Coordination of MRNA 3' end formation and nuclear export by a nuclear poly(A)-Binding protein


Material Information

Coordination of MRNA 3' end formation and nuclear export by a nuclear poly(A)-Binding protein
Physical Description:
viii, 167 leaves : ill. ; 29 cm.
Nykamp, Keith Robert, 1974-
Publication Date:


Subjects / Keywords:
RNA, Messenger -- genetics   ( mesh )
RNA Polymerase II -- genetics   ( mesh )
RNA-Binding Proteins -- genetics   ( mesh )
Heterogeneous-Nuclear Ribonucleoproteins -- genetics   ( mesh )
Polyadenylation -- genetics   ( mesh )
Yeasts   ( mesh )
Research   ( mesh )
Department of Molecular Genetics and Microbiology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Medicine -- Department of Molecular Genetics and Microbiology -- UF   ( mesh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D)--University of Florida, 2003.
Bibliography: leaves 138-166.
Additional Physical Form:
Also available online.
Statement of Responsibility:
by Keith Robert Nykamp.
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 - 030585692
oclc - 79277005
System ID:

Full Text





Dedicated to my grandparents, parents and brothers,
my wife, Dawn, and son, Jonas.


I would like to thank my mentor, Maurice Swanson, for providing me with the resources and intellectual support necessary to complete my doctoral dissertation. During my graduate studies, he has challenged me to defend my hypotheses with unequivocal data and has equipped me with the tools needed for a lifetime of scientific inquiry. I also thank my committee members, Alfred Lewin, Stephen Sugrue, and Thomas Yang, for their valuable assistance at important times throughout my graduate career. I extend a special thanks to James Dahlberg, from the University of Wisconsin-Madison, for taking time out of his hectic schedule to serve as my outside examiner in Gainesville. Importantly, I thank past and present members of the Swanson Lab, in particular James Anderson, Ron Hector, and Carl Urbinati, for their help with many of the experiments described in this manuscript. I would also like to thank Lionel Minvielle-Sebastia for providing the in vitro polyadenylation results and John Aitchison for identifying purified proteins by mass spectrometry. Finally, my wife, Dawn, deserves special mention because without her love and support none of this would have been possible.


ACKNOW LEDGEMENTS ........................................................................ iii

A B S T R A C T .......................................................................................... vii

INTRODUCTION ................................................................................... 1

Integrating RNAP II Transcription and Nuclear Processing Events .............. 6
Synthesis of Pre-mRNA ................................................................. 6
Transcription Elongation and Genome Maintenance ............................ 9
RNAP II Regulates Pre-mRNA Capping .......................................... 10
Pre-mRNA Splicing and Recruitment of the Spliceosome .................... 12
Regulation of Alternative Splicing by SR Proteins and hnRNPs .............. 15
ATP-dependent Remodeling of the Spliceosome during Splicing ............ 17
Terminal Exon Definition Requires the Cap and 3' Cleavage Site ........... 19
RNAP II and the Spliceosome Connect Splicing to Transcription ............ 21
3' End Cleavage and Polyadenylation ............................................ 23
Monitoring 3' End Integrity during Transcription Elongation .................. 27
Coordinating Nuclear Processing Events and mRNA Export ..................... 29
General Mechanisms of Nucear/Cytoplasmic Transport ...................... 29
mRNA Export Requires a Complex Array of Factors ........................... 31
Pre-mRNA Splicing Factors Are Required for mRNA Export ................. 33
TREX Links Transcription, Splicing and mRNA Export ........................ 37
Connections between Polyadenylation and mRNA Export ................... 38

MATERIALS AND METHODS .................................................................. 43

Yeast and Bacterial Culture Media ..................................................... 43
Yeast Strains and Plasmids ............................................................... 44
Nucleic Acid Isolation Procedures ....................................................... 47
Cell Transformation .......................................................................... 49


Yeast Genetic Manipulations ............................................................. 50
Yeast Total Cell Protein Isolation ....................................................... 50
Fluorescence In Situ Hybridization and Cellular Immunofluorescence .......... 51
In Vitro 3' End Processing Assays ..................................................... 52
Filter Binding Assays ........................................................................ 53
Tandem Affinity Purification ................................................................. 53
Preparation of Polyclonal Antisera and Monoclonal Antibodies ................. 55
Poly(A) Tail Length Determination ..................................................... 56
In Vitro GST Pull-down Experiments ..................................................... 56
Phosphatase Treatment with X Phosphatase .......................................... 58

R E S U LT S ....................................................................................... .. 59

Research Objectives ........................................................................ 59
Nab2p Binds Poly(A) RNA and Limits Poly(A) Tail Length In Vitro .............. 61
Expression and Purification of Recombinant GST-Nab2p-His6 Protein ..... 62
rNab2p Binds with High Affinity to Poly(A) RNA Homopolymers ............. 62
Nab2p Restricts Poly(A) Tail Length In Vitro ..................................... 65
NAB2 Limits Poly(A) Tail Lengths and Promotes mRNA Export In Vivo ......... 72
Nuclear Targeting of Pabl p Suppresses the nab2A Growth Defect ......... 73
Nab2p Is Required For Poly(A) Tail Length Control and mRNA Export .... 76
Nab2p Associates with Factors Required for mRNA Export ...................... 85
Purification of Nab2pT-associated Proteins ....................................... 85
Nab2p and Kapl04p Form an Abundant Complex in Yeast Extracts ........ 89
Nab2p Co-purifies with Nuclear mRNA-binding Proteins ..................... 95
mRNA Export Is Inhibited by a Mutation in the NES of Nab2p ................ 100
Mex67p Interacts with the N-terminus of Nab2p .................................. 101
nab2-20 Stabilizes Mex67p-Nab2p Interactions In Vitro ......................... 104
Nab2p Is Phosphorylated in Strains Defective for mRNA Export ............. 110

D IS C U S S IO N ....................................................................................... 115

Nab2p Is Required for Poly(A) Tail Length Restriction ............................... 115
Potential Role for Nab2p in Preventing Nucleolar Retention of mRNA .......... 121 Nab2p Interacts with Import and Export Receptors ................................... 126
Lim itatio ns ....................................................................................... 130


C o nclusio ns ..................................................................................... 132

A P P E N D IX .......................................................................................... 133

R E F E R E N C E S ..................................................................................... 138

BIOGRAPHICAL SKETCH ...................................................................... 167


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


Keith Robert Nykamp

December 2003

Chair: Maurice S. Swanson
Major: Molecular Genetics and Microbiology

Eukaryotic messenger RNA is transcribed in the nucleus and translated into protein in the cytoplasm. Prior to export from the nucleus, mRNAs must be capped at the 5' end with a 7-methylguanylate (m7G), introns must be removed via splicing, and a polyadenylate tail must be added to the 3' end. Splicing and polyadenylation are carried out by large macromolecular machines and are coordinately regulated by RNA polymerase II (RNAP II) transcription. During transcription, members of a large family of nuclear RNA-binding proteins, the heterogeneous nuclear RNA binding proteins (hnRNPs), bind to nascent RNAP II transcripts. Co-transcriptional association of hnRNPs with nascent pre-mRNA is thought to regulate the specificity and timing of subsequent RNA processing events by influencing the recruitment of the spliceosome and cleavage/polyadenylation factors. Following pre-mRNA processing, export from


the nucleus is promoted by interactions between the mRNA and specialized nuclear export factors.

Although hnRNPs associate with nuclear poly(A) RNA and are required for efficient mRNA export in vivo, their role in mRNA export has remained controversial. Suggestions have been made that hnRNPs bind non-specifically to pre-mRNA and do not directly recruit mRNA export receptors. Alternatively, hnRNPs may act at the interface between mRNA processing events and nuclear export, orchestrating the temporal recruitment of mRNA export factors after processing has occurred in vivo. The goal of the research presented in this report was to test the latter hypothesis using the yeast hnRNP Nab2p. My results demonstrate that Nab2p is a nuclear poly(A)-binding protein, required for both termination of polyadenylation and mRNA export. Surprisingly, these two processes can be uncoupled in nab2 mutant strains, and Nab2p interacts with the nuclear mRNA export factor Mex67p. Based on these results, the proposition is made that Nab2p coordinates the termination of polyadenylation with mRNA export in vivo.



The synthesis of messenger RNA (mRNA) in a eukaryotic cell occurs in the nucleus and involves transcription of pre-mRNA by RNA polymerase II (RNAP II) followed by several processing steps, including capping, splicing, and polyadenylation (see Figure 1). The translation of mRNA into proteins, however, occurs in the cytoplasm. Importantly, nuclear pore complexes (NPCs) allow for the translocation of RNAs and proteins through the nuclear envelope (NE) (reviewed in Ryan and Wente, 2000).

Compartmentalization of transcription and pre-mRNA processing in the nucleus and protein translation in the cytoplasm allows for much greater regulation of gene expression than could be attained otherwise (reviewed in Hood and Silver, 1999; Komeili and O'Shea, 2000). For example, p53 normally shuttles between the nucleus and cytoplasm, but localizes to the nucleus during stress conditions where it increases the transcription of stress response genes (Middeler et al., 1997). Interestingly, the inability of p53 to localize in the nucleus correlates with the proliferation of several tumors emphasizing the importance of protein import regulation (Moll et al., 1995; Shlamp et al., 1997).

The addition of a 7-methylguanylate cap to the 5' end and a polyadenylate tail to the 3' end of mRNA provides additional opportunities for the eukaryotic cell to regulate gene expression (Shatkin and Manley, 2000). Under normal



conditions, the cytoplasmic cap-binding protein (elF4E) and poly(A) binding protein (PABP), protect against degradation and aid in protein synthesis through interactions with the cap and poly(A) tail, respectively (reviewed in Wilusz et al., 2001). AU-rich element (ARE) RNA-binding proteins, such as AUFI/hnRNP D and tristetraprolin (TTP) bypass PABP-dependent mRNA stabilization by promoting mRNA decay when bound to 3' untranslated (UTR) regions of protooncogene and cytokine mRNAs (Lai et al., 1999; Laroia et al., 1999; Loflin et al., 1999). Alternatively, HuR antagonizes the binding of AUF1/hnRNP D to AREs and augments PABP-dependent stabilization of these mRNAs (Fan and Steitz, 1998; Gallouzi et al., 2000). Other ARE-binding proteins (TIAR and TIA-1) have been demonstrated to inhibit PABP-dependent re-initiation of protein synthesis (Gueydan et al., 1999; Piecyk et al., 2000). Importantly, ARE mutations or aberrant levels of ARE-binding proteins correlate with a variety of diseases, including autoimmunity, arthritis, myeloid hyperplasia, and tumorigenesis (Gouble et al., 2002; Kontoyiannis et al., 1999; Taylor et al., 1996). Properly regulated mRNA turnover is clearly very important for cell viability.

Alternative pre-mRNA splicing also gives rise to extraordinary levels of gene regulation by increasing the number of different proteins that can be produced by a single gene (Smith and Valcarcel, 2000). The D. melanogaster Down's Syndrome cell adhesion molecule (Dscam) gene provides an extreme example of this phenomenon (Schmucker et al., 2000). The authors predict that

-38,000 distinct protein isoforms are generated by alternative pre-mRNA splicing. Given so much diversity and complexity, it is not surprising that


symptoms of many diseases have been correlated with splicing errors (reviewed in Faustino and Cooper, 2003). For example, mutations that eliminate or alter constitutive cis-acting splicing signals often result in abnormal mRNAs due to aberrant intron inclusion or exon exclusion. In most cases, these abnormal mRNAs are susceptible to degradation by nonsense-mediated decay (NMD) and result in lost function (Culbertson, 1999). If aberrantly spliced mRNAs are exported, however, they may generate truncated proteins with additional, possibly deleterious, functions.

While pre-mRNA processing events and nucleocytoplasmic transport provide many opportunities for gene regulation, they also create logistical problems for eukaryotic organisms. What mechanisms are in place to ensure that only mature, and not incompletely processed, mRNAs are exported to the cytoplasm? It turns out that proteins essential for transcription are also important for accurate pre-mRNA processing and factors vital for pre-mRNA processing events are necessary for efficient mRNA export (for reviews see Hirose and Manley, 2000; Jensen et al., 2003; Maniatis and Reed, 2002; Proudfoot et al., 2002; Steinmetz, 1997). Integrating transcription, pre-mRNA processing and mRNA export makes perfect sense because this linkage provides a mechanism for the eukaryotic cell to assess mRNA quality at each step of the gene expression pathway before proceeding to the next step.

E o CL
cm Q) 0
L- 4,
0 c 0 M
ZZ o

c C 0

0 cn

=5 Z
CL a)
>< E
0 U)



CL ch
> %
a) 0 O)o

cn 0) -C
CL 0
0 a 0

z E

c E
Co -0
a) CL -0 t-5 0 o 0
C -0


c (n
=3 o E -2




E' 72
a o

l dqcd U(v)-l 0
000 (" co co

C uj 'o


'x In
Z4,N U


Inte-grating RNAP II Transcription and Nuclear Processing Events Synthesis of Pre-mRNA

RNAP II transcribes all eukaryotic pre-mRNA and is the primary target for gene regulation as well as the principal conductor of pre-mRNA processing events. Importantly, the mechanisms of pre-mRNA transcription are highly conserved among eukaryotes allowing researchers to study transcription initiation and elongation in a variety of model organisms. Genetic screens in D. melanogaster and S. cerevisiae, as well as biochemical strategies in human and yeast cells, have identified a holoenzyme consisting of five major protein complexes (RNAP II, General Transcription Factors (GTF), Srb-mediator; SWISNF, and CDK) required for transcription from chromatin-associated DNA templates (for reviews, see Conaway and Conaway, 1993; Kornberg et al., 1992; Lemon and Tjian, 2000; Myer and Young, 1998; Woychik and Hampsey, 2002). The core enzyme consists of twelve subunits all of which are required for enzymatic activity. In yeast, these subunits are encoded by genes RPB1 through RPB12. An interesting feature of RNAP II is a disordered COOH-terminal repeat domain (CTD) consisting of twenty-six consensus YSPTSPS heptad repeats encoded by the yeast RPB1 gene (Cramer et al., 2001). The CTD is unique to RNAP II and is central to the regulation of transcriptional initiation and elongation (Myer and Young, 1998; Price, 2000; Woychik and Hampsey, 2002).

Before transcription is initiated and synthesis of pre-mRNA begins, the holoenzyme assembles on the DNA at the site of transcription. RNAP II alone does not recognize specific DNA sequences and is unable to respond to


transcriptional activators (Myer and Young, 1998). Promoter recognition requires the GTF complex (TFIID, TFIIB, TFIIF, TFIIH, and TFIIE) (Roeder, 1996; Woychik and Hampsey, 2002). DNA-binding activities associated with TFIID and TFIIB nucleate the formation of a pre-initiation complex (PlC) immediately upstream of the transcription start site. Addition of the remaining GTFs (TFIIF, TFIIE, and TFIIH) completes PIC assembly on the promoter. TFIIF plays a structural role in stabilizing PIC assembly while TFIIE and TFIIH eliminate transcription initiation obstacles, such as supercoiled DNA and DNA-bound nucleosomes, through ATP-dependent unwinding and remodeling activities (Dvir et al., 2001). SWI-SNF also utilizes ATP hydrolysis to facilitate initiation and elongation in the midst of chromatin barriers (Lemon and Tjian, 2000). Srbmediator receives signals from enhancer-bound activators and modulates PIC formation and transcription initiation through interactions with RNAP II and TFIIH (Woychik and Hampsey, 2002). Finally, the cyclin dependent kinase (CDK) and its cyclin partner regulate PIC assembly (Liao et al., 1995). Yeast cells lacking the CDK homologue (SrblOp) exhibit de-repression of highly repressed genes such as SPO013, GAL1, SUC2, PHO5, and MFA2 (Kuchin et al., 1995; Wahi and Johnson, 1995). SrblOp exerts its effect on transcription initiation by phosphorylating the CTD of RNAP II on Serine 5 (Ser5) of the heptad repeats (Hengartner et al., 1998). These findings, combined with the fact that RNAP II purified from PIC preparations does not contain phosphates (Kang and Dahmus, 1993), suggest a model in which SrblOp suppresses transcription from a subset of genes by blocking PIC formation. How is Srbl Op activity controlled to allow for


transcription initiation of highly active genes? A phosphatase, Fcplp in yeast, interacts with TFIIF and has the ability to dephosphoyrlate the CTD at both Ser5 and Serine 2 (Ser2) of the heptad repeats (Cho et al., 2001). Fcplp may stimulate initiation by antagonizing the effects of SrblOp on Ser5. In addition, dephosphorylation of Ser2 by Fcpl may be required for recycling RNAP II for another round of transcription (see below).

After PIC formation, RNAP II escapes from the promoter and enters an early elongation phase. These steps require ATP-dependent helicase and CTD kinase activities associated with the transcription factor TFIIH (Dvir et al., 2001). In yeast, the product of the KIN28 gene co-purifies with TFIIH and kin28 mutations correlate with general transcription defects (Cismowski et al., 1995; Valay et al., 1995). Hengartner et al. (1998) have further demonstrated that Kin28p is responsible for phosphorylating Ser5 after PIC formation and may provide a signal for elongation to commence (Hengartner et al., 1998). Remarkably, the timing of Ser5 phosphorylation can have profound effects on pre-mRNA transcription. SrblOp mediated phosphorylation of Ser5 prior to PIC formation blocks transcription initiation while Ser5 phosphorylation by Kin28p after PIC formation stimulates elongation. Clearly, promoter escape and early elongation are highly regulated processes and these steps play an important role in regulating gene expression (Kugel and Goodrich, 1998; Kumar et al., 1998). For example, heat shock protein genes have stalled RNAP II complexes at their promoters with nascent (20-40 nucleotides) transcripts waiting to elongate in response to heat stress (Uptain et al., 1997). Defective elongation, as observed


in Xeroderma Pigmentosum patients, gravely affects cell viability (Liu et al., 2001).

The phosphorylation status of Ser2 plays an important role in the transition from early to productive elongation. A Positive Transcription Elongation Factor b (P-TEFb) is required for the synthesis of long transcripts in mammals (Reines et al., 1999). P-TEFb is composed of a CDK subunit (Cdk9) and several cyclins and phosphorylates Ser2 of the CTD after initiation (Price, 2000) providing evidence that Ser2 phosphorylation is essential for efficient elongation. In yeast, the Cdk9 homolog (Ctkl p) is not essential for vegetative growth and presumably not required for transcription elongation (Sterner et al., 1995). Corden and colleagues have found that Ctklp regulates transcription during heat shock and under nutritional stress (Patturajan et al., 1999; Patturajan et al., 1998). The authors propose that Ctkl p may be recruited to specific promoters in conjunction with RNAP I1. Transcription elongation of these genes is then regulated by Ctklp-dependent phosphorylation of Ser2. Importantly, these results provide evidence that transcription is regulated not only before initiation begins but also during elongation.

Transcription Elonqation and Genome Maintenance

Productive elongation is a very complex process and other factors have been identified that facilitate elongation by removing protein obstacles from the DNA, alleviating torsional strain caused by DNA supercoiling, or by interacting with RNAP II to suppress transcriptional pausing and arrest (Reines et al., 1999). SWI-SNF is essential for the formation of the PIC, promoter clearance and


elongation of certain genes (Wolffe, 1994). FACT was purified by virtue of its ability to facilitate transcriptional elongation from chromatin templates in vitro using only RNAP II and GTFs (Orphanides et al., 1998). SII (PPR2 in yeast) directly interacts with RNAP II to activate cleavage of stalled transcript so that the enzyme can re-extend, allowing repeated attempts at elongation. In yeast, the THO complex also plays a role in elongating through transcriptional pauses (Chavez et al., 2000). Mutations in members of the THO complex, such as Hprl p and Tho2p, result in defective elongation of a lacZ reporter gene, although elongation through several other yeast genes is unaltered (Piruat and Aguilera, 1998). Mutant hprl and tho2 strains are also hypersensitive to the elongation inhibitor, 6-azauracil, indicating THO facilitates productive elongation through transcriptional blocks. Interestingly, hprl mutants also exhibit hyperrecombination phenotypes (Chavez and Aguilera, 1997). The authors believe that when transcription is hindered due to elongation impediments, such as long DNA repeats, the transcribed region becomes highly susceptible to recombination. Indeed, a stalled RNAP II may provide a signal to the DNA repair machinery that a region of the genome is unstable and needs to be repaired by homologous recombination.

RNAP II Regulates Pre-mRNA Capping

The addition of a 7-methylguanylate (m7G) to the 5' end of a pre-mRNA is the first step in the production of mature mRNA (for reviews see Shatkin, 1976; Shuman, 2001). Cap addition is a three-step process carried out by three distinct enzymes in yeast. An RNA triphosphatase (Cetl p) hydrolyzes the y-phosphate


from the 5' end of the pre-mRNA and a guanyltransferase (Cegip) links a GMP to the resulting diphosphate end via a 5' to 5' phospoanhydride bond (Shibagaki et al., 1992; Tsukamoto et al., 1997). A methyltransferase (Abdlp) rounds out the capping process by transferring a methyl group to the guanine on position 7 (Mao et al., 1995).

It has been known for a long time that capping occurs co-transcriptionally when nascent RNA chains are between 25 and 30 nucleotides in length (Coppola et al., 1983; Jove and Manley, 1984). How is the mRNA capping process coordinately regulated with transcription? Several studies have provided evidence that all three capping enzymes selectively associate with RNAP II only when the heptad repeats are phosphorylated at the Ser5 position (Komarnitsky et al., 2000; McCracken et al., 1997; Rodriguez et al., 2000; Shroeder et al., 2000). Further, Ceglp activity is dependent on the presence of Cetlp and increases with higher levels of phosphorylation at Ser5, but not Ser2 (Cho et al., 2001; Ho et al., 1998; Ho and Shuman, 1999). Thus, the capping apparatus is recruited by RNAP II to the site of transcription and allosterically activated before transcription elongation commences.

The benefits of coupling transcription and capping are not biased toward the capping reaction. Interactions among Cetlp, Ceglp and the CTD also serve as a checkpoint for productive elongation. Myers et al. (2002) identified the Cetlp-Ceglp heterodimer as a repressor of transcription elongation and reinitiation in vitro (Myers et al., 2002). Apparently, the transcription machinery is unable to carry on with transcription if Abdlp is not present to methylate the


guanylate cap. In support of these results, Jove and Manley demonstrated twenty years ago that inhibiting transmethylation by adding Sadenosylhomocysteine to transcription reactions in vitro suppressed initiation (Jove and Manley, 1982).

Proper cap formation is essential for cellular viability and has been demonstrated to be important for mRNA stability and turnover, pre-mRNA processing, and translation initiation (Gallie, 1998; Lewis and Izaurralde, 1997; Shatkin, 1985; Shatkin and Manley, 2000; Tucker and Parker, 2000; Wilusz et al., 2001). These cellular functions can be attributed to proteins that specifically recognize the m7G cap in the nucleus and the cytoplasm (Marcotrigiano et al., 1997; Mazza et al., 2002). In the nucleus, a heterodimeric complex of 80 (CBP80) and 20 (CBP20) kDa subunits binds the cap and stimulates splicing and polyadenylation (Cooke and Alwine, 1996; Izaurralde et al., 1994) (also see below). In the cytoplasm, the cap is bound by a single 25 kDa polypeptide (elF4E) that stimulates protein translation through its interactions with the cap, PABP and several translation initiation factors (Sonenberg, 1996). Pre-mRNA Splicinq and Recruitment of the Spliceosome

In an astonishing discovery twenty-six years ago, mRNA sequences (exons) of eukaryotic genes were found to be interrupted throughout the DNA by non-coding segments (introns) (Berget et al., 1977; Chow et al., 1977). Exons are joined together in a contiguous manner via two sequential transesterification reactions (Burge et al., 1999; Kramer, 1996; Sharp, 1994). In the first reaction, the 2' hydroxyl of a conserved branchpoint adenosine attacks the phosphate


group of the 5' splice-acceptor site. This results in two molecules: (1) exon 1

(El) with a free 3' hydroxyl and (2) a lariat intermediate formed by a unique 5' to 2' linkage between the first nucleotide of the intron and the branchpoint adenosine. In the second reaction, the 3' hydroxyl of El attacks the 3' splicedonor site of exon 2 (E2) generating a ligated mRNA and a free intronic lariat. The first transesterification reaction has been shown to occur in vitro in the absence of proteins and ATP (Valadkhan and Manley, 2001; Valadkhan and Manley, 2003) providing evidence that catalysis is RNA-mediated. However, five small nuclear RNAs (Ul, U2, U4, U5, U6) and approximately 100 proteins assemble into a large, energy consuming, splicing machine (spliceosome) on the pre-mRNA and increase the efficiency and fidelity of the transesterification reactions both in vitro and in vivo (for reviews see Adams et al., 1996; Burge et al., 1999; Hastings and Krainer, 2001; Kramer, 1996; Madhani and Guthrie, 1994; Staley and Guthrie, 1998; Will and Luhrmann, 2001).

Spliceosome assembly begins with the selection of a 5' splice site and the formation of a commitment complex (CC) (Michaud and Reed, 1991; Seraphin and Rosbash, 1989). Commitment complex formation specifies the boundaries between introns and exons and is possibly the most important step in the regulation of splicing. As mentioned earlier, inappropriate splice site selection can be very detrimental to the health of a eukaryotic cell. How are relatively short splicing signals recognized within a sea of RNA? In yeast, CC assembly is initiated by the formation of a duplex between UI snRNA and a conserved GUAUGU sequence adjacent to the 5' splice site (Zhang and Rosbash, 1999).


Concurrent with U1 binding, a branchpoint binding protein (BBP) binds the branchpoint adenosine sequence (UACUAAC) (Berglund et al., 1997) and Mud2p binds a polypyrimidine tract located between the branchpoint and 3' splice site (AG dinucleotide) (Abovich et al., 1994). Importantly, interactions among the U1 snRNP, BBP, and Mud2p define introns within the pre-mRNA and provide a scaffold for the recruitment of U2 snRNP (Abovich and Rosbash, 1997) (also see below).

In metazoan organisms, highly degenerate splice site and branchpoint sequences amid very long introns (> 1000 nucleotides) make intron recognition much more difficult (Berget, 1995; Reed, 1996). Berget and colleagues proposed that vertebrate splice site recognition requires interactions between the 3' splice site of the upstream intron and the 5' splice site of the downstream intron for the definition of exons (Robberson et al., 1990). In support of this model, SR proteins, characterized by RNA recognition motifs (RRM) and arginine/serine-rich (RS) domains, bind exonic splicing enhancers (ESEs) and participate in CC formation (Fu, 1995; Graveley, 2000). Both SF2/ASF and SC35 interact with the U1-70K U1 snRNP protein through RS domains (Jamison et al., 1995; Kohtz et al., 1994; Staknis and Reed, 1994) suggesting these proteins facilitate the early recognition of 5' splice sites. Furthermore, U1 depletion and U1 snRNA antisense experiments have demonstrated that SR proteins, in particular SC35, promote CC formation even in the absence of functional U1 snRNPs (Crispino et al., 1993; Tarn and Steitz, 1994).


Recognition of the mammalian branchpoint region also requires branchpoint and pyrimidine tract binding proteins, mBBP and U2 auxilliary factor (U2AF), respectively (Abovich et al., 1994; Arning et al., 1996; Berglund et al., 1997; Ruskin et al., 1988). U2AF is composed of two subunits, U2AF65 and U2AF35 (Zamore and Green, 1989). U2AF65 is most similar to yeast Mud2p and binds directly to the polypyrimidine tract and 3' splice site region of the intron (Abovich et al., 1994; Singh et al., 1995). U2AF35 is unique to metazoan organisms and physically interacts with both SC35 and SF2/ASF (Fu and Maniatis, 1992; Wu and Maniatis, 1993) providing an explanation for the observation that both 5' and 3' splice sites are required for CC formation in mammals (Michaud and Reed, 1993; Reed, 1996). Thus, SR proteins bind to sequences within mammalian exons and bridge 3' and 5' splice sites thereby facilitating exon recognition.

Regulation of Alternative Splicing by SR Proteins and hnRNPs

Given their function in 5' splice site recognition and exon definition, it is not surprising that SR proteins also influence the selection of alternative splice sites (Black, 2003; Faustino and Cooper, 2003; Grabowski and Black, 2001; Smith and Valcarcel, 2000). In addition, heterogeneous nuclear RNA binding proteins (hnRNPs) associate with the pre-mRNA during transcription and, in several cases, antagonize the binding of SR proteins to the pre-mRNA (Dreyfuss et al., 2002; Dreyfuss et al., 1993; Krecic and Swanson, 1999). When one of two identical 5' splice sites needs to be paired with a single 3' splice site in vitro, high concentrations of hnRNP Al in splicing extracts inhibits the binding of Ul snRNP


to both 5' splice sites (Eperon et al., 2000). However, an ESE placed near one 5' splice site results in the recruitment of U1 snRNP to that site and precipitates the formation of a CC on the pre-mRNA (Eperon et al., 2000). Importantly, ESE effects are dependent on the addition of recombinant SF2/ASF to extracts. These findings have led to the proposal that hnRNP Al, and possibly all hnRNPs, bind indiscriminately to pre-mRNA, block binding of sequence-specific RNA-binding proteins (i.e. U1 snRNP and SR proteins) and inhibit the inclusion of alternatively spliced exons (Hastings and Krainer, 2001). In support of this model, SF2/ASF displaces hnRNP Al from ESEs when present in extracts at high concentrations (Eperon et al., 2000; Zhu et al., 2001). Although this model adequately explains the experimental observations of the Krainer group, it is not in accordance with findings reported by Black and colleagues. For example, Rook et al. (2003) described a purine-rich ESE required for inclusion of the neuronal-specific exon (Ni) of the c-src gene (Rooke et al., 2003). UVcrosslinking and immunoprecipitation experiments have indicated that SF2/ASF, hnRNP Al, H and F all interact with the NI ESE. Consistent with previous studies, SF2/ASF stimulates splicing and hnRNP Al represses the inclusion of Ni. However, hnRNP H and F are both components of the enhancer complex and required for N1 inclusion (Chou et al., 1999). In addition, a C to T mutation in exon 7 of the SMN2 gene converts the surrounding sequence into an hnRNP Al binding site leading to exclusion of exon 7 from SMN2 mRNA (Kashima and Manley, 2003). Taken together, these studies provide evidence that hnRNPs are not to be dismissed as generic RNA-binding proteins and general repressors of


sequence-specific RNA-binding proteins. Rather, they bind defined sequences and perform specialized roles in the regulation of alternative splicing.

It is important to point out that very few yeast genes contain introns. Of the genes that do yield spliced mRNAs, most have consensus 5' splice-sites, very few have multiple introns and only one or two genes encode alternatively spliced mRNAs (Davis et al., 2000). This may explain why only a few hnRNP and SR proteins have been discovered in yeast and suggests that these proteins will play important roles in other pre-mRNA processing events, besides splicing (Anderson et al., 1993b; Wilson et al., 1994; Windgassen and Krebber, 2003; Zenklusen et al., 2001).

ATP-dependent Remodeling of the Spliceosome during Splicinq

After 5' splice site recognition, the pre-spliceosome is formed by a duplex between the U2 snRNA and branchpoint sequence (Kramer, 1996; Staley and Guthrie, 1998). Before the U2 snRNP can bind to the branchpoint, however, two energy-dependent, structural rearrangements must occur. First, the branchpoint base-pairing sequence of U2 snRNA must be prepared for binding. Second, Mud2p/U2AF65 and BBP need to be removed from the U2 snRNA pre-mRNA binding site. In yeast, Prp5p modulates the former rearrangement and the latter is coordinated by Sub2p (Kistler and Guthrie, 2001; Libri et al., 2001; O'Day et al., 1996; Zhang and Green, 2001). Both of these proteins are members of the DExD/H-box RNA helicase superfamily characterized by conserved ATPase, helicase and RNA interaction motifs (de la Cruz et al., 1999; Tanner and Linder, 2001). Temperature-sensitive mutations in prp5 result in pre-spliceosome


formation defects in vitro and are lethal in combination with mutant U2 alleles (Ruby et al., 1993; Wells and Ares Jr., 1994; Yan and Ares Jr., 1996). Subsequent experiments have demonstrated that Prp5p is responsible for an ATP-dependent structural change in the intact U2 snRNP prior to branchpoint binding (O'Day et al., 1996). Furthermore, mutations that lead to misfolding of U2 snRNA are suppressed by gain-of-function mutations in cus2 while loss-offunction alleles of cus2 are lethal in combination with both U2 and prp5 mutations (Perriman and Ares Jr., 2000; Yan et al., 1998). Remarkably, pre-spliceosome assembly does not require ATP in cus2A extracts indicating that Cus2p is responsible for rendering pre-spliceosome formation dependent on ATP (Perriman and Ares Jr., 2000; Yan et al., 1998). The authors of these studies propose that Cus2p binds U2 snRNA and inhibits branchpoint binding until Prp5p removes Cus2p through ATP-dependent remodeling activities. SUB2 is homologous to the human gene UAP56 encoding the U2AF65-interacting protein and genetically interacts with MUD2 (Fleckner et al., 1997; Kistler and Guthrie, 2001). Although SUB2 is essential for cell viability, deletion of MUD2 allows for growth in the absence of SUB2. Cus2p and Mud2p appear to provide ATPdependent checkpoints during the splicing process allowing splicing to be halted, if necessary, through regulatory control of Prp5p and Sub2p, respectively.

Four more ATP-dependent remodeling and assembly steps are required for the transesterification steps to proceed. The U1 snRNA interaction with the 5' splice-site must be exchanged for a mutually exclusive interaction with the U6 snRNA (Kandels-Lewis and Seraphin, 1993; Konforti et al., 1993). The U6


snRNA is not recruited to the spliceosome alone. Rather, the U4/U6oU5 trisnRNP escorts U6 to the pre-spliceosome (Kramer, 1996). For the spliceosome to become catalytically active, an extremely stable base-pair interaction between U4 and U6 snRNAs must be disrupted (Nilsen, 1998). After activation of the spliceosome, both transesterification reactions require extensive remodeling of the pre-mRNA in order to proceed in vivo (Staley and Guthrie, 1998). Prp28 removes Ul from the 5' splice site (Staley and Guthrie, 1999), Brr2 unwinds the U4/U6 RNA duplex (Raghunathan and Guthrie, 1998), and Prp2 restructures the pre-mRNA during the first transesterification reaction while Prp16 does so during the second reaction (Kim and Lin, 1996; Schwer and Guthrie, 1991). Terminal Exon Definition Requires the Cap and 3' Cleavage Site

As discussed above, accurate splicing requires the efficient assembly of a commitment complex on the pre-mRNA that distinguishes exons from introns. According to the exon definition model, proteins bind to introns near the 5' and 3' splice sites and interact with ESE bound proteins to form a network of interactions across the exon. For internal exon definition, the supporting cast of characters include U1 snRNP, U2AF, SR proteins and hnRNPs. Two additional exons exist in the pre-mRNA and require a different collection of proteins for their definition. A 5' terminal exon is defined by the 5' cap and 5' splice site within the first intron and a 3' terminal exon is demarcated by a 3' splice site of the last intron and a 3' end cleavage site. In order for the exon definition model to be correct, cap-binding proteins and 3' end formation factors must also be required for splicing fidelity and efficiency. In fact, an interaction between the human


CBC and hnRNP F has been described (Gamberi et al., 1997). HnRNP F preferentially associates with CBC-RNA complexes rather than naked RNA and depletion of hnRNP F from HeLa cell nuclear extracts reduces splicing efficiency in vitro. How does the CBC-hnRNP F interaction influence splicing activity? Black and colleagues have demonstrated that hnRNP F forms a complex with hnRNP H and SF2/ASF on the N1 exon of the c-src pre-mRNA to promote exon inclusion (Chou et al., 1999). Possibly, when bound to ESEs, hnRNP F and SF2/ASF form a network of interactions to define the 5' terminal exon: hnRNP F interacts with CBC and SF2/ASF recruits the Ul snRNP to the 5' proximal splice site. Studies in yeast also indicate that connections between the CBC and U1 snRNP are necessary for delineating the cap-proximal exon. Using in vivo reporter assays, Fortes et al. (1999) demonstrated that LUC7 is required for nonconsensus splice site recognition and luc7 mutant strains fail to utilize capproximal splice sites if two competing 5' splice sites are present (Fortes et al., 1999a). Luc7p is a U1 snRNP-associated protein with an SR-like domain and is indispensable for CC formation in vitro as well as Ul snRNP-CBC interactions in vivo (Fortes et al., 1999a). In a second study by the same group, cell growth was compromised in a mud2A strain if both cbp80 and cbp20 were also deleted from the genome (Fortes et al., 1999b). Importantly, growth is only mildly affected in either mud2A or cbp8OAcbp2OA strains.

For 3' terminal exon definition, an interaction between U2AF65 and poly(A) polymerase (PAP) in HeLa cells has been described (Vagner et al., 2000). PAP stimulates the binding of U2AF65 to the pyrimidine tract of the


terminal intron. These interactions appear to be specific for metazoan organisms because no interactions between Mud2p and yeast PAP have been described. Furthermore, Dower and Rosbash (2002) observed efficient splicing in vivo in the absence of both pre-mRNA capping and polyadenylation. Since yeast introns are small, splicing signals are highly conserved, and most introns are near the 5' end of the gene (Davis et al., 2000), interactions between the spliceosome and 3' end formation complex may not be essential for efficient splicing in yeast. In fact CBC-spliceosome interactions only appear to be essential when non-consensus signals are utilized lending credence to the idea that exon definition is critical for recognition of weak, alternatively used splice sites, such as those primarily found in metazoan organisms.

RNAP II and the Spliceosome Connect Splicingq to Transcription

Considering synthesis of pre-mRNA is carried out solely by RNAP II, it is not surprising that RNAP II has been reported to interact with components of the spliceosome. For example, reporter genes containing introns are not spliced in metazoan cells if transcribed by RNAP I or RNAP III (Sisodia et al., 1987; Smale and Tjian, 1985). Early in situ studies using Chironimus tentans Balbiani ring genes to visualize pre-mRNA processing events during transcription discovered that genes were spliced concurrently with transcription (Bauren and Wieslander, 1994; Kiseleva et al., 1994). In addition, phosporylated RNAP II (RNAP 110) increases the efficiency of splicing both in vivo and in vitro (Fong and Bentley, 2001; Fong et al., 2003; McCracken et al., 1997; Misteli and Spector, 1999). Alternatively, the addition of hypophosphorylated RNAP II (RNAP IIA) inhibits


splicing in vitro (Hirose et al., 1999; Zeng and Berget, 2000). In yeast, Prp40p is recruited to RNAP IIO, but not RNAP IIA (Morris and Greenleaf, 2000). Prp40p is a component of the U1 snRNP and directly interacts with BBP (Abovich and Rosbash, 1997). During transcription elongation, the U1 snRNP-Prp40p complex is tethered to RNAP II increasing the likelihood of 5' splice site recognition. Interactions between RNAP IIO and U1 snRNP also serve to retain the 5' splice site near elongating RNAP II, allowing branchpoint and 3' splice site sequences to be recognized by BBP and Mud2p immediately after synthesis. After binding to the pre-mRNA, U1 snRNP, Mud2p and BBP precipitate the formation of a commitment complex.

Interactions between RNAP II and the spliceosome also influence transcription re-initiation and elongation. TFIIH interacts with U1 snRNA and promoter-proximal 5' splice sites stimulate transcription re-initiation in a Uldependent manner (Kwek et al., 2002). Two factors required for efficient elongation, SII and TAT-SF1, also co-purify with U2 snRNP and RNAP IIO in vivo (Fong and Zhou, 2001; Robert et al., 2002). TAT-SF1 stimulates transcription elongation in vitro through interactions with P-TEFb (Price, 2000). In yeast, the TAT-SF1 homolog (CUS2) plays a mechanistic role in pre-spliceosome formation (Yan et al., 1998). Altogether, these findings provide for an appealing model in which Cus2p/TAT-SF1 and U2 snRNP are recruited to RNAP IIO and stimulate transcription elongation most likely through interactions with Ctklp/P-TEFb. Importantly, Cus2p/TAT-SF1 may inhibit association of U2 snRNP with the premRNA until BBP and Mud2p/U2A65 bind to the branchpoint region of the intron.


Assuming commitment complex formation occurs without difficulty, Prp5p/hPRP5 remodels the U2 snRNP by removing Cus2p/TAT-SF1 and Sub2p/UAP56 dissociates Mud2p/U2A65 from the branchpoint allowing for assembly of the prespliceosome.

3' End Cleavaqe and Polyadenylation

Termination of RNAP II transcription coincides with cleavage of the premRNA at sequences downstream of the translation stop codon. Cleavage is followed by the addition of a polyadenylate tail of 70-90 nucleotides in yeast and 200-300 nucleotides in mammals (Proudfoot et al., 2002; Zhao et al., 1999). Like splicing, large protein complexes execute the cleavage and polyadenylation reactions with precision (Colgan and Manley, 1997; Wahle and Keller, 1996). Much has been learned about the factors responsible for cleavage and polyadenylation because many of these proteins are conserved from yeast to humans (see Figure 2) (Keller and Minvielle-Sebastia, 1997; Manley and Takagaki, 1996). In yeast, an A/U-rich Efficiency Element (EE) and an A-rich Positioning Element (PE) found upstream of the cleavage site recruit Cleavage Factor IB (CF IB) to the pre-mRNA. A second factor, Cleavage and Polyadenylation Factor I (CPF) associates with sequences surrounding the Pyrimidine-Adenosine (PyA) cleavage site (Proudfoot and O'Sullivan, 2002; Wahle and Ruegsegger, 1999). Interestingly CF IB is a single subunit factor composed of the yeast hnRNP Nab4p/Hrplp (Kessler et al., 1997). Nab4p controls cleavage site selection in vitro and in vivo and increases the efficiency of polyadenylation in vitro by interacting with the EE and recruiting CF IA and CPF

_0 "0 c 0 w c o
- CU -= W cn cu (1) cu
E u) cu 0)
cr E a)
m m E E cn >,
T 0) -,- 0 c Z D- (D


'a (n 0 a C: co CL 4- CU (D -o
0 0
LL: C,
0) w c m C) 0 w o > -0 c -0 U- o m
U) 0 c
cn L
a) = CD x o
E 0 cm "a tw m CL a) a) E LE x SE 1 0 -a co 0 L) UE o
(D 0
0)= -0
m m
E > -5,
0 M C:

CD -C >.
C CL 75 C 0 CL M cn

cn > 0 E E c

C: M E uS
m 0)
E to > -c
C: .- U) a) (D c E Co c, 0
C a) LL 0 0
(D 4x

m (D
E :3 < u)
a) U- _0 C: U) L) U_ CD cu
L) M
LZ =3

ca C:



w ui U.

uj Cl) ui S
cu CL C,
a) 0lw




cu U.



to the mRNA (Chen and Hyman, 1998; Gross and Moore, 2001; MinvielleSebastia et al., 1998; Valentini et al., 1999). Thus, analogous to their role in selection of alternative splice sites, hnRNPs also control the fidelity of 3' end cleavage site selection. After CF IB, CF IA and CPF bind to the pre-mRNA, cleavage is carried out by an unknown endonuclease (CF II) providing a free 3' end for the addition of adenosines by PapIlp (Wahle and Ruegsegger, 1999).

Poly(A) addition in mammalian cells is initially a slow process with PAP dissociating from the growing tail after the addition of each nucleotide. After 10 or more adenylate residues have been added, however, a nuclear poly(A)binding protein (PABPN1, formerly PAB II) interacts with the nascent poly(A) tail, the cleavage and polyadenylation specificity factor (CPSF), and PAP to form a quaternary structure which transforms PAP into a processive enzyme (Bienroth et al., 1993; Murthy and Manley, 1995; Wahle, 1991). After approximately 200 nucleotides, the interaction between PABPN1 and CPSF is disrupted, thus terminating the polyadenylation reaction (Wahle, 1995). What is the mechanism of CPSF-PABPN1-PAP complex disruption? Using electron microscopy (EM), an oligomeric PABPN1 particle has been observed to co-localize with the growing poly(A) tail (Keller et al., 2000). Attractively, the maximum diameter observed for a PABPN1 particle (-21 nm) correlates with the particle size of a PABPN1poly(A) complex when the poly(A) molecule is 200-300 nucleotides in length. Longer poly(A) molecules do not result in larger PABPN1 particles suggesting the maximum PABPN1 particle size restricts poly(A) tail lengths to 300 nucleotides in vivo. These authors propose a model whereby PABPN1 initially


converts PAP into a processive enzyme via interactions with CPSF, but these interactions are not feasible when the PABPN1 particle reaches a maximum size, thus converting PAP back into a distributive enzyme.

In yeast, an essential poly(A)-binding protein (Pabip) has also been described, although Pablp shares minimal sequence similarity with PABPN1 (Sachs et al., 1986; Wahle, 1991). Pablp interacts in the yeast two-hybrid system with the CF IA component Rnal5p (Amrani et al., 1997) and recombinant Pablp limits poly(A) tail lengths in vitro (Minvielle-Sebastia et al., 1997). The role of Pablp in poly(A) tail length control in vivo, however, is questionable because the steady state distribution of Pablp is cytoplasmic (Anderson et al., 1993a). Furthermore, the examination of poly(A) tail lengths of total RNA in pablA cells reveals modest increases (-40 nucleotides) in vivo, whereas precursor RNAs have poly(A) tails greater than 200 nucleotides in pablA extracts (MinvielleSebastia et al., 1997; Sachs and Davis, 1989). Together, these results suggest that poly(A) tail length control in vivo is likely to require additional factors. Monitoring 3' End Integrity during Transcription Elongation

How are transcription and 3' end processing events coupled in vivo? The same general theme holds true for 3' end processing as described previously for capping and splicing. The CTD of mammalian RNAP II is essential for efficient cleavage at poly(A) sites and yeast strains containing CTD truncation mutants exhibit reduced cleavage efficiency (Hirose and Manley, 1998; Licatalosi et al., 2002; McCracken et al., 1997; Ryan et al., 2002). Components of the cleavage and polyadenylation machinery in yeast, including Rnal5p of CF IA and


Yhhlp/Cftlp of CPF, co-purify with RNAP IIO in vivo and directly interact with the phosphorylated CTD in vitro (Barilla et al., 2001; Dichtl et al., 2002b).

A conserved network of interactions between several GTFs and factors involved in cleavage and polyadenylation also exists to link transcription to 3' end formation. Dantonel et al. (1997) immunopurified CPSF with TFIID from HeLa cells. After transcription initiates, however, CPSF dissociates from TFIID and remains associated with RNAP IIO. In addition, PC4, a mammalian transcription co-activator, binds double-stranded DNA, interacts with TFIIA and RNAP IIO, and is phosphorylated by TFIIH in vitro (Malik et al., 1998; Werten et al., 1999). PC4 also associates with the 64 kDa subunit of CstF suggesting that CstF is recruited to the PIC as well (Calvo and Manley, 2001). These interactions have been conserved throughout evolution because the yeast homologue of PC4 (Sublp) binds to the CstF-64 counterpart (Rnal5p) in vitro (Calvo and Manley, 2001). SUB1 and SSU72 interact with TFIIB to assist in PIC assembly (Sun and Hampsey, 1996). Interestingly, mutations in these genes disrupt the fidelity of start site selection (Wu et al., 1999). Ssu72p also associates with two subunits of CPF (Ptalp and Ydhlp/Cft2p) and is required for CPF-dependent cleavage of the 3' end (Dichtl et al., 2002a; He et al., 2003). Together, Ssu72p and Sublp provide transcription start site specificity for RNAP II and recruit CF IA and CPF to the site of transcription. In addition, the presence of CPF at the site of transcription may provide a signal for RNAP II to begin elongation because Ptalp also interacts with Kin28p (Rodriguez et al., 2000). Phosphorylation of the CTD at Ser5 by Kin28p triggers the commencement of transcription elongation


(Hengartner et al., 1998). In the absence of efficient cleavage and polyadenylation by CPF, Kin28p may not be able to phosphorylate RNAP II, thereby halting transcription elongation. Thus, Ssu72p and Sublp ensure that newly synthesized mRNAs are correctly cleaved and polyadenylated

Coordinating Nuclear Processinq Events and mRNA Export General Mechanisms of Nuclear/Cytoplasmic Transport

After the 3' end has been cleaved and a poly(A) tail has been added the mature mRNA is ready for export. How are molecules transported into and out of the nucleus? The current model for general nuclear transport proposes that an adapter binds both cargo and a receptor. Several adapters and receptors have been identified in a number of different organisms that also associate with integral NPC proteins, or nucleoporins (Nups) (reviewed in Gorlich and Kutay, 1999; Komeili and O'Shea, 2001; Nakielny and Dreyfuss, 1999; Ohno et al., 1998; Pemberton et al., 1998). In addition to adapters and receptors, protein transport requires the GTPase Ran (Kuersten et al., 2001). Typically for protein import, an adapter (importin a) binds to a nuclear localization sequence (NLS) exemplified by the single SV40 large T antigen (PKKKRKV) or bipartite nucleoplasmin (KRPAAIKKAGQAKKKK) sequences and forms a quaternary structure with an import receptor (importin P), the NLS-containing protein, and RanGDP (Gorlich and Mattaj, 1996; Weis, 1998). A cytoplasmic GTPase Activating Protein (RanGAP) facilitates hydrolysis of GTP by Ran ensuring that RanGDP molecules accumulate in the cytoplasm (reviewed in Azuma and Dasso, 2000; Kuersten et al., 2001). Removal of GDP from Ran, in turn, requires


a nuclear guanine exchange factor (RanGEF) resulting in the accumulation of RanGTP in the nucleus. Assembly of an import complex is strengthened in the cytoplasm by RanGDP and recruited to the NPC by interactions between importin 13 and nucleoporins. After translocation through the nuclear pore, exchange of GDP for GTP in the nucleus releases the imported NLS protein from importin a, importin 13, and Ran (Ohno et al., 1998). Although several importin a and importin P3 proteins have been identified in metazoan organisms, the yeast genome encodes one importin a (Srpl) and one importin 13 (Kap95) (Gorlich and Kutay, 1999). Kap95p physically interacts with Nup116p and co-localizes with Srplp near the NPC in vivo (lovine et al., 1995; Koepp et al., 1996). Proper localization of Kap95p, Srplp, and NLS-containing proteins require a functional GTPase activating protein (Rnalp) and a functional guanine exchange factor (Prp20p) (Koepp et al., 1996).

Similar principles apply for export of proteins from the nucleus. Coprecipitation experiments first identified a human nuclear factor, homologous to importin 13, associated with NUP214/CAN and NUP88 (Fornerod et al., 1997b). The human gene was named hCRM1 because, in addition to having similarity to importin 13, the open reading frame was homologous to a previously identified gene in Schizosaccharomyces pombe (CRM1+). Early studies of CRMI+ indicated that it encoded a nuclear protein required for efficient DNA, RNA, and protein synthesis and condensation of chromosomal regions was observed in CRM1+ mutants at non-permissive temperatures (Adachi and Yanagida, 1989). Increased expression of CRM1+ was subsequently found to confer resistance to


leptomycin B (LMB), an agent known to induce cell cycle arrest at G1 and G2 phases of the cell cycle (Nishi et al., 1994). Analogous to protein import, a leucine-rich nuclear export sequence (NES) was discovered in the HIV-1 Rev and kinase A inhibitor PKI proteins (Fischer et al., 1995; Wen et al., 1995). In Xenopus oocytes and permeabilized HeLa cells, CRM1 has been shown to mediate the export of reporter proteins containing an NES while LMB strongly inhibited export of NES peptides (Fornerod et al., 1997a; Fukada et al., 1997; Ossareh-Nazzari et al., 1997). CRM1 exhibited cooperative interactions in vitro with both NES peptides and RanGTP providing evidence that CRM1 behaves like a traditional transport receptor (Fornerod et al., 1997a). mRNA Export Requires a Complex Array of Factors

In the same study, Fornerod et al. (1997) demonstrated that U snRNA export, but not mRNA export, was stimulated in Xenopus oocytes by increased levels of CRM1 and inhibited by LMB. In metazoan organisms U snRNAs are exported to the cytoplasm after transcription, assembled into functional snRNPs, and imported into the nucleus (reviewed in Paushkin et al., 2002; Will and Luhrmann, 2001). Four of the five spliceosomal U snRNAs (U1, U2, U4, and U5) are transcribed by RNAP II and have a m7G-cap to which the nuclear CBC is able to bind. In addition to CRM1, the CBC also plays an important role in the export U snRNAs (Hamm and Mattaj, 1990; Jarmolowski et al., 1994). In an elegant study, the Mattaj group provided the final piece in the puzzle of U snRNA export by identifying an adaptor protein (PHAX) which has the ability to bridge CRM1 and CBC interactions (Ohno et al., 2000). PHAX contains a leucine-rich


NES and formation of an snRNA-CBC-PHAX-CRM1 export complex required RanGTP. Remarkably, PHAX was unable to bind to CRM1, even in the presence of U snRNP, CBC, and RanGTP proteins, if it was not phosphorylated. These results provided evidence for an additional level of regulation for U snRNP export.

At the same time CRM1 was discovered in metazoans, Stade et al. (1997) reported that Crmlp was responsible for the export of NES peptides in yeast. Intriguingly, these authors also observed a block to poly(A) RNA export in a conditional crml mutant strain suggesting that CRM1 may serve as a receptor for mRNA export in yeast. A subsequent study by Neville and Rosbash (1999), however, demonstrated that a strain with an LMB-sensitive Crmlp continued to synthesize proteins in the presence of LMB arguing against a direct role for Crmlp in mRNA export. The discrepancy between these two reports can be explained by the finding that increased expression of a cytoplasmic DExD/H-box helicase (DBP5), required for proper poly(A) RNA localization, suppresses the crml-dependent mRNA export block (Hodge et al., 1999). Thus, mutations in CRM1 may indirectly affect mRNA export by altering the sub-cellular localization of cytoplasmic mRNA export factors.

What are the receptors and adapters for mRNA export? A synthetic lethal screen using a nup85 mutant allele identified MEX67 as a gene required for mRNA export in yeast (Segref et al., 1997). Mex67p interacts with Mtr2p to form a heterodimer localized to the nuclear periphery in vivo (Santos-Rosa et al., 1998). In addition to Nup85p, Mex67p-Mtr2p physically associates with several


other nucleoporins in vitro and mutations in mtr2 profoundly affect the localization of both Mex67p and poly(A) RNA in vivo (Santos-Rosa et al., 1998; Strasser et al., 2000). A human protein (TAP), with similarity to Mex67p, was discovered to bind the constitutive transport element (CTE) of type D retroviruses and mediate the export of unspliced viral RNAs containing a CTE (Braun et al., 1999; Gruter et al., 1998). Similar to HIV, the type D retroviruses hijack the human cellular export machinery to export their own RNAs. Unlike CRM1, however, TAP is an mRNA export receptor because injecting CTE RNA into the nucleus of Xenopus oocytes profoundly inhibits the export of cellular mRNAs (Pasquinelli et al., 1997). Importantly, co-expression of TAP and the human Mtr2p counterpart (p15) in the mex67Amtr2A yeast strain restored growth and stimulated mRNA export in vivo (Katahira et al., 1999).

Pre-mRNA Splicinq Factors Are Required for mRNA Export

Although both Mex67p and TAP have been demonstrated to crosslink in the presence of UV to poly(A) RNA in vivo, neither of these proteins bind very strongly to non-CTE RNA in vitro (Katahira et al., 1999; Segref et al., 1997). Analogous to importin a for protein import, an hnRNP-like protein (Yralp in yeast; Aly/REF in humans) serves as an export adapter, bridging the connection between mRNA and export receptors (Mex67p or TAP) (Portman et al., 1997; Strasser and Hurt, 2000; Stutz et al., 2000; Zenklusen et al., 2001) (also see Figure 3). Other mRNA export adapters have also been described in human cells (for review see Moore and Rosbash, 2001). Using cell permeable peptides to selectively inhibit the association of hnRNP Al and HuR with their respective

0) c n f -- U) c 0 C < CL Z X

0-0 C: 0 a) a)
0 cn (D a.
< 0 <
z 0 IX
E 0- u) > (D c
CL cn
CU 0
0 c) E
a) cn
0 o C: cu
-a.., Cl "a
cn X 0 0 U) CL
_0 0 cu cl- E EL w lu cu a)
70 U M a) CD C
cn L
Co Q)-o 0 0 = 0 X a) CL
0 x o 0
a) Z5 CL
c cn
< = < < c: Z cu Z W E X
E 0 E a)
L- E
4- =
0 c
0,.o'5 o
(n C.) U) E x
CD cn (D c c: 'a. 2 E 0 Z E
0 a 0
E 0
X 0
w x 0
X CD a W x
a) LL E
> 70 0 m
0) 4i> 0

0 0 0 c
a- C: C: CL x 0 0
0 0 c
< wooo
__ x
z =3.CU
X >, (D c 0
E -0 0
" c m
0 0.0 M:a

75 CL.. U)
*C: o
_0 0 m 40 (n cm E r CD
co 0 0 E < L



4) 0
p %
06 u
a 0 W
(D in

E z

dc==h. CL

z =1 0.

E cc
0 CU dZqeN u(V)

-0CN Lo
t CD

OD C14 IL a.
ca CD

C14 0
m CL
w x
z r M X

(uD < x z CL
E O)

0 OL


export receptors (Trn 1 and Trn 2) Gallouzi and Steitz (2001) demonstrated that these RNA-binding proteins mediate the export of two distinct mRNAs. The Steitz group also provided evidence that SR proteins bind to ESEs and directly interact with TAP to stimulate the export of spliced mRNAs (Huang et al., 2003). In humans, several pathways may exist to export different sub-classes of mRNAs, although the primary export pathway likely involves the cooperative recruitment of TAP by REF/Aly and various SR proteins. What about yeast? Have other adapter proteins been described? To date, Yralp, and Yra2p are the only RNA-binding proteins known to functionally interact with Mex67p. Npl3p (also known as Nablp) and Gbp2p contain SR-like domains, associate with nuclear poly(A) RNA and shuttle between the nucleus and cytoplasm (Lee et al., 1996; Wilson et al., 1994; Windgassen and Krebber, 2003). Mutations in NPL3 and increased expression of GBP2 lead to nuclear accumulation of poly(A) RNA suggesting these genes may play a role in the export of mRNA (Lee et al., 1996; Windgassen and Krebber, 2003). However, a direct role for Npl3p and Gbp2p in the recruitment of export receptor(s) to the mRNA has not been demonstrated.

Surprisingly, formation of the Mtr2p-Mex67p-Yral p-mRNA export complex does not require RanGTP (Katahira et al., 1999). ATP-hydrolysis is believed to fuel association and dissociation of mRNA export complexes because several DExD/H-box helicases are required for mRNA export (reviewed in Linder and Stutz, 2001). Yralp has an affinity for the spliceosomal ATPase Sub2p (Strasser and Hurt, 2001). Importantly, Sub2p and Mex67p interact with overlapping domains of Yral p and these interactions are mutually exclusive in vitro (Strasser


and Hurt, 2001). These results suggest that Yralp-mRNP export takes place in two distinct steps (Figure 3). First, Yralp and Sub2p associate with pre-mRNA during transcription and splicing. Second, Sub2p is exchanged for Mex67p so that the mRNA can be released from RNAP II and targeted to the NPC by Mex67p-Mtr2p. It is important to point out that it is not clear if Sub2p ATPase activity is required for mRNA export or if Sub2p plays a structural role in recruiting Yralp to the pre-mRNA during splicing. Nevertheless, these interactions are conserved in metazoans (Luo et al., 2001) and provide a mechanism for coordinating pre-mRNA splicing with mRNA export (reviewed in Reed and Hurt, 2002). A second DExD/H-box helicase (Dbp5p) is also required for efficient export of poly(A) RNA (Snay-Hodge et al., 1998). Dbp5p exhibits RNA-dependent helicase activity in vitro (Tseng et al., 1998) and associates with Rat7p/Nupl59p on the cytoplasmic side of the nuclear envelope in vivo (Hodge et al., 1999; Schmitt et al., 1999; Strahm et al., 1999). These results have led to the proposal that Dbp5p awaits the exported mRNP on the cytoplasmic side of the nuclear pore, exchanging nuclear-associated proteins for cytoplasmic mRNA binding proteins immediately after export (Lei and Silver, 2002; Reed and Hurt, 2002).

TREX Links Transcription, Splicingq and mRNA Export

Sub2p and Yralp are both recruited during transcription by members of the THO complex (Tho2, Hprl, Mftl and Thp2) (Strasser et al., 2002). Importantly, Sub2p and Yralp associate with highly active genes during transcription elongation and temperature-sensitive mutations in any of the THO


genes result in nuclear accumulation of poly(A) RNA export at restrictive growth temperatures (Strasser et al., 2002; Zenklusen et al., 2002). Mutations in sub2, yral, mex67 and mtr2 also lead to defects in transcription elongation and exhibit hyper-recombination phenotypes as previously described for THO mutants (Jimeno et al., 2002). In addition, Mex67p-Mtr2p interacts genetically and physically with the Sac3p-Thplp complex (Fischer et al., 2002). THP1 is required for efficient transcription elongation (Gallardo and Aguilera, 2001) and SAC3 has an essential role in both mRNA export and progression through the cell cycle (Jones et al., 2000; Lei et al., 2003). The precise mechanisms linking transcription, splicing and mRNA export have not been fully elucidated yet, but one can suppose that THO recruits Yralp and Sub2p during transcription elongation (see Figure 3). If Yralp and Sub2p are not present, THO components inhibit elongation so that mRNA synthesis is halted in the absence of functional export factors. If Yralp and Sub2p associate with the THO components, however, efficient transcription elongation and pre-mRNA splicing proceeds without further incident. Exchanging Sub2p-Yralp interactions for Mex67pMtr2p-Yralp interactions leads to a TREX-released mRNP and interactions between Mex67p-Mtr2p and Sac3p-Thplp mediate the translocation of the mRNP to the cytoplasm (Figure 3).

Connections between Polyadenylation and mRNA Export

Splicing is unlikely to provide the only link between pre-mRNA processing and mRNA export because only a small percentage of yeast genes contain introns (Davis et al., 2000). In addition, the D. melanogaster and C. elegans


REF/Aly genes are dispensable for mRNA export and cell viability (Gatfield and Izaurralde, 2002; Longman et al., 2003). In contrast, UAP56 is essential for all stages of C. elegans development and RNAi experiments reveal an essential role for the helicase in mRNA export. Aside from the previously described role in splicing, Sub2p/UAP56 activity may also be required for release of mRNPs from the site of transcription after polyadenylation because mutations in sub2 lead to an accumulation of intronless RNAs at or near the site of transcription (Jensen et al., 2001a).

It has been known for several years that polyadenylation can stimulate the export of mRNA (Eckner et al., 1991; Huang and Carmichael, 1996). Mutations in components of the cleavage and polyadenylation machinery block poly(A) RNA export (Brodsky and Silver, 2000; Hammell et al., 2002; Libri et al., 2002) and mRNA export mutants accumulate hyperadenylated RNA in nuclear foci (Hilleren and Parker, 2001; Jensen et al., 2001b; Kadowaki et al., 1994; Tseng et al., 1998). Nuclear accumulation of poly(A) RNA in mRNA export mutants (rat71 and ripIA) and 3' end formation mutants (rnal5-1 or papl-1) requires the function of nuclear exosome components (Rrp6p, Rrp4p, and Mtr4p) (Hilleren et al., 2001; Libri et al., 2002). These authors envisage the exosome as a gatekeeper for mRNA export, responsible for retaining and degrading improperly cleaved or hyperadenylated RNAs in the nucleus (see Figure 4). The exosome consists of at least ten different proteins with 3' to 5' exonuclease activity (Allmang et al., 1999b; Mitchell et al., 1997) and was first described as a conserved protein complex required for processing 3' ends of several small

cn 4
C: :3 0 (D CL 0 E :3 0 C: 0 a)

(D C:
U) <
_0 z
c CD
a (0 C)


75 0
CL :3

C: cu

cn < z Co 4C: E (D '0 -0 0
cu .
x cn a) a) (n CL 4-,
>% CD
0 L- 4
cl 0
c 0
C 2 (D CL E


o 0
a) c
o (D
E 0
E in o o o CL cn x x o a)
o E


E 8 0

Na 0
0 t5 T


z 12
< ;E

T C14 0



0 CL X

u E 0


nuclear RNAs including the 5.8S ribosomal RNA (rRNA), small nucleolar RNAs (snoRNAs) and U4 snRNA (Allmang et al., 1999a; Mitchell et al., 1997). More recently, it has been demonstrated that normal mRNAs are also degraded in the nucleus by the exosome (Das et al., 2003; Jacobs Anderson and Parker, 1998). Deletion of nup116 leads to the retention and RRP6-dependent degradation of normal poly(A) RNA in the nucleus (Das et al., 2003). In addition to Rrp6p, the yCBC and a 5' to 3' nuclear exonuclease (Ratlp) are required for degradation of nuclear poly(A)* RNA in the nup116A strain. These authors propose that a nuclear mRNA degradation (DRN) system, consisting of the CBC and nuclear exosome, is responsible for general mRNA turnover.

How are polyadenylated mRNAs released from the site of transcription and exported to the cytoplasm? Is there an export adapter for the poly(A) tail, analogous to Yralp, that actively recruits an export receptor to the mRNA? Does a protein bind the poly(A) tail and inhibit CBC-Rrp6p recruitment to the exosome for retention and degradation in the nucleus? Answers to these questions may provide a more complete understanding of the mechanisms regulating polyadenylation, transcript release and mRNA export.


Yeast and Bacterial Culture Media

Yeast cells were cultured at specified temperatures in YPD media consisting of 1% Bacto-yeast extract (Fisher), 2% Bactopeptone (Fisher), 2% of dextrose (Fisher), and 2% Difco-Agar (for plates), unless otherwise indicated. For growth of the nab2ApGAL:NAB2 strain, 2% galactose (Sigma) was substituted for dextrose. Gene expression from plasmids was achieved by growing yeast cells in synthetic media (.67% Bacto-yeast nitrogen base without amino acids [Difco], supplemented with the appropriate L-amino acids [Sigma] as needed) with dextrose (SD) or galactose (SG) as previously described (Rose et al., 1990). The addition of 5-flouro-orotic acid (5-FOA) (Diagnostic Chem. Ltd.) to SD media allowed for the eviction of URA3 marked plasmids when necessary (Sikorski and Boeke, 1991). Diploid cells were induced to sporulate by growth overnight on Pre-Sporulation (Pre-Spo) plates (0.8% Bacto-yeast extract, 0.003% Bacto-peptone, 10% dextrose) followed by 2-3 days of growth on sporulation (SpoX) plates (1% KOCH3, 0.025 mg/ml ZnOCH3).

Plasmid propagation in bacteria was carried out by growth at 370 C in Luria-Bertani media (1% Bacto-tryptone, 0.5% Bacto-yeast extract, 1% NaCI, pH 7.5, and 2% Difco-agar for plates) supplemented with 100 pg/ml ampicilin (Sigma) or 40 pg/ml kanamycin (Roche) as required.



Yeast Strains and Plasmids

The yeast strains and plasmids described in this study are listed in the appendix. For preparation of full-length recombinant Nab2p, primer set MSS1131/MSS1132 was designed to amplify the NAB2 gene from L4717 genomic DNA with sequence encoding six histidine residues (His6) fused to the 3' end of the open reading frame. The PCR product was cloned into pGEX-4T-1 (Amersham Pharmaciea Biotech) between the EcoRI and Sail sites. The resulting pGST-Nab2-His6 plasmid was transformed into BL21-CodonPlus (DE3) cells (Stratagene). The expressed GST-Nab2-His6 protein was purified using standard gluatathione-Sepharose (Amersham Pharacia Biotech) and His-Bind (Novagen) chromatographic techniques.

The nab2ApPAB1NL strain was constructed by amplification of the PAB1 open reading frame from L4717 genomic DNA using MSS1261 and MSS1259 followed by insertion into Xhol-BstEll-cut pYES2/SPB4 (Invitrogen) to generate an NLS-PAB1-V5 gene fusion (pPAB1.3). NLS-PAB1-V5 was excised from pPAB1.3 by Smal-Xbal digestion and inserted into pRS315 (LEU2, CEN, [(Sikorski and Hieter, 1989)]) to generate pPAB1.4. The PAB1 5' and 3' untranslated regions (UTRs) were amplified by PCR from L4717 genomic DNA using MSS1287/MSS1263 and MSS1268/MSS1288 primer pairs, respectively. The 3' UTR PCR fragment was inserted between Xbal and Sacl sites and the 5' UTR between Sail and Smal sites of pPAB1.4 to generate the pPAB1.6 plasmid. The mutant NLS (mnl) sequence was fused to the 5' of the PAB1 ORF by PCR amplication from L4717 genomic DNA using primers MSS1262 and MSS1259.


The PABmni, DNA fragment was inserted into Smal-Sphl-cut pPAB1.6, resulting in pPAB1.7. The pPAB1.6 and pPAB1.7 plasmids were transformed into the nab2ApGAL::NAB2 shuttle strain (YRH201c) to create the nab2ApPAB1NL (YKN105) and nab2ApPABlmnl (YKN106) strains, respectively. Transformed cells were initially plated on SG media without leucine (SG-Leu) and then on YPD to test for nab2A suppression. For YKN105, cells were also streaked to 5FOA plates to select for the loss of pGAL::NAB2 (URA3, CEN, (Anderson, 1995)).

The NAB2 allele for tandem affinity purification (NAB2T) was constructed by amplification of the calmodulin binding peptide (CBP), tabbaco etch virus (TEV) protease site, protein A (ProtA) sequences from pBS1479 (Rigaut et al., 1999) using primers MSS811 and MSS812. The resulting PCR fragment also contained 50 nucleotides of homology to the 3' end of the NAB2 open reading frame fused in-frame to the TAP tag followed by sequence encoding the TRP1 gene and another 50 nucleotides of homology to the 3' UTR of NAB2. After transformation of the DNA fragment into L4717 cells, tryptophan auxotrophs were selected and expression of the Nab2pT fusion protein (YKN206) was confirmed by immunoblot analysis using aNab2 3F2 (1:1000).

For tandem affinity purification of the Nab2NpT and nab2-20NpT peptides, the 5' and 3' UTRs of NAB2 were amplified from L4717 genomic DNA with primer sets MSS1356/MSS1573 and MSS1384/MSS1358, respectively. Insertion of the 5' UTR fragment into Sall-Pstl-cut and the 3' UTR fragment into BamHI-Sacl-cut pRS315 yielded plasmid pNAB2.46. The TAP tag was excised from pBS1479


(Rigaut et al., 1999) using BamHI and Hindlll restriction enzymes and inserted into pNAB2.46 to generate pNAB2.47. Primers with sequence encoding the SV40 NLS and c-Myc epitope in-frame with the open reading frame of NAB2 (MSS1574 and MSS1575) were created and used to amplify the wild-type NES of NAB2 from L4717 genomic DNA and the mutant NES of nab2-20 from pnab220. The pNAB2.47 plasmid was digested with Pstl and BamHI and the wild-type NES (pNAB2.48) and mutant NES (pNAB2.54) fragments were inserted into the digested plasmid. The NAB2NT (YKN284) and nab2-2ONT (YKN287) strains were created by transformation of pNAB2.48 and pNAB2.54 into L4717, respectively. An immunoblot analysis of the YKN284 and YKN287 strains was performed using cNab2 3F2 to confirm expression of the tagged peptides in vivo (data not shown).

Strains expressing the AD-NT1 and AD-NT2 peptides for epitope mapping experiments were prepared by transforming pNAB2.42 and pNAB2.41 into AH109, respectively. The first N-terminal truncation of NAB2 (NT1) was amplified from L4717 genomic DNA using primer set MSS1423/MSS1444 and the second N-terminal trunction (NT2) was amplified from L4717 genomic DNA using primers MSS1423 and MSS1450. Both fragments were inserted into EcoRl-BamHI-cut pGADT7 (Clontech) to give plasmids pNAB2.42 (NT1) and pNAB2.41 (NT2).

The in vitro glutathione S-transferase pull-down experiments were performed by creating plasmids containing GST-LRR (pMEX67.5), GST-SUB2 (pSUB2.5), GST-PAB1 (pPAB1.8), GST-YRA1 (pYRA1.17), and GST-NPL3 (pGST-NAB1) gene fusions followed by transformation into BL21-CodonPlus


(DE3) cells for inducible expression of the fusion proteins. Briefly, the LRR of MEX67 was amplified from L4717 genomic DNA with primers MSS1661 and MSS1662 and inserted into EcoRI-Sall-cut pGEX-4T-1. The open reading frames of SUB2 and PAB1 were amplified from L4717 genomic DNA using primer sets MSS1729/MSS1730 and MSS1727/MSS1728, respectively. SUB2 and PAB1 were inserted into pGEX-4T-1 after digestion with BamHI and Xhol. Since the YRA1 gene contains a single intron, El was amplified from L4717 genomic DNA using primer set MSS1539/MSS1540 and E2 with primers MSS1541 and MSS1542. The El PCR fragment was inserted into BamHIEcoRI-digested pGEX-4T-1 to give pYRA1.16 and E2 was inserted into EcoRIXhol-cut pYRA1.16, creating pYRA1.17. The NAB2 N-terminal fragment was excised from pNAB2.41 by digestion with Ncol and BamHI and inserted into pET15b (Novagen) to create pNAB2.58. The N-terminal region of nab2-20 was amplified from pnab2-20 (Anderson, 1995) with MSS1423 and MSS1444 and inserted into EcoRI-BamHI-cut pGADT7 to create pNAB2.45. The mutant Nterminal fragment was excised from pNAB2.45 by digestion with Ncol and BamHI and inserted into pET15b to create pNAB2.59. The pNAB2.58 and pNAB2.59 plasmids were transformed into BL21-CodonPlus (DE3) cells for inducible expression of the Nab2pN and Nab2-20pN fragments, respectively.

Nucleic Acid Isolation Procedures

Yeast genomic DNA was isolated from 5 ml of logarithmically growing (OD600 = 0.5 2 U/ml) L417 cells. Cells were pelleted by centrifugation at 1,500 x g for 5 minutes. The pellet was re-suspended in 0.5 ml of cell wall digestion


buffer (800 mM Sorbitol, 100 mM EDTA, 30 mM DTT) and 50 pl of 5 mg/ml Zymolyase 100T (Seikagaku Corporation). Cell walls were digested for 1 hour at 370 C and spheroplasts were harvested by centrifugation at 3,000 x g for 1 minute. Spheroplasts were re-suspended in 0.5 ml of lysis buffer (50 mM Tris, 20 mM EDTA, 1% SDS) and incubated at 650 C for 30 minutes with mixing every 10 minutes. Proteins and cell debris were precipitated by incubating on ice in the presence of 1.8 M KOAC and pelleted at 40 C by centrifugation at 14,000 x g for 15 minutes. The supernatant was transferred to a fresh tube and DNA was precipitated in 50% isopropanol for 10 minutes at 240 C. The DNA was pelleted at 14,000 x g for 1 minute, washed and dried in a SpeedVac (Savant). After resuspension of the pellet in 250 p of TE buffer (10 mM Tris-HCI pH 7.5, 1 mM EDTA), RNA was digested at 370 C for 1 hour using 200 ng of RNase A. Proteins were extracted with phenol and genomic DNA was precipitated using 0.3 M sodium acetate. DNA was recovered by centrifugation at 14,000 x g, washed, dried and resuspended in 100 pl of TE. The concentration of genomic DNA was determined by measuring the absorbance of 260 nm light (OD260). Previously described methods were used to recover plasmids from yeast cells (Strathern and Higgins, 1991) and bacterial cells (Sambrook et al., 1989).

Total RNA was extracted from yeast cells using 650 C acid equilibrated phenol as previously described (Guthrie and Fink, 1991). Briefly, logarithmically growing cells were harvested by centrifugation at 2,000 x g for 5 minutes and resuspended in AE lysis buffer (50 mM NaOC3H, pH 5.3, 10 mM EDTA, 1% SDS). Proteins and cell debris were extracted from the supernatant at 650 C with AE


equilibrated phenol. RNA was precipitated at -700 C for 1 hour using 2 M ammonium acetate. After centrifugation, pelleted RNA was re-suspended in diethyl pyrocarbonate (DEPC) treated d2H20 and the concentration was determined by measuring the OD260.

Cell Transformation

Transformation of yeast cells was performed by treatment with lithium acetate (Ito et al., 1983). Cells were grown in YPD or SD to an OD6o0 = 0.5 U/ml. A total of 10 ml for each transformation was harvested by centrifugation at 1,500 x g for 5 minutes. Cells were washed with d2H20 and harvested again by centrifugation. The cell pellet was re-suspended in 700 pI of buffer (Tris-EDTA, 0.1 M lithium acetate, 43% polyethylene glycol molecular weight 3350) containing 5 pg of denatured calf thymus DNA and 1-5 pg of plasmid DNA. Incubation was carried out at 240 C for 30 minutes, rotating end over end. At last, cells were heat-shocked at 42' C for 15 minutes, washed, and plated on SD plates supplemented with the appropriate amino acids.

Electro-competent bacterial cells (DH5a) were transformed with plasmid DNA (1-100 ng) using an electric current generated by a BioRad GenePulser (2.5 kV, 200 Q, 25 pFd). After electro-poration and DNA uptake, cells were allowed to recover in LB media at 370 C for 1 hour. The cell suspension was plated on LB supplemented with appropriate antibiotics and incubated overnight at 370 C to allow for colony formation.


Yeast Genetic Manipulations

Mating, sporulation, and tetrad dissections were all performed as described (Guthrie and Fink, 1991). Haploid cells of complementing mating type (mat a and mat a) strains were patched on YPD plates at 240 C for 1-2 days. Diploids were selected on SD media containing the appropriate amino acids. Spore formation was achieved by incubating diploids on Pre-Spo plates for one day and SpoX plates for two days. Cell walls were digested for 10 minutes in 1.0 M sorbitol and 0.5 mg/ml Zymolyase 20T (Seikagaku Corporation) releasing individual spores for micromanipulation. Dissected spores were allowed to germinate and form colonies on YPD plates at 240 C unless otherwise indicated. Colonies were tested for various auxotrophic markers and growth at specified temperatures. The inability of individual spores to form colonies was also noted.

Yeast Total Cell Protein Isolation

Five OD600 units of yeast cells were harvested by centrifugation at 2,000 x g. Cells were washed once with d2H20 and re-suspended in 200 pl of 10% trichloroacetic acid (TCA) (Fisher) in a microfuge tube. Acid-washed beads were added to the cell suspension. The tubes were vortexed vigorously for 10 seconds and chilled on ice for 15 seconds. Vortexing was repeated for a total of six times. The lysate was removed from the beads and subjected to centrifugation at 14,000 x g for 10 minutes. The protein pellet was recovered after centrifugation and re-suspended in 100 pl of Laemmli cocktail (10 mM TrisHCI pH 6.8, 4% SDS, 20% glycerol, 1.3 M 1-mercaptoethanol, 4 mg/ml Bromphenol blue). The protein cocktail was neutralized with 1 M Tris-base and


boiled for 3 minutes prior to fractionation on 12.5% polyacrylamide-1% SDS gels, unless otherwise noted.

Fluorescence In Situ Hybridization and Cellular Immunofluorescence

In order to determine the subcellular localization of proteins and poly(A)* RNA, yeast cells were grown to an OD600 = 0.1 0.5 U/ml. After harvesting by centrifugtion at 1,000 x g for 10 minutes, cells were fixed in freshly prepared 4% formaldehyde (from EM grade paraformaldehyde) for 2 hours with gentle rocking. Fixed cells were pelleted at 800 x g for 2 minutes and washed twice with WB1 (100 mM KH2PO4, pH 6.5) and once with WB2 (100 mM KH2PO4, pH 6.5, 1.2 M Sorbitol). Cells were harvested as before and re-suspended in 1 ml of WB2 plus 30 mM P-mercaptoethanol. Spheroplast formation was achieved by the addition of 300 ng of Zymolyase 100T and incubation at room temperature for 30-45 minutes. After greater than 90% of the cells no longer contained cell walls as determined by phase-contrast microscopy (400X magnification), cells were collected by centrifugation at 2000 x g for 1 minute. Cells were washed with WB2 and attached to polylysine-coated 10-well HTC Blue Slides (Cell-Line Associated). Slides were washed once in ice-cold PBS (150 mM NaCI, 3 mM KCI, 10 mM Na2HPO4, 2 mM KH2PO4, pH 7.4) and dipped in 100% methanol for 5 minutes at -20o C followed by 100% acetone for 30 seconds. After three washes with PBS, slides were placed in PBS plus 0.1% Triton-X (Sigma) for 5 minutes and washed three more times with PBS. Slide wells were blocked for non-specific antibody interactions by the addition of 3 % BSA/PBS for 30 minutes at 240 C. Primary antibodies were diluted in 3% BSA/PBS (ctNab2, 1:500;


cPabl, 1:5000; aNopl, 1:500; aV5, 1:500; OaDig, 1:100) and added to respective wells for a 60 minute incubation at 240 C. Poly(A)* RNA was detected by fluorescence in situ hybridization (FISH) using a d(T)50-digoxigenin conjugated probe according to a previously published procedure (Amberg et al., 1992). The Alexa Fluor 488 goat anti-mouse IgG, secondary antibody was used to detect aNab2, aPabl, aV5 and aDig. For co-localization of poly(A) RNA and Noplp, the anti-Nopl A66 antibody (gift from John Aris, University of Florida, Gainesville, FL) was detected by the Alexa Fluor 555 goat anti-mouse IgG2A secondary antibody (Molecular Probes). DNA was detected by 0.5 pg/ml 4'6-diamidino-2phenylindole (DAPI). Cells were visualized using an Axiophot (Zeiss) microscope equipped with a X100 fluorescence/differential contrast objective and images were captured by either monochrome (Zeiss) or color (Spot) digital cameras.

In Vitro 3' End Processinq Assays

Standard 3'-end processing assays were performed essentially as described (Minvielle-Sebastia et al., 1997) with 2 pl of either YAS394 (pablA) or BMA64 (PAB1) cell extracts. Reconstitution assays with purified factors were performed with 0.5 pl of partially purified CF I (Mono Q fraction from CFII purification; (Minvielle-Sebastia et al., 1998)) and 0.5 pl of CPF (affinity purified from strain YSD10 expressing a TAP-tagged Fiplp fusion protein; unpublished data). Reactions were supplemented with either 25-800 ng of GST-Nab2p-His6, 50-800 ng of GST-Nab4p (Minvielle-Sebastia et al., 1998), or 50-1000 ng of purified bovine PABPN1 (gift of Elmar Wahle, University of Halle-Wittenberg,


Halle Saale, Germany). The polyadenylation time course experiment was performed with a 12-fold reaction mixture of YAS394 extract supplemented with 9.6 pg of GST-Nab2p-His6 and incubated at 300 C for 15 minutes without RNA substrate. Following addition of labeled transcript, aliquots of 20 pl were withdrawn at the specified times. Reactions were analyzed on 40 cm long 6% polyacrylamide-8.3 M urea gels.

Filter Bindinq Assays

Standard filter binding assays with r(A)25 (Dharmacon Research) and either GST-Nab2p-His6 or GST-His6 were performed as previously described (Wong and Lohman, 1993). Binding reactions were incubated at 300 C for 30 minutes in 20 mM Tris-HCI pH 7.4, 100 mM KCI, 2.5 mM MgCI2, 0.5 U/pl Superase-In (Ambion) with 5 pM 32P end-labeled A25 RNA and increasing amounts of purified GST-Nab2p-His6 or GST-His6 fusion proteins. Protein-RNA complexes were filtered on 0.45 micron nitrocellulose (Osmonics) and bound versus unbound RNA fractions were quantified using a Storm phosphorlmager (Molecular Dynamics). Each point on the curve represents the mean of three independent experiments.

Tandem Affinity Purification

Total protein extracts of YKN206 were prepared by growing cells at 300 C in 2.5 L of YPD medium to an OD600 = 2-4 U/ml. Cells were harvested by centrifugation and resuspended in 1 volume of Buffer A(TAP) (10 mM K-HEPES, pH 7.9, 10 mM KCI, 1.5 mM MgCI2, .5 mM DTT) with protease inhibitors (.5 mM PMSF, 2 mM Benzamidine, 1 [tM leupeptin, 2 RM pepstatin, 4 RM chymostatin,


2.6 pM aprotinin). Cells were lysed by two subsequent passages through a French-press at 1,200 PSI. After lysis, KCI was added to a final concentration of 200 mM and the supernatant was clarified by centrifugation at 60,000 x g for 30 minutes and 130,000 x g for 90 minutes. The supernatant was dialyzed against 1000X volume of Buffer D(TAP) (20 mM K-HEPES, pH 7.9, 50 mM KCI, .2 mM EDTA, .5 mM DTT, 20% v/v glycerol) with protease inhibitors (.5 mM PMSF, 2 mM Benzamidine). Protein extracts were added to 200 pl of IgG-Sepharose beads (Amersham) and incubated at 40 C for 2 hours. The beads were washed three times with IPP150 Buffer (10 mM Tris-CI, pH 8.0, 150 mM NaCI, .1% NP40) and then incubated with 1 ml of TEV Cleavage Buffer (10 mM Tris-CI, pH 8.0, 150 mM NaCI, .5 mM EDTA, .1% NP40, 1 mM DTT) and 100 units of TEV protease at 160 C for 2 hours. Three volumes of calmodulin binding buffer (10 mM Tris-CI, pH 8.0,150 mM NaCI, 1 mM MgOAC, 1 mM imidazole, 2 mM CaCI2, .1% NP40, 10 mM P-mercaptoethanol) and 200 pl of calmodulin affinity resin (Stratagene) was added to the TEV eluate and incubated at 40 C for 1 hour. After three washes with calmodulin binding buffer, Nab2pCBP was eluted using 800 pil of calmodulin elution buffer (10 mM Tris-CI, pH 8.0, 150 mM NaCI, 1 mM MgOAC, 1 mM imidazole, 2 mM EGTA, .1% NP40, 10 mM P-mercaptoethanol). For protein analysis, 400 pl of the calmodulin eluate was precipitated using 10% TCA, pelleted at 14,000 x g, and re-suspendend in 10 pl of Laemmli cocktail. The protein cocktail was neutralized with 1 M Tris-base and boiled for 3 minutes prior to fractionation on a 12.5% polyacrylamide-1% SDS gel. Proteins were visualized using either Coomassie blue stain (2.5 mg/L Coomassie blue dye


[BioRad], 50% methanol, 10% acetic acid) or silver stain (0.1% Silver Nitrate). Proteins were identified by either matrix assisted laser desorption time-of-flight mass spectrometry (MALDI-TOF MS) in collaboration with John Aitchison and David Dilworth (Institute for Systems Biology, Seattle, WA) or immunoblot analysis as previously described (Anderson et al., 1993b) using the following antibodies and dilutions: aNab2 3F2 (1:1000), czNab2 1F2 (1:500), aKapl04 1D12 (1:1000), aPabl 1G1 (1:1000), aNab3 2F12 (1:250), aNab4 3H1 (1:500), oNpl3 (1:500) (gift from Pamela Silver, Harvard Medical School, Boston, MA), aMex67 (1:500) (gift from Ed Hurt, University of Heidelberg, Heidelberg, Germany), aYral (1:1000) (gift from Douglas Kellogg, University of California, Santa Cruz, CA) and horseradish peroxidase-conjugated sheep anti-mouse secondary (1:5000) (Amersham Pharmacia Biotech).

Preparation of Polyclonal Antisera and Monoclonal Antibodies

The University of Florida Interdisciplinary Center for Biotechnology Research (ICBR) hybridoma laboratory was responsible for preparation of the Nab2pTAP polyclonal antiserum and aKapl04 1D12 monoclonal antibody. Two injections (5 hg) of purified Nab2pTAP were performed on BALB/c mice (Harlow and Lane, 1988). Ten days after the second injection, a preliminary bleed was performed and dilutions (1:100, 1:250, 1:500) of the blood were tested for immunoreactivity against Nab2pT-associated proteins. The polyclonal antiserum (1:500) was used to screen a Xgt 11 expression library for recombinant yeast genes as previously described (Snyder et al., 1987). Positive plaques were purified, cloned and


sequenced. Identification of genes was accomplished by searching the Saccharomyces Genome Database (SGD) for matching DNA sequences.

To isolate the aKapl04 monoclonal antibody, spleen cells of the mouse were fused with SP2/O myeloma cells as previously described (Harlow and Lane, 1988). The 1D12 monoclonal antibody reacted against Xgt 11 phage expressing KAP104.

Poly(A) Tail Length Determination

Total RNA was prepared from whole cells and poly(A) tail lengths were determined as previously described (Minvielle-Sebastia et al., 1998). Briefly, total RNA (1 pg) was 3'-end labeled at 0o C for 20 hours with [32P]pCp (NEN) using 20 U of T4 RNA ligase (New England Biolabs) in a 30 pl reaction mix (50 mM HEPES pH 8.3, 5 pM ATP, 10 mM MgCI2, 3.3 mM DTT, 10% DMSO, 300 pg/ml acetylated BSA). Non-poly(A) RNA was digested at 370 C for 2 hours using RNases A (50 ng/pl) and T1 (1 U/pl) in 10 mM Tris-HCI pH 7.5 and 300 mM NaCI. The digestions were stopped by the addition of EDTA, proteins were extracted with phenol and RNA was precipitated using 2 M ammonium acetate and 5 pg of RNase-free glycogen (Boehringer Mannheim). Poly(A) tails were fractionated on 8% polyacrylamide-8.3 M urea gels and visualized by autoradiography.

In Vitro GST Pull-down Experiments

Direct interactions between the N-terminus of Nab2p and either the LRR of Mex67p, Pablp or Sub2p were demonstrated using the ProFoundM Pull-Down GST Protein:Protein Interaction Kit (Pierce Biotechnology). Induction of


recombinant proteins in strains carrying pNAB2.58 (EKN102), pNAB2.59 (EKN103), pMEX67.4 (EKN104), pSUB2.4 (EKN111), pPAB1.8 (EKN110), pGST-NAB1 (EKN109), pYRA1,17 (EKN114) or pGEX-4T-1 (EKN101) plasmids was achieved by adding 1 mM IPTG (Sigma) to TB+K (1.2% Bacto-tryptone, 2.4% Yeast Extract, 17 mM KH2PO4, 85 mM K2PO4, 0.4% glycerol, pH 7.0) growth media and growing cells at 370 C for 2 hours. Cells were pelleted and resuspended in 200 pl of TBS (25 mM Tris-HCI pH 7.2, 150 mM NaCI) with protease inhibitors (5 pg/ml pepstatin A, 1 pg/ml chymostatin, 1 mM Eaminocaproic acid, 1 mM p-aminobenzamidine, 1 pg/ml leupeptin, 2 pg/ml aprotinin). Cells were lysed by adding 200 pl of ProFound'" Lysis Buffer (Pierce Biotechnology) and the supernatant was clarified by centrifugation at 14,000 x g for 5 minutes. GST fusion proteins were bound to glutathione-Sepharose beads (Amersham Pharmacia Biotech) by incubating 800 pl of supernatant with 30 pl of beads at 40 C for 2 hours. After three washes with 800 pl of wash buffer (50% ProFound'TM Lysis Buffer, 12.5 mM Tris-HCI pH 7.2, 75 mM NaCI), an equal volume of either EKN102 or EKN103 cell lysate was added to GST pre-bound Sepharose beads. Binding reactions were carried out at 40 C for 3 hours. Beads were washed three times with wash buffer and proteins were eluted by boiling at 1000 C for 3 minutes in 20 pl of Leammli cocktail. One-half of each binding reaction was loaded onto a 12.5% polyacrylamide-1% SDS gel and proteins were detected using either aNab2 1F2 or aGST (gift from the University of Florida ICBR hybridoma core).


Phosphatase Treatment with X Phosphatase

For phosphatase treatment of Nab2p in mRNA export mutant strains, cells (5 OD600 units) were lysed by glass bead vortexing in 250 pi of lysis buffer (50 mM Tris-HCI pH 7.5, 100 mM NaCI, 10 mM imidazole, 1 mM EDTA, 0.2 mM Na3VO4, 10 mM NaF, 0.5 mM PMSF, 5 pg/ml pepstatin A, 1 pg/ml chymostatin, 1 mM E-aminocaproic acid, 1 mM p-aminobenzamidine, 1 pg/ml leupeptin, 2 pg/ml aprotinin). De-phosphorylation was performed by adding 400 units of X phosphatase (New England Bio-labs) to 5 pi of lysate according to the manufacturer's instructions. Reactions were stopped by precipitation of proteins in 10% TCA. Proteins were re-suspended in Laemmli cocktail and fractionated on 8% polyacrylamide-1% SDS gels. The Nab2p isoforms were detected by immunoblot analysis using aNab2 3F2. UV-light induced photocrosslinking was performed as described (Wilson et al., 1994).


Research Obiectives

Efficient and reliable gene expression in eukaryotic cells requires precise temporal and spatial coordination of many nuclear and cytoplasmic events. Nuclear events include mRNA transcription, capping, splicing, polyadenylation, and nuclear export whereas protein translation, and mRNA turnover occur in the cytoplasm. RNAP II coordinates nuclear processing events with transcription by recruiting the appropriate protein complexes to the 5' ends, introns and 3' ends of pre-mRNA (reviewed in Bentley, 2002; Hirose and Manley, 2000; Proudfoot et al., 2002; Steinmetz, 1997). In addition, proteins that bind to the 7methylguanylate cap stimulate splicing, polyadenylation and nuclear export of mRNA (Lewis and Izaurralde, 1997; Shatkin and Manley, 2000). Factors required for efficient splicing, in turn, interact with proteins required for polyadenylation and mRNA export (Lei and Silver, 2002; Luo et al., 2001; Lutz et al., 1996; Strasser and Hurt, 2001; Takagaki et al., 1996; Vagner et al., 2000) providing further evidence that nuclear processing events, from transcription to nuclear export, are intimately linked.

During transcription an abundant class of heterogeneous RNA binding proteins (hnRNPs) also associate with pre-mRNA and are believed to play a role in the coordination of pre-mRNA processing events and nuclear mRNA export by



binding to specific RNA sequences and serving as export adaptors for karyopherins (Dreyfuss et al., 2002; Dreyfuss et al., 1993; Krecic and Swanson, 1999; Lei and Silver, 2002; Nakielny and Dreyfuss, 1999). This view of hnRNPs as active participants in pre-mRNA processing, particularly mRNA export, is controversial. Reed and Magni (2002) have proposed that hnRNPs bind primarily to introns and need to be removed from the mRNA-Particle (mRNP) before export while splicing-dependent SR proteins and conserved mRNA export factors (Yralp and REF/Aly) actively recruit nuclear export receptors (Mex67p and TAP) to the mRNP. An alternative possibility is that hnRNPs act at the interface between nuclear RNA processing events and mRNA export to precisely orchestrate the dissociation of the RNA processing machinery prior to the recruitment of export receptors. The goal of my thesis research has been to test the latter hypothesis using the yeast hnRNP Nab2p.

Nab2p was first identified as a nuclear polyadenylated RNA binding protein required for vegetative growth in yeast (Anderson et al., 1993b). In addition to loss of viability, depletion of Nab2p resulted in an accumulation of hyperadenylated RNA in the nucleus (Hector, 2000; Hector et al., 2002). These phenotypes mirror those for several mRNA export mutants, including prp20, mtr2, rat7, rat8, mex67, glel, and rip1 (Forrester et al., 1992; Hilleren and Parker, 2001; Jensen et al., 2001b; Kadowaki et al., 1994; Tseng et al., 1998). Unique to NAB2, however, a mutant was isolated that uncouples the polyadenylation defect from the export block. At temperatures permissive for growth, nab2-21 cells exhibit hyperadenylated tails but do not accumulate


poly(A) RNA in the nucleus, whereas at temperatures restrictive for growth, poly(A) RNA export is also blocked. Interestingly, PABI was isolated as a high copy suppressor of nab2A growth defects but increased expression of Pabl p did not resolve the hyperadenylation defect and only mildly restored poly(A) RNA export to wild-type levels (Hector, 2000; Hector et al., 2002). Altogether, these results suggest Nab2p performs dual functions in vivo. First, Nab2p terminates polyadenylation, possibly analogous to PABPN1, by directly binding to the polyadenylate tail. Secondly, after binding and restricting tail length, Nab2p coordinates the release from RNAP II and serves as an adapter for the nuclear export of poly(A) RNA. The following specific aims were designed to provide evidence for this model: (1) demonstrate Nab2p, like PABPN1, binds to poly(A) RNA homopolymers and limits poly(A) tail length in vitro; (2) determine if NAB2, but not PABI, is responsible for limiting poly(A) tail lengths and promoting mRNA export in vivo; (3) identify the factors, in combination with Nab2p, involved in the nuclear export of polyadenylated mRNA.

Nab2p Binds Poly(A) RNA and Limits Poly(A) Tail Length In Vitro

The maturation of eukaryotic mRNA 3' ends first requires cleavage at specific sequences upstream of the transcriptional termination site by a group of cleavage factors, followed immediately by the addition of adenosine residues to the cleaved pre-mRNA (for reviews see Colgan and Manley, 1997; Proudfoot and O'Sullivan, 2002; Wahle and Keller, 1996; Zhao et al., 1999). While the number of adenosines added to a given mRNA can vary slightly depending on the cell type, developmental stage and growth conditions, poly(A) tails are generally


found to be between 200-300 nucleotides in mammals and 70-90 nucleotides in Saccharomyces cerevisiae (Wahle and Ruegsegger, 1999). When cells are depleted of Nab2p, however, poly(A) tail lengths increase dramatically up to 600 nucleotides (Hector et al., 2002).

Expression and Purification of Recombinant GST-Nab2p-His6 Protein

In order to determine if aberrant tail length regulation observed in nab2A cells is directly related to Nab2p function, a recombinant Nab2p (rNab2p) was produced and used for in vitro studies. Briefly, DNA sequence encoding a glutathione S-transferase (GST) protein was cloned 5' of the NAB2 start codon to generate a GST-Nab2p fusion product. The GST-Nab2p was expressed in E. col cells and purified using a glutathione column. Unfortunately, a majority of the proteins eluted from the glutathione column were breakdown products of Nab2p (Figure 5A, GST-Nab2p lane). In an attempt to purify full-length rNab2p away from the degradation products, six histidine (His6) residues were fused to the carboxyl terminus of GST-Nab2p to allow for an additional purification step using a nickel column. The use of both tags for purification purposes greatly increased the amount of full-length rNab2p that was recovered (Figure 5A, GST-Nab2p-His6 lane).

rNab2p Binds with Higqh Affinity to Poly(A) RNA Homopolymers

The affinity of Nab2p for poly(A) RNA was determined using a standard filter-binding assay and (rA)25 (Wong and Lohman, 1993). As shown in Figure 5B, the observed dissociation constant (Kd) was 10.5 nM (blue squares). Importantly, GST-His6 protein had little or no affinity for poly(A)Y RNA (Figure 5B,

cn s- (D &C M 0 0 E 75 IF =3
CD CL 04 0 -0 cm CU Z CD
.C Im
C: cu

04 CU >
_0 M
cu Z E
U) C/)

4-' a) 0 c:
0 cu

-C (D

M CU >, (n 0 CU CD
Lo "
cn CU
cn c

E 04 :t:! z >..p =
6 CU

E -co
0 LU
-C o
U) a

z L) ly- "0 00 C)0 04
CL -0 -r- z 7Z- 4CL 0) C-) c : O 0

0 C/)
Cl) CU
Cl) M
U) LZ "a c
_0 0) a cu
CL U) (N = = cu Lo
=3 a) C14
cn >



U? C>


M Ir
ca C:)
Z z


C cc
Z C:)

Co 0 C 0
co 0 lw 04

punog VNN %

.. ... ... .

W) LO 0 Ln m


green circles) and bovine poly(A)-binding protein (PABPN1) has been shown previously to have an affinity for (rA)25 (Kd = 7 nM; Gorlach et al., 1994) comparable to Nab2p. Considering Nab2p binds strongly to poly(A) RNA and is localized to the nucleus of yeast cells (Anderson et al., 1993b), we reasoned that Nab2p functions to restrict poly(A) tail lengths. In vitro assays were utilized to assess the role of Nab2p in restricting poly(A) tail length during polyadenylation. Nab2p Restricts Poly(A) Tail Lenqth In Vitro

Although cleavage and polyadenylation reactions are tightly coupled in vivo, these two activities can be separated in vitro. A labeled iso-1-cytochrome c (CYC1) RNA is cleaved and polyadenylated when added to yeast whole cell protein extracts (WCE) (Butler and Platt, 1988). Remarkably, the addition of cordycepin triphosphate or substituting CTP for ATP in WCE inhibits polyadenylation but not cleavage of the precursor CYC1 RNA (Butler et al., 1990). Alternatively, if the precursor CYC1 RNA is pre-cleaved (CYC1pre) at the normal cleavage position using a convenient restriction site, this CYC1pre RNA is still polyadenylated (Lingner et al., 1991a; Lingner et al., 1991b). If pablA extracts are used, tail lengths increase to about 200 nucleotides. The addition of recombinant Pablp to these extracts restricts poly(A) tail lengths to about 60 nucleotides (Amrani et al., 1997; Minvielle-Sebastia et al., 1997). It has been demonstrated that Pablp binds to poly(A) RNA and recruits poly(A) nucleases (Pan2p and Pan3p) to the tail which degrades the poly(A) RNA in a 3' to 5' manner (Boeck et al., 1996; Brown et al., 1996). These authors propose poly(A) tails are normally synthesized very long but Pablp gradually deadenylates tails


through the activity of Pan2p and Pan3p (Brown and Sachs, 1998). Maximum poly(A) tail length (80-100 nucleotides) for pabI and pan mutants observed in vivo, however, does not correspond to the maximum tail length exhibited in vitro (greater than 200 nucleotides) (Brown and Sachs, 1998; Minvielle-Sebastia et al., 1997) suggesting an additional factor(s) is required for poly(A) tail length regulation in vivo. To determine if Nab2p can restrict poly(A) tail length in vitro, in collaboration with Lionel Minvielle-Sebastia (University of Bordeux, France), increasing amounts of Nab2p was added to YAS394 (pablA/rrp46A) extracts and 32P-labeled CYC1pre RNA poly(A) tail lengths were examined. Note that in the presence of Pabip (Figure 6A, PAB1 control lane), poly(A) tail lengths were restricted to about 60 nucleotides but there was relatively inefficient conversion of precursor to oligo(A) and from oligo(A) to poly(A) tails. In the pablA strain, polyadenylation was unrestricted with very long tails (greater than 200 nucleotides (Figure 6A, pablA control lane). Importantly, the addition of Nab2p (Figure 6A, 400-800 ng), restricted tail lengths to approximately 90 nucleotides. Surprisingly, the addition of low levels of Nab2p stimulated polyadenylation (note the low levels of precursor and oligo(A) RNA up to 200 ng) (Figure 6A). At 400 ng and greater, Nab2p inhibited polyadenylation. This could be due to the sequestration of a factor required for efficient polyadenylation in vitro. Indeed, Nab2p interacts with the CF IB (Nab4p) in the yeast two-hybrid system (Krecic, 1998; Uetz et al., 2000). Nab4p is required for cleavage site selection and stimulates polyadenylation in combination with CF IA and PAP (Kessler et al., 1997; Minvielle-Sebastia et al., 1998). As shown in Figure 6B, the addition of

(1) 70 L- cn cn
0 4- 4C-) Z
Co (U
:E CL A?
c a) C x c co CL cu LL,

0 > 0
o > 0 0 0 C: = .0
< cu cn
a C
CD 0 C: M :3
CU (D 0
o L- CU -0
(n -0 >, CU M 0
c: E (n
:3 -0 a) 0 0 0 0-0
U) -0 a 0
a) cu -r, CL =3 cn n
cu a) 0
v- (D CL L) c >.
E c <
u) E c)
U) (D 0
CL _0 I- 0
> (D
cu CCU) w E z Lo Z 4-- a0 L) (n

UJ -0 -0
0 0) CU CU
z rn
> cn <
0 co E z
-0 cu CL (D O
a) -:= "It -a .0
x 10.0
-C E a) U) cu 0. a)
Z :3
C: M
0 C\j a)
-0 U) ('I-)
29 (D V- CU 0 :;t -0
00 Z cu
0 U)
< CL
U) 0) LU a
-0 a C:) (L) C
0 cu
c cu 0) 00 U) 70
C: CU (1)
C) Lo >- cu C) 0
U) 0 co "a
a) 0
> N- 'r.- a)
4 o C:,.o L-C (1) c cu C: C2
a) 0 -C
-o E '= c))
cr 0) a) (D Co a) co -5, c: U) c: a C:
< a) Cl a)
>- E =3 -o cu m CL CU a) (n CU 04 (D 0) cl. (D a) a) CU
-0 0. = C- 0.75 -0 CU 5, -E :3 5, a)
z -r- a) U) Z> U)



/0 c
%0 CO
0 AW

/0 09

04 00 r- Cl
C) C) q1t CV)
ce) 04 04 C14 CN


Cc 06
/0 0 9

04 r- Nt 0) C14 00 0 C:)
CN 04 C) C) It m C) 0) 00
Co LO ct) CN C14 C14 CA



CcaCz oe CL
-/0- VIAC14 t- -It m CNCO r- a C)
cli N C) Cl .tce) C) 0) 00
co V) 4qr C") C14 C*4 04 CN -


Nab4p to pablA extracts stimulates hyperadenylation in vitro suggesting Nab2p and Nab4p may have antagonistic functions in polyadenylation. Remarkably, the addition of low levels of bovine PABPN1 (up to 200 ng) restricts tail lengths in the yeast system while inhibiting polyadenylation at higher concentrations (greater than 400 ng) (Figure 6C).

These results indicate that Nab2p is necessary for proper tail length control in vitro using whole-cell extracts, but it was also of interest to determine if Nab2p is sufficient for tail length restriction in the absence of other cellular proteins. To accomplish this, rNab2p was added to highly purified cleavage and polyadenylation factors, CF I and CPF. As before, hyperadenylation of the CYC1pre RNA was observed in the absence of Pablp (Figure 7A) (Amrani et al., 1997; Minvielle-Sebastia et al., 1997). The addition of low levels (25-50 ng) of Nab2p increased polyadenylation efficiency and limited tails to -90 nucleotides in length while higher levels of Nab2p (100-200 ng) profoundly inhibited polyadenylation (Figure 7A). Importantly, this activity of Nab2p was not specific for the uncoupled polyadenylation reaction because using a non-cleaved CYC1 RNA Nab2p promoted polyadenylation while restricting tail lengths (Figure 7B).

The mechanism for poly(A) restriction by Pablp has been proposed to involve hyperadenylation followed by deadenylation (Brown and Sachs, 1998) whereas PABPN1 is believed to bind to the poly(A) tract during synthesis and terminate polyadenylation at a specific length (Wahle, 1995). To discriminate between these two potential models for Nab2p activity, the effect of 400 ng of Nab2p on polyadenylation in pablA extracts at various time points was

0 I 1 -0 < C) .s.. 0 0 c 00
CL LL U) cu
04 Q_ (!)
-0 1- -z
Ca (.) 4--
0 Lo
+ Cl)
o LL 0 (L)
0 >
C) C.) co >,
04 co a) C CL a) 04
70 04 CU CD
co C L- Z
a) =) 'a O- a)
8. c:
CU ca
>-. U)
(D a >.
-C qT C C)
:tt M a) 0
-0 CD cu -o "0 0) 0
O)Z CU (D C 0)
I >% .- c
c: > E
:E 0 co C)
cS (U.) c)- -a) E q
cn 0
(n co c:
0 >
c E
0 L
0 C a) cn
0 cn
0 0
E co
a) w :t-- -a Na 2 x 'a : =) .0 0. a)
0 CU *
3:-o Z 0
a) c
5% c C;)
0 W (D c:L
E 0
CD 0- c 0 CN cu
0-0 -0
cn c CU =3 cn
C6 CU z L) cu
-C 0 4- 0
C: 4- 0 q (D
>, 0 00 U)
cn cu I ,
4- cn CU -0-0 cm &- L- a)
0 CD a) 0 cu E
Co Co 0 C:

C: cn a)
a) Q) :tl. 70 0.
co 'C: C) CU (n 4-cu cu o
0 cn
_0 E .9 c: (D
co C%4 a) cn a)
L- o -0a) c co CL < E =3
Co CL 0
0. z :3
>% CD 0 U) CL cn
.r cn 0
U) CU (D X 0
> L- QD 4) 5
co 0 a) a)
0 C: :3
.c x
-0 a) 0 o
CL V- 0. 0 -0 L) a) C\l -m C14 m m 0
mu h,-- -2az IOU ;" -a>' E
z cL CL Z W a. i= E





06 09
OA co

L 0

-M CN 00 1,- C)
C) nt M a)

wkob ca

C*4 0
Z 06

A .4. 0) C14CO rCN 0 C) mt M C) 0) 00
LO m C14 C14 CN CN


Q) C: I

LL 0
z 00, + Z 09

u -,o T /0,,'

cv- It 0) CNOO r- C)C) CIJN 0 0 'Wrcl) C) 0)00 (OLO It ce) C1404 C\l C%4


determined (Figure 7C). Remarkably, even at the earliest time point (2 min), Nab2p restricted poly(A) tail length to 90 nucleotides. Over time (up to 180 minutes), no further decrease in poly(A) tail length was observed. These results are contrary to those obtained with 250 ng of Pablp in which poly(A) tails successively shorten over time (Minvielle-Sebastia et al., 1997) suggesting the mechanisms of poly(A) tail length regulation for Nab2p and Pablp are distinct. It is important to note that wild type extracts (Figure 6A and Figure 6C, PABI lanes) exhibit tail lengths about 20 nucleotides shorter than pablA extracts that have been augmented with 400 ng of Nab2p. This is likely to reflect the PablPan2/3 activity present in wild type extracts, but not in pablA extracts (Brown and Sachs, 1998; Brown et al., 1996). We conclude from these results that Nab2p is an essential nuclear poly(A)-binding protein required for efficient polyadenylation and poly(A) tail length restriction in vitro.

NAB2 Limits Poly(A) Tail Length and Promotes mRNA Export In Vivo

To explore the possibility that Nab2p is also responsible for poly(A) tail length restriction in vivo, a high-copy suppressor screen was previously employed to isolate proteins that could compensate for loss of Nab2p (Hector, 2000). Surprisingly, high-copy PAB1 was the only identified suppressor of the nab2A growth defect. Initially these results argued that the essential function of Nab2p is in poly(A) tail length control. Closer inspection of the nab2ApPAB1 phenotypes, however, revealed that increased PABI expression did not correlate with significantly shortened poly(A) tails (Hector, 2000; Hector et al., 2002). Poly(A) tail lengths were approximately 140 nucleotides in length in nab2ApPAB1


cells, a mere 20-30 nucleotides shorter than tail lengths observed in nab2ApGAL::NAB2 when grown in glucose (Hector et al., 2002). Pablp also regulates mRNA turnover (Tucker and Parker, 2000; Wilusz et al., 2001) and promotes translation initiation through interactions with the poly(A) tail and elF4E (Sachs and Davis, 1989). Thus, higher levels of Pablp in the cytoplasm may lead to increased mRNA stability and a coincident upsurge in translation initiation thereby alleviating the deleterious effects of the export block in the nab2A strain. Arguing against this explanation, Pabip was found to be present in both the nucleus and cytoplasm by immunofluorescence microscopy when PAB1 was expressed from a high-copy plasmid (Hector et al., 2002). Nuclear Targetingq of Pab lp Suppresses the nab2A Growth Defect

To prove Pablp is required in the nucleus to compensate for the loss of Nab2p, DNA encoding an SV40 nuclear localization (NL) sequence (PKKKRKV) and a mutant NL (mnl) sequence (PKTKRKV) was cloned 5' of the PAB1 start codon in a low-copy plasmid (pPAB1NL and pPAB1mni, respectively). In addition, as a means of discriminating between endogenous Pablp and the PablNL or Pablmnl fusion proteins, the stop codon of PAB1 was replaced by sequence encoding the V5 epitope (Invitrogen). These plasmids were transformed into the nab2ApGAL::NAB2 shuttle strain (YRH201c). When grown on galactose at 240 C, both the nab2ApPAB1NL and nab2ApPABlmni strains grow (Figure 8A) because Nab2p is still present (Figure 8B and data not shown). When cells were streaked to glucose, only the nab2ApPAB1NL and NAB2 strains show growth (Figure 8A). Importantly, the Pab1NL and Pablmn proteins were expressed at

70 .-L

>- -i
0 c T- U')

04 >
(u c
7E cu
0 0)
c co 0 <
(n 70
Q) 4-CL
CL co U) a) Z U) a) -0 -0
C: Co (D CU C L-i Co (1) z .. 3:
:B cn o C:
a. c w
CL =3 '6 C6
E w
cu 0
CO cn
64 C4 0
4- CU
-Q CO 0
cu < CL
Z CO 04
-j co
cn -0
cn (D CN
C/) 0 C/) LL
-t 0 cy) CL 04
0. C,4
=3 co 0).0 Cl) C\j C: CD
(/) ci Z
M lc: W Z 0 75 c
4- 0 cu
:3 0 :3 LC)) -0
-C c
(j) CU
3: 0 c 0 CL FU CL
co V'a- co
V) (D <
C z o
< cu cli
co a- cou) -C23 -Coo




1/ .i4, f

n .0
a. z


z z

cc 0
tv co%


,a < C4 C


levels similar to the endogenous Pablp as indicated by immunoblot analysis using aPabl 1G1 (Figure 8B). Unlike endogenous Pablp, however, the PablNL protein, as visualized by cell immunofluorescence using the cV5 antibody (Invitrogen), accumulated in the nucleus (Figure 9A). Surprisingly, import of the PablNL protein was incomplete (Figure 9A) suggesting either import by the SV40 NLS in yeast is relatively inefficient or PablNL is also rapidly exported. Nevertheless, these results clearly demonstrate Pablp must cycle through the nucleus to compensate for the loss of Nab2p. Nab2p Is Required for Poly(A) Tail Lenqth Control and mRNA Export

How does nuclear Pablp suppress the growth defect of a nab2A strain? Since Pabl p has been implicated in poly(A) tail trimming in vitro and in vivo (Amrani et al., 1997; Brown et al., 1996; Minvielle-Sebastia et al., 1997), poly(A) tail lengths were assayed in nab2ApPABlNL (data not shown). Poly(A) tails were actually 10-20 nucleotides longer than previously described for nab2ApPAB1 (Hector et al., 2002) indicating, in the absence of Nab2p, targeting Pablp to the nucleus is insufficient to limit poly(A) tail length in vivo. The localization of poly(A) RNA in both the nab2ApPAB1NL and nab2ApPAB1 strains was also determined by fluorescence in situ hybridization (FISH) using a d(T)50 probe. A major portion of the poly(A)* RNA signal was present in the cytoplasm of both strains (Figure 9A, nab2ApPAB1NL and nab2ApPAB1 rows), although a significant nuclear accumulation of poly(A)* RNA was still observed (Figure 9A, compare poly(A)* RNA fluorescence in nab2ApPAB1NL to that of NAB2). Remarkably, nuclear accumulation of poly(A)Y RNA in nab2ApPAB1NL cells

Figure 9. PablNL accumulates in the nucleus of nab2ApPAB1NL. A) SV40 NLStagged Pablp and poly(A) RNA concentrate in the nucleus of a nab2A strain. The localizations of V5-tagged Pablp, poly(A)* RNA and DNA are compared in NAB2, nab2ApPAB1NL and nab2ApPABl1mni strains.
B) Suppression of low-copy pPAB1NL is restricted to 240 C. Serial tenfold dilutions of NAB2 and nab2ApPAB1NL strains were spotted on YPD plates. Plates were incubated for either 3 days (240 C and 370 C)
or 7 days (140 C).



(xV5 poly(A)+ RNA DNA




14-C 24-C 37--C



correlated with an inability to grow at high (370 C) and low (140 C) temperatures (Figure 9B). Two competing hypotheses can be proposed to explain these findings: (1) the accumulation of Pablp in the nucleus (Figure 9A) inhibits mRNA export as well as growth at temperatures other than 240 C or (2) Nab2p functions in some capacity to stimulate mRNA export and Pablp is unable to replace Nab2p in this role. I favor the latter hypothesis because the RNA localization and polyadenylation phenotypes of nab2ApPAB1NL resemble those observed in connection with mutations in known mRNA export factors. Poly(A) tail lengths were assayed in the temperature-sensitive mRNA export mutant (mex67-5) and the RanGAP temperature-sensitive mutant (mal-1). A severe hyperadenylation defect was observed in mex67-5 at the non-permissive temperature (370 C), but not the permissive temperature (240 C), while mnal-1 only exhibited a slight increase in poly(A) tail length at 370 C (Figure 10). Importantly, RNA1 is required for protein import but not mRNA export (Koepp et al., 1996), whereas MEX67 is required for efficient mRNA export but not protein transport (Segref et al., 1997). Therefore, a tight correlation between nuclear mRNA export and poly(A) tail length restriction was observed. These results parallel those described for several other mRNA export mutant strains, including mtr2, dbp5, and rat7 (Hilleren and Parker, 2001; Jensen et al., 2001b). Jensen et al. (2001) reported that hyperadenylated HSP104 transcripts accumulated in nuclear foci when rat7I cells were shifted to non-permissive temperatures (Jensen et al., 2001b). The nuclear exosome, in particular Rrp6p, is required for foci retention of both hypoadenylated SSA4 RNA in pap1-1 and hyperadenylated HSPI04 RNA in

Figure 10. Hyperadenylation and mRNA export defects correlate in mex67-5.
Poly(A) tail analysis of MEX67, mex67-5 and rnal-1 grown in YPD at either 240 C or shifted to 370 for 30 minutes. Each lane represents equal amounts of labeled RNA. The indicated size markers are
pBR322 Mspl fragments.



b\ -C 150




rat7-1 (Hilleren et al., 2001). Intriguingly an accumulation of poly(A)* RNA in sub-nuclear foci was also observed for nab2AGAL::NAB2 when grown in glucose for 2 hours and nab2-21 when shifted to 140 C for 30 minutes (Anderson, 1995; Hector, 2000). Considering the nuclear exosome and Rrp6p is responsible for processing 3' ends of rRNA and snoRNA (Allmang et al., 1999a) and GFP-Rrp6p localizes predominately to the nucleolus (Huh et al., 2003), it is possible that hyperadenylated RNAs are trafficked to the nucleolus for retention and degraded by the exosome when Nab2p is not functional. Furthermore, Pablp may allow for growth when targeted to the nucleus in a nab2A strain by inhibiting nucleolar retention of poly(A) RNA. Providing support for this model, poly(A) RNA colocalized with an abundant nucleolar protein (Noplp) in both nab2AGAL::NAB2 and nab2-21 under non-permissive growth conditions (Figure 11) but poly(A)* RNA was observed throughout the nucleus and cytoplasm in the nab2ApPAB1NL strain (Figure 9A). Thus in each nab2 mutant strain examined, the lethal growth phenotype correlated with nucleolar localization of poly(A) RNA. Based on these results I propose that Nab2p normally functions to inhibit nucleolar retention, and possibly degradation, of polyadenylated RNA while at the same time facilitates the export of poly(A) RNA to the cytoplasm. Targeting Pablp to the nucleus inhibits nucleolar retention but does not restrict tail lengths or promote mRNA export in the absence of functional NAB2. Further evidence is needed to support the idea that Nab2p and Rrp6p have antagonistic roles in nuclear mRNA export/retention and poly(A)Y RNA stabilization/degradation, but

Figure 11. Loss of nab2 function leads to nucleolar accumulation of poly(A)'
RNA in vivo. Co-localization of poly(A) RNA (green) and Nopl p (red) in nab2ApGAL:NAB2 cells shifted to glucose for 2 hours and nab2-21 cells shifted to 140 C for 30 minutes. FISH analysis of poly(A) RNA distribution was performed using conventional immunoflourescence
microscopy. DNA was detected with DAPI.


poly(A)+ RNA Nopl p merge/DAPI nab2ApGAL:NAB2
glucose 2h

140 C


this model also predicts that Nab2p recruits a nuclear export receptor to the poly(A) tail which stimulates the nuclear export of mRNA.

Nab2p Associates with Factors Required for mRNA Export

Several recent studies have implicated Nab2p in mRNA export (Duncan et al., 2000; Green et al., 2003; Green et al., 2002; Hector et al., 2002). Loss of NAB2 in vivo or shifting mutant nab2 strains to non-permissive growth temperatures results in rapid accumulation of poly(A) RNA in the nucleus (Green et al., 2002; Hector et al., 2002). Intriguingly, Nab2p and poly(A) RNA co-localize near nuclear pores in a mutant tomi strain (Duncan et al., 2000) and Nab2p interacts with a NPC-associated protein (MIplp) in the yeast two-hybrid system (Green et al., 2003). The authors of these studies propose that Nab2p functionally interacts with export receptors and the NPC to mediate the export of poly(A) RNA. In addition, examination of the Nab2p domain structure revealed a nuclear export sequence (NES) near the N-terminus of Nab2p that was required for efficient poly(A) RNA export (Marfatia et al., 2003). Purification of Nab2pT-associated Proteins

The founding member of the RNA export factor (REF) family of proteins (Yralp) binds mRNA in the nucleus and forms a bridge between the site of transcription and nuclear pore through interactions with the nuclear export factor Mex67p (Portman et al., 1997; Strasser and Hurt, 2000; Stutz et al., 2000; Zenklusen et al., 2001). In metazoan organisms, REF/Aly, SR proteins and hnRNPs all serve as adapters for mRNA export, channeling mRNA through the nucleus to the NPC via interactions with TAP (Gallouzi and Steitz, 2001; Huang


et al., 2003; Longman et al., 2003; Luo et al., 2001; Moore and Rosbash, 2001; Zhou et al., 2000). It seemed likely that Nab2p may also promote mRNA export of poly(A) RNA in yeast by serving as an adapter protein, recruiting an export receptor to the poly(A) tail. The tandem affinity purification (TAP) protocol (Rigaut et al., 1999) was employed to purify proteins that specifically associate with Nab2p.

By virtue of homologous recombination, the wild-type NAB2 stop codon was replaced with sequences encoding a calmodulin binding protein (CBP) and protein A (ProtA) to generate the NAB2T strain (YKN206) (Figure 12A). A protein that migrated more slowly (Nab2pT) than the wild type protein (Nab2p) was detectable by immunoblot analysis using aNab2p 3F2 (Figure 12B) indicating recombination was successful. Importantly, levels of Nab2pT corresponded to that of Nab2p (Figure 12B) and growth rates of the two strains were comparable (data not shown). The dual CBP-ProtA tag allows for purification of native Nab2pT, and any associated proteins, by two successive steps (Figure 13). Briefly, whole cell extracts were prepared from YKN206 and incubated at 40 C with IgG Sepharose (Pharmacia). The ProtA peptide has a relatively high affinity for Immunoglobulin G (lgG) protein allowing for retention of Nab2Tp-associated proteins (Nab2pTAP) on the Sepharose beads. After several washes to remove nonspecific proteins, the addition of TEV released Nab2pTAP from the IgG Sepharose by cleavage between the CBP and ProtA peptides (see Figure 12A). The TEV eluate was subsequently incubated at 40 C with calmodulin affinity resin (Stratagene). In this case, the CBP tag is indispensable

C CL CD cn
0 -C
0 0 0
E C: C: E Co Co 0 co


0 c
-0 0 a)
0 a)
U) E.u N 0-0 ca 0 m

4- 0 a)
E 0 Ua) CO
E -',,
a) a) CL

a) a) z
a) r U) C < CU
0 C -0 L- a)
CL (n
< >
2 0 C
-0 co
c cn 0
a) : a) CL
a) -0 0) cu
-0 CL
C: -o E cu 2
-0 CU -0
co C: a)
N 0 c: a) m -0 -(n 'a):g ,::c cu c: C;)- z
a) 0
0_,2 c: L.2 cu CL
E m (n O.S C: CL
0 CN
c 70 -0 -02 a) C: 0 tv 4(DZ OZ o



C'4 .0
m cc z z 90a

co co





because 2 mM CaCI2 induces binding of Nab2p to the calmodulin resin and 2 mM EGTA chelates the CaCl2 releasing Nab2pT and all associated proteins (Nab2pTAP) under mild, non-denaturing, conditions. Nab2p and Kap104 Form an Abundant Complex in Yeast Extracts

The Nab2pTAP was first used to generate antibodies against the most abundant proteins. Standard procedures were performed to inject 5 [tg of purified proteins into BALB/c mice (Harlow and Lane, 1988). After an appropriate immune response, the polyclonal antiserum was used to screen a Xgtl 1 genomic expression library as previously described (Snyder et al., 1987). Seventeen phage clones positively reacted against the Nab2pTAP antiserum. Of these clones, thirteen were also immunoreactive to aNab2 3F2 and aNab2 1F2. The remaining clones were sequenced and discovered to encode for Nab2p (2 clones) and Kap104p (2 clones). Interestingly, both NAB2 clones initiated translation near the QP region of Nab2p suggesting 3F2 and 1F2 recognize sequences within the NES region of Nab2p (see Figure 14A). To explore this possibility further, two N-terminal truncations (NT1 and NT2) of Nab2p (Figure 14A) were fused to the GAL4 activation domain (AD) peptide (Clontech) (Figure 14B) and tested for 3F2 and 1F2 immunoreactivity. Monoclonal antibody aNab2 1F2 recognizes only AD-NT2 while aNab2 3F2 recognizes both AD-NT1 and ADNT2 peptides (Figure 14C), restricting the 3F2 epitope to the first 42 amino acids of Nab2p.

Kapl04p, a member of the importin P/karyopherin P family of transport receptors, is responsible for importing both Nab2p and Nab4p into the nucleus

c IL

04 :3 75 E -0-o
0 0 0

Co o
L- Co uj
--v (D

(n 0 C)
cn U)
x -o E u)
CU C 04 >-'
> %4.- -0 __ _0 L) a)
0 in (D o.L- (D 0 cu CL 0
L- 0
CU (D 0
= U)
a CU W
a) CL
cn 0

C: cu

L) 3: 0
C: (D IL) W
0 a) 0) CvJ CU C%4
a) > -o E
-0 a- CU cu
CD a) Z N
a) -6 4-cu __ 70 0
>, a) c
cQ 0 cn -0 m .= cn -0 CD (D co a)
r 75 70 70 co Q m
ucc -a >
0 E U-j IS

(n (D z c m
C) o
cn Z r y 0 a F 0 'o
CL .0

E 'E
:3 45 0 0
a- Z 4--- cu m





< c E

2 cc

2 6 CL




N4 2
C13 +
N .0
m c
z .0



(Aitchison et al., 1996). Kapl04p interacts with RG-rich domains present in both Nab2p and Nab4p and mediates import through additional interactions with GLFG-repeat containing nucleoporins (Nup116, Nupl00, Nupl45, and Nup57) (Aitchison et al., 1996). Together, RanGTP and RNA abrogate the binding of Kapl04p to Nab2p and Nab4p, providing a mechanism for Kapl04p-Nab2/4 dissociation once in the nucleus (Lee and Aitchison, 1999). In the cytoplasm, hydrolysis of RanGTP is believed to release Nab2p and Nab4p from mRNA and induce association with Kapl04p for another round of import because Kapl04p, but not Kap95p, displaces both Nab2p and Nab4p from single stranded DNA (ssDNA) cellulose (Lee and Aitchison, 1999). Nab4p is most similar to metazoan hnRNP Al and the Kapl04p homologue (Trn 1) imports hnRNP Al into the nucleus of COS7 cells (Siomi et al., 1997). Although a metazoan protein with significant similarity to Nab2p has not been described, the RGG-domain of Nab2p interacts with Trn 1 in vitro and is sufficient for nuclear import of reporter proteins in human cells (Siomi et al., 1998; Truant et al., 1998). Recently, the Trn 1-hnRNP Al heterodimer has also been implicated in the export of specific mRNAs (Gallouzi and Steitz, 2001; Moore and Rosbash, 2001). However, a direct role for the yeast Kapl04p-Nab2p complex in mRNA export is unlikely because kap104 mutants exhibit a delayed (> 3 hours) accumulation of poly(A) RNA in the nucleus (Aitchison et al., 1996). Nevertheless, discovery of Kapl04p immunoreactivty in the polyclonal antiserum suggests Kapl04p is an abundant component of the purified Nab2pTAP complex. To substantiate this claim, the