a RESEARCH ARTICLE

TEG-1 CD2BP2 Regulates Stem CellProliferation and Sex Determination in theC. elegans Germ Line and Physically InteractsWith the UAF-1 U2AF65 Splicing FactorChris Wang,1 Laura Wilson-Berry,2 Tim Schedl,2 and Dave Hansen1*

Background: For a stem cell population to exist over an extended period, a balance must be maintainedbetween self-renewing (proliferating) and differentiating daughter cells. Within the Caenorhabditis ele-gans germ line, this balance is controlled by a genetic regulatory pathway, which includes the canonicalNotch signaling pathway. Results: Genetic screens identified the gene teg-1 as being involved in regulat-ing the proliferation versus differentiation decision in the C. elegans germ line. Cloning of TEG-1 revealedthat it is a homolog of mammalian CD2BP2, which has been implicated in a number of cellular processes,including in U4/U6.U5 tri-snRNP formation in the pre-mRNA splicing reaction. The position of teg-1 in thegenetic pathway regulating the proliferation versus differentiation decision, its single mutant phenotype,and its enrichment in nuclei, all suggest TEG-1 also functions as a splicing factor. TEG-1, as well asits human homolog, CD2BP2, directly bind to UAF-1 U2AF65, a component of the U2 auxiliary factor.Conclusions: TEG-1 functions as a splicing factor and acts to regulate the proliferation versus meiosisdecision. The interaction of TEG-1 CD2BP2 with UAF-1 U2AF65, combined with its previously describedfunction in U4/U6.U5 tri-snRNP, suggests that TEG-1 CD2BP2 functions in two distinct locations in thesplicing cascade. Developmental Dynamics 241:505–521, 2012. VC 2012 Wiley Periodicals, Inc.

Key words: CD2BP2; TEG-1; stem cell proliferation; pre-mRNA splicing; UAF-1; U2AF65

Key findings:� TEG-1 inhibits stem cell proliferation in the C. elegans germ line.� TEG-1 likely functions as a splicing factor.� TEG-1 CDBP2 physically interacts with UAF-1 U2AF65.

Accepted 3 January 2012

INTRODUCTION

Stem cells provide the material neces-sary for tissue generation during devel-opment, and tissue replacement andrepair in adult animals. The capacity ofstem cells to contribute to these proc-esses over an extended time period

is tied to their ability to produceboth self-renewing and differentiatingdaughter cells. Self-renewing daughtercells retain the stem cell characteristicsof the parental cell; therefore, a pool of

stem cells is maintained over time. Thedifferentiating daughter cells are used

to make the desired tissue. Maintain-ing the balance between self-renewingand differentiating daughter cells iskey to proper stem cell function since adisruption in this balance prevents

long term tissue generation; too littleself-renewal results in the eventual

Dev

elop

men

tal D

ynam

ics

1University of Calgary, Department of Biological Sciences, Alberta, Calgary, Canada2Department of Genetics, Washington University School of Medicine, St. Louis, MissouriGrant sponsor: Canadian Institutes of Health Research (CIHR); Natural Sciences and Engineering Research Council of Canada(NSERC); National Institutes of Health (NIH); Grant number: GM63310.*Correspondence to: Dave Hansen, Department of Biological Sciences, University of Calgary, Calgary, Alberta, CanadaT2N 1N4. E-mail: dhansen@ucalgary.ca

DOI 10.1002/dvdy.23735Published online 30 January 2012 in Wiley Online Library (wileyonlinelibrary.com).

DEVELOPMENTAL DYNAMICS 241:505–521, 2012

VC 2012 Wiley Periodicals, Inc.

depletion of the stem cell population,while too little differentiation results innot enough cells being made for propertissue generation. Understanding howthis balance between self-renewal anddifferentiation is maintained is anessential step in understanding stemcell biology and function.

The germ line of the nematodeCaenorhabditis elegans has emergedas a powerful model to study how astem cell population maintains thebalance between self-renewal and dif-ferentiation (Hansen and Schedl,2006; Kimble and Crittenden, 2007).In the C. elegans hermaphrodite, thegerm line is contained in two gonadarms that meet at a common uterus.Populations of mitotically dividing(proliferating) cells exist at the endsof the two gonad arms distal from theuterus (Fig. 1A). In the adult, eachpopulation consists of �200–250 cells(Lamont et al., 2004; Killian and Hub-bard, 2005), with cells progressingthrough the cell cycle approximatelyevery �6.5–8 hr (Fox et al., 2011). Atleast some of these proliferating cellsare considered stem cells. Cells arekept in the proliferative state due totheir close proximity to the Distal TipCells (DTCs), with a single DTC cap-ping the distal end of each gonad arm.The proliferation promoting influenceof the DTC is primarily accomplishedthrough the conserved GLP-1/Notchsignaling pathway (Hansen andSchedl, 2006; Kimble and Crittenden,2007; Racher and Hansen, 2010). TheDSL (Delta/Serrate/LAG-2) LAG-2and APX-1 ligands are expressed onthe surface of the DTC (Hendersonet al., 1994; Tax et al., 1994; Nadara-jan et al., 2009). When they bind theGLP-1/Notch receptor (Yochem andGreenwald, 1989), which is expressedon the surface of the germ cells (Crit-tenden et al., 1994), the cleaved intra-cellular portion of GLP-1 (referred toas GLP-1(INTRA)) is thought to enterthe nucleus and bind to the CSL(CBF1/Suppressor of Hairless/LAG-1)LAG-1 transcription factor (Christen-sen et al., 1996; Mumm and Kopan,2000) and the SEL-8/LAG-3 co-activa-tor (Doyle et al., 2000; Petcherski andKimble, 2000). This complex then acti-vates genes necessary for the prolifer-ative fate. As cells move along thegonad arm, away from the DTC, Notchsignaling levels are thought to

decrease, allowing cells to enter mei-otic prophase. The first obvious indica-tion that cells have entered into mei-otic prophase is when the DNA takeson a crescent moon shape as chromo-somes pair and synapse (Francis et al.,1995a; Dernburg et al., 1998; Mac-Queen and Villeneuve, 2001). Cellscontinue to move proximally whiletransiting through the various stagesof meiotic prophase, eventually form-ing fully differentiated gametes(sperm or oocytes depending on theage of the hermaphrodite).

The GLP-1/Notch signaling path-way promotes the proliferative fate,at least in part, by inhibiting theactivities of two redundant meioticpromoting genetic pathways, referredto as the gld-1 and gld-2 pathways(Kadyk and Kimble, 1998) (Fig. 1B).Within the gld-1 pathway are the gld-1 and nos-3 genes (Francis et al.,1995b; Hansen et al., 2004b); withgld-1 encoding a KH domain contain-ing RNA binding translational regula-tor (Jones and Schedl, 1995), and nos-3 encoding a protein with similarityto Drosophila Nanos (Kraemer et al.,1999), also a translational regulator.Within the gld-2 pathway are the gld-2 and gld-3 genes, encoding the cata-lytic portion of a poly(A) polymerase(Kadyk and Kimble, 1998), and ahomolog to Drosophila BicC, respec-tively (Eckmann et al., 2002). Thespatial organization of the activity ofthe GLP-1/Notch signaling pathwayrelative to the activities of the gld-1and gld-2 pathways is key to main-taining the balance between prolifera-tion and differentiation. Loss orreduction of GLP-1/Notch signaling inthe distal end results in ectopic acti-vation of the gld-1 and gld-2 path-ways in this region, causing cells toenter meiosis prematurely and a lossof the proliferative population (Austinand Kimble, 1987; Eckmann et al.,2004; Hansen et al., 2004b). Con-versely, ectopic activation of GLP-1/Notch signaling beyond its normalregion of activity, or a combinedreduction in the activities of theredundant gld-1 and gld-2 pathways,results in excess proliferation (Berryet al., 1997; Kadyk and Kimble, 1998;Pepper et al., 2003; Hansen et al.,2004b). Indeed, in certain mutantgene and/or allele combinations, allcells within the gonad arm have the

proliferative fate, resulting in a com-plete germline tumor (Berry et al.,1997; Hansen et al., 2004a). Othergenes have been identified that func-tion with the GLP-1/Notch signalingand gld-1/gld-2 pathways, or parallelto them, to regulate the balancebetween proliferation and differentia-tion. These genes function in manydifferent cellular processes. For exam-ple, mutations that decrease the func-tion of the proteasome enhance over-proliferation mutants, indicating afunction to promote meiotic develop-ment (MacDonald et al., 2008). Asanother example, the cell-cycle regu-lators Cyclin E (CYE-1) and CDK2(CDK-2) promote the proliferativefate (Fox et al., 2011; Jeong et al.,2011). A number of splicing factorshave also been identified as beinginvolved in regulating the prolifera-tion versus differentiation decision(Belfiore et al., 2004; Mantina et al.,2009; Kerins et al., 2010; Zanettiet al., 2011). Genetically, these splic-ing factors appear to function largelyin the gld-1 pathway, functioningdownstream of GLP-1/Notch signal-ing (Fig. 1B). A screen of splicing fac-tors found that proteins functioningthroughout the splicing cascade par-ticipate in the proliferation/differen-tiation decision, suggesting that anoverall decrease in splicing or spliceo-some activity contributes to over-pro-liferation (Kerins et al., 2010).Pre-mRNA splicing involves the re-

moval of introns from the pre-mRNAbetween the exons prior to nuclearexport and translation. The splicingreaction is accomplished through theactivity of the spliceosome, which is alarge RNA-Protein complex that isassembled in a stepwise manner(Wahl et al., 2009; Valadkhan andJaladat, 2010). Some of the earlieststeps in spliceosome assembly includerecognition of the 50 and 30 splice siteson the pre-mRNA by protein or RNA/protein complexes. The U1 snRNPparticle, consisting of the U1 snRNAand a number of proteins, recognizesand binds to the 50 splice site (Sera-phin and Rosbash, 1989; Michaudand Reed, 1991). The U2 AuxiliaryFactor (U2AF) recognizes and bindsto the 30 splice site (Wahl et al., 2009;Valadkhan and Jaladat, 2010).Human U2AF consists of two pro-teins, U2AF65 and U2AF35 (Zamore

Dev

elop

men

tal D

ynam

ics

506 WANG ET AL.

Dev

elop

men

tal D

ynam

ics

Fig. 1.

Fig. 2.

Fig. 2. teg-1(oz230) enhances the over-prolif-erative phenotype in different genetic back-grounds. Dissected gonad arms from adulthermaphrodite animals one day past the fourthlarval stage. Gonad arms were stained withDAPI (blue) to show nuclear morphology, anti-REC-8 antibodies (green) to mark mitotic cells,and anti-HIM-3 antibodies (red) to mark meioticcells. Both teg-1(oz230) (A) and glp-1(ar202gf)(B) animals, grown at 15�C, show proliferationrestricted to the distal part of the gonad arm(although some anti-REC-8 cross-reactivity withsperm is observed in the proximal ends of thegonad arms); however, mitotic cells are foundthroughout the gonad arms of glp-1(ar202gf)teg-1(oz230) (C) animals (100% of gonad armsshow this over-proliferation phenotype, n>20),although clusters of cells that have entered intomeiosis are observed (red). D,E: teg-1(oz230)also enhances the over-proliferative phenotypeof gld-2(q497) single mutants (D), with gld-2(q497); teg-1(oz230) double mutants display-ing extensive proliferation in both the distal andproximal regions of the gonad arms (E) (100%of gonad arms show this over-proliferation phe-notype, n>20). All strains were also markedwith unc-32(e189). Scale bar ¼ 20 mm. F: Theoverall size of the distal mitotic zone is statisti-cally larger in glp-1(ar202gf) teg-1(oz230) ani-mals than in either single mutant, as measuredby the number of cell diameters from the distalend of the gonad arm that are mitotic (anti-REC-8(þ) [green] and distal to any meioticnuclei, anti-HIM-3(þ); [red]). P < 9.7 � 10-5 t-test. Error bars ¼ 1 SD. n¼ 20 gonad arms.

Fig. 1. The proliferation/differentiation bal-ance and sex determination in the C. elegansgerm line. A: A wild-type adult (one day pastthe fourth larval stage) hermaphrodite dissectedgonad arm stained with DAPI (blue) to visualizenuclear morphology, anti-REC-8 antibodies(green) to mark mitotically cycling cells, andanti-HIM-3 antibodies (red) to mark meioticcells. Distal is to the left, with mitotic or prolifer-ative cells existing in the distal end of the gonadarm. Cells move proximally, entering meiosis inthe transition zone, and eventually forming firstsperm, then oocytes. Scale bar ¼ 20 mm. B: Asimplified genetic pathway showing genesinvolved in regulating the proliferation versusdifferentiation decision in the C. elegans germline. Genes that promote proliferation areshown in green, while genes that inhibit prolifer-ation and/or promote meiotic entry are shownin black. lag-2, glp-1, and lag-1, which encodecomponents of the core Notch signaling path-way, through a number of intermediates, inhibitthe activities of genes in two redundant down-stream pathways, referred to as the gld-1 andgld-2 pathways. As cells move away from thedistal end of the gonad, the gld-1 and gld-2pathways become active, thereby promotingmeiotic entry (differentiation) and/or inhibitingproliferation. Adapted from Hansen and Schedl(2006). C: A simplified genetic pathway showingthe genetic relationships between genesinvolved in regulating sex determination in theC. elegans germ line. Genes that promote themale fate are shown in blue, while genes thatpromote the female fate are shown in black.Hermaphrodites first produce sperm, and thena switch is made during development such thatadult hermaphrodites produce oocytes.Adapted from Ellis and Schedl (2006).

and Green, 1989), which in C. elegansare UAF-1 and UAF-2, respectively.U2AF65 binds to a polypyrimidinetract that is found at the 30 end ofintrons, while U2AF35 binds to the 30

splice site (Wahl et al., 2009; Valad-khan and Jaladat, 2010). Although inhumans the polypyrimidine tract canvary somewhat in sequence and posi-tion relative to the 30 splice site, in C.elegans the conserved UUUUCAG/Rsequence is found immediately adja-cent to the 30 splice site (Blumenthaland Steward, 1997). Both UAF-1 andUAF-2 bind to this conserved sequence(Zorio and Blumenthal, 1999a). Bind-ing of U2AF to the 30 splice site helpsin the recruitment of the U2 snRNP(Wahl et al., 2009; Valadkhan andJaladat, 2010). In addition to recogni-tion of the 50 and 30 splice sites by U1and U2AF, respectively, SF1/BBP(SFA-1 in C. elegans; Mazroui et al.,1999), binds to the branch point (Abo-vich and Rosbash, 1997; Berglundet al., 1997), which is just upstreamfrom the polypyrimidine tract. Subse-quent steps in spliceosomal assemblyinclude replacement of SF1/BBP atthe branch point by U2, recruitment ofthe U4/U6.U5 tri-snRNP, and disasso-ciation of U1 and U2 from the spliceo-some. This stepwise assembly contrib-utes to the activated spliceosome(complex B*), which allows for the twotransesterification reactions of intronremoval to occur. The first transesteri-fication reaction results in a free 50

exon and the 50 end of the intronattached to the branch site in a lariatformation. The second transesterifica-tion reaction ligates the two exons andliberates the intron (Wahl et al., 2009;Valadkhan and Jaladat, 2010).

Here we demonstrate that reductionin teg-1 activity causes a disruption inthe balance between proliferation anddifferentiation in the C. elegans germline. In two different sensitized geneticbackgrounds, teg-1 mutants were iden-tified by their ability to enhance over-proliferation, resulting in a germ linetumor. We identified teg-1 as encodinga homolog of proteins from yeast tohumans, including CD2BP2 and yeastLin1p, which have been implicated innumerous cellular functions, includingsplicing. Within the splicing reaction,CD2BP2 has been suggested to beinvolved in U4/U6.U5 tri-snRNP for-mation (Laggerbauer et al., 2005). We

found that TEG-1 binds UAF-1, theworm homolog of the U2AF65 splicingfactor, and that this interaction is con-served in humans. Therefore, TEG-1CD2BP2 likely is involved in two dis-tinct steps in the splicing reaction.

RESULTS

teg-1 Alleles Identified in Two

Over-Proliferation Enhancer

Screens

In an effort to better understand howthe balance between proliferation anddifferentiation is controlled in theC. elegans germ line, we sought toidentify genes involved in regulatingthis process. To identify these genes,we performed two genetic screens formutations that enhance germlineover-proliferation. In the first screen,we mutagenized animals that con-tained the glp-1(oz112oz120) weakgain-of-function allele (Berry et al.,1997). These animals are relativelywild-type with respect to germlineproliferation at 15�C, with only 0.05%of glp-1(oz112oz120) gonad arms con-taining an over-proliferative germline (4/8,000 gonad arms) (Berry,1998). We screened for recessiveenhancers of the over-proliferativephenotype by cloning F1 progeny ofmutagenized animals and screeningfor those that showed �25% of prog-eny (F2) with tumorous germ lines.From screening 8,200 haploidgenomes, we identified 12 enhancingmutations, with the strongest 7 corre-sponding to 4 complementationgroups. We have named these teg-1(oz189), teg-2(oz192, oz194, oz216,oz218), teg-3(oz190), and teg-4(oz210)(tumourous enhancer of glp-1(oz112oz120)). We have previouslyreported the cloning of teg-4, whichencodes a homolog of the humanSAP130 splicing factor (a subunit ofthe SF3b splicing complex) (Mantinaet al., 2009). teg-2 and teg-3 are invarious stages of characterization,mapping, and cloning, which we willdiscuss elsewhere. Here we report thecharacterization and cloning of teg-1.

In the second screen, we identifiedmutations that have a synthetic over-proliferation phenotype with the gld-2(q497) allele. Since gld-2 pathwaygenes function redundantly with gld-1 pathway genes in regulating the

balance between proliferation and dif-ferentiation (Kadyk and Kimble,1998), and since the activities of boththe gld-1 and gld-2 pathways must bereduced in order for a germlinetumour to be formed, this screen wasspecifically designed to identify genesthat function in the gld-1 geneticpathway, although mutations in genesthat function upstream of Notch sig-naling, or in a parallel pathway, mayalso cause over-proliferation in a gld-2(q497) background. Since gld-2(q497) homozygous animals are ster-ile due to defects in gametogenesis(Kadyk and Kimble, 1998), we muta-genized gld-2(q497) homozygotes thatwere fertile because they carried thegaDp1 free duplication, which con-tains a wild-type copy of gld-2. Theprogeny (F2) of cloned F1 animalswere screened for those with germlineover-proliferation. These tumorousanimals presumably lost the gaDp1free duplication carrying the wild-type copy of gld-2, and were homozy-gous for the newly generated muta-tion. Since these animals were sterile,we recovered the induced mutationfrom siblings that carried gaDp1[gld-2(þ)], and were either homozygous orheterozygous for the new mutation.We called genes identified in thisscreen syt (synthetic tumorous). Wepreviously reported two genes identi-fied in this screen; syt-1, which isallelic to nos-3 (Hansen et al., 2004b),and pas-5, which encodes a subunit ofthe proteasome (MacDonald et al.,2008). From this screen, we also iso-lated the teg-1(oz230) allele; there-fore, we isolated two alleles of teg-1,each from independent screens withdifferent genetic backgrounds.

Analysis of teg-1

Over-Proliferation

Since teg-1 mutant alleles were iso-lated from two different screens, weknow that teg-1 mutants can cause/enhance over-proliferation in differ-ent sensitized genetic backgrounds.To more closely examine the nature ofthese over-proliferative tumors, westained dissected gonads withmarkers for mitotic and meiotic cells;anti-REC-8 antibodies mark mitoticzone cells (including those in pre-mei-otic S-phase) and anti-HIM-3 antibod-ies mark meiotic prophase cells

Dev

elop

men

tal D

ynam

ics

508 WANG ET AL.

(Zetka et al., 1999; Pasierbek et al.,2001; Hansen et al., 2004a). In per-forming the characterization, we usedthe glp-1(ar202gf) allele for the sensi-tized background to be consistentwith the analyses of other over-prolif-eration enhancing mutations (Hansenet al., 2004a,b; MacDonald et al.,2008; Mantina et al., 2009). We foundthat glp-1(ar202gf) teg-1(oz230) germlines were over-proliferative, withexcess REC-8(þ) HIM-3(�) cells ascompared to wild type or either singlemutant (Fig. 2). However, the excessproliferation was primarily found inthe distal and proximal regions of thegonad, with meiotic cells found inbetween. The average size of the dis-tal mitotic zone in glp-1(ar202gf) teg-1(oz230) double mutant animals was35 cell diameters from the distal end,while the average sizes for the teg-1(oz230) and glp-1(ar202gf) singlemutants were 14 and 22 cell diame-ters, respectively (Fig. 2C, F). There-fore, the double mutant displays alate-onset-tumorous phenotype inwhich the size of the distal mitoticzone is larger than wild type, whichhas a mitotic zone �20 cell diametersin length (Crittenden et al., 1994;Hansen et al., 2004a).

As has been observed with othermutations that are synthetic tumor-ous with gld-2(q497), gld-2(q497); teg-1(oz230) animals, although over-pro-liferative, have some cells that entermeiotic prophase (Fig. 2E). The pro-portion of cells that appear meioticdiffer from animal to animal; how-ever, all gonad arms have some cellsthat have entered into meiosis (anti-HIM-3(þ)). Therefore, the combinedremoval of gld-2 and teg-1 functiondoes not completely shift the balancetowards proliferation.

teg-1 Encodes a Homolog of

Human CD2BP2

To determine how teg-1 may be func-tioning at the molecular level, wemapped and cloned the gene. teg-1(oz230) was localized to a 0.4-cMregion between spe-16 and dpy-18 onlinkage group III by standard three-factor mapping. Single nucleotidepolymorphism and deficiency map-ping further narrowed the criticalregion to �25 kb containing threegenes. Sequencing revealed that one

of these genes, Y47D3A.27, containedpremature stop codons associatedwith each of the two teg-1 alleles (Fig.3). An integrated transgene contain-ing the Y47D3A.27 gene rescues themutant phenotype of teg-1(oz230) (seebelow). Additionally, antibodies raisedagainst the predicted Y47D3A.27 pro-tein, which detects a protein of thecorrect size in wild-type extracts, failsto detect a protein in teg-1(oz230) mu-tant extracts (see below). Together,these data support teg-1 as beingY47D3A.27.

Sequencing of a cDNA correspond-ing to teg-1, yk82g9, as well assequencing RT-PCR generated cDNAusing primers that are predicted toamplify the entire coding region,revealed a gene model differingslightly from the predicted genemodel for Y47D3A.27 (http://ws220.wormbase.org/). Sequencing ofthese cDNAs suggests a gene modelthat contains five exons and is pre-dicted to produce a protein of 353amino acids (as compared to the pre-dicted gene model that includes sixexons and a 370–amino acid protein;Fig. 3). The oz230 and oz189 allelescontain premature stop codons atamino acid positions 53 and 142,respectively. Therefore, oz189 andoz230 are likely strong loss-of-func-tion or null alleles.

BLAST alignment of the predictedTEG-1 protein identifies homologousproteins from a variety of species,including CD2BP2 from humans,DmLIN1 from Drosophila mela-nogaster, and Lin1p from Saccharo-myces cerevisiae (Fig. 3). Althoughconserved regions are found over theentire lengths of the proteins, thelevel of similarity is not overly high;TEG-1 is 42% similar and 31% identi-cal to human CD2BP2, 42% similarand 31% identical to DrosophilaDmLIN1, and 26% similar and 15%identical to yeast Lin1p. CD2BP2homologs have been suggested tofunction in a number of processes,including pre-mRNA splicing, DNAreplication, chromosome segregation,and the mammalian immuneresponse (Nishizawa et al., 1998;Freund et al., 1999). CD2BP2 homo-logs do not have any obvious motifs,other than a GYF motif in the carboxyend. The GYF domain was first iden-tified in CD2BP2, and binds proline-

rich sequences (Nishizawa et al.,1998; Freund et al., 1999). Alignmentof proteins from numerous speciesrevealed a sequence signature for thedomain; W-X-Y-X6-11-G-P-F-X4-M-X2-W-X3-G-Y-F, where X denotes anyamino acid (Kofler and Freund, 2006).TEG-1 does contain some sequencesimilarity to the GYF signature in itscarboxy end; however, it does not com-pletely match the signature (Fig. 3).Therefore, based solely on sequencesimilarity to the characterized GYFdomain, it is not clear if TEG-1 con-tains this functional protein interac-tion domain in its carboxy terminus(see below).

TEG-1 Controls Germline

Development in a Similar

Fashion to Splicing Factors

As mentioned, TEG-1 is homologousto CD2BP2 homologs, which havebeen suggested to have many func-tions. Perhaps the best-characterizedfunction of these homologs is theirrole in pre-mRNA splicing, withCD2BP2 functioning in the formationof the U4/U6.U5 tri-snRNP (Lagger-bauer et al., 2005). Interestingly,splicing factors have been shown to beinvolved in regulating the prolifera-tion versus differentiation decision inthe C. elegans germ line (Puoti andKimble, 1999, 2000; Belfiore et al.,2004; Konishi et al., 2008; Mantinaet al., 2009; Kerins et al., 2010;Zanetti et al., 2011); therefore, teg-1’sinvolvement in this decision could bethrough a role in pre-mRNA splicing.However, the overall low level of simi-larity between TEG-1 and its homo-logs, as well as the other cellular func-tions for CD2BP2 homologs, preventus from assuming that TEG-1 has arole in splicing. We reasoned that ifTEG-1 is functioning as a splicing fac-tor, and if it is a loss of this functionin teg-1 mutants that contributes tothe over-proliferation defect, teg-1mutants should show similar pheno-types and epistatic relationships asother splicing factor mutants. To testthis, we first made double mutantswith teg-1(oz230) and mutants of thegenes functioning in the gld-1 andgld-2 pathways regulating the prolif-eration versus meiotic entry decision.As with other splicing factor mutants,

Dev

elop

men

tal D

ynam

ics

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 509

teg-1(oz230) forms synthetic tumorswith both of the gld-2 pathway genes(gld-2 and gld-3), but not with thegld-1 pathway genes (gld-1 and nos-3;Table 1), suggesting that teg-1 likelyfunctions in the gld-1 pathway. Addi-tionally, the gld-3(q730); teg-1(oz230)tumor is epistatic to glp-1(q175null)(Table 1), suggesting that teg-1 func-tions downstream of Notch signaling,

which is also consistent with whathas been found with other splicingfactor mutants (Mantina et al., 2009;Kerins et al., 2010).

Additionally, we also observed a dif-ference in epistasis with gld-2 andgld-3, suggesting that their functionsare not equivalent in the gld-2 path-way. While gld-3(q730); glp-1(q175)teg-1(oz230) animals have a tumorous

germ line, gld-2(q497); glp-1(q175)teg-1(oz230) animals are not tumor-ous, but rather display the under-pro-liferative Glp phenotype associatedwith glp-1(q175) (Table 1). Similarepistasis results were observed forthe teg-4 and prp-17 splicing factors(Mantina et al., 2009; Kerins et al.,2010). Tumor formation in gld-3(q730); glp-1(q175) teg-1(oz230),

Dev

elop

men

tal D

ynam

ics

Fig. 3. TEG-1 is homologous to proteins from yeast to humans. A: Sequencing of the yk82g9 cDNA, analysis of contig MM454_contig00336,as well as sequencing reverse transcribed PCR amplification products, revealed a gene model for teg-1 that includes five exons. Boxes repre-sent exons, while joining lines represent introns. Blackened portions of the exons represent non-coding regions of the exons (50 and 30 UTRs).From the A in the initiator ATG, the first exon includes nucleotides from position 1–138, exon two 239–544, exon three 1,948–2,220, exon four3,106–3,315 (the gene model contained in wormbase places exon four at 3,106–3,323), exon five 4,218–4,352 (Note: only includes coding por-tions of exons, including taa stop codon). 50RACE and 30RACE revealed a 7-bp 50UTR and a 479-bp 30UTR, and also revealed that the teg-1transcript is SL1 trans-spliced. The positions of the mutations associated with the two alleles of teg-1 are shown, with the oz230 mutationchanging AAG to UAG (amber) at codon 53 (position 257 on genomic DNA from ATG). The oz189 mutation changes UGG to UGA (opal) atcodon 142 (position 526 on genomic DNA from ATG). B: An alignment of amino acid sequences between the TEG-1 predicted protein andhomologs from Saccharomyces cerevisiae, Drosophila melanogaster, and humans. Alignment was performed using ClustalW, with similar aminoacids shown with black text on a grey background, and identical amino acids shown with white text on a black background. Amino acids wereconsidered similar or identical if three or four of the four proteins contained a similar or identical amino acid at a given position. The thick blackunderline at the end of the proteins represents the regions deleted in the DGYF constructs. Asterisks are below each of the amino acids inCD2BP2 that are part of the GYF signature (W-X-Y-X6-11-G-P-F-X4-M-X2-W-X3-G-Y-F) (Kofler and Freund, 2006). The predicted locations of thepremature stop codons introduced by the oz189 and oz230 alleles are shown. TEG-1 is 14.8, 30.9, and 30.7% identical to ScLin1p, DmLIN1,and HsCD2BP2, respectively; and 25.6, 42.2, and 41.7% similar to these same proteins. Genbank accession numbers: ScLin1p (NP_012026.1),DmLin1 (Q9VKV5), and HsCD2BP2 (NP_006101.1).

510 WANG ET AL.

with a complete absence of GLP-/Notch signaling, strongly supportsthe placement of teg-1 downstream ofNotch signaling in the genetic path-way regulating the proliferationversus differentiation decision. Theabsence of tumor formation ingld-2(q497); glp-1(q175) teg-1(oz230)animals could still be consistent withteg-1 functioning downstream of GLP-

1/Notch signalling, if teg-1 functionsredundantly with another gene/pathway in inhibiting proliferation,for example, if removing teg-1 func-tion does not completely remove theactivity of the gld-1 pathway. Thatgld-2(q497); glp-1(q175) teg-1(oz230)animals are Glp, while gld-3(q730);glp-1(q175) teg-1(oz230) animals aretumorous, suggests that removal of

gld-3 reduces the activity of the gld-2pathway to a greater extent than theremoval of gld-2 activity, or that gld-3has functions in addition to its role inthe gld-2 pathway. In any event, thefact that a synthetic tumor is formedin gld-3(q730); teg-1(oz230) animals,even in the absence of GLP-1/Notchsignaling, strongly supports theplacement of teg-1 downstream ofGLP-1/Notch signaling. Additionally,the fact that teg-1 shows the samecomplex epistatic relationships as theteg-4 and prp-17 splicing factors sup-ports teg-1 functioning as a splicingfactor.

Characterization of teg-1

Single Mutants

To further compare teg-1(0) pheno-types with those observed in othersplicing factor mutants, we analyzedteg-1’s single mutant phenotypes. Aswith other splicing factor mutants(Belfiore et al., 2004; Mantina et al.,2009; Kerins et al., 2010; Zanettiet al., 2011), teg-1 has a role in germ-line sex determination; loss of teg-1function results in a cold sensitiveMog (masculinization of the germline) phenotype. At 15�C, teg-1 her-maphroditic animals fail to switchfrom the production of sperm tooocytes, resulting in over �600 spermmade per gonad arm (6326156, n¼5),as compared to �150 in wild-type ani-mals (Hirsh et al., 1976) (Fig. 4). Wealso found that fog-1(q241), fog-3(q443), and fem-3(e1996) are all epi-static to teg-1(oz230) (Table 2), sug-gesting that teg-1 likely functionsupstream of these genes. Epistasis ofgermline sex determination geneswith other splicing factors, as well assomatic reporter assays with fem-3,suggest that the splicing factors couldbe functioning on the fem-3 gene, ulti-mately regulating its translation(Graham and Kimble, 1993; Grahamet al., 1993; Ellis and Kimble, 1995;Gallegos et al., 1998; Kerins et al.,2010). A complexity to this is that wefound that both fog-2(null) and tra-2(gain-of-function) are, at least par-tially, epistatic to teg-1(oz230) (Table2). However, the epistasis with fog-2is incomplete (88% of teg-1(oz230);fog-2(q71) animals are Fog [Feminiza-tion of the germ line], with theremaining animals showing some

Dev

elop

men

tal D

ynam

ics

TABLE 1. Epistasis Analysis of teg-1 With Genes Regulating the

Proliferation Vs. Differentiation Decision

Genotype

Proliferation vs. differentiation

phenotype (20�C)a

teg-1(oz189 and oz230) Wild-typegld-1(q485) Wild-typeb

nos-3(oz231) Wild-typec

gld-2(q497) Wild-typed

gld-3(q730) Wild-typee

glp-1(q175) Glpf

gld-2(q497) gld-1(q485) Tumorousg

gld-1(q485); teg-1(oz230) Wild-typeh

nos-3(oz231); teg-1(oz230) Wild-typei

gld-2(q497); teg-1(oz230) Tumorousj

gld-3(q730); teg-1(oz230) Tumorousgld-2(q497) gld-1(q485); glp-1(q175) Tumorousk

gld-2(q497); glp-1(q175) teg-1(oz230) Glpl

gld-3(q730); glp-1(q175) teg-1(oz230) Tumorousm

a100% of animals show the phenotype described, unless otherwise stated. For eachgenotype, a minimum of 30 gonad arms was analyzed.

b(Francis et al., 1995a). gld-1(q485) XX animals have tumorous germ lines; however,this has to do with female germ cells entering into meiotic prophase, but failing toprogress through meiosis. These cells then return to mitosis, forming a tumor. Theinitial entry into meiosis in these animals is very similar to wild-type; therefore,they do not have a defect in the balance between proliferation and meiotic entry.

c(Hansen et al., 2004b).d(Kadyk and Kimble, 1998).e(Eckmann et al., 2004).f(Austin and Kimble, 1987). Germ line proliferation abnormal. The proliferative pop-ulation of cells is depleted, resulting in gonads containing only a limited number ofdifferentiated cells. In glp-1 animals, only �16 sperm are formed per gonad arm.g(Kadyk and Kimble, 1998).hVab progeny from gld-1(q485)/hT2; vab-7(e1562) teg-1(oz230) were examined byNomarski optics and dissected gonads were analyzed using fluorescent microscopy.Both XX and XO animals were examined.

iUnc progeny from nos-3(oz231); unc-32(e189) teg-1(oz230) were examined byNomarski optics.jVab or Unc self or cross progeny from gld-2(q497)/hT2; vab-7(e1562) teg-1(oz189 oroz230)/hT2 or gld-2(q497)/hT2; unc-32(e189) teg-1(oz189 or oz230)/hT2 wereexamined by Nomarski optics. Also, dissected gonads were DAPI stained and ana-lyzed using fluorescent microscopy.k(Kadyk and Kimble, 1998).lPremature entry to meiosis phenotype similar to a glp-1(null) phenotype. Unc prog-eny from gld-2(q497)/hT2; unc-32(e189) glp-1(q175 and q172) teg-1(oz230)/hT2were examined by Nomarski optics and dissected gonads were analyzed using fluo-rescent microscopy.mUnc progeny from gld-3(q730)/mIn1; unc-32(e189) glp-1(q175) teg-1(oz230)/hT2were examined by Nomarski optics and dissected gonads were analyzed using fluo-rescent microscopy.

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 511

sperm production); therefore, teg-1could still function with the othersplicing factors in regulating fem-3,and the epistasis with fog-2(null) andtra-2(gf) could reflect a limited rolefor teg-1 in regulating fem-3. It mustbe noted that teg-1’s role in sex deter-mination is not essential; homozy-gotes are not Mog at 25�C, eventhough the teg-1(oz189 & oz230) al-leles are likely null (see below).

Therefore, even at 15�C, knockingout teg-1 function likely does notcompletely block the sex determin-ing signaling pathway at the posi-tion where teg-1 functions. Anincrease in a feminization signalupstream of teg-1, such as with fog-2(null) and tra-2(gf), could stillresult in sufficiently decreased ac-tivity of the fem genes so as tocause feminization.

teg-1 single mutants also displayother phenotypes. At 20�C, teg-1(0) ani-mals are sterile, producing sperm andsmall, abnormally shaped oocytes (Fig.4), while at 25�C, they produce func-tional sperm and oocytes, and thereforeare fertile, provided that their motherhad one wild-type copy of teg-1 (inother words, mþz� animals are fertile).Homozygous progeny of homozygousmothers (m�z�), grown at 25�C, aresterile with no differentiated germ cellsbeing formed (Fig. 4). Therefore, inaddition to the cold-sensitive Mog phe-notype, which is consistent with TEG-1functioning as a splicing factor, TEG-1also has functions in germline prolifer-ation and oogenesis.

TEG-1 Is Enriched in the

Nucleus

Splicing occurs in the nucleus of cells;therefore, most splicing factors are nu-clear enriched. To determine if TEG-1is also nuclear enriched, supporting arole for TEG-1 as a splicing factor, aswell as to gain additional insight intoTEG-1 function, we raised antibodiesagainst TEG-1 (see Experimental Pro-cedures section). These antibodies arespecific to TEG-1 as they detect an�65-kD protein on Western blots thatis absent in teg-1(oz230) homozygotes(Fig. 5). Staining revealed that TEG-1is enriched in the nuclei of all germ-line and somatic cells of the gonad(Fig. 5); however, staining above back-ground was also detected in the cyto-plasm of the gonad and intestine, sug-gesting that TEG-1 is also present inthe cytoplasm (Fig. 5). Within thenuclei, TEG-1 is found throughout thenucleoplasm, but appears to be absentin the regions containing DNA. TEG-1levels are not uniform throughout thecells of the gonad. TEG-1 levelsappear highest in nuclei of theoocytes, the somatic Distal Tip Cell(DTC), and sheath cells. TEG-1enrichment in the nucleus is consist-ent with a role in splicing.

TEG-1 Physically Interacts

With Known Splicing Factor

UAF-1

The genetic and expression analysesdescribed above suggest that TEG-1is functioning, at least in part, as asplicing factor. Additionally, the

Dev

elop

men

tal D

ynam

ics

Fig. 4. teg-1(oz230) single mutant animals display temperature-specific germline phenotypes.Germline phenotypes of teg-1(oz230) homozygous animals were analyzed using Nomarski opticsin adult animals one day past the fourth larval stage. Animals were grown at (B) 15�C, (C) 20�C,and (D, E) 25�C. A wild-type animal (A) grown at 20�C is provided for comparison. For each ani-mal, only one gonad arm is shown, with the distal arm on top, and the distal end to the left. Thegonad arm is outlined with a dashed line, from the distal end to the spermatheca. Strains weregrown as teg-1(oz230)/hT2 heterozygotes at the specified temperature for multiple generationsprior to analyzing teg-1(oz230) homozygous animals for their germ line phenotype, except forteg-1(oz230 m-z-) animals (E), in which the mother of the analyzed animal was also homozygousfor teg-1(oz230). B: Gonads in animals grown at 15�C are somewhat smaller and exclusivelyproduce sperm, even in older animals. C: At 20�C, animals produce sperm and smaller mis-shapen oocytes; however, they are sterile. D: At 25�C, animals produce sperm and oocytes, aswell as viable progeny; however, the progeny (E) are sterile, with no obvious oocyte production,although a limited number of sperm are observed. For each temperature >100 animals were an-alyzed, with all animals grown at the same temperature exhibiting similar phenotypes. Actual ge-notype unc-32(e189) teg-1(oz230). Scale bar ¼ 20 mm.

512 WANG ET AL.

human and yeast TEG-1 homologsphysically interact with known splic-ing factors (see Discussion), althoughthe overall similarity between TEG-1and its homologs is not overly high, soit cannot be assumed that TEG-1would have the same binding part-ners in C. elegans (Fig. 3). To increaseour understanding of how TEG-1 mayfunction in the splicing reaction, wesought to identify proteins with whichit binds. To accomplish this, weexpressed a TAP-tagged version ofTEG-1 in animals and immunopreci-pitated (IP’d) TAP::TEG-1 and associ-ated proteins. Importantly, thistagged version of TEG-1 rescues teg-1(oz230) mutant phenotypes, and isexpressed at a similar level as endoge-nous TEG-1, suggesting that TAP::-TEG-1 is functioning similarly to the

wild-type protein (Fig. 5). After TEG-1 and its associated proteins wereIP’d, we subjected the samples tomass-spectometry analysis to deter-mine the identities of the interactingproteins, and compared these proteinsto control IPs. UAF-1, a conservedsplicing factor that is homologous tomammalian U2AF65, was identifiedin two independent TAP::TEG-1 IPs,but was not detected in either of thetwo control IPs. A full description ofproteins found to bind TEG-1 will bedescribed elsewhere. In order to verifythe TEG-1–UAF-1 interaction, weperformed reciprocal IPs from wholeworm lysates. UAF-1 was co-IP’d withTEG-1 (using TEG-1-specific antibod-ies), and TEG-1 was co-IP’d withUAF-1 (using UAF-1-specific antibod-ies) (Fig. 6A).

These reciprocal co-IP experimentsprovide strong evidence that at leasta proportion of TEG-1 and UAF-1exist in a complex within the cell;however, this does not demonstratethat these two proteins directly inter-act. To test if the TEG-1–UAF-1 inter-action is direct, we expressed bothproteins in bacterial cells, wheremRNA splicing does not occur andsplicing factors are not present.Therefore, if TEG-1 and UAF-1 arepart of the same multi-protein com-plex, the other proteins of the complexlikely do not exist within the bacterialcell, and only direct interactionswould be detected. We found that bac-terially expressed TEG-1 and UAF-1co-IP (Fig. 6B), suggesting that thetwo proteins directly bind. Further-more, when we expressed a truncatedversion of TEG-1, which lacks apotential (although quite degenerate)GYF protein binding motif (calledTEG-1DGYF; Fig. 3), an interactionbetween TEG-1DGYF and UAF-1 wasnot detected (Fig. 6B). The lack ofinteraction between TEG-1DGYF andUAF-1 provides an important nega-tive control, suggesting that the bac-terial TEG-1–UAF-1 co-IPs are spe-cific. Furthermore, the lack ofdetectable binding between TEG-1DGYF and UAF-1 suggests that thedegenerate GYF domain is necessaryfor the TEG-1–UAF-1 interaction.While reducing uaf-1 activity throughRNAi results in an embryonic lethalphenotype (Zorio and Blumenthal,1999b), partially reducing uaf-1 activ-ity through RNAi in a glp-1(gf) back-ground results in a significantincrease in proliferation in the germline (Table 3), suggesting that, liketeg-1, uaf-1 is also involved in regulat-ing the proliferation versus differen-tiation decision in the C. elegans germline. Additionally, since a completereduction of teg-1 activity in teg-1 nullmutants still allows for animals toreach adulthood, while severelyreducing uaf-1 activity results in em-bryonic lethality, suggests that theTEG-1 function may be modulatory.

The TEG-1–UAF-1 Interaction

Is Conserved in Humans

An interaction between homologs ofTEG-1 and UAF-1 has not been previ-ously reported in any system. Given

Dev

elop

men

tal D

ynam

ics

TABLE 2. Interaction of teg-1 With Genes in the Germ Line

Sex Determination Pathway

Genotype Germline phenotype (15�C)

teg-1(oz189 and oz230) Moga

fog-1(q241)b Fogc

fog-2(q71)d Foge

fog-3(q443)f Fogfem-3(e1996)g Fogtra-2(e2020gf)h Fogfog-1(q241); teg-1(oz230)i Fogteg-1(oz230); fog-2(q71)j Fogfog-3(q443); teg-1(oz230)k Fogteg-1(oz189); fem-3(e1996)l Fogtra-2(e2020gf); teg-1(oz230)m Fog

aMasculinization of the germ line.b(Barton and Kimble, 1990).cFeminization of the germ line.d(Schedl and Kimble, 1988).eXX only; XO animals are wild type.f(Ellis and Kimble, 1995).g(Hodgkin, 1986).h(Doniach, 1986).iAll Vab Unc progeny of fog-1(q241) unc-13(e51)/ hT2; vab-7(e1562) teg-1(oz230)/hT2 were Fog.jfog-2(q71) rol-9(sc148)/ unc-51(e369) fog-2(q71) males mated with vab-7(e1562) teg-1(oz230)/hT2 hermaphrodites. L4 hermaphrodites cloned and put at 15�C. Of thosethat segregated both Rol and Vab animals, 21/24 Rol Vab progeny (vab-7(e1562)teg-1(oz230); fog-2(q71) rol-9(q71) contained small oocytes and no sperm; 3/24 con-tained sperm. 21/21 non-Rol non-Unc Vab animals (teg-1(oz230); fog-2(q71)/þ) wereMog.kAll Unc progeny of teg-1(oz230)/hT2; unc-13(e51) fog-3(oz230) were Fog.lvab-7(e1562) teg-1(oz189); unc-24(e138) fem-3(e1996) dpy-20(e1282) females wereobtained from an unbalanced strain.mtra-2(e2020) males were mated with vab-7(e1562) teg-1(oz230)/hT2 purged her-maphrodites. Single male progeny were then mated with single female progeny.From matings that failed to segregate hT2, L4 (XX) Vab animals were cloned andscored the next day. 19/25 animals had small oocytes and no sperm. The remain-ing six animals were Mog (presumably tra-2(þ)/tra-2(þ)).

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 513

the overall low level of similaritybetween TEG-1 and its homologs (Fig.3), it is possible that the TEG-1–UAF-1 interaction is unique to nematodes,and not present in other systems. Totest if the TEG-1–UAF-1 interactionis present in humans, we tested if thesame interaction could be detected inHeLa cells. A GFP-tagged version of

CD2BP2, the human homolog of TEG-1, was transformed into Hela cells,followed by IP with anti-GFP antibod-ies (Fig. 6C). As a control, a constructencoding only GFP was transformedinto cells and IP’d. Using antibodiesspecific to U2AF65, the human homo-log of UAF-1, we detected U2AF65 co-IP with GFP::CD2BP2, but not with

the GFP-negative control (Fig. 6C). Wealso detected the interaction by per-forming the reciprocal IP; immunopre-cipitate obtained using U2AF65 anti-bodies contained GFP::CD2BP2, asdetected using anti-GFP antibodies.Additionally, CD2BP2 and U2AF65expressed in bacterial cells were co-IP’d, suggesting that the interactionbetween the two proteins is direct(Fig. 6D). Together, these results dem-onstrate that the TEG-1–UAF-1 inter-action, which we first identified inC. elegans, is also present in humans.

DISCUSSION

We isolated teg-1 alleles in two differ-ent mutant screens, each designed toidentify genes that function in theproliferation versus differentiationdecision in the C. elegans germ line.We found that, in addition to its role inthe proliferation versus differentiationdecision, teg-1 also functions in regulat-ing germline sex determination, germ-line proliferation, and oogenesis. Clon-ing of teg-1 revealed that it encodes ahomolog of human CD2BP2, and thatit is enriched in the nucleoplasm of

Dev

elop

men

tal D

ynam

ics

Fig. 5. TEG-1 is nuclear enriched in all gonadal cells. A: A dissected gonad arm from a wild-type (N2) adult hermaphrodite, one day past thefourth larval stage grown at 20�C, stained with DAPI (blue) to visualize nuclear morphology, and anti-TEG-1 specific antibodies (red). The distalend of the gonad is to the left. Highest levels of TEG-1 enrichment are observed in nuclei of the distal tip cell (asterisk), sheath cells (arrowhead),and oocytes (large nuclei in the proximal half of the gonad arm). The inset shows a blow-up of a portion of the pachytene region of the gonadarm, with the DAPI and anti-TEG-1 channels merged. The regions of the nuclei occupied by DNA and TEG-1 appear to be mutually exclusive. In-tensity comparisons of oocyte and intestine cytoplasm in wild-type and teg-1(oz230) animals demonstrates that some TEG-1 is present in thecytoplasm (oocyte intensity measurements: 1,695 6 428 [n¼10] in N2 vs. 4566258 [n¼10] in teg-1(oz230); Intestine intensity measurements:1,504 6 218 [n¼15] in N2 vs. 596 6 233 [n¼17] in teg-1(oz230)). Scale bar ¼ 20 mm. B: Western blot showing relative size of the TEG-1 protein,and the specificity of the anti-TEG-1 antibodies. A TAP-tag (HA-His8x-TEV-Myc) was fused to the N-terminus of teg-1 genomic DNA to generatean integrated transgenic line. Proteins at 65 kDa (endogenous TEG-1) and at 72 kDa (TAP-TEG-1) are detected in the extract of the transgenicstrain. The comparable expression levels between these two proteins indicates that the TAP-TEG-1 is expressed at a level similar to endogenousTEG-1. Approximately 80 adult hermaphorodites from each C. elegans strain were loaded onto each lane of the SDS-PAGE gel. The wild-type andteg-1(oz230) strains are also marked with unc-32(e189). Paramyosin, detected with MH16 antibodies, was used as a loading control.

TABLE 3. Interaction of uaf-1 With glp-1(ar202gf)

Genotype

Germline phenotype

at 18�C (%)

naWild type Mogb Proc

rrf-1d; glp-1(ar202gf); pL4440(RNAi)e 96 0 4 54rrf-1; uaf-1(RNAi)f 49 51 0 49rrf-1; glp-1(ar202gf); uaf-1(RNAi) 32 15 53 34

aNumber of gonad arms.bMasculinization of the germ line.cProximal proliferation. Gonads contain over-proliferative tumor in the proximalend of the gonad.drrf-1(pk1417).epL4440 is an empty vector, used for control RNAi.fRNAi was performed by placing gravid hermaphrodites on RNAi plates and scoringthe progeny, ensuring that progeny were not exposed to RNAi until after hatching.If uaf-1 RNAi was performed by placing L4 animals on the RNAi plate, thisresulted in most of the progeny arresting as embryos, and a very small brood size.

514 WANG ET AL.

cells. We also demonstrated that TEG-1 physically interacts with UAF-1, theC. elegans homolog of human U2AF65.Finally, we demonstrated that this is aconserved interaction, with the humanhomologs of TEG-1 and UAF-1 inter-acting in human cells, revealing apotential new role for TEG-1 CD2BP2in the splicing reaction.

Splicing and the Balance

Between Proliferation and

Differentiation

A number of genes encoding splicingfactors have been identified that,when mutated, result in similargermline phenotypes in C. elegans,

including masculinization of the germline, and over-proliferation in sensi-tized genetic backgrounds (Puoti andKimble, 2000; Belfiore et al., 2004;Konishi et al., 2008; Mantina et al.,2009; Kerins et al., 2010; Zanettiet al., 2011). Indeed, a comprehensiveexamination of splicing factors andthe ability of a reduction in theiractivity to cause over-proliferation indifferent genetic backgrounds hasrecently been performed (Kerinset al., 2010). Of 114 splicing factorstested by RNAi, 11 disrupted germ-line sex determination resulting in aMog phenotype, while 31 enhancedthe over-proliferation phenotype of aglp-1(gf) mutant. Importantly, the

splicing factors that resulted in thesephenotypes are found throughout thevarious steps of the splicing reaction.Therefore, it is not just one step of thesplicing reaction that can be dis-rupted to cause these phenotypes, noris it likely that a subset of splicingfactors have been co-opted into a func-tion separate from the other splicingfactors. Rather, it appears as thougha general disruption in splicing, orthe splicing machinery, results in theMog and over-proliferation pheno-types. Attempts at identifying poten-tial target mRNAs whose mis-splicingresults in a Mog or over-proliferationphenotype have thus far been unsuc-cessful (Puoti and Kimble, 1999; Bel-fiore et al., 2004), although defects insplicing due to a reduction in one ofthese factors has recently beendescribed (Zanetti et al., 2011). As wehave previously suggested, a failurein detecting causative splicing defectscould be due to the reduction in prop-erly spliced product (or accumulationof mis-spliced product) being too smallto detect using conventional techni-ques, or that the proper targets maynot have yet been tested (Mantinaet al., 2009). An alternate explanationcould be that the reduction in thesplicing machinery that causes theover-proliferation phenotype may notbe due to a defect in splicing, butrather that a reduction in splicing fac-tor activity could interfere with someother aspect of RNA metabolism,which results in over-proliferation.

Function of TEG-1/CD2BP2 in

the Splicing Reaction

The splicing of introns from pre-mRNAs is a multistep process thatinvolves many protein and RNA com-plexes. The U2 Auxiliary Factor(U2AF) is involved in the formation ofthe splicing commitment complex,including being necessary for properU2 snRNP binding to the pre-mRNA(Ruskin et al., 1988; Zamore andGreen, 1989). U2AF binds to the 30

splice site of the intron, whichinvolves U2AF65, the large subunit ofU2AF, binding to a polypyrimidinetract, and the small U2AF subunit,U2AF35, binding to the AG/ splicesite (Merendino et al., 1999; Wu et al.,1999; Zorio and Blumenthal, 1999a;Wahl et al., 2009; Valadkhan and

Dev

elop

men

tal D

ynam

ics Fig. 6. TEG-1 directly interacts with UAF-1 in C. elegans extracts, and this interaction is con-

served in humans. A: Reciprocal co-immunoprecipitations (IPs) were performed using UAF-1 spe-cific antibodies (left) and TEG-1 specific antibodies (right), followed by SDS electrophoresis andWestern blotting. In the UAF-1 IP, TEG-1-specific antibodies detect a band of the correct size,whereas an IP using non-specific pre-immunized rabbit serum did not result in detection of TEG-1. In the TEG-1 IP, a band of the correct size was detected with UAF-1-specific antibodies, butnot when pre-immune serum was used to perform the IP. B: E. coli expressed GST-TEG-1 and6X-His-UAF-1 co-IP. 6X-His-UAF-1 was expressed in E. coli cells with GST, GST-TEG-1, or GST-TEG-1 with the GYF domain removed (GST-TEG-1(DGYF); Fig. 3B). When UAF-1-specific antibod-ies were used to IP, only GST-TEG-1(full length) was co-IP’d (top). Likewise, when GST-specificantibodies were used to IP, only the strain containing GST-TEG-1(full length) was able to co-IPUAF-1, as detected by UAF-1-specific antibodies. C: Co-IP of U2AF65 and CD2BP2 from HeLaextracts. GFP or GFP-CD2BP2 were transiently expressed in HeLa cells and proteins were IP’dusing anti-GFP-specific antibodies (left) or anti-U2AF65-specific antibodies (right). U2AF65 andU2AF35 were detected in the anti-GFP IP on a Western blot, while the negative control, Hsp90heat shock protein, was not co-IP’d. Likewise, GFP-CD2BP2 was co-IP’d with U2AF65, whileGFP and Hsp90 were not (right). U2AF35 was co-IP’d with U2AF65, whether using lysate fromGFP-expressing cells or GFP-CD2BP2-expressing cells. D: E. coli expressed GST-CD2BP2 andU2AF65 co-IP. U2AF65 was expressed in E. coli cells with GST, GST-CD2BP2(full length) or GST-CD2BP2(DGYF). Only GST-CD2BP2(full length) was co-IP’d with U2AF65. Likewise, when IP wasperformed using anti-GST antibodies, U2AF65 was only detected when lysate from cells express-ing GST-CD2BP2(full length) was used; U2AF65 could not be detected when lysate from cellsexpressing GST or GST-CD2BP2(DGYF) was used for the IP.

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 515

Jaladat, 2010). In vertebrates, boththe length of the polypyrimidine tract,and the distance between the tractand the splice site, can vary betweenintrons (Wahl et al., 2009; Valadkhanand Jaladat, 2010). However, in C. ele-gans most introns have a highly con-served consensus, UUUUCAG/R, withno spacing between the tract and thesplice site (Zhang and Blumenthal,1996; Blumenthal and Steward, 1997;Kent and Zahler, 2000). Both UAF-1,the C. elegans homolog of U2AF65(Zorio et al., 1997), and UAF-2, the C.elegans homolog of U2AF35 (Zorio andBlumenthal, 1999b), bind to this con-sensus sequence (Zorio and Blumen-thal, 1999a), and are likely responsiblefor the high degree of sequence conser-vation in the consensus C. elegans 30

splice site (Hollins et al., 2005).Homologs of TEG-1 have been impli-

cated in splicing. Human CD2BP2 co-purifies with the U5 snRNP, bindingthe 15- and 102-kDa proteins; there-fore, CD2BP2 has also been calledU5-52K (Bach et al., 1989; Behrensand Luhrmann, 1991; Laggerbaueret al., 2005). During the stepwiseassembly of the spliceosome, theU5 snRNP associates with the U4/U6di-snRNP to form the U4/U6.U5 tri-snRNP (Behrens and Luhrmann,1991). Importantly, even thoughCD2BP2(U5-52K) binds to the U5snRNP, it is not found in the U4/U6.U5 tri-snRNP, suggesting thatCD2BP2(U5-52K) may be involved inthe assembly of the tri-snRNP, but dis-sociates from U5 during tri-snRNPformation. (Laggerbauer et al., 2005).The ortholog of TEG-1 in Saccharomy-ces cerevisiae, Lin1p, also binds to theU5 snRNP, suggesting that this role intri-snRNP assembly is conserved(Bialkowska and Kurlandzka, 2002).

We propose that the interaction wehave detected between TEG-1CD2BP2(U5-52K) and UAF-1 U2AF65reveals a function for TEG-1CD2BP2(U5-52K) that is distinct fromits role in tri-snRNP formation.Indeed, proteomic analyses of Com-plex A and Complex B support a func-tion for TEG-1 CD2BP2 that is inde-pendent of tri-snRNP formation. TheU4/U6.U5 tri-snRNP binds to the spli-ceosome as a preassembled complex(Bindereif and Green, 1987; Chengand Abelson, 1987; Konarska andSharp, 1987). However, CD2BP2(U5-

52K) was detected in experiments inwhich Complexes A and B were iso-lated and their protein constituentsdetermined through proteomic analy-sis (Hartmuth et al., 2002; Deckertet al., 2006). If CD2BP2(U5-52K) isonly present in the U5 snRNP prior toU4/U6.U5 tri-snRNP formation, and ifthe tri-snRNP is pre-assembled priorto becoming part of the spliceosome,then CD2BP2(U5-52K) should neverbe part of the spliceosome as it isbeing assembled on the pre-mRNA,including as part of Complex A orComplex B. Furthermore, the U4/U6.U5 tri-snRNP has not yet becomepart of the splicesome during ComplexA formation. Therefore, even ifCD2BP2(U5-52K) does have some af-finity for U5 after tri-snRNP forma-tion, it should not be isolated withComplex A. Our identification of TEG-1 CD2BP2(U5-52K) as binding toUAF-1 U2AF65 is consistent with itspresence in Complexes A and B, andsuggests an additional function forTEG-1 CD2BP2(U5-52K).

We do not know when or why TEG-1 CD2BP2(U5-52K) binds to UAF-1U2AF65 during spliceosome assem-bly. However, U2AF65 has beenshown to bind other proteins duringspliceosome assembly. For example,in both yeast and humans U2AF65binds to the branch point binding pro-tein SF1, presumably to aid in com-mitment complex formation (Abovichand Rosbash, 1997). Additionally,SF3b155, which is part of the U2snRNP, binds to U2AF65 as the U2snRNP replaces SF1 at the branchpoint (Gozani et al., 1998). Two-hybridanalysis also identified multiple otherU2AF65 splicing factor binding part-ners (Prigge et al., 2009); however, thefunctions of these interactions areunknown. Therefore, U2AF65 bindsdifferent proteins as formation of thespliceosome progresses. TEG-1CD2BP2(U5-52K) binding to UAF-1U2AF65 may be another such interac-tion functioning in spliceosome forma-tion. Although we do not yet knowwhat the consequences of this interac-tion may be, it is intriguing that bothUAF-1 U2AF65 and the U5 snRNP,with which TEG-1 CD2BP2(U5-52K)was previously shown to bind (Lagger-bauer et al., 2005), associate with the30 splice site. As mentioned, U2AF65binds to the polypyrimidine tract near

the 30 splice site, with UAF-1 bindingto the analogous UUUUCAG/Rsequence adjacent to the 30 splice sitein C. elegans. In mammalian cells,three proteins of the U5 snRNPreplace U2AF in this region duringspliceosome assembly (Chiara et al.,1997). Even though these three U5proteins (110, 116, and 220 kDa) arenot the same U5 snRNP proteins withwhich CD2BP2(U5-52K) interacts (15and 102 kDa), it is possible that func-tions that CD2BP2(U5-52K) provideswhile binding to the U5 snRNP and toUAF-1 U2AF65 are related. For exam-ple, it is possible that TEG-1CD2BP2(U5-52K) is involved in assist-ing both UAF-1 U2AF65 and the U5snRNP to bind to the 30 end of theintron. It is also possible that TEG-1CD2BP2(U5-52K) is involved in facili-tating the replacement of UAF-1U2AF65 by the U5 snRNP. Differenti-ating between these and other possi-bilities will require studying the tim-ing of the TEG-1/CD2BP2 interactionswith UAF-1/U2AF65 and the U5snRNP during spliceosome assembly.

EXPERIMENTAL

PROCEDURES

C. elegans Growth Conditions

and Strains

Maintenance and manipulation of C.elegans was performed as previouslydescribed (Brenner, 1974). Bristol N2was the wild-type strain used and thefollowing mutants were used in thisstudy. LGI: fog-1(q241), gld-2(q497),rrf-1(pk1417), gld-1(q485), fog-3(q443); LGII: tra-2(e2020gf), gld-3(q730), nos-3(oz231); LGIII: unc-32(e189) glp-1(ar202gf), glp-1(oz112oz120), spe-16(hc54), teg-1(oz189), teg-1(oz230), dpy-18(e364) LGIV: fem-3(e1996) LGV fog-2(q71).

Mapping teg-1

Using standard three-factor mapping,we mapped both the over-prolifera-tion and Mog phenotypes of teg-1(oz189) to a �0.5 cM region between,or very close to, spe-16 and dpy-18 onlinkage group III. We further nar-rowed the critical region through acombination of single nucleotide poly-morphism (SNP) mapping and defi-ciency mapping. The right boundary

Dev

elop

men

tal D

ynam

ics

516 WANG ET AL.

of the critical region was determinedthrough SNP mapping using teg-1(oz189) dpy-18(e364)/HA-8. Thecold-sensitive Mog phenotype wasmapped by picking Dpy non-Teg(Mog) animals. A recombinant wasobtained that retained N2 DNA fromdpy-18 to a SNP in T28D6.5, makingthis SNP the right boundary of thecritical region (SNP between11337877 and 11339172-wormbasefreeze WS220; information providedby Kate Hill). The left boundary of thecritical region was narrowed by map-ping the endpoint of the deficiencytDf2, which complements teg-1(oz189).We found that tDf2 deletes part ofY47D3A.25, but not the gene immedi-ately to the right, Y47D3A.27. There-fore, the left boundary of the criticalregion is in Y47D3A.25. These left andright boundaries narrowed the criticalregion to a portion of chromosome IIIcontaining only three genes:Y47D3A.27, Y47D3A.28, and T28D6.6.Sequencing revealed that Y47D3A.27contained nonsense mutations corre-sponding to both the teg-1(oz189) andteg-1(oz230) alleles. Sequencing of theyk82g9 cDNA, as well as RT-PCRamplified cDNA (using primers R10-CCGAAACGCGCCGTTTCTTT andR11-TTACAAATAAAGCTCGAAATCA)was used to determine the inton/exonboundaries.

Generation of Anti-TEG-1

Antibodies

Two synthetic peptides, one corre-sponding to the TEG-1 amino-terminalsequence (CVKERPGNADEDEEEKKKFHT), and one to the carboxyl-termi-nal sequence (CAGAPMEQEEEEADDSVKWEY), were conjugated by Key-hole-limpet hemocyanin (KLH) andinjected into rats and rabbits for anti-body production by Covance (CovanceInc., Princeton, NJ). Antibodies fromeach species of the antisera were affin-ity purified against the correspondingpeptides coupled to the SulfoLink Cou-pling Gel (Pierce, Rockford, IL) accord-ing to the manufacturer’s instruction.

Gonad Dissections and

Indirect Immunofluorescence

Staining

Gonad dissections and antibody stain-ing were performed as previously

described (Jones et al., 1996). Briefly,dissected gonads were fixed with 3%paraformaldehyde for 10 min, fol-lowed by fixation/permeabilizationwith �20�C, 100% methanol for > 30min. The gonads were blocked in 3%BSA. Affinity purified rat anti-TEG-1(N-terminal) antibodies were used at1:500 dilution, anti-REC-8 antibodieswere used at 1:200 dilution, and anti-HIM-3 antibodies (Zetka et al., 1999)were used at 1:500 dilution. NuclearDNA was visualized by DAPI. Imageswere taken using a Zeiss (Zeiss,Thornwood, NY) Imager Z1 micro-scope equipped with an AxiocamMRM digital camera.

Plasmid Construction

TAP-tag teg-1 (pDH122)

construction.

The teg-1 promoter region, a 522-bpfragment containing sequences down-stream of rab-35 stop codon andupstream of teg-1 start codon, wasPCR amplified (teg-1 Pro F primer:tctgaaaattgaacaatttgtg; teg-1 Pro Rprimer: tttttggctgaaaatgagtgaa) andsub-cloned into the pBluescript II KS(þ) vector (Stratagene, La Jolla, CA)at NotI and BamHI sites. The tandemaffinity purification (TAP) tag encodesfor HA-8xHis-TEV-Myc epitopes andwas amplified by PCR (TAP F primer:atgtacccatacgatgtcccaga; TAP R pri-mer:caaatcctcctcgctgatc) using thepSB_GW::TAG vector as template(Polanowska et al., 2004; Walhoutand Boulton 2006). The amplifiedTAP-tag was then sub-cloned into thevector containing the promoter at theBamHI and EcoRI sites, where theEcoRI site was filled-in with Klenowenzyme. The teg-1 coding region wasamplified, using teg-1 Gen F primer(ccgaaacgcgccgtttctt) and teg-1 Gen Rprimer (tgatgttgggttgcagaagcct), andsub-cloned into the EcoRV site of thepCRII-TOPO vector (Invitrogen,Carlsbad, CA). The sequence contain-ing both the teg-1 promoter and theTAP-tag was sub-cloned into the KpnIand SpeI sites of this pCRII-TOPOvector, resulting in the completed res-cuing construct (pDH122).

pL4440-uaf-1 Construct (pDH182).

The uaf-1 cDNA sequence (368–951)was sub-cloned into pL4440 vector at

XbaI and HinDIII sites and used totransform the E. coli HT115(DE3)strain, which were able to producedsRNA (Timmons and Fire, 1998).

Transgenic Array and

Integration

The TAP-tag teg-1 construct(pDH122) was injected into N2 her-maphrodite gonads, together withpJM67 (an elt-2::GFP constructkindly provided by J. M. McGhee,University of Calgary) and the pRF4plasmid as visible transformationmarkers (Mello et al., 1991; Fukushigeet al., 1998). After following the select-able co-injection markers for two gen-erations, the extrachromosomal arrayswere randomly integrated into the ge-nome by gamma-irradiation (137Cs)with a total dose of �3,000 rads. Twoindependent integrants, designatedugIs3 [tap::teg-1] (strain XB307) andugIs4 [tap-teg-1] (strain XB308), wereisolated and out-crossed three times.The integrated array corresponding tougIs3 was crossed into teg-1(oz230)and was able to rescue the sterile phe-notype of teg-1(oz230) allele.

Preparation of C. elegans

Extracts

To generate sufficient starting mate-rial for purification processes, proteo-mic analyses, and immunoprecipita-tion studies, the wild-type orXB307(ugIs3) transgenic line from 20cleared, 35-mm plates was used to in-oculate 250 ml S medium supple-mented with a E. coli OP50 pellet con-centrated from 2 L of LB broth(Sulston and Hodgkin, 1988). Theworm culture was grown for 3–5 daysat 20�C on a shaking platform (250rpm) until a freshly cleared culturewas generated. The live worms wereharvested by sucrose gradient fol-lowed by three washes with 0.1 MNaCl (Polanowska et al., 2004). Thepellet was flash frozen in liquid nitro-gen and stored at �80�C. The flash-frozen worm pellet was resuspendedin 5 volumes of CSK lysis buffer (100mM PIPES, pH 6.0, 100 mM NaCl, 3mM MgCl2, 1 mM EGTA, 1 mM dithi-othreitol [DTT], 0.3 M sucrose, 0.5%Triton X-100, 1 mM PMSF, andEDTA-free complete protease inhibi-tor tablet [Roche, Montreal, QC])

Dev

elop

men

tal D

ynam

ics

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 517

(Polanowska et al., 2004). The wormsuspension was disrupted by FrenchPress at 20,000 psi and the solublelysate was obtained by centrifugationat 10,000g for 20 min. The proteinconcentration was determined by Bio-Rad Protein Assay (Bio-Rad, Hercu-les, CA) according to the manufac-turer’s instruction.

Affinity Purification of

TAP-TEG-1

A two-step affinity purification wasused to prepare samples for massspectrometry identification. To pre-pare pre-immune and anti-TEG-1antibodies-coupled protein A beads,0.5 ml of agarose protein A beads(Sigma, St. Louis, MO) were incu-bated with 0.5 ml of each pre-immunesera or affinity purified rabbit anti-TEG-1 (C-terminal) antibodies over-night at 4�C. The pre-immune andantibody-coupled beads were cross-linked as described in Polanowskaet al. (2004). Approximately 6–8 ml ofpacked XB307(ugIs3) worm pelletwas lysed in CSK lysis buffer asdescribed above. The extract was af-finity purified through a 2-ml Ni-NTAagarose column (Qiagen, German-town, MD). After washing the columnwith 5 volumes of PBS containing 10mM imidazole, the coupled proteinswere eluted with 10 ml of 100 mM im-idazole in PBS (137 mM NaCl, 10 mMNa2PO4-12H2O, 2.5 mM KCl, 1.8 mMKH2PO4, pH 7.2). In the secondimmunoaffinity step, the imidazoleeluates were equally divided andbatch absorbed with pre-immune oranti-TEG-1 antibody-coupled proteinA at 4�C for 2–3 hr. Beads were thentransferred to a Poly-Prep column(Bio-Rad) and washed three timeswith 6 volumes of PBS. The boundproteins were eluted three times with100 ml/fraction of 0.1 M glycine, pH2.5, and then neutralized with 1 MTris, pH 8.0. The protein sampleswere concentrated by speed-vac,resolved on 12.5% Laemmli SDS-PAGE gels, and then stained with Bio-Safe Coomassie (Bio-Rad). Bands wereexcised and samples were analyzed bythe SAMS Centre for Proteomics atUniversity of Calgary by LC-MS/MSfor protein identification. Peptide hitsthat were found in the pre-immunecontrol were subtracted from the anti-

TEG-1 antibody data. Common pro-teins found between two independentaffinity purified experiments were an-alyzed for their mRNA expression pat-terns using the NEXTDB in situ data-base to determine if they were germline expressed (Kohara, 2001).

Co-Immunoprecipitation

(Co-IP)

For UAF-1 co-IP, extracts from 1.5 mlof packed N2 pellet was obtained asdescribed above. For each immuno-precipitation, 10 mg of the extractwas incubated overnight at 4�C withof �20 mg of pre-immune sera, anti-UAF-1 antibodies, or affinity purifiedrat anti-TEG-1 (C-terminal) antibod-ies coupled to protein G beads(Sigma). The beads were washedthree times with CSK lysis buffer andresuspended in 2� SDS-PAGE load-ing buffer. Bound proteins were ana-lyzed by immunoblots.

Bacterial Pull-Down

Experiments

TEG-1 (2-353) and UAF-1 (1-496)were sub-cloned into pGEX-4T1 (GEHealthcare, Waukesha, WI) and pET-28a (Novagen, Darmstadt, Germany),respectively. A truncated teg-1 lackingthe GYF domain (amino acid residues296–353) was also sub-cloned intopGEX-4T1. Both GST- and His6x-tagconstructs were co-expressed in the E.coli BL21(DE3) strain (Tolia andJoshua-Tor, 2006). For pull-downassays, His6x-UAF-1 fusion proteinsthat bind to GST-TEG-1 recombinantproteins were co-immunoprecipitatedby immobilizing GST, GST-TEG-1,and GST-TEG-1DGYF on glutathioneagarose beads (GE Healthcare, Wau-kesha, WI). On the other hand, GSTrecombinant proteins that interactwith the His6x-UAF-1 fusion proteinswere purified on Ni-NTA beads (Qia-gen). The beads were washed fivetimes with PBS containing 0.3 MNaCl. Proteins were eluted by adding2� SDS-PAGE loading buffer into thebeads and analyzed by immunoblots.

Immunoblotting

Proteins were resolved on a 10% SDS-PAGE gel and transferred to Trans-Blot nitrocellulose membrane using

Mini-Trans Blot Cell (Bio-Rad). Themembrane was blocked with 5% Car-nation milk and incubated overnightat 4�C with either: (1) affinity purifiedTEG-1 (N-terminal) antibodies at1:1,000; (2) anti-UAF-1 antibodies(Zorio and Blumenthal, 1999a) at1:20,000; (3) anti-U2AF65 antibodies(Sigma) at 1:2,000; (4) anti-U2AF35antibodies (Novus Biologicals; Little-ton, CO) at 1:2,000; (5) anti-GFP(Rockland Immunochemicals, Gil-bertsville, PA) at 1:2,000; (6) anti-Hsp90a (Stressgen, San Diego, CA) at1:1,000; or (7) anti-paramyosin(MH16, hybridoma cell line from De-velopmental Studies HybridomaBank, University of Iowa, Iowa City,IA) tissue culture supernatant at 1:1,in TBS (137 mM NaCl, 25 mM Tris-HCl, pH 7.2, 2.5 mM KCl)-5% milk so-lution. The membrane was incubated2 hr at room temperature with HRP-conjugated secondary antibodies(Jackson Laboratories, Bar Harbor,ME) diluted at 1:5,000 and thenvisualized by SuperSignal WestPicosubstrate (Thermo Scientific, Wal-tham, MA) or Amersham ECLAdvance kit (GE Healthcare, Wauke-sha, WI). Images were collected onKodak (Rochester, NY) BioMax MRfilms.

HeLa Tissue Culture and

Transfection

HeLa-CCL2 cell line (a gift from M. J.Lohka, University of Calgary) wasgrown in Dulbecco’s modified Eagle’smedium supplemented with 3.7 g/mlNaHCO3, pH 7.2, 10% fetal bovin se-rum, 4 mM L-glutamine, 75 mg/mlPenicillin and Streptomycin at 37�Cwith 5% CO2. The plasmids express-ing GFP-CD2BP2 (Kofler et al., 2004)were prepared by Qiagen EndoFreemaxi plasmid kit (Qiagen) and trans-fected into HeLa-CCL2 using Lipo-fectamine 2000 (Invitrogen). At 24 hrafter transfection, cells were washedthree times with ice-cold PBS and lysedin HEPES lysis buffer (40 mM HEPES,pH 7.5, 120 mM NaCl, 1 mM EDTA,1% Triton X-100, 10 mM pyrophos-phate, 50 mM NaF, 1.5 mM Na3VO4,and protease inhibitor cocktail [Roche])on ice for 30 min with intermittent vor-texing. Whole cell lysate was centri-fuged at 10,000g for 10 min at 4�C andthe protein concentration of the soluble

Dev

elop

men

tal D

ynam

ics

518 WANG ET AL.

fraction was determined. For eachimmunoprecipitation, 2 mg of theextract was pre-cleared with the pro-tein A beads and then incubated over-night at 4�C with 20 mg of either anti-U2AF65 monoclonal antibodies (Sigma)or anti-GFP antibodies (RocklandImmunochemicals) that were coupledto protein A beads. The beads weresubsequently washed five times withthe lysis buffer and dissolved in 2�SDS-PAGE loading buffer for immuno-blotting analysis.

ACKNOWLEDGMENTSWe thank Kate Hill for providing valu-able information on single nucleotidepolymorphisms in the teg-1 region. Wethank Pavel Pasierbek and JosephLoidl for anti-REC-8 antibodies, Moni-que Zetka for anti-HIM-3 antibodies,and Tom Blumenthal for UAF-1 anti-bodies and advice. We thank Ted Han-sen for generating the two TEG-1-specific peptides, Christian Freund forproviding the GFP::CD2BP2 con-struct, Adrian Krainer for the pET9c-U2AF65 construct, and Carrie She-manko for anti-Hsp90a antibodies anduse of her tissue culture facility. Wethankmembers of the Hansen lab, andanonymous reviewers, for criticalreading of the manuscript. Some nem-atode strains used in this work wereprovided by the CaenorhabditisGenetics Center, which is funded bythe NIH National Center for ResearchResources (NCRR). This work wasfunded by a grant from the NationalInstitutes of Health (NIH GrantGM63310) to T.S., and grants from theCanadian Institutes of HealthResearch (CIHR) and the Natural Sci-ences and Engineering ResearchCouncil of Canada (NSERC) to D.H.

REFERENCES

Abovich N, Rosbash M. 1997. Cross-intronbridging interactions in the yeast com-mitment complex are conserved inmammals. Cell 89:403–412.

Austin J, Kimble J. 1987. glp-1 isrequired in the germ line for regulationof the decision between mitosis andmeiosis in C. elegans. Cell 51:589–599.

Bach M, Winkelmann G, Luhrmann R.1989. 20S small nuclear ribonucleopro-tein U5 shows a surprisingly complexprotein composition. Proc Natl Acad SciUSA 86:6038–6042.

Barton MK, Kimble J. 1990. fog-1, a reg-ulatory gene required for specification

of spermatogenesis in the germ line ofCaenorhabditis elegans. Genetics 125:29–39.

Behrens SE, Luhrmann R. 1991. Immu-noaffinity purification of a [U4/U6.U5]tri-snRNP from human cells. GenesDev 5:1439–1452.

Belfiore M, Pugnale P, Saudan Z, Puoti A.2004. Roles of the C. elegans cyclophi-lin-like protein MOG-6 in MEP-1 bind-ing and germline fates. Development131:2935–2945.

Berglund JA, Chua K, Abovich N, ReedR, Rosbash M. 1997. The splicing factorBBP interacts specifically with the pre-mRNA branchpoint sequenceUACUAAC. Cell 89:781–787.Berry LW. 1998. Regulation of the Mi-totic/Meiotic cell fate decision in Caeno-rhabditis elegans. In: Department ofGenetics. St. Louis: Washington Univer-sity School of Medicine. p 327.

Berry LW, Westlund B, Schedl T. 1997.Germ-line tumor formation caused byactivation of glp-1, a Caenorhabditiselegans member of the Notch family ofreceptors. Development 124:925.

Bialkowska A, Kurlandzka A. 2002. Pro-teins interacting with Lin 1p, a putativelink between chromosome segregation,mRNA splicing and DNA replication inSaccharomyces cerevisiae. Yeast 19:1323–1333.

Bindereif A, Green MR. 1987. An orderedpathway of snRNP binding duringmammalian pre-mRNA splicing com-plex assembly. EMBO J 6:2415–2424.

Blumenthal T, Steward K. 1997. RNAprocessing and gene structure. In: Rid-dle DL, Blumenthal T, Meyer B, PriessJR, editors. C. elegans II. Cold SpringHarbor, NY: Cold Spring Harbor Labo-ratory Press. p 117–145.

Brenner S. 1974. The genetics of Caeno-rhabditis elegans. Genetics 77:71–94.

Cheng SC, Abelson J. 1987. Spliceosomeassembly in yeast. Genes Dev 1:1014–1027.

Chiara MD, Palandjian L, Feld KramerR, Reed R. 1997. Evidence that U5snRNP recognizes the 30 splice site forcatalytic step II in mammals. EMBO J16:4746–4759.

Christensen S, Kodoyianni V, BosenbergM, Friedman L, Kimble J. 1996. lag-1, agene required for lin-12 and glp-1 signal-ing in Caenorhabditis elegans, is homolo-gous to human CBF1 and DrosophilaSu(H). Development 122:1373–1383.

Crittenden SL, Troemel ER, Evans TC,Kimble J. 1994. GLP-1 is localized tothe mitotic region of the C. elegansgerm line. Development 120:2901.

Deckert J, Hartmuth K, Boehringer D,Behzadnia N, Will CL, Kastner B,Stark H, Urlaub H, Luhrmann R. 2006.Protein composition and electron mi-croscopy structure of affinity-purifiedhuman spliceosomal B complexes iso-lated under physiological conditions.Mol Cell Biol 26:5528–5543.

Dernburg AF, McDonald K, Moulder G,Barstead R, Dresser M, Villeneuve AM.1998. Meiotic recombination in C. ele-

gans initiates by a conserved mecha-nism and is dispensable for homologouschromosome synapsis. Cell 94:387–398.Doniach T. 1986. Activity of the sex-determining gene tra-2 is modulated toallow spermatogenesis in the C. eleganshermaphrodite. Genetics 114:53–76.

Doyle TG, Wen C, Greenwald I. 2000.SEL-8, a nuclear protein required forLIN-12 and GLP-1 signaling in Caeno-rhabditis elegans. Proc Natl Acad SciUSA 97:7877–7881.

Eckmann CR, Kraemer B, Wickens M,Kimble J. 2002. GLD-3, a bicaudal-Chomolog that inhibits FBF to controlgermline sex determination in C. ele-gans. Dev Cell 3:697–710.

Eckmann C, Crittenden SL, Suh N, KimbleJ. 2004. GLD-3 and control of the mitosis/meiosis decision in the germline of Caeno-rhabditis elegans. Genetics 168:147.

Ellis R, Schedl T. 2006. Sex determinationin the germ line WormBook, ed. TheC. elegans research community, Worm-Book, doi/10.1895/wormbook.1.82.1, http://www.wormbook.org.

Ellis RE, Kimble J. 1995. The fog-3 geneand regulation of cell fate in the germline of Caenorhabditis elegans. Genetics139:561–577.

Fox PM, Vought VE, Hanazawa M, LeeMH, Maine E, Schedl T. 2011. Cyclin Eand CDK-2 regulate proliferative cellfate and cell cycle progression in the C.elegans germline. Development 138:2223–2234.

Francis R, Barton MK, Kimble J, SchedlT. 1995a. gld-1, a tumor suppressorgene required for oocyte development inCaenorhabditis elegans. Genetics 139:579–606.

Francis R, Maine E, Schedl T. 1995b.Analysis of the multiple roles of gld-1in germline development: interactionswith the sex determination cascade andthe glp-1 signaling pathway. Genetics139:607–630.

Freund C, Dotsch V, Nishizawa K, Rein-herz EL, Wagner G. 1999. The GYF do-main is a novel structural fold that isinvolved in lymphoid signaling throughproline-rich sequences. Nat Struct Biol6:656–660.

Fukushige T, Hawkins MG, McGhee JD.1998. The GATA-factor elt-2 is essentialfor formation of the Caenorhabditis ele-gans intestine. Dev Biol 198:286–302.

Gallegos M, Ahringer J, Crittenden S,Kimble J. 1998. Repression by the 30UTR of fem-3, a sex-determining gene,relies on a ubiquitous mog-dependentcontrol in Caenorhabditis elegans.Embo J 17:6337–6347.

Gozani O, Potashkin J, Reed R. 1998. Apotential role for U2AF-SAP 155 inter-actions in recruiting U2 snRNP to thebranch site. Mol Cell Biol 18:4752–4760.

Graham PL, Kimble J. 1993. The mog-1gene is required for the switch fromspermatogenesis to oogenesis in Caeno-rhabditis elegans. Genetics 133:919–931.

Graham PL, Schedl T, Kimble J. 1993.More mog genes that influence theswitch from spermatogenesis to

Dev

elop

men

tal D

ynam

ics

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 519

oogenesis in the hermaphrodite germline of Caenorhabditis elegans. DevGenet 14:471–484.

Hansen D, Hubbard EJ, Schedl T. 2004a.Multi-pathway control of the prolifera-tion versus meiotic development deci-sion in the Caenorhabditis elegansgermline. Dev Biol 268:342–357.

Hansen D, Schedl T. 2006. The regulatorynetwork controlling the proliferation-meiotic entry decision in the Caeno-rhabditis elegans germ line. Curr TopDev Biol 76:185–215.

Hansen D, Wilson-Berry L, Dang T,Schedl T. 2004b. Control of the prolifer-ation versus meiotic development deci-sion in the C. elegans germline throughregulation of GLD-1 protein accumula-tion. Development 131:93–104.

Hartmuth K, Urlaub H, Vornlocher HP,Will CL, Gentzel M, Wilm M, LuhrmannR. 2002. Protein composition of humanprespliceosomes isolated by a tobramy-cin affinity-selection method. Proc NatlAcad Sci USA 99:16719–16724.

Henderson ST, Gao D, Lambie EJ, KimbleJ. 1994. lag-2 may encode a signalingligand for the GLP-1 and LIN-12 recep-tors of C. elegans. Development 120:2913–2924.

Hirsh D, Oppenheim D, Klass M. 1976.Development of the reproductive systemof Caenorhabditis elegans. Dev Biol 49:200–219.Hodgkin J. 1986. Sex determination inthe nematode C. elegans: analysis oftra-3 suppressors and characterizationof fem genes. Genetics 114:15–52.

Hollins C, Zorio DA, MacMorris M, Blu-menthal T. 2005. U2AF binding selectsfor the high conservation of the C. ele-gans 30 splice site. RNA 11:248–253.

Jeong J, Verheyden JM, Kimble J. 2011.Cyclin E and Cdk2 control GLD-1, themitosis/meiosis decision, and germlinestem cells in Caenorhabditis elegans.PLoS genetics 7:e1001348.

Jones AR, Schedl T. 1995. Mutations ingld-1, a female germ cell-specific tumorsuppressor gene in Caenorhabditis ele-gans, affect a conserved domain alsofound in Src-associated protein Sam68.Genes Dev 9:1491–1504.

Jones AR, Francis R, Schedl T. 1996.GLD-1, a cytoplasmic protein essentialfor oocyte differentiation, shows stage-and sex-specific expression during Cae-norhabditis elegans germline develop-ment. Dev Biol 180:165.

Kadyk LC, Kimble J. 1998. Genetic regu-lation of entry into meiosis in Caeno-rhabditis elegans. Development 125:1803–1813.

Kent WJ, Zahler AM. 2000. Conservation,regulation, synteny, and introns in alarge-scale C. briggsae–C. elegansgenomic alignment. Genome Res 10:1115–1125.

Kerins JA, Hanazawa M, Dorsett M,Schedl T. 2010. PRP-17 and the pre-mRNA splicing pathway are preferen-tially required for the proliferation ver-sus meiotic development decision andgermline sex determination in Caeno-

rhabditis elegans. Dev Dyn 239:1555–1572.

Killian DJ, Hubbard EJ. 2005. Caeno-rhabditis elegans germline patterningrequires coordinated development of thesomatic gonadal sheath and the germline. Dev Biol 279:322–335.

Kimble J, Crittenden SL. 2007. Controlsof germline stem cells, entry into meio-sis, and the sperm/oocyte decision inCaenorhabditis elegans. Annu Rev CellDev Biol 23:405–433.

Kofler MM, Freund C. 2006. The GYF do-main. FEBS J 273:245–256.

Kofler M, Heuer K, Zech T, Freund C.2004. Recognition sequences for theGYF domain reveal a possible spliceoso-mal function of CD2BP2. J Biol Chem279:28292–28297.

Kohara Y. 2001. [Systematic analysis ofgene expression of the C. elegans ge-nome]. Tanpakushitsu Kakusan Koso46:2425–2431.

Konarska MM, Sharp PA. 1987. Interac-tions between small nuclear ribonucleo-protein particles in formation ofspliceosomes. Cell 49:763–774.

Konishi T, Uodome N, Sugimoto A. 2008.The Caenorhabditis elegans DDX-23, ahomolog of yeast splicing factor PRP28,is required for the sperm-oocyte switchand differentiation of various cell types.Dev Dyn 237:2367–2377.

Kraemer B, Crittenden S, Gallegos M,Moulder G, Barstead R, Kimble J,Wickens M. 1999. NANOS-3 and FBFproteins physically interact to controlthe sperm-oocyte switch in Caenorhab-ditis elegans. Curr Biol 9:1009–1018.

Laggerbauer B, Liu S, Makarov E, Vorn-locher HP, Makarova O, Ingelfinger D,Achsel T, Luhrmann R. 2005. Thehuman U5 snRNP 52K protein(CD2BP2) interacts with U5–102K(hPrp6), a U4/U6.U5 tri-snRNP bridg-ing protein, but dissociates upon tri-snRNP formation. RNA 11:598–608.

Lamont LB, Crittenden SL, Bernstein D,Wickens M, Kimble J. 2004. FBF-1 andFBF-2 regulate the size of the mitoticregion in the C. elegans germline. DevCell 7:697–707.

MacDonald LD, Knox A, Hansen D. 2008.Proteasomal regulation of the pro-liferation vs. meiotic entry decision inthe C. elegans germ line. Genetics 180:905–920.

MacQueen AJ, Villeneuve AM. 2001. Nu-clear reorganization and homologouschromosome pairing during meiotic pro-phase require C. elegans chk-2. GenesDev 15:1674–1687.

Mantina P, Macdonald L, Kulaga A, ZhaoL, Hansen D. 2009. A mutation in teg-4,which encodes a protein homologous tothe SAP130 pre-mRNA splicing factor,disrupts the balance between prolifera-tion and differentiation in the C. elegansgerm line. Mech Dev 126:417–429.

Mazroui R, Puoti A, Kramer A. 1999.Splicing factor SF1 from Drosophilaand Caenorhabditis: presence of an N-terminal RS domain and requirementfor viability. RNA 5:1615–1631.

Mello CC, Kramer JM, Stinchcomb D,Ambros V. 1991. Efficient gene transferin C. elegans: extrachromosomal main-tenance and integration of transformingsequences. Embo J 10:3959–3970.

Merendino L, Guth S, Bilbao D, MartinezC, Valcarcel J. 1999. Inhibition of msl-2splicing by Sex-lethal reveals interac-tion between U2AF35 and the 30 splicesite AG. Nature 402:838–841.

Michaud S, Reed R. 1991. An ATP-inde-pendent complex commits pre-mRNA tothe mammalian spliceosome assemblypathway. Genes Dev 5:2534–2546.

Mumm JS, Kopan R. 2000. Notch signal-ing: from the outside in. Dev Biol 228:151–165.

Nadarajan S, Govindan JA, McGovern M,Hubbard EJ, Greenstein D. 2009.MSP and GLP-1/Notch signaling coor-dinately regulate actomyosin-dependentcytoplasmic streaming and oocytegrowth in C. elegans. Development 136:2223–2234.

Nishizawa K, Freund C, Li J, Wagner G,Reinherz EL. 1998. Identification of aproline-binding motif regulating CD2-triggered T lymphocyte activation. ProcNatl Acad Sci USA 95:14897–14902.

Pasierbek P, Jantsch M, Melcher M,Schleiffer A, Schweizer D, Loidl J.2001. A Caenorhabditis elegans cohe-sion protein with functions in meioticchromosome pairing and disjunction.Genes Dev 15:1349–1360.

Pepper AS, Killian DJ, Hubbard EJ.2003. Genetic analysis of Caenorhabdi-tis elegans glp-1 mutants suggests re-ceptor interaction or competition.Genetics 163:115–132.

Petcherski AG, Kimble J. 2000. LAG-3 isa putative transcriptional activator inthe C. elegans Notch pathway. Nature405:364–368.

Polanowska J, Martin JS, Fisher R, ScopaT, Rae I, Boulton SJ. 2004. Tandemimmunoaffinity purification of proteincomplexes from Caenorhabditis elegans.Biotechniques 36:778–780, 782.

Prigge JR, Iverson SV, Siders AM,Schmidt EE. 2009. Interactome for aux-iliary splicing factor U2AF(65) suggestsdiverse roles. Biochim Biophys Acta1789:487–492.

Puoti A, Kimble J. 1999. The Caenorhab-ditis elegans sex determination genemog-1 encodes a member of the DEAH-Box protein family. Mol Cell Biol 19:2189–2197.

Puoti A, Kimble J. 2000. The hermaphro-dite sperm/oocyte switch requires theCaenorhabditis elegans homologs ofPRP2 and PRP22. Proc Natl Acad SciUSA 97:3276–3281.

Racher H, Hansen D. 2010. Translationalcontrol in the C. elegans hermaphroditegerm line. Genome 53:83–102.

Ruskin B, Zamore PD, Green MR. 1988.A factor, U2AF, is required for U2snRNP binding and splicing complexassembly. Cell 52:207–219.

Schedl T, Kimble J. 1988. fog-2, agerm-line-specific sex determinationgene required for hermaphrodite

Dev

elop

men

tal D

ynam

ics

520 WANG ET AL.

spermatogenesis in Caenorhabditis ele-gans. Genetics 119:43–61.

Seraphin B, Rosbash M. 1989. Identifica-tion of functional U1 snRNA-pre-mRNAcomplexes committed to spliceosome as-sembly and splicing. Cell 59:349–358.

Sulston J, Hodgkin J. 1988. Methods. In:Wood WB, editor. The nematode Caeno-rhabditis elegans. Cold Spring Harbor,NY: Cold Spring Harbor Laboratory.p 587–606.

Tax FE, Yeargers JJ, Thomas JH. 1994.Sequence of C. elegans lag-2 reveals acell-signalling domain shared withDelta and Serrate of Drosophila. Na-ture 368:150–154.

Timmons L, Fire A. 1998. Specific interfer-ence by ingested dsRNA. Nature 395:854.

Tolia NH, Joshua-Tor L. 2006. Strategiesfor protein coexpression in Escherichiacoli. Nature methods 3:55–64.

Valadkhan S, Jaladat Y. 2010. The spli-ceosomal proteome: at the heart ofthe largest cellular ribonucleoproteinmachine. Proteomics 10:4128–4141.

Wahl MC, Will CL, Luhrmann R. 2009. Thespliceosome: design principles of adynamic RNP machine. Cell 136:701–718.

Walhout AJM, Boulton SJ. 2006. Bio-chemistry and molecular biology. In:Community TCeR, editor. WormBook.[http://www.wormbook.org].

Wu S, Romfo CM, Nilsen TW, Green MR.1999. Functional recognition of the 30splice site AG by the splicing factorU2AF35. Nature 402:832–835.

Yochem J, Greenwald I. 1989. glp-1 andlin-12, genes implicated in distinct cell-cell interactions in C. elegans, encodesimilar transmembrane proteins. Cell58:553–563.

Zamore PD, Green MR. 1989. Identifica-tion, purification, and biochemical char-acterization of U2 small nuclearribonucleoprotein auxiliary factor. ProcNatl Acad Sci USA 86:9243–9247.

Zanetti S, Meola M, Bochud A, Puoti A.2011. Role of the C. elegans U2 snRNPprotein MOG-2 in sex determination,

meiosis, and splice site selection. DevBiol 354:232–241.

Zetka MC, Kawasaki I, Strome S, Muller F.1999. Synapsis and chiasma formation inCaenorhabditis elegans require HIM-3, ameiotic chromosome core component thatfunctions in chromosome segregation.Genes Dev 13:2258–2270.

Zhang H, Blumenthal T. 1996. Functionalanalysis of an intron 30 splice site inCaenorhabditis elegans. RNA 2:380–388.

Zorio DA, Blumenthal T. 1999a. Both sub-units of U2AF recognize the 30 splicesite in Caenorhabditis elegans. Nature402:835–838.

Zorio DA, Blumenthal T. 1999b. U2AF35is encoded by an essential gene clus-tered in an operon with RRM/cyclophi-lin in Caenorhabditis elegans. RNA 5:487–494.

Zorio DA, Lea K, Blumenthal T. 1997.Cloning of Caenorhabditis U2AF65: analternatively spliced RNA containing anovel exon. Mol Cell Biol 17:946–953.

Dev

elop

men

tal D

ynam

ics

TEG-1 CD2BP2 INHIBITS STEM CELL PROLIFERATION 521