FI to the 8

subunit, parts of the b subunit must be in contact with the a or p subunits. Cross-linking

studies have shown that the b subunit interacts with both the a and P subunits through

parts of the dimerization domain as well as the 8 binding domain (149, 191).

The second stalk is believed to act as a stator during rotational catalysis to hold

the a3p3 hexamer in place against the rotation of the c subunit ring and central stalk.

Evidence exists to support this model, but some data seem to conflict with it. These

considerations were eloquently explained by Dunn et al. (63). The first issue is that of

the stability of the second stalk and its relationship to the a3P3 hexamer. Once F, is

bound to Fo through the second stalk, the stator is expected to remain associated with one

a subunit or ap pair during catalysis. In support of this idea, Ogilvie et al. (173)

demonstrated that cross-linking the 8 subunit to one of the a subunits at an introduced

aCys-2 did not affect the coupling activity of the FIFo ATP synthase. Linking two b

subunits together through disulfide bonds also had no effect on coupling (191, 192).

These results suggest that the second stalk is a permanent element of the enzyme that is

connected to one a subunit or ap pair. However, Rodgers and Capaldi (191) identified a

disulfide bond between the C-terminus of the b subunit and (cys-90o that did block

coupling, implying that some flexibility or movement is necessary in the b-a association.

If the second stalk remains associated with only one a subunit or ac3 pair during

catalysis, then one would expect a difference among the three ap pairs, and this

difference would presumably affect activity of the enzyme. In light of this idea, Kersten

et al. (124) used electron spin resonance spectroscopy to examine the effects the b

subunit has on structural changes within the catalytic nucleotide-binding sites of the P

subunits. The data demonstrated that the hydrophilic region of the b subunit affects the

conformation of the catalytic sites and puts the sites in a conformation that seems to be

consistent with the open state. This result implies that the b subunit may play a role in

the release of ATP from the catalytic site.

The second issue is the tensile strength of the second stalk. Rotation of the central

stalk applies a torque on the a3133 hexamer that must be opposed by the second stalk so

that the hexamer does not rotate with the y subunit. A rigid rod-like b subunit would

serve this function well. However, insertion and deletion analyses suggested that the

second stalk was a flexible structure (217, 219). The interaction of the b subunit with FI

may add some strength to the second stalk so that it can function as a stator.

The final issue is the elasticity of the second stalk. The flexible b subunit, along

with ye, may transiently store energy during rotational catalysis. Junge and colleagues

(38) proposed a model where the b subunit would tolerate stepwise elastic deformation as

protons moved through Fo. The stored energy would then be expended in one step to

drive conformational changes which allow the release of ATP from the catalytic site. On

the other hand, Oster and Wang (175) have emphasized the importance of the b subunit

as a flexible, elastic connection coupling F1 and F0 so that any torque produced from

either proton movement or ATP hydrolysis can be distributed more smoothly to the rest

of the enzyme, resulting in increased efficiency. Both models require that the b subunit is

elastic, although there is no direct evidence to support the elasticity of the b subunit. The

putative coiled-coil segment in the hydrophilic region could provide several methods for

elastic deformation, including bending, stretching, and compression. How energy is

stored in the second stalk may become clear once more structural information is



The H,K-ATPase utilizes the energy stored in ATP to transport H+ outward in

an electroneutral exchange for the inward transport of K with a stoichiometry of

2H+:2K+:1ATP (187). The enzyme was first identified in the stomach where its high

expression is responsible for luminal gastric acid secretion. Additional H,K-ATPase

isoforms have since been identified in other organs, including the colon, kidney, heart,

and skin. The H ,K -ATPase is a Type IIB P-type ATPase and thus consists of a large a

subunit and a smaller P3 subunit. The enzyme most likely exists as an (c/IP)2 tetramer

(186). The a subunit contains the binding sites for ATP, phosphorylation, and inhibitors

and is designated the catalytic subunit (99, 244). The P subunit is necessary for efficient

targeting of the a subunit from the endoplasmic reticulum to the plasma membrane and

also modulates enzymatic activity (43). The Na,K-ATPase is the only other known

Type IIB P-type ATPase, and its a and P subunits are structurally and functionally

related to the subunits from the H+,K+-ATPase. In fact, much of the information

currently available on the H+,K-ATPase has been inferred from studies of the

prototypical Na+,K -ATPase.

Enzyme Structure

Hydropathy plots have indicated that the HKa subunit has eight to ten membrane

spanning regions with a large cytoplasmic loop (140, 210). Both the N- and C- termini

appear to be cytoplasmic, indicating an even number of transmembrane domains between

them (204). Limited proteolytic cleavage data showed the presence of at least eight

transmembrane segments (20, 207), while in vitro translation experiments suggested the

presence of ten transmembrane domains (17). Efforts to crystallize the H+,K+-ATPase

have produced only low-resolution two-dimensional crystals (95, 184, 266) that have not

provided any useful understanding of the overall structure of the a subunit. However,

high-resolution structural data have been determined for two Type IIA P-type ATPases.

Two-dimensional crystals of the H+-ATPase of Neurospora crassa (N. crassa) yielded a

three-dimensional map of the enzyme at a resolution of 8 A (16). Both an 8 A resolution

three-dimensional map (271) and a 2.6 A resolution atomic model (240) were determined

for the sarcoplasmic reticulum (SR) Ca2+-ATPase. The Ca2+-ATPase and the H+-ATPase

had similar structural characteristics. Both enzymes had an extensive mass on the

cytoplasmic side, a membrane sector, and a small extracellular region. There were ten

clearly defined and tightly packed transmembrane domains, and the arrangement of the

transmembrane helices of Ca2+-ATPase was established (240). Although there is

generally a low degree of sequence homology across the family of P-type ATPases, they

all perform similar functions of ion transport with ATP hydrolysis. Moreover, highly

conserved regions most likely related to function or conformation are found in most of

the members. Therefore, it seems logical that the common principles determining the

relationship between molecular structure and functional properties are preserved within

the family, and the HKa subunit almost certainly has ten transmembrane domains with a

similar arrangement to that of the Ca2+-ATPase (Figure 1-2A).

All of the known isolated P subunit isoforms from Na+,K -ATPase and H ,K+-

ATPase share a common molecular architecture: a short N-terminal cytoplasmic

segment, a single transmembrane domain, and a large C-terminal extracellular domain



Intracellular Side

SCH-28080 Omeprazole

Extracellular Side C

Figure 1-2. Transmembrane organization of H,K-ATPase. A, The spatial
arrangements of the ten transmembrane domains of the HKo subunit, shown
in red, have been assigned based on structural data from Ca2+-ATPase. The
likely position of the HKP subunit transmembrane domain is shown in blue.
The cation binding sites are represented by black circles. B, The topological
structure of the HKa subunit is shown in red with the ten transmembrane
domains numbered. The ATP and inhibitor binding sites are indicated.
The HKp subunit is shown in blue. Glycosylation sites are represented by
orange circles, and the disulfide bonds are depicted by black lines.

(Figure 1-2B). The extracellular domain is glycosylated and contains six completely

conserved cysteine residues that are linked by three internal disulfide bonds (43). The

cysteine residues form the bonds in a sequential arrangement (41, 126, 156). Both

glycosylation and disulfide bond formation are important in a/p assembly, as well as in

the activity of an intact enzyme complex (43).

Wheat germ agglutinin chromatography was used to analyze the areas of

interaction between the a and P subunits. Intact gastric vesicles were trypsinized which

left the P subunit largely intact. The vesicles were solubilized and passed over a wheat

germ agglutinin column to retain the P subunit. The transmembrane domain pair 7 and 8

was retained on the column, suggesting an interaction between a and P within these

transmembrane domains (208). The yeast-two hybrid system identified a similar area of

interaction between the a and P subunits, namely within RArg-898 and aThr-928 (151). This

region is flanked by transmembrane domains 7 and 8. Interestingly, a chimera of the

Na+,K -ATPase a subunit containing the region from UOGn-905 to aVal-930 of the H+,K -

ATPase preferentially assembled with the P subunit of the H ,K -ATPase (254). When

comparing the amino acid sequence similarity within the region of both a subunits, a

homologous 16 amino acid sequence was identified. There were several sequence

differences in the region, such as the position of charged amino acids. The homologous

amino acid sequence of the a subunit of the two enzymes is likely a point of contact with

the 0 subunit, and the slight differences in the sequences of this region may explain the

selective assembly of the P subunits with their a subunits. Yeast-two hybrid analysis

also identified interactions of the a subunit with two areas within the P subunit between

3GIn-64 to PAsn-130 and PAla-156 to PArg-188 (151). Taken together, the data suggest that the

extracellular loop before transmembrane domain 8 of the a subunit interacts with the

extracellular domain of the P subunit. Because the single transmembrane domain of the

P subunit ends at position 62 and position 64 is included in the interaction between a and

3, it is expected that the transmembrane domain of the P subunit is next to

transmembrane domain 8 of the a subunit (Figure 1-2B).

Enzyme Mechanism

The mechanism by which the H+,K+-ATPase utilizes ATP hydrolysis to

translocate H+ and K+ across the membrane is not clearly understood. Many experiments

have been conducted on other P-type ATPases, including the Na+,K+-ATPase and the

Ca2+-ATPase, to characterize the transitional steps and mechanism of ion translocation.

The most popular model for the catalytic mechanism of all P-type ATPases is the E -E2

model (51, 108, 117, 206). This mechanism assumes two major conformational states, El

and E2, which are distinguished by their affinity for ATP and Pi, respectively. Each state

has ion binding sites on opposite sides of the membrane. El binds cytoplasmic ions and

E2 binds luminal ions.

The reaction mechanism of ion transport for the H+,K+-ATPase is depicted in

Figure 1-3. The enzyme probably transports a proton as a hydronium ion (H30+). H30+

binds with high affinity to the EI-ATP form. The phosphate from ATP is transferred to a

specific Asp residue on the a subunit and transforms the enzyme to the intermediate E -

Pi-H30+ in which H30 is occluded. This form is rapidly converted to E2-Pi-H30O. The

E2-Pi-H30+ enzyme has a low affinity for H30+ and a high affinity for K. Hence, the E2-

Pi-K enzyme is formed, and H30+ is released into the lumen. E2-Pi-K+ is rapidly

dephosphorylated causing the occlusion of K in the intermediate E2-K+ form. ATP

Intracellular Side


H3O+ K+








Extracellular Side

Figure 1-3. Enzymatic mechanism of H+,K-ATPase. Each step of
the reaction is described in detail within the text. Adapted from van
Driel and Callaghan (244).


binds to E2-K+, and the enzyme is converted back to Ei, which releases K+ into the

cytoplasm. El binds to H30, thereby completing the reaction cycle (244).

The atomic model of SR Ca2+-ATPase was determined with two Ca2+ bound in

the membrane domain (240). When comparing this structure to that obtained from

tubular crystals (271), it was noticed that the cytoplasmic region was more compact in the

maps from tubular crystals. The authors fitted the atomic model to the low-resolution

map from tubular crystals and concluded that concerted domain motions occur during

Ca2+ binding. Indeed, direct evidence for the conformational changes between the El and

E2 states has been obtained with extrinsic fluorescent probes. Fluoroscein isothiocyanate,

which labels (XLys-516 on the cytoplasmic side of the hog H+,K+-ATPase, exhibits a marked

decrease in fluorescence in the presence of K+ and enhanced fluorescence in the presence

of Na+ as a surrogate for H+ (109, 185). Therefore, the El and E2 states change the

environment of the fluorescein bound in the cytoplasmic domain. The K+-competitive

antagonist arylquinoline MDPQ binds in the region of transmembrane domains 1 and 2

on the extracellular surface of the H+,K+-ATPase. Upon binding to the enyzme, the

fluorescence of this reagent is enhanced. The phosphorylation of the H+,K+-ATPase

further increases the fluorescence of MDPQ (183). Consequently, conformational

changes occur on both sides of the membrane during the cycling between E, and E2


Gastric H+,K+-ATPase

During the 1960's researchers were interested in the mechanism of hydrochloric

acid production in the stomach. One theory, the redox theory, was in parallel with the

chemiosmotic hypothesis proposed by Mitchell for mitochondria (158). The redox theory

suggested that a system of cytochromes separated protons from electrons and released

protons to the outside of the cell. The other theory, the high-energy phosphate theory,

was modeled after the Na+,K-ATPase explained by Skou (213). This theory proposed

that the chemical energy of ATP was used to drive the uphill transport of protons. The

fact that ATP-dependent acidification by membrane vesicles from the stomach was

observed offered validation to the ATPase theory. During the 1970's the search began

for a proton-transporting ATPase in the stomach. A gastric-specific ATPase was soon

identified that was activated by K%, but not Na+ and was not inhibited by ouabain, a

selective Na+,K+-ATPase inhibitor (78, 194). Because maximal transport activity by the

ATPase required the presence of K* ionophores such as valinomycin (71, 194), the

enzyme was distinguished as catalyzing an electroneutral exchange of H+ for K+ and

consequently called the H+,K+-ATPase.

Gastric acid secretion is a regulated process resulting from a complex set of

mechanisms acting at the central, peripheral, and cellular levels. A gastric parietal cell is

an elaborate arrangement of secretary membranes, where the HI,K+-ATPase is highly

expressed. The secretary membranes are networks of cytoplasmic helically coiled

tubules called tubulovesicles (73, 181). The H,K -ATPase is inactive in the

tubulovesicles due to the impermeability of the vesicles to K+. When a parietal cell is

stimulated, the tubulovesicles incorporate into the apical membrane of the cell to form the

secretary canaliculus. The cytoskeleton reorganizes to form microvilli at the canalicular

surface. Translocation of the H,K-ATPase from the tubulovesicles to the canaliculus

activates the enzyme because the canaliculus is permeable to K (99).

HKaj Subunit Isoform

HKai is the gastric a subunit isoform. The subunit is made up of 1033 amino

acids and is encoded by the mouse gene Atp4a, which was localized to chromosome 7

(143). The cDNA for HKai has been cloned from a number of species with

approximately 90% primary sequence identity. The subunit contains conserved

sequences including the phosphorylation site at Asp-386 (250), ATP-protected binding

sites for pyridoxal 5-phosphate at Lys-497 (141, 231), and fluoroscein isothiocyanate at

Lys-518 (66). These sites are located on the large cytoplasmic loop, which links

transmembrane domains 4 and 5 of the HKai subunit. Although HKai is sometimes

referred to as the gastric a subunit isoform, HKai has been identified in other organs,

including the kidney.

Selective inhibitors of the gastric H+,K+-ATPase have been identified.

Omeprazole is a substituted pyridyl methylsulfinyl benzimidazole, an acid-activated

thiophilic compound, that forms stable disulfide bridges with accessible cysteine residues

on the extracellular side of the enzyme. It covalently inhibits the gastric H+,K+-ATPase

and is often used for the treatment of acid-related diseases. Once administered, the drug

becomes protonated in the lumen of the stomach. After protonation, omeprazole is

converted to a thiophilic-reactive sulfenamide that can react with accessible cysteine

residues of the H+,K+-ATPase and inhibit the enzyme. The area of inhibition has been

mapped to acys-813 (21) (Figure 1-2B).

SCH-28080 is an imidazo-pyridine and inhibits the H+,K+-ATPase by competing

with K at the extracellular binding site (115, 153, 166). Since the drug is strictly K+-

competitive, it is a reversible inhibitor. The region of inhibition of SCH-28080 was

believed to be within the extracellular loop between transmembrane domains 1 and 2

(165) (Figure 1-2B). A mutation at aMet- 13, which lies within this region, did indeed

have a large effect on SCH-28080 binding affinity. In contrast, Asano et al. (14)

demonstrated that the entire extracellular loop between transmembrane domains 1 and 2

of HKc could be replaced with the same region from the Na+,K+-ATPase a subunit with

no loss of SCH-28080 binding. The Na+,K+-ATPase is not inhibited by SCH-28080 so

the chimeric HKal subunit was not acquiring a SCH-28080 binding site from the NaKa

subunit sequence. At this time, it is unclear why there is a disagreement in the

importance of this loop in SCH-28080 binding.

Interestingly, binding of SCH-28080 prevents inhibition by omeprazole,

suggesting an area of overlap of binding sites for the inhibitors (100). Indeed, mutations

in transmembrane domain 4 toward the extracellular side of the segment had more

significant effects on the affinity of SCH-28080 than at (Met- 13 (164), and this surface of

the loop had been proposed to bind SCH-28080 (135). Nevertheless, there were no

effects on the affinity of the competing ion K+, and therefore argues that the outer region

of transmembrane domain 4, although binding SCH-28080, is not involved in binding K.

However, binding of K inhibits SCH-28080 binding, signifying a change in this region

induced by K binding. Taken together, the data suggest that the outer region of

transmembrane domain 4 is a major binding area for SCH-28080, and acys-813 at the end

of transmembrane domain 5 must also be included in the binding region. This area most

likely represents the K+ binding site. In agreement with this hypothesis, the Ca2+-ATPase

crystal structure shows the two Ca2+ bound surrounded by transmembrane domains 4 to 6

and transmembrane domain 8 (240).

The function of H+,K+-ATPase in the acid secretion of the stomach has been well

established. However, studies have suggested that the enzyme may be responsible for the

viability and normal development of parietal cells (119). Animals treated with inhibitors

of acid secretion normally have parietal cells with dilated canaliculi (94, 119, 120, 137).

Treatment with omeprazole caused deterioration of parietal cells and an increase of the

number of preparietal cells (119). To study the importance of the H+,K+-ATPase in the

development, maintenance, and function of the gastric mucosa, mice with a null mutation

in the HKao subunit gene (Atp4a) were generated through homologous recombination

(220). Sequences in exon 8 that encoded the phosphorylation site were deleted and

replaced with the neomycin resistance gene. Absence of HKai transcript and protein

were confirmed by Northern analysis and immunohistochemistry. There were no

distinguishable differences in physical features or mortality among Atp4a+'+, Atp4a+/', and

Atp4a''" mice. Although HKci was not produced, the HKO subunit mRNA was

increased, and the protein was readily detected in the Atp4a'" mice. However, Atp4ad'

mice had a gastric luminal pH of approximately 7, whereas the pH from Atp4a+'+ and

Atp4a+'" mice were considerably lower at about pH 3.2 and 3.5, respectively. The

secretary canaliculus was much larger and had few microvilli, and the tubulovesicles

were larger in diameter in the parietal cells of the Atp4a'- mice. Nevertheless, there was

no indication of an increase in the degeneration of parietal cells in those mice. The

number of parietal cells was essentially the same in mutant and wild-type mice. The

study demonstrated that expression of HKa was required for gastric acid production, but

the H+,K+-ATPase probably does not play a role in parietal cell differentiation.

HK1 Subunit

Although it was known that the Na+,K-ATPase contained a p subunit, the 1

subunit of the gastric H+,K-ATPase eluded researchers because it stained weakly with

Coomassie Blue and went undetected for some time. Lectin affinity chromatography

could be used to isolate the HKai subunit, which indicated that the a subunit was closely

associated with a 60 to 80 kD glycoprotein (33, 174). Deglycosylation of the protein

resulted in a 35 kD core protein which was partially sequenced (91, 189, 238). The

sequence shared homology with the (3 subunits of the Na+,K-ATPase, and the HK3

subunit has since been cloned and sequenced from a variety of species. The HKO subunit

consists of 294 amino acids and is encoded by the mouse gene Atp4b located on

chromosome 8 (37). The protein has six to seven possible N-linked glycoslyation sites,

depending on the species (162, 189, 209, 238), and most of the possible sites are used

(33, 174, 238)

Co-immunoprecipitation studies using monoclonal antibodies specific for HKP

have shown a physical interaction between HK3 and HKal in gastric membranes (112).

The a/p association is strong but not covalent. Nonionic detergents solubilize the

complex but do not dissociate the subunits, while treatment with more chaotropic

detergents results in dissociation of the subunits. The HKO subunit is necessary for the

function of the H+,K+-ATPase. Many studies involving the expression of H+,K+-ATPase

or Na ,K -ATPase in heterologous systems have demonstrated that neither ca nor 3

subunits alone possess ion-transport activity. Only a subunits that assemble with 3

subunits in the endoplasmic reticulum are stably expressed, acquire their functional

properties, and are transported to the plasma membrane. Indeed, the HKp subunit may

contain trafficking signals required for correct membrane localization in polarized cells.

Interestingly, HKp can act as a surrogate for the proper assembly of the NaKa

subunit isoforms (103), and any of the NaKa subunits can assemble properly with any of

the NaKp or the HKp subunit isoforms (65). These observations suggest that the a/p

interactions must involve homologous regions of the two subunits, and this association

must confer a structure which is necessary for function of the enzyme. The fact that

mismatching of subunits has not been observed in vivo suggests that correct assembly of

the enzyme complex involves factors other than the primary amino acid sequence.

The role of glycosylation on the extracellular domain of the HKp subunit is not

completely understood. An early hypothesis suggested that the oligosaccharides offered

protection against the acidic gastric conditions (72). A preliminary study demonstrated

that glycosylation of HKP provided resistance against pepsinolysis in vitro but to what

degree this functions in vivo remains uncertain (40). Incubation of HKP with

tunicamycin completely abolished the K+-ATPase activity and SCH-28080-sensitive

phosphorylation of the enzyme when expressed in Sf9 cells (127). However, the

assembly of the protein was not affected, suggesting that the sugars were required only

for biosynthesis of a functional enzyme. It should be noted, however, that functional

wild-type H+,K+-ATPase did not traffic properly in Sf9 cells. Catalytically active

enzyme complexes were localized to intracellular membranes and not on the plasma


Recently, Asano et al. (13) created a series of mutants that progressively deleted

all of the glycosylation sites in HKp and expressed the constructs in HEK293 cells.

Removing any one of the glycosylation sites had no effect on assembly or enzymatic

activity. Progressive removals of the sites on activity were cumulative, and deleting all

seven sites resulted in a loss of ao/ assembly and function. Interestingly, assembly was

not affected when three glycosylation sites were removed, but the surface delivery of the

0 subunit was hindered, indicating that glycosylation played a role in the delivery

mechanism of the enzyme. These results suggested that glycosylation of HKp plays a

role in activity, assembly, and the surface delivery of the enzyme. In contrast, if the

glycosylation sites of the NaK1i subunit, which has three N-linked sites, were abolished

by mutagenesis (19), or if the sugars were removed by tunicamycin (228, 269), the

unglycosylated NaKpI3 subunit still associated with the NaKoa subunit and produced a

fully functional enzyme at the cell surface. However, the P subunit was more susceptible

to proteolysis and had a decreased efficiency in assembling with a subunits (19). The

observations suggested that the glycosylation sites on the NaKP i subunit were not

required for enzyme assembly, function, or protein trafficking but played an important

role in the stability and proper folding of the protein.

The six cysteine residues in the extracellular domain of HKP subunit are

completely conserved among all known P subunit isoforms. The residues form disulfide

bonds in a sequential arrangement (41, 126, 156). The importance of the disulfide bond

formation has been studied with reducing agents. The reduction of the disulfide bonds in

HKp inactivated the H+,K -ATPase,. Similar results were observed with NaK131 (123).

Binding of K+ increased the resistance of disulfide bond reduction in the H+,K+-ATPase

(41). High concentrations of Na+ is known to shift the Na+,K-ATPase or H*,K+-ATPase

to the El state (214), and high concentrations of Na+ caused the disulfide bonds of HKP

to be more susceptible to reduction in the presence of K+ (41). It seems that the strength

of disulfide bonds in HKp increased as the enzyme changed from the El to the E2K+

conformation, suggesting that the disulfide bonds offered conformational stability during

K+ transport. A study with a monoclonal antibody that inhibits enzymatic activity

provided alternative support for HKp subunit participation in enzymatic function.

Monoclonal antibody 2G11 I binds to an epitope within the first 36 N-terminal amino acids

of HKp and inhibits H+,K+-ATPase activity (42). The antibody binding altered K+

binding affinity, thereby supporting the hypothesis that the P subunit aids in the apparent

stability of the E2K+ conformation.

Each disulfide bond may play a different role in an intact enzyme complex.

Noguchi et al. (171) reported that eliminating the most N-terminal disulfide bond of

Torpedo NaKPi produced an inactive ca/P complex, and abolishing either of the two most

C-terminal disulfide bonds prevented assembly of the a/P complex. Beggah et al. (19)

conducted a similar study with Xenopus laevis NaKp i and confirmed that the two most

C-terminal disulfide bonds were important in the assembly of the enzyme complex.

However, eliminating only the N-terminal disulfide bond allowed the formation of a

small number of functional complexes at the cell surface, which was in disagreement

with the data from Noguchi et al. (171). The authors believed that the inconsistency in

the data was likely due to the sensitivity of the assays used. It appears that the domain

comprised between the two most C-terminal disulfide bonds is able to adopt a structure

that partially permits correct a/P subunit interaction.

The cytoplasmic tail of the HKp subunit has a four-residue amino acid sequence

that is highly homologous to tyrosine-based endocytosis signals. The signal could be

responsible for the endocytosis and retrieval of the H+,K+-ATPase from the secretary

canalicular membranes to the tubulovesicles. To determine if the signal was important in

the cycling of the enzyme in parietal cells, Courtois-Coutry et al. (48) generated

transgenic mice expressing HKp in which the motifs tyrosine residue was mutated to an

alanine. The parietal cells from mice expressing the mutant HKp subunit incessantly

secreted acid and constitutively expressed H+,K+-ATPase at their cell surfaces. The

results suggested that the tyrosine-based signal is required for the endocytosis of the

enzyme and for the termination of acid secretion.

The biological role of HKp was analyzed by a null mutation into the HKp subunit

gene (Atp4b) through homologous recombination (200). The mutation disrupted exon 1

so that a functional protein could not be produced. Absence of the HKp transcript and

protein in the stomach were confirmed by RT-PCR, Western analysis, and

immunohistochemistry. The protein levels of HKao in the Atp4b'/' mice were lower than

in Atp4b+'+ and Atp4b+- mice. This was in contrast to what was observed in the Atp4a'/

mice (220). There were no distinguishable differences in physical features or mortality

among the Atp4b+l+, Atp4b+', and Atp4b-'" mice. However, Atp4b-'1 mice had a gastric

luminal pH of about 7, whereas the pH from Atp4b+/+ and Atp4b+/" mice were

considerably lower at pH 3.6 and 4.2, respectively. Sections of the gastric mucosa were

examined at various ages of mice, and the appearance of parietal cells occurred more

slowly in the Atp4b"'1 mice than in normal mice. In the Atp4b'"' mice, the secretary

canaliculus was much larger and had few microvilli, and the tubulovesicles were larger in

diameter. The results demonstrated that HKp was required for the acid secretary activity

of the stomach. The fact that the secretary membranes of parietal cells were altered in

the absence of the HKP subunit implied a membrane structural, and possibly, a

biosynthetic role of the H+,K+-ATPase.

Renal H+,K+-ATPase

Potassium is critical for many cell functions, and its concentration in cells and

extracellular fluid is kept constant despite wide variations in dietary K intake. The

normal concentration of K in the extracellular fluid is within a narrow range of 3.5 to 5.0

mEq/L. Plasma K levels rising above or dropping below this narrow range produces

hyperkalemia or hypokalemia, respectively. Both conditions cause cardiac arrhythmias

resulting in life-threatening situations. The kidney plays the major role in K

homeostasis by regulating K+ excretion and functions to conserve K+ when dietary K

intake is reduced. However, the ion pumps and channels that were responsible for K

conservation remained obscure into the late 1980's. Doucet and Marcy (57) measured [y-

32P] ATP hydrolysis and detected a K-activated ATPase activity in the connecting

segment, cortical collecting duct, and outer medullary collecting duct of rabbits and rats.

The ATPase had a high affinity for K+ (Km=0.2-0.4 mM). The renal ATPase was not

inhibited by ouabain, but two known gastric H+,K -ATPase inhibitors, omeprazole and

vanadate, were able to inhibit the ATPase activity. Vanadate inhibition indicated that the

enzyme was a P-type ATPase. In addition, K+ depletion increased the K -ATPase

activity in the outer medullary collecting duct and the cortical collecting duct, making the

enzyme a reasonable candidate for active K absorption. Similar results were observed

by Garg and Narang (80). Moreover, yet another gastric H,K+-ATPase inhibitor, SCH-

28080, also abolished K -ATPase activity.

Wingo (262) analyzed dietary K+ influence on K+ transport in the kidney. He

established rabbits on one of three diets: a K-replete diet, a K-deplete diet, or a K-

deplete Na+-supplemented diet. Both K -deplete diets are referred to as K-restricted

diets. Plasma K+ levels were significantly less in K+-deplete animals when compared to

K+-replete and K+-deplete Na+-supplemented animals. However, muscular K+ levels

were similar in all animals, demonstrating that significant changes in K+ excretion

occurred without any indication of muscle K+ depletion. Plasma aldosterone levels were

significantly lower in the K+-deplete Na+-supplemented rabbits (2.88 0.57 ng/dl) when

compared to levels from K+-deplete rabbits (15.6 5.3 ng/dl) and K+-replete animals

(44.7 14.0 ng/dl). Therefore, any changes in K+ transport could not be accounted by

the effects of aldosterone.

Wingo observed that perfused outer medullary collecting duct from the inner

stripe (OMDCi) of the three groups had similar rates of fluid reabsorption and similar

transepithelial voltages. However, there was a significant change in K+ reabsorption,

measured as a flux of 42K+, in the three groups. K+ reabsorption nearly tripled in the K+-

restricted rabbits (17.2 2.0%, K+-deplete; 20.8 4.2% K+-deplete Na+-supplemented)

when compared to the K+-replete rabbits (5.9 0.8%). He concluded that K+

reabsorption in the medullary collecting duct is similar to K+ secretion in the cortical

segments at typical fluid flux rates and could have significant effects on the amount of K+

excreted in urine. In a later paper, Wingo (261) demonstrated that the OMCDi had

active, omeprazole-sensitive acidification and K+ absorptive properties and was the first

to suggest the presence of an H+,K+-ATPase in the kidney.

Chevel et al. (39) monitored the 86Rb flux (a radioactive K+ congener) and K+-

ATPase activities in the cortical and medullary collecting ducts. SCH-28080 inhibited

both activities with an IC50 of 0.5 gM. Studies were conducted in the presence of 2.5

mM ouabain to inhibit Na+,K+-ATPase activity. Evidently, Rb uptake occurred in a

manner other than by the Na ,K+-ATPase. In fact, structural, functional, and biochemical

studies have implicated the H+,K+-ATPase in the final regulation of K+ excretion and

absorption in the kidney (263, 265).

Renal HKa Subunit Isoforms

One of the a subunits present in the kidney is the prototypical gastric HKao. Ahn

and Kone (5) used the rat HKai sequence to generate a probe to study HKai1 mRNA

distribution along the collecting duct by reverse-transcriptase polymerase chain reaction

(RT-PCR) on tubule segments and in situ hybridization. Signals showing the presence of

HKai mRNA were observed from the connecting segment through the inner medullary

collecting duct. A similar distribution had been observed by immunohistochemistry

using antibodies directed against the HKai subunit (264).

The second class of H+,K+-ATPases resident in the kidney are the HKa2 subunit

containing enzymes. Crowson and Shull (49) generated the first cDNA for a rat HKa

colonic subunit. Referred to here as the HKa2a subunit, it is sometimes called the colonic

HKa subunit. They noted a weak hybridization signal from the kidney in a tissue

distribution Northern analysis, and evidence from other laboratories corroborated this

observation (133). The deduced polypeptide consisted of 1036 amino acids and had 63%

amino acid identity to HKal and 63% identity to the three NaKa subunit isoforms. The

cDNA for ATP1AL1 was cloned from human skin axilla with mRNA expression in brain

and kidney (89, 160). The cDNA encoded a 1039 amino acid protein that shared 86%

identity to the rat HKa2a protein and 65% amino acid identity to human HKai. Both the

human and rat cDNAs encoded ouabain-sensitive H/K exchange activities when

expressed in Xenopus laevis oocytes (47, 159) and HEK293 cells (87) with the HKP


Using 5' rapid amplification of cDNA ends (RACE) from rat kidney and distal

colon mRNA with reverse primers for HKa2a, Kone and Higham (131) identified two

distinct HKac2 mRNAs that were products of the rat HKa2 gene. The first variant,

HKa2a, was identical to the previously described sequence (49), and the 1036 amino acid

protein was encoded by a 4.0 kb mRNA transcribed from the 5'-most promoter. The

second variant, HKa2b, was a 929 amino acid protein encoded by a 4.0 kb mRNA

transcribed from an internal promoter within intron 1 of the HKa2 gene. The HKa2b

mRNA contained an unusually large 5' UTR with eight upstream open reading frames.

The HKa2b mRNA sequence was identical to HKa2a starting at the codon for Lys-4 of

HKia2a. However, the first AUG triplet of the HKa2b mRNA that was in a favorable

context for translation initiation corresponded to Met- 109 of HKa2a. Therefore, the

predicted HKa2b protein of 929 amino acids lacked the first 108 amino acids of the

HKa2a sequence, putting the N-terminus of the HKa2b protein immediately before the

first predicted transmembrane domain. The truncation deleted putative protein kinase A

and protein kinase C phosphorylation sites present in the rat HKa2a protein. Northern

analysis demonstrated that both mRNAs (HKa2b > HKa2a) were expressed abundantly in

distal colon and less in the proximal colon and kidney. The levels of the two mRNAs

increased with K' deprivation. Stable expression of HKa2b with the HKI subunit in

HEK293 cells revealed a K+ uptake that was SCH-28080-resistant and partially sensitive

to ouabain.

Campbell et al. (35) identified a different alternatively spliced HKct2 isoform,

HKa2c, in rabbit renal cortex using RT-PCR and 5' RACE. In contrast to the rat, the

alternative splice product of the rabbit yielded an mRNA with a longer open reading

frame. The deduced amino acid sequence of HKO2c encoded a protein that was 61 amino

acids longer at the N-terminus than the rabbit HKa2a protein. Moreover, there were three

short upstream open reading frames in the sequence that may play a role in translational

regulation. The extended N-terminal region of the putative HKa2c-encoded protein has a

casein kinase II phosphorylation motif and a protein kinase a motif. These sites, not

present in HKa2a, offer potential regulatory sites in the HKa2c. Northern and Western

analysis demonstrated the presence of both HKa2a and HKa2c mRNAs and protein

expression in the rabbit renal cortex. HKa2c protein has been localized to the apical

membrane of a small population of cells in the connecting segment, cortical collecting

duct and OMCDi (247). Studies in our laboratory by Deborah Zies demonstrated that

both isoforms are the product of a single gene and appear to be under control of one

promoter (unpublished observations). Studies of plant Ca2+-ATPase provided evidence

for a role of an N-terminal domain of a P-type ATPase in the regulation of pump

function. The N-terminal end of the protein was shown to function as an autoinhibitory

domain modulated by calmodulin (92). Whether the markedly different N-termini on the

rat HKa2b and rabbit HKa2c subunits impart properties differing from the enzymes

containing the HKo2a subunits remains to be determined.

It was not immediately clear whether the human ATP 1AL1 and the rat HKca2a

should be viewed as homologous to one another. Modyanov and his colleagues (89) took

the position that it was indeed homologous to rat HKa2a. The alternative interpretation

was that the sequence identity between ATP1AL1 and rat HKC2a was too low to justify

classifying the two genes as homologous. Indeed, the extent of identity between

homologous mammalian a subunits of P-type ATPases is usually much higher. For

example, the amino acid identity between the human, rat, and rabbit HKaI subunits

exceeds 97%. However, the cloning of the rabbit HKa2a cDNA has helped to clarify the

gene homology issue (35). Comparisons of the deduced primary sequences of either the

human ATP1AL1 or rat HKa2a to rabbit HKa2a yielded about 87% identity, and much of

the sequence diversity between these proteins clustered in the N-terminal 45 amino acids

(Figure 1-4). Distance analysis of the complete deduced sequences of the HKa2a and

ATPIAL1 subunits suggested that these proteins are more closely related to each other

than to either HKa( or any of the NaKa subunits (Figure 1-5). This supports the

argument in favor of assigning the three subunits as the products of homologous genes.

Moreover, the 5' mRNA sequences that give rise to the rat HKct2b and rabbit HKa2c

sequences appeared to be transcribed from genomic DNA contained in the introns

between the first two exons encoding the HKa2a mRNAs in both species. This similar

processing of transcripts provided added support for the argument that the rat and rabbit

genes are homologous.

To assess the role of the HKa2a subunit in K homeostasis in vivo, Meneton et al.

(154) generated a mouse model carrying a disruption of the HKa2 gene (cHKa) and

analyzed the ability of the animals to conserve K under normal and K-deplete


E-i I




1) I I
0411 1

c M


01 I




9 4 I -

(a (N
.H ., < a 3

-i *L L Lo


z 34
-4 04
Q ** I

z ***
0 *Z z
(9. *2


I *l 9
c; *(>I 9

<~i *0HI

~ > ***I
r 4

U). *I

Qz: .1

H >
cr *~ *



P-4 a4 E-4


F-4 I I I

:1. cli N0 .

33 33H CM Cd
s- -I -i a
4-) 41 -
-H -H X 33 3

,Q .Q CL 4- 4J1
Mf (13 Ei rO (0
4 4 4 i

--! r-I w-l 1


1* I1



z ::

M *

1 4.
6 .
c .o

aM 4J ) 4J

'r n %. 0o oo

. .



> > >

> H

< : ,',

> : C/

e 44



H d a


~.) -

8 Cd
.- a 0

U o
-U 5
6pj -a 5~ U

w 0
U. 0 "a-

.-CA I:P

0 c O U

; 4- 4

0 0 U
tltr- bt'

to to 0
2 ci

C 43 +3


- n H -
jo 003 c =
E- Ej.' c
to "'. EA(-1
U En :3

rat NaKa2

rat NaKa1

rat NaKa3

human NaKa3


rabbit HKa2a

human NaKa2

rabbit HKa ,

rat HKa human HKaI

rat HKo2a

Figure 1-5. Distance analysis of HKa and NaKa subunits. Protein sequences
were aligned with CLUSTALW (237). The resulting alignments were used by
PHYLIP to determine the relatedness among the subunits. TreeView (176) was
used to draw the final tree. The tree is not rooted. The forks indicate closest
primary sequence relationships without bias on the length of branches.

conditions. Critical transmembrane domain sequences required for transport activity

were replaced with the neomycin resistance gene through homologous recombination.

The replacement was confirmed by RT-PCR and sequence analysis. Homozygous

cHKaa- mice fed K -replete diets had no apparent difficulty with K+ homeostasis when

compared to wild type cHKa+'+ mice. Urinary K+ excretion levels were also similar

between wild type and homozygous mice fed K-replete diets. However, urinary

excretion of animals on K+-restricted diets had dropped over 100-fold, and to similar

levels, in both wild type and homozygous mice. The cHKad' mice suffered a loss in body

weight, plasma K+, and muscle K+ levels. Although fecal K+ excretion was decreased in

both sets of mice on K+-restricted diets, cHKad' mice had a fourfold greater reduction in

fecal K+ excretion. The results indicate that HKc2a plays an important role in K+

homeostasis during K deprivation, although the effect was primarily in the colon and not

in the kidney. It is likely that the gastric HKai subunit isoform in the kidney was able to

compensate for the loss of HKt2a activity.

Renal p Subunit Isoforms

The identity of the P subunits in renal H ,K+-ATPases is much less clear owing to

uncertainty about the identity of the 0 subunits(s) associated with the HKac2 subunits.

Three distinct P proteins have been implicated directly or indirectly in renal H+,K -

ATPases. These are the HKP subunit, the NaKP31 subunit, and the NaKi33 subunit.

Although it has not been rigorously tested in kidney tissue, probably the safest

assumption is that the HKp subunit is the mate to the HKoi subunit. These two proteins

have been conclusively shown to constitute the H+,K+-ATPase of the parietal cell (99).

Indeed, HKp subunit mRNA was readily detectable in the kidney by RT-PCR, Northern

analysis, and in situ hybridization (35, 36). The distribution of HK3 mRNA in the

connecting segment through the collecting duct was similar to the HKai distribution.

Transgenic animals were used to show that the HKp gene promoter directed reporter

gene expression in the collecting duct (32). The HKP subunit also supports H+,K -

ATPase activity in Xenopus laevis oocyte and tissue culture expression systems for

HKoC2a subunit pumps (12, 45, 87). However, the HKp subunit has never been shown to

associate with HKa2a in vivo.

Two p subunits, which were clearly the products of different genes, have been

detected in apparent association with the HKa2a subunit. Two groups independently

showed that antibodies directed against the HK(2a subunit co-immunoprecipitated the

NaKp1 subunit from both colon and kidney tissues (44, 132). The NaKp1 subunit was

reported to be capable of supporting H+,K+-ATPase activity for the HKa2a subunit in an

in vitro expression system (45, 197). Given the relatively high abundance of the NaKp i

subunit in the kidney and its ability to productively associate with HKO2a in vitro, it was

perhaps not surprising that NaKp1 was found in association with HKa2a subunit.

Sangan et al. (196) introduced the HKcp subunit as a potential candidate to serve

as a mate for the HKa2a subunit in the colon by using an immunoprecipitation strategy.

The HKcp subunit sequence was identical to the previously identified NaK[3 sequence

(142) and will be referred to as NaK[3 throughout this document. Sangan et al. (196)

noted the presence of a hybridization signal for NaK33 in a Northern analysis of mRNA

from kidney among several other tissues. NaKi3 protein expression was up-regulated in

the apical membrane of the distal colon in response to K+ depletion. In a later report

Sangan et al. (197) stably expressed rat HKa2a and NaKP3 in HEK293 cells. The

subunits formed a functional enzyme complex and yielded expression of both ouabain-

insensitive H+,K+-ATPase activity and 86Rb uptake.

Very recently, Geering et al. (83) investigated the association efficiency of human

HKa2a (ATP1AL1) with all known P subunit isoforms (HKP, NaKPi, NaKP2, NaKP3).

The authors expressed the subunits in Xenopus laevis oocytes and monitored their

resistance to cellular and proteolytic degradation and exit from the endoplasmic reticulum

(ER). Their results indicated that certain p subunits associated preferentially with

ATP1AL1, but that probably none of the known P subunits seemed to fulfill all of the

requirements to serve as a real mate to HKa2a. ATP 1AL1 was well protected against

cellular degradation by HKp and NaKp2, and these complexes were able to leave the ER.

However, NaKp i interaction was much less efficient and produced few stable enzyme

complexes that left the ER. Even though HKp and NaKp2 protected ATP1AL1 from

cellular degradation, these enzyme complexes were less resistant to trypsinolysis than

authentic 3/P complexes such as NaKca/NaKpi. The authors do point out the possibility

that ATP1AL1, when associated with p subunits, may be more sensitive to trypsin than

NaKai/NaKp3i complexes. NaKp3 stably associated with ATP 1 AL1, but the enzyme

complexes were moderately trypsin-sensitive and were impaired in their ability to leave

the ER. The authors concluded that the incomplete stabilization of NaKPil, the trypsin

sensitivity of complexes with HKP or NaK32, and the impaired ER exit of NaKP3

complexes suggest that none of the known 0 subunits is the true mate to the HKa2

subunits. However, at this time no other P subunit candidates have been identified.

H+ Versus Na.+ K+-ATPase

Experimental evidence exists which supports the theory that the colonic H+,K+-

ATPase can also act as a Na ,K -ATPase. Grishin and Caplan (88) analyzed HEK293

cells that stably expressed ATPIAL1 and HKp. This cell line was relatively resistant to

ouabain with a Ki of about 50 uM, whereas only 200 nM ouabain inhibited growth of

control cells. Flame photometry was used to determine the steady state levels of

intracellular Na+ levels. Intracellular Na+ content in cells stably expressing ATPlAL1

and HKp was half that of control cells. Furthermore, the rate of Na+ efflux in cells

expressing the two subunits was differentially sensitive to the presence of 1 utM ouabain

or 1 mM ouabain. However, 1 jM and 1 mM ouabain concentrations inhibited Na+

efflux to the same extent in control cells. The data strongly suggested that the

ATP1AL1/HKp enzyme complex was able to function as a Na+,K+-ATPase. Using an

inhibitory antibody to rat HKa2a, Codina et al. (46) were able to inhibit Na+-dependent

K+-ATPase activity in native rat distal colon membranes, supporting the position that the

colonic H+,K+-ATPase can act as a Na+,K+-ATPase in vivo.

Our laboratory cloned the unique HKa2c subunit isoform from rabbit renal cortex

(35). The encoded protein was greater than 90% identical to the published HKo2a

sequence. The only difference was a 61 amino acid extension at the N-terminus of

HKa2c. We had an interest in understanding the importance of the N-terminal extension

on the function of the enzyme as an H+,K+-ATPase. To analyze its importance, I set out

to express both rabbit HKoC2a and HKa2c subunit isoforms in a heterologous expression

system with the rabbit NaKP1 subunit, and later with the NaKP3 subunit. This


dissertation describes the expression of these subunits, and the observations that the

proteins likely did not form stable enzyme complexes.


Recombinant DNA Techniques

Plasmid DNA was purified by CsCl-ethidium bromide gradient

ultracentrifugation (81), by the Qiagen Maxi Prep kit, or by the Qiagen Mini Prep kit

(Qiagen, Valencia, CA). Restriction endonuclease reactions, ligation reactions, and

transformations were performed according to the recommendations of the suppliers (New

England Biolabs, Beverly, MA and Life Technologies, Grand Island, NY). DNA

fragments and PCR products were separated by agarose gel electrophoresis, eluted from

the agarose, and purified using the QIAquick gel extraction kit (Qiagen). Determination

of nucleotide sequences was conducted by the core facilities of the University of Florida

Interdisciplinary Center for Biotechnology Research (ICBR). Oligonucleotide synthesis

was performed by either Life Technologies or by ICBR.

Plasmid Constructions

bArg-36 Mutagenesis and Strain Construction

Plasmid pKAM14(b) (147) was used to construct mutations of the uncF(b) gene

at the codon for Arg-36. Mutations were generated using PCR mutagenesis essentially as

described by Ho et al. (101). The primers used to amplify a segment of DNA using

pKAM14(b) as a template are depicted in Table 2-1. In each case, a silent mutation

generating a MunI site was included in the mutagenic primers for screening purposes.

The PCR reactions were carried out in a Coy TempCycler II model for 30 cycles. The

- I I Y I I -

EC .

T3 3



>1 r-H

rA .1.1
0 E--



CE- i

n 0

0 0
< p
UU e
OU 1


1a ^
4 {aI


S E-4

0 u

0 u
< E-
u> P

(N 0 t1: H r- w C3) O0
c 0 0 OO 0 0 0 0
S (N (N H (N (' i)^ f (m )
u 0 u 00 U U U u
M I M 1 C Q 2Q Q M Q

a .-

Z 0


0 0

* 21

o U


.i2 ri


o *^


o *
g '5
-5 T


oo n
*a u

PCR products were digested with AatlI and PpuMI and then ligated to the 2.8 kilobase

pair (kbp) AatlI/PpuMI fragment of plasmid pKAM14(b). Plasmid DNA was prepared

from transformants, and recombinant plasmids were selected for the presence of a MunI

site. The mutations were identified through nucleotide sequencing.

Plasmids encoding the bArg-36 mutations in pKAM14(b), as well as control

plasmids pKAM14(b) and pBR322 (25), were transformed into E. coli strain KM2(Ab)

which has been deleted for the uncF(b) gene from the chromosome (146). The strains

were also transformed with plasmid pKAM16(lacI) to provide improved regulation of

the plasmid-encoded uncF(b) genes (147).

Episomal Expression Plasmid Constructions

The pREP4(Hygr) and pREP8(His') episomal expression vectors were purchased

from Invitrogen (Carlsbad, CA). The trans-acting factor Epstein Barr Virus Nuclear

Antigen 1 (EBNA-1) required for the Epstein Barr Virus' (EBV) origin of replication was

present in both plasmids. Each one of the rabbit HKa2a and HKa2c cDNAs was cloned in

the pREP4 plasmid. Rabbit HKP and NaKP31 cDNAs were each cloned in the pREP8


The construction of HKoX2a in pREP4 was as follows. The entire coding region of

HKc2a (nt 16-4079) had been cloned in the TOPO TA cloning vector (Invitrogen) by Dr.

W.G. Campbell, and the plasmid was named pHKa2a53. The plan was to move the

coding region of HKo2a, including 5' and 3' UTRs, into pREP4 as a 4.1 kb Notl/BamHI

fragment. Because of a BamHI site within the nucleotide sequence of HKa2a, conditions

which resulted in a partial digest of BamHI were utilized. Plasmid pHKa2a53 was

digested to completion with NotI and BssHII and partially with BamHI. BssHII was used

to digest the vector backbone to aid in gel purification of the 4.1 kbp Notl/BamHI

fragment. The fragments were separated by agarose gel electrophoresis, and the 4.1 kbp

fragment containing the entire coding region of HKa2a was gel purified and ligated into

pREP4. The final plasmid was named pTLC 19 (Figure 2-1). There were two start

codons present in the multiple cloning region of the TOPO TA cloning vector that were

carried with the HKa2a cDNA into pTLC 19. These start codons caused the protein to be

out of frame. Therefore, pTLC19 did not produce a functional HKc2a protein.

A plasmid containing the entire coding region of rabbit HKa2c did not exist in our

laboratory. Because HKa2a and HKt2c only differed at their 5' regions, I decided to use

RT-PCR to amplify the unique region of HKa2c, including an overlapping region with

HKa2a, and splice the fragment into pTLC 19 to create HKcX2c in pREP4. RT-PCR was

conducted using RNA isolated from rabbit distal colon with sense primer BC366 and

antisense primer BC265 (Table 2-2). These primers amplified a 656 bp fragment of

HKa2c with a 260 bp overlap with HKa2a cDNA sequence. The fragment was cloned in

TOPO TA cloning vector and named pHKa2c.9 (Figure 2-2). The HKa2c nucleotide

sequence was confirmed by nucleotide sequencing. Plasmid pHKa2c.9 was digested to

completion with AfllI and Notl. The 530 bp fragment was separated by agarose gel

electrophoresis, gel purified, and ligated into pTLC 19 to create pTLC24 (Figure 2-2).

This plasmid also had the two start codons from the TOPO TA cloning vector so it did

not produce a functional HKat2c protein.

The rabbit HKp cDNA had been previously amplified in our laboratory by RT-

PCR as two overlapping fragments. Each fragment was cloned into TOPO TA cloning

vector by Dr. W.G. Campbell. Plasmid pHKp5H contained the 5' region of HKp and

BamHI (2815)

EcoRI (4407)
/ NotI (4434)

EcoRI ('
BamHI (29'


BssHII (5927)

NotI (4553)

Figure 2-1. Rabbit HKc2a, cDNA construction in pREP4. Plasmid
pHKa2a53 was digested to completion with NotI and BssHII and
partially with BamHI. The 4.1 kb Notl/BamHI fragment was gel
purified and ligated into pREP4 to create pTLC 19.


0 z
0 -- -

0 1 0m ?n u





O H U 0
Sp 0 U
0 H 0 U
H < HO
U H 0 C.
U 0 U
H U < u
H U H 0
0 0 0 0
U < 0 0
H- H H *<
H 0 U H
U H U <
< H H H
H H 0 U
< U 0 <
H 0 H H
,0 U .< <

.C U U U fr U U U iA
cQ C m < ( c a <



EcoRI (329)
BamHI (295) N

BamHI (407

NotI (1027)

Not (4853)

Figure 2-2. Rabbit HKa,, cDNA construction in pREP4. Plasmid
pHKa2c.9 was digested to completion with NotI and AflllI. The 0.5 kb
Notl/AflllI fragment was gel purified and ligated into pTLC 19 to create

pHKp3C contained the 3' region of HKp. The plan was to construct the full-length

cDNA of HKp in pUC18 and clone the full-length cDNA in pREP8. Plasmid pHKp3C

was digested to completion with BamHI and Pstl. The 708 bp fragment was isolated by

agarose gel electrophoresis and ligated in pUC 18. The plasmid was called pTLC21. The

BamHI site in the multiple cloning region of pHKp5H was destroyed by digesting the

plasmid with BamHI, filling in the overhanging ends using Klenow, and religating the

plasmid to create pTLC20. Plasmid pTLC20 was digested to completion with PstI and

partially with HindIII. The resulting 734 bp fragment containing the 5' region of HKp

was isolated and ligated in pTLC21 to generate the full-length HKP cDNA. This plasmid

was named pTLC22. Plasmid pTLC22 was digested with KpnI and BamHI, the 1.4 kb

band containing the full-length HKI was isolated and ligated into pREP8 to create

pTLC23 (Figure 2-3).

The rabbit NaKpi cDNA was amplified by RT-PCR using primers with unique

restriction sites engineered to aid in the cloning in pREP8. This is discussed in more

detail in Chapter 4. The unique restriction sites KpnI and Nhel were engineered at the 5'

and 3' ends of NaKPI, respectively. The fragment was ligated in TOPO TA cloning

vector and named p3Al. QuikChange Mutagenesis (Stratagene, La Jolla, CA) was

conducted to correct the mutations at the 3' end of NaKp1 using sense primer BC467 and

antisense primer BC466 (Table 2-3). The plasmid with the corrected 3' end of rabbit

NaKpi was named pNaKpi. pNaKpi was digested with KpnI and Nhel, the 0.9 kbp

fragment was isolated and ligated in pREP8 to create pTLC25 (Figure 2-4).

PstI (1115)

BamHI (,

nI (405)
EcoRI (444)
I NotI (457)
HindIII (837)

PstI (1139)

SNotI (1820)
EcoRI (1829)
BamHI (1862)

pnI (1807)

Figure 2-3. Rabbit HKP cDNA construction in pREP8. Plasmid
pTLC22 was digested to completion with KpnI and BamHI. The
1.4 kb KpnI/BamHI fragment was gel purified and ligated into
pREP8 to create pTLC23.

*as M .8 a'IE

eona ,.co
mI 1- 03

U .,; &% 0

0 U

0 H

U u








0 S
i O

0 I
< p

U o '0 (00 0o 'O vo


KpnI (1229)
NotI (1271)

KpnI (1362)

Figure 2-4. Rabbit NaKp, cDNA construction in pREP8. Plasmid
pNaKp, was digested to completion with KpnI and Nhel. The 0.9 kb
KpnI/Nhel fragment was gel purified and ligated into pREP8 to create

BamHI (4

Stable/Transient Plasmid Construction

The plasmid used for the stable and transient vector construction was pBudCE4

(Invitrogen). This plasmid was unique in that it contained two promoters for cloning two

cDNAs, as well as allowing the researcher the ability to clone each cDNA in frame with

epitope tags present after each promoter. The HKi2a and HKa2c cDNAs were cloned

behind the human EF-la promoter and in frame with the V5 epitope and polyhistidine

(6xHis) metal-binding tag. The NaKi cDNA was cloned behind the human CMV

promoter, but not in frame with the epitope tags since a good commercial anti-rabbit

NaKP1 antibody was available.

To clone HKa2a and HKa2c in frame with the V5 epitope and 6xHis tag, a BstBI

site was engineered at the stop codon of each of the HKca2 cDNAs. The BstBI site was

engineered using QuikChange Mutagenesis. pHKa2a53 was digested with AatII and

KpnI and the 2 kbp fragment containing the 3' end of HKa2a was ligated in pUC 18 to

create pTLC29. Primer pairs TC 12 and TC 13 were used for the QuikChange

Mutagenesis to create pTLC30.1 (Table 2-3). The BstBI site in each primer is

underlined. The mutations were confirmed by nucleotide sequencing. The DNA

fragment with the BstBI site was moved back to pHKa2a53 as an Aatll/KpnI fragment to

regenerate the full-length HKa2a cDNA with the BstBI site. The plasmid was named

pTLC31. Plasmid pTLC31 was digested with NotI and BstBI, the 3.1 kbp band was gel

purified and cloned in pBudCE4 to generate pTLC34. The HKac2a cDNA is in frame with

the V5 and polyhistidine epitope tags.

An early hypothesis as to the failure of the episomal expression was the large 5'

UTR of HKcc2c. This was later determined to be due to the start codons carried over from

the TOPO TA cloning vector. Before cloning HKa2c in pBudCE4, the 5' UTR was made

shorter through cassette mutagenesis. pHKa2c.9 was digested with Xbal and Hpal. The

Hpal site was within the 5' UTR of HKa2c, and the Xbal site was within the multiple

cloning region of TOPO TA vector. Primers TC 18 and TC 19 were designed with Xbal

and Hpal complementary ends (Table 2-3). The cassette mutagenesis was confirmed by

a loss of an Nsil site. The plasmid was called pTLC32. To create HKa2c in pBudCE4,

pTLC32 was digested with EcoRV and Notd, the 300 bp band was gel purified and ligated

into pTLC34. The resulting plasmid was named pTLC35 (Figure 2-5). The HKa2c

cDNA is in frame with the V5 and polyhistidine epitope tags..

I determined that two ATG start codons were present on the lower strand in the

multiple cloning region of the TOPO TA cloning vector. This would have caused major

translational problems. The two start sites carried over during the cloning of HKa2a in

pBudCE4. However, the sites were eliminated in HKc2c during the shortening of the 5'

UTR. To correct the problem in HKa2a, pTLC34 was digested to completion with NotI

and partially with EcoRI. Cassette mutagenesis was conducted with primer pairs TC24

and TC25 with proper overhangs. A Hpal site was engineered for screening purposes

(Table 2-3). The plasmid was called pTLC46 (Figure 2-6).

To clone NaKI31 into pBudCE4, pNaKp i was digested with BamHI and XbaI, the

0.9 kbp fragment was gel purified and cloned into pBudCE4 to create pTLC33 (Figure 2-

7). To clone HKa2a and HKa2c into pTLC33, pTLC46 and pTLC35 were each digested

with NotI and BstBI, the 3.1 and 3.3 kbp fragments were gel purified, respectively and

cloned into pTLC33 which was digested with BstBI and partially with NotI to create

pTLC47(HKCt2a/NaKpI) (Figure 2-8A) and pTLC37(HKc2c/NaKpI) (Figure 2-8B).

EcoRV (573)

EcoRI (329)
BamHI (295-

BstBI (3607)-

EcoRV (3360)--
Nod (3064)---1 M 1

Figure 2-5. Rabbit HKao cDNA construction in pBudCE4. Plasmid
pTLC32 was digested to completion with EcoRV and Nod. The 0.3 kb
EcoRV/NotI fragment was gel purified and ligated into pTLC34 to create
pTLC35. Figure was adapted from Invitrogen catalog.

BstBI (3407)-

EcoRV (3160)---- 7.7 kb
EcoRI (3106)--- 7.
H al (3096)- EcoRI (798)

Figure 2-6. Rabbit HKa,, cDNA construction in pBudCE4. Plasmid
pTLC34 was digested to completion with Nod and partially with EcoRI.
Primer pairs TC24 and TC25 were ligated into pTLC34 to create
pTLC35. Figure was adapted from Invitrogen catalog.

---XbaI (707)


-- amHI (1613)

Figure 2-7. Rabbit NaKp, cDNA construction in pBudCE4. Plasmid
pNaK1, was digested to completion with BamHI and Xbal. The 0.9 kb
fragment was gel purified and ligated into pBudCE4 to create pTLC33.
Figure was adapted from Invitrogen catalog.

BstBl (459z

EcoRV (434'
NotI (405

*Xbal (707)

BamHI (1613)

Xbal (707)

BamHI (1613)

Abf I

Figure 2-8. Plasmids pTLC47 and pTLC37 expression constructs. A, Plasmid
pTLC47 encodes the HKa,. cDNA ,in frame with V5 epitope tag, and NaKP, cDNA.
B, Plasmid pTLC37 encodes the HKa, cDNA, in frame with V5 epitope tag, and
NaKp, cDNA. Figures were adapted from Invitrogen catalog.

F F0 ATP Synthase Methods

Bacterial Growth Conditions

Growth media were either Luria broth supplemented with 0.2% (w/v) glucose

(LBG) or minimal medium consisting of A salts supplemented with 0.2% (w/v) succinate

or 5 mM glucose. Isopropyl-1-thio-P-D-galactoside (IPTG) (0.4-40 utg/ml), ampicillin

(Ap) (100 jpg/ml), and chloramphenicol (Cm) (30 j.g/ml) were added to media as

appropriate. Growth of aerobic liquid cultures was performed with constant agitation and

monitored using a Klett-Summerson colorimeter. Anaerobic conditions were achieved

using a BBL "Gas Pak" (Beckton Dickinson Microbiolobgy Systems, Cockeysville, MD)

anaerobic jar which had been flushed with N2 prior to establishing the anaerobic

atomsphere. Unless otherwise indicated, incubations of cultures were at 37C.

Preparation of Cell Fractions

Inverted membrane vesicles were prepared from cultures grown in 500 ml LBG-

Ap-Cm medium containing IPTG (40 jig/ml). Cells were harvested and washed in

STEM buffer (0.1 M N-tris[hydroxymethyl]-methyl-2-aminoethanesulfonic acid, 20 mM

magnesium acetate, 0.25 M sucrose, 0.25 mM

[ethylenebis(oxyethylenenitrilo)]tetraacetic acid, 40 mM e-amino-n-caproic acid, pH

7.0), and then resuspended in STEM buffer containing 5 mMp-aminobenzamidine, 1

mM dithioerythritol, 1 mM PMSF, and 10 ug/ml deoxyribonuclease. Cells were

disrupted using a French Press at 14,000 psi. Unbroken cells and debris were pelleted by

two successive centrifugations (8,000 x g, 10 min), and the membranes were collected by

ultracentrifugation (150,000 x g, 1.5 h) in a Beckman Ti-70.1 rotor. Membranes were

washed in B2 buffer (50 mM Tris-HCl, 5 mM MgSO4, 1 mM dithioerythritol, 6 mM p-

aminobenzamidine, 10% (v/v) glycerol, pH 7.5). The final membrane pellet was

resuspended in B2 buffer and stored at 40C. Membrane vesicles stripped of Fi were

prepared essentially as described previously by Hartzog and Cain (93).

Determination of Protein Concentration

Total membrane protein concentration was determined by a modified Lowry

procedure (144). Buffer C was made by adding 100 parts of Buffer A (2% Na2CO3, 0.4%

NaOH, 0.16% sodium potassium tartrate, 1% sodium dodecyl sulfate (SDS)) to one part

Buffer B (4% CuSO4). Aliquots from prepared membranes in a total volume of 1 ml

were incubated in 3 ml of freshly made Buffer C. After a 15 min incubation at room

temperature, 0.3 ml Folin-Ciocalteu phenol reagent, freshly diluted 1:1 with distilled

water, was added to the mixture. The solution was incubated for 45 min at room

temperature, and OD660 was recorded with a Pharmacia spectrophotometer. Each sample

was analyzed in triplicate. Bovine serum albumin with a concentration range of 0 to 100

jtg of protein was used to generate a standard curve. A correlation coefficient of 0.99

was routinely obtained for the standard curve. Protein concentrations were calculated

from the linear range of the standard curve by linear regression.

ATP Hydrolysis Assay

ATP hydrolysis activity of membrane fractions was assayed by the acid

molybdate method (211). This assay measures the release of inorganic phosphate from

ATP. The reaction consisted of 60 tg membrane protein in a total volume of 4 ml buffer

(50 mM Tris-HCl, 1 mM MgCl2, pH 9.1) incubated at 370C. A 0.435 ml aliquot was

removed from the reaction before the addition of ATP so that the background phosphate

could be subtracted from the measurement. 150 mM ATP (80 gL) in 25 mM Tris-HCl

(pH 7.5) was added to start the reaction. 0.435 ml aliquots of the reaction mixture were

taken at 2, 5, 7, 10, and 15 min, added to 2 ml acid/molybdate solution and were stored

on ice. The acid/molybdate solution stopped the reaction and prevented further ATP

hydrolysis from occurring. Inorganic phosphate was developed using 0.1 ml of a 1:10

dilution of Eikonogen solution. After incubation for 30 min at room temperature, the

absorbance of each sample was determined at 700 nm. A phosphate standard curve was

used in the concentration range of 0 to 0.6 utmol. The specific activity was expressed as

pmol of ATP hydrolyzed/min/mg membrane protein. Membranes were assayed for

determinations of linearity with respect to both time and enzyme concentration.

The ATP hydrolysis data described in Table 3-3 was determined by Dr. C.

Ketchum in the laboratory of Dr. R. Nakamoto (University of Virginia). ATP hydrolysis

rates ([tmol/min/mg) were measured at 30C in a buffer containing 25 mM HEPES-KOH

(pH 7.5), 200 mM KCI, 8 mM MgSO4 (3.1 mM free Mg2+), 10 mM glucose, 1 mM

phosphoenolpyruvate, 5 mM ATP, 5 gM carbonyl cyanide-m-chlorophenylhydrazone

(CCCP), 50 pg/ml pyruvate kinase. The reactions were started with the addition of 0.10

to 0.15 mg/ml E. coli membranes. Time points were taken every 2 min up to 8 min and

stopped by the addition of 5% SDS. Inorganic phosphate was determined by the method

of Taussky and Shorr (234).

Assays of F F0 ATP Synthase Activity

Membrane energization of vesicle preparations was detected by the fluorescence

quenching of 9-amino-6-chloro-2-methoxyacridine (ACMA) (9). 250 gg of membrane

protein was assayed in buffer (50 mm MOPS, 10 mM MgC12, pH 7.3). ACMA was

added to the reaction mixture at a final concentration of 1 gM. ATP was added to a final

concentration of 0.4 mM. Fluorescence was recorded with an excitation at 410 nm and

an emission at 490 nm. Treatment of membrane fractions with 50 pM

dicyclohexylcarbodiimide (DCCD) was performed by pre-incubating the membrane

fractions for 15 min at 370C.

Rates of ApH+-driven ATP synthesis in Table 3-3 were determined by Dr. C.

Ketchum from the laboratory of Dr. R. Nakamoto (University of Virginia) as described

by Al-Shawi et al. (7). ATP synthesis rates (pmol/min/mg) were measured at 300C with

vigorous shaking in a buffer containing 25 mM HEPES-KOH, 200 mM KC1, 5 mM

MgSO4 (3.0 mM free Mg2+), 10 mM glucose, 1 mM ADP, 10 mM [32P]Pi, 50 units/ml

hexokinase, and 0.12 mg/ml E. coli membranes at pH 7.5. The reactions were started

with the addition of 2 mM NADH. Time points were taken every 2 min up to 8 min and

stopped with ice cold 10 mM H2SO4. Inorganic phosphate was preciptated by the method

of Sugino and Miyoshi (226), and the radioactivity remaining in the supernatant was

determined by Cherenkov counting. Control experiments to determine ApH+-

independent incorporation of [32P]Pi into the nonprecipitable material was determined in

the same conditions but with the presence of 5 pM CCCP and subtracted from the rates

obtained in the absence of CCCP.

Immunoblot Analysis

Membrane fractions were suspended in buffer (62.5 mM Tris-HCI, 10% (v/v)

glycerol, 5% (v/v) 2-mercaptoethanol, 3% (w/v) SDS, pH 6.8), incubated for 5 min at

950C, and subjected to polyacrylamide gel electrophoresis on 15% Tris-Glycine BioRad

Ready Gels. Proteins were electroblotted using transfer buffer (25 mM Tris-Base, 192

mM glycine, 20% (v/v) methanol, pH 8.3) onto nitrocellulose (0.22 pm). The blot was

blocked for 1 h at room temperature or overnight at 4C with 5% non-fat dry milk

(BioRad) in 10 mM Tris-HCl, 150 mM NaCI, pH 7.2 (TBS). Polyclonal antibodies

against SDS-denatured b subunit (52, 54) were kindly provided by Drs. K. Altendorf and

G. Deckers-Hebesteit (Univeristat Osnabriick). The primary antibody incubation was

performed using a 1:5,000 dilution of b subunit-specific antibodies in 2.5% non-fat dry

milk for 1 h. Secondary antibody incubation was conducted with horseradish peroxidase-

linked donkey anti-rabbit antibodies (Amersham Pharmacia Biotech, Piscataway, NJ) at a

dilution of 1:50,000 for 1 h. After each antibody incubation, blots were washed three

times in TBS, 0.1% (v/v) Tween-20 (TBS-T). Antibody binding was detected by

chemiluminescence using the ECL system. Signals were visualized on Hyperfilm-ECL

using a Kodak X-OMAT instrument. Strength of signals was quantified with a Kodak

image digitizing system (DC 120).


3' RACE reactions were performed using the Marathon cDNA Amplification Kit

(Clontech, Palo Alto, CA) according to manufacturer's instructions with the following

modifications. The reverse transcriptase reactions were conducted with 5 pg total RNA

from rabbit renal medulla. The PCR reactions were performed with degenerate primer

BC424 and anchor primer API (Table 2-2). The thermal conditions for the PCR reaction

were 940C X 1 min presoak followed by 5 cycles of 940C X 30 s denaturation and 72C

X 3 min extension, 5 cycles of 94C X 30 denaturation and 700C X 3 min extension, 25

cycles of 94C X 30 s denaturation and 680C extension with a final 3 min extension at

680C. Fragments were cloned in TOPO TA cloning vector for sequence analysis.


5' RACE reactions were performed using the Marathon cDNA Amplification Kit

according to manufacturer's instructions with the following modifications. The reverse

transcriptase reactions were carried out with 5 utg total RNA from rabbit renal medulla.

Gene-specific primer BC464 was used to create the cDNA library. The PCR reactions

were performed with gene-specific primer BC448 and anchor primer API (Table 2-2).

The thermal conditions for the PCR were 94C X Imin presoak followed by 25 cycles of

940C X 30 s denaturation, 650C X 1 min annealing, and 720C x 2 min extension, with a

final 10 min extension at 72C. One jtl of the PCR reaction was used for nested PCR

with BC465 and nested anchor primer AP2 (Table 2-2). The thermal conditions for the

nested PCR were 94C X 1 min presoak followed by 25 cycles of 94C X 30 s

denaturation, 56C X 1 min annealing, and 720C X 1 min extension, with a final 7 min

extension at 720C. Fragments were cloned in TOPO TA cloning vector for sequence


Tissue Culture

Three different cell lines were used throughout this dissertation. HEK293 cells

are a permanent line of primary human embryonal kidney cells transformed by human

adenovirus type 5 DNA. HEK293cl8 cells constitutively express EBNA-1 gene product

and demonstrate an increase in transfection efficiency of episomal expression vectors

over the parent HEK293 cell line. This cell line carries neomycin resistance for the

EBNA-1 gene. These cells were a kind gift of Dr. M. Kilberg (University of Florida).

COS-1 cells are a SV40 transformed kidney cell line from African green monkey. All

cell lines were grown in Dulbecco's modified Eagle medium (DMEM) supplemented

with 10% fetal bovine serum (DMEM-10). All media was filter sterilized with a 0.22 pm

cellulose acetate bottle top filter. Cells were maintained in tissue culture flasks in 5%

CO2 atmosphere in a humidified incubator at 370C.

Episomal Transfection

HEK293cl8 cells were grown to 80% confluency in DMEM-10, 250 mg/L G418

in 60 mm dishes. On the day of transfection, 5 pg DNA was diluted to a final volume of

150 ul in serum-free DMEM. Twenty ul Superfect transfection reagent (Qiagen) was

added, and the reaction mixture was allowed to incubate at room temperature for 10 min.

Cells were rinsed once with PBS, and the PBS was aspirated off the dish. The DNA-

Superfect mixture was mixed with 1 ml DMEM-10 and was added dropwise to the cells.

Cells were kept in a humidified incubator at 370C in 5% CO2 atmosphere for 3 h. After

the 3 h incubation, cells were washed four times with PBS, and fresh DMEM-10 was

added. After cells had grown for 24 h, DMEM-10 with appropriate antibiotic (10 mM

histidinol; 500 gg/ml hygromycin) was added. Selection was complete when cells on

control plate had died.

Stable Transfection

HEK293 cells were grown to 80% confluence in 100 mm dishes. On the day of

transfection, 10 ul of FuGENE 6 Transfection Reagent (Roche, Indianapolis, IN) was

added to 150 pl serum-free DMEM. Linearized plasmid (3.3 jig) was added to the

mixture. After a 20 min incubation at room temperature, the DNA-FuGENE mixture was

added dropwise to the cells in dishes, and the cells were incubated overnight in the 37C

incubator. Twenty-four hours after transfection, the media was removed and replaced

with DMEM-10 containing 100 [ig/ml Zeocin (Invitrogen). Clones began to appear

about two weeks post transfection. Single cell clones were isolated with cloning rings

and allowed to grow in 24-well plates until cells reached confluency. The clones were

expanded into T-25 flasks and ultimately into T-75 flasks for membrane preparation and

Western analysis. Clones never appeared on control plates.

Transient Transfection

COS-1 cells were grown to 70% confluence in 150 mm dishes. On the day of

transfection, 54 ptl of FuGENE was diluted into 1.8 ml serum-free DMEM. Eighteen .tg

DNA was added to the tube and was incubated for 20 min at room temperature. After the

incubation, the DNA-FuGENE mixture was added dropwise to the cells. The cells were

placed in the 37C incubator for 24 h. After the 24 h incubation, the cells were split into

three 150 mm dishes and were grown for another 24 h before membrane preparation and


H+,K+-ATPase Biochemical Assays

Plasma Membrane Preparation

Six 150 mm dishes were normally used for a large-scale plasma membrane

preparation. Cells were washed two times with 10 ml PBS. The cells were then scraped

from dish with 5 ml PBS. This cell solution was added to the next plate, and the

procedure was repeated until the last plate had been scraped. This process was repeated

two more times resulting in a total volume of 15 ml. Cells were pelleted by

centrifugation (1,000 x g, 10 min). The cell pellet was resuspended in 2 ml buffer (10

mM Tris, pH 7.5, 1 mM PMSF) and was kept on ice for 10 min. Cells were broken by

homogenization with 25 strokes. Two ml homogenization buffer (0.5 M sucrose, 10 mM

Tris, pH 7.5, 1 mM PMSF) was added and 25 more strokes were performed. Unbroken

cells and nuclei were pelleted by centrifugation (1,000 x g, 10 min). Depending on the

size of the pellet, homogenization of the pellet was repeated. Normally, homgenization

was repeated two more times. The supernatant was spun at 12,000 x g for 15 min to

pellet mitochondria and other organelles. The resulting supernatant was spun at 100,000

x g for 90 min to recover low density membranes. The final membrane pellet was

resuspended in buffer (250 mM sucrose, 5 mM Tris, pH 7.4).

Cell Lysate Preparation

Cells in a T-75 flask were washed once with PBS. Cells were then scraped into 1

ml PBS and pelleted at 1500 x g for 5 min. The cell pellet was ressupended in 50 ml Cell

Lysis Buffer (50 mM Tris, 150 mM NaCl, 1% Nonidet P-40, pH 7.8). The cell

suspension was incubated at 370C for 10 min to completely lyse the cells. The cell lysate

was centrifuged at 10,000 x g for 10 min at room temperature to pellet nuclei. The cell

lysate was transferred to a new tube.

Bicinchoninic Acid Assay

The bicinchoninic acid assay was used to determine protein concentration as

described previously (215). Solution A contained 1% 4,4'-dicarboxy-2,2'-biquinoline,

2% Na2CO3-H20, 0.16% sodium tartrate, 0.4% NaOH, and 0.95% NaHCO3. Solution B

was 4% CuSO4*5H20. The standard working solution was made fresh by mixing 100

volumes of solution A to 2 volumes of solution B with a final concentration of 1% ultra-

pure SDS. Samples were brought to a total volume of 100 pl, and 2 ml of the standard

working solution was added to each tube. Samples were incubated in a 370C water bath

for 30 min. After the 30 min incubation, the tubes were kept at room temperature for 10

min. The OD562 was recorded with a Pharmacia spectrophotometer. Each sample was

analyzed in duplicate. Bovine serum albumin with a concentration range of 0 to 80 pg of

protein was used to generate a standard curve. A correlation coefficient of 0.99 was

routinely obtained for the standard curve. Protein concentrations were calculated from

the linear range of the standard curve by linear regression.

ATP Hydrolysis Microassay

ATP hydrolysis activity of plasma membrane fractions was assayed by a sensitive

acid molybdate method that can detect nanomolar concentrations of phosphate in a

solution (245). The reaction consisted of 50 pg plasma membrane protein in a total

volume of 1 ml buffer (40 mM Tris-HCl, 3 mM MgC12, +/- 15 mK KC1, pH 7.4)

incubated at 37C. A 50 pL aliquot was removed from the reaction before the addition of

ATP so that the background phosphate could be subtracted from the measurement.

Twenty pl of 150 mM ATP (Tris salt or SigmaUltra disodium salt ATP) in 25 mM Tris-

HCI (pH 7.5) was added to start the reaction. Fifty pil aliquots of the reaction mixture

were taken at 2, 5, 7, 10, and 12 min and added to a tube containing 200 pl 1.75%

ammonium heptamolybdate'4H20 in 6.3 N H2SO4 and 950 pLL H20. The acid/molybdate

solution stopped the reaction and prevented further ATP hydrolysis from occurring.

Inorganic phosphate was developed using 200 pl of a malachite green solution (0.035%

malachite green, 0.35% polyvinyl alcohol). After incubation for 30 min at room

temperature, the absorbance of each sample was determined at 610 nm. A phosphate

standard curve was used in the concentration range of 0 to 20 nmol. Each reaction,

including the standard curve, was done in duplicate. The specific activity was expressed

as pmol of ATP hydrolyzed/h/mg membrane protein.

Northern Analysis

Total RNA was isolated from tissue culture cells with the RNeasy kit (Qiagen).

Northern blots were conducted according to Davis et al. (50). Total RNA (10 tg) was

separated by electrophoresis on a 1% agarose, 0.22 M formaldehyde denaturing gel. The

gel was soaked in 50 mM NaOH for 30 min to depurinate RNA. The gel was neutralized

in 100 mM Tris, pH 7 for 30 min. Finally, the gel was equilibrated in 10X SSC (1.5 M

NaCl, 150 mM sodium citrate, pH 7) for 40 min. Capillary transfer to a nylon membrane

(Amersham) was performed overnight in 10X SSC. RNA was immobilized on the

membrane by baking 1 h at 800C in a vacuum oven. 32P-labeled probes were prepared by

random primer extension with 25 ng DNA according to the RTS RadPrime DNA

Labeling System (Life Technologies). Membranes were pre-hybridized for at least 20

min and hybridized with labeled probe for 18 h in hybridization solution (250 mM

Na2HPO4, 10 mg/ml BSA, 7% SDS) at 650C. The membranes were washed in buffer (20

mM Na2HPO4, 1% SDS, pH 7.2) at 650C. After washing, membranes were exposed to

Kodak BioMax MR film. Exposure was normally 18 h. Glyceraldehyde-3-phosphate

dehydrogenase (GAP3DH) was used as a loading control.

Immunoblot Analysis

Membrane fractions were resuspended in buffer (62.5 mM Tris-HCl, 10% (v/v)

glycerol, 5% (v/v) 2-mercaptoethanol, 3% (w/v) SDS, pH 6.8) and were incubated for 5

min at 950C before SDS-PAGE on 10% Tris-glycine gels. Proteins were electroblotted

using transfer buffer (25 mM Tris-Base, 192 mM glycine, 20% (v/v) methanol, pH 8.3)

onto nitrocellulose (0.22 gm). All antibody incubations were performed at room

temperature unless otherwise indicated. Antibody binding was detected by

chemiluminescence using the ECL system. Signals were visualized on Hyperfilm-ECL

using a Kodak X-OMAT instrument.

Anti-V5 Antibodies

Membrane fractions were separated by SDS-PAGE using 10% Tris-glycine gels.

After the proteins were transferred to nitrocellulose, the blot was blocked for 1 h at room

temperature or overnight at 4C with 5% non-fat dry milk in TBS. The primary antibody

incubation was performed using a 1:2,500 dilution of V5-specific monoclonal antibodies

(Invitrogen) in 2.5% non-fat dry milk for 1 h. Secondary antibody incubation was

conducted with horseradish peroxidase-linked sheep anti-mouse antibodies (Amersham

Pharmacia Biotech) at a dilution of 1:10,000 in 2.5% non-fat dry milk for 1 h. After each

antibody incubation, blots were washed three times for 5 min in TBS-T.

Anti-Rabbit NaKI31 Subunit Antibodies

Membrane fractions were separated by SDS-PAGE on 10% Tris-glycine gels.

After proteins were transferred to nitrocellulose, the blot was blocked for I h at room

temperature or overnight at 4C with 5% non-fat dry milk in TBS. The primary antibody

incubation was performed using 0.5 pg/ml rabbit NaKI3 subunit specific mouse

monoclonal antibodies (Upstate Biotechnology, Lake Placid, NY) in 2.5% non-fat dry

milk overnight at 4C. Secondary antibody incubation was conducted with horseradish

peroxidase-linked sheep anti-mouse antibodies at a dilution of 1:10,000 in 2.5% non-fat

dry milk for I h. After each antibody incubation, blots were washed three times for 5

min in TBS-T.

Anti-myc Antibodies

Membrane fractions were separated by SDS-PAGE on 10% Tris-glycine gels.

After proteins were transferred to nitrocellulose, the blot was blocked for I h at room

temperature or overnight at 4C with 5% non-fat dry milk in TBS. The primary antibody

incubation was performed using 2.5 jig/ml myc-specific mouse monoclonal antibodies

(Oncogene Research Products, Boston, MA) in 2.5% non-fat dry milk for I h. Secondary

antibody incubation was conducted with horseradish peroxidase-linked sheep anti-mouse

antibodies at a dilution of 1:10,000 in 2.5% non-fat dry milk for 1 h. After each antibody

incubation, blots were washed three times for 5 min in TBS-T.

Immunofluorescence Microscopy

Glass coverslips (Coming 22 x 22 mm) were sterilized by placing them in them in

a glass petri dish and autoclaving them. The steam cycle of the autoclave was not used,

as this caused the coverslips to stick to each other. A sterile Pasteur pipette connected to

a vacuum line was used to pick up the coverslip and place it in a 6-well plate. Cells were

added to the well and were allowed to adhere for 24 h. The cells were transfected using

FuGENE as described in an earlier section. Briefly, plasmid DNA (1 jIg) was diluted in

100 pl DMEM containing 3 pl FuGENE. The complex was allowed to form for 20 min

and then added to the wells.

After a 24 h incubation time for protein expression, the wells were rinsed three

times with PBS. The cells were fixed for 20 min in 3% paraformaldehyde. After fixation

was complete, the wells were washed three times (5 min each) with PBS. Free aldehydes

resulting from the paraformaldehyde incubation can have fluorescent properties, so cells

were incubated for 30 min with 50 mM L-glycine to quench free aldehydes. Wells were

washed three times (5 min each) with PBS. The V5-epitope tag was attached to the C-

terminus of the HKO2 subunits, which is inside the cell. Hence, the cells were

permeabilized for 15 min in 0.1% Triton-X- 100, 0.1% non-fat dry milk. Cells were

blocked overnight at 4C in 20% normal goat serum, 0.1% Triton-X-100.

A humid box was used for the primary and secondary antibody incubations. A

soaked paper towel was placed in the bottom of a small Tupperware container. The paper

towel was covered with Parafilm. V5-specific monoclonal antibodies were diluted

1:5000 in 20% normal goat serum, 0.1% Triton-X-100. 100 tl of the diluted antibody

solution was placed on the parafilm in the Tupperware container. The coverslips were

carefully picked up by the edge with tweezers. The edge of the coverslips were blotted

on a Kimwipe and placed with the cell-side down atop the drops of antibody on the

parafilm. The lid was placed on the container, and the cells were incubated for 1 h at

room temperature. Using tweezers, the coverslips were placed cell-side up back into the

6-well tray and washed three times with 0.1% Triton-X-100, 0.1% non-fat dry milk.

Polyclonal goat anti-mouse IgG antibodies conjugated to Texas Red (Molecular Probes,

Eugene, OR) were diluted to 1 ig/ml in 20% normal goat serum, 0.1% Triton-X-100.

Using the same procedure as the primary antibody incubation, the coverslips were placed

cell-side down atop the drops of antibody and were incubated for 1 h at room

temperature. From this point forward, the coverslips were shielded from light to avoid

bleaching the fluorescence. The coverslips were placed cell-side up back into the 6-well

tray and washed three times with 0.1% Triton-X- 100, 0.1% non-fat dry milk. The edges

of the coverslips were blotted on a Kimwipe and placed cell-side down on a drop of

Fluoromount G mounting medium (Fisher Scientific) on glass slides. The slides were


dried for 1 h in the dark. Once dried, the edges of the coverslip were sealed with a layer

of nail polish. Confocal images were generated on a BioRad model MRC-1024 confocal

laser scanning system. The xy sections were generated with a 1.5 gm motor step.