BIOLOGY OF REPRODUCTION 78, 472–482 (2008)Published online before print 21 November 2007.DOI 10.1095/biolreprod.107.064774

Pre-Messenger RNA Cleavage Factor I (CFIm): Potential Role in AlternativePolyadenylation During Spermatogenesis1

Becky L. Sartini,3,5 Hang Wang,4,5 Wei Wang,5 Clarke F. Millette,6 and Daniel L. Kilpatrick2,5

Department of Molecular and Cellular Physiology,5 University of Massachusetts Medical School, Worcester,Massachusetts 01655Department of Cell and Developmental Biology and Anatomy,6 University of South Carolina School of Medicine,Columbia, South Carolina 29208

ABSTRACT

A hallmark of male germ cell gene expression is thegeneration by alternative polyadenylation of cell-specificmRNAs, many of which utilize noncanonical A(A/U)UAAA-independent polyadenylation signals. Cleavage factor I (CFIm), acomponent of the pre-mRNA cleavage and polyadenylationprotein complex, can direct A(A/U)UAAA-independent polyad-enylation site selection of somatic cell mRNAs. Here we reportthat the CFIm subunits NUDT21/CPSF5 and CPSF6 are highlyenriched in mouse male germ cells relative to somatic cells. Bothsubunits are expressed from spermatogenic cell mRNAs that areshorter than the corresponding somatic transcripts. Comple-mentary DNA sequencing and Northern blotting revealed thatthe shorter Nudt21 and Cpsf6 mRNAs are generated byalternative polyadenylation in male germ cells using proximalpoly(A) signals. Both sets of transcripts contain CFIm bindingsites within their 30-untranslated regions, suggesting autoregu-lation of CFIm subunit formation in male germ cells. CFImsubunit mRNA and protein levels exhibit distinct developmentalvariation during spermatogenesis, indicating stage-dependenttranslational and/or posttranslational regulation. CFIm bindingsites were identified near the 30 ends of numerous male germ celltranscripts utilizing A(A/U)UAAA-independent sites. Togetherthese findings suggest that CFIm complexes participate inalternative polyadenylation directed by noncanonical poly(A)signals during spermatogenesis.

alternative polyadenylation, CFIm, gamete biology, generegulation, spermatogenesis, testis

INTRODUCTION

Spermatogenesis is a dynamic cell-differentiation processthat produces a morphologically distinct haploid sperm cell.Although supported by the Sertoli cell in the seminiferoustubule, male gamete development is driven to a great degree byan intrinsic germ cell gene regulatory program that is

characterized by the expression of cell-specific transcripts thatare developmentally controlled and stage specific [1].

A central feature of male germ cell gene expression is thegeneration of cell-specific mRNAs by alternative polyadenyl-ation, many of which use noncanonical, A(A/U)UAAA-independent polyadenylation signals [2]. Alternative 30-endformation controls cell-specific and developmental expressionof numerous male germ cell-enriched mRNAs, including thosefor the transcription factors SREBF2_V1 (SREBP2GC) [3],CREMs [4], and NR6A1 (GCNF) [5]. Spermatogenic cellsappear to use alternative polyadenylation more than do somaticcells, and this happens through the selection of noncanonical,A(A/U)UAAA-independent polyadenylation signals in the 30-untranslated region (UTR) [6, 7]. Further, an increasedproportion of male germ cell transcripts have truncated 30UTRscompared to somatic cell transcripts [8], indicating thepreferential use of more proximal polyadenylation signals inthe male germ line. Thus, noncanonical poly(A) site selectionappears to be an important regulatory mechanism for malegerm cell gene expression and development.

Pre-mRNA cleavage and polyadenylation site selectioninvolves the interaction of the cleavage/polyadenylationcomplex with multiple cis-acting sequence elements locatedwithin the transcribed gene [9] and occurs cotranscriptionally[10]. Most somatic mRNAs utilize a canonical A(A/U)UAAAhexamer polyadenylation signal found upstream of the RNAcleavage site, whereas UG- and U-rich sequences are locateddownstream [11]. The cleavage/polyadenylation complexconsists of cleavage polyadenylation specificity factor (CPSF),cleavage factors Im and IIm (CFIm and CFIIm), cleavagestimulatory factor (CstF), poly (A) polymerase (PAP), and poly(A) binding protein II. The downstream U-rich (and possiblyGU-rich) sequences are bound by CstF, which, together withCFIm and CFIIm, is specifically required for the pre-mRNAcleavage step [11]. CPSF binds to the hexamer sequence andinteracts directly with PAP, and these two proteins are requiredfor both the cleavage and polyadenylation steps [12].

Additional upstream elements also function as enhancers of30-end processing [9] and are evolutionarily conserved [13].These include ‘‘U-rich’’ sequences as well as UGUANelements [14]. Several trans-factors are present on upstreamU-rich elements as part of a large complex composed of RNAsplicing factors and components of the 30-end processingfactors Cstf and CPSF [15]. The CPSF subunit FIP1L1 directlybinds to upstream U-rich sites [16]. The UGUA motif canoccur upstream as well as downstream of the RNA processingsite often as multiple copies, and was identified as a bindingsite for cleavage factor CFIm using SELEX analysis [17]. GC-rich sequences tend to occur just 30 of the UGUA element,which may contribute to CFIm binding by creating a favorablesecondary structure [17].

1Supported by Public Service grant R01 HD45723. Core resourcessupported by the Diabetes Endocrinology Research Center grantDK32520 were also used.2Correspondence: FAX: 508 856 5997;e-mail: daniel.kilpatrick@umassmed.edu3Current address: Department of Fisheries, Animal and VeterinaryScience, University of Rhode Island, Kingston, RI 02881.4Current address: Department of Urology, The Johns Hopkins Univer-sity School of Medicine, Baltimore, MD 21287.

Received: 3 August 2007.First decision: 28 August 2007.Accepted: 13 November 2007.� 2008 by the Society for the Study of Reproduction, Inc.ISSN: 0006-3363. http://www.biolreprod.org

472

CFIm facilitates assembly of the 30 processing complex andenhances both the rate and efficiency of poly(A) site cleavagein vitro [18]. This factor binds to pre-mRNA at a very earlystep and facilitates the binding of other 30-end processingfactors via direct interaction with and recruitment of CPSF, butnot CstF or PAP [18, 19]. Multiple UGUAN sites flanking thehexamer site contribute to CFIm binding based on gel shiftanalysis [14]. Further, these sites promote both pre-mRNAcleavage and poly(A) addition using in vitro reconstitutionassays [14, 17]. CFIm is a heterodimer consisting of a 25-kDaprotein (NUDT21/CPSF5) and a larger subunit. A complexconsisting of NUDT21 and a 68-kDa subunit protein (CPSF6)are sufficient for RNA binding [20], and this heterodimerreconstitutes CFIm processing activity [18]. Additional largersubunits (59 and 72 kDa) also have been identified [18, 19],but their significance remains uncertain.

Several components of the pre-mRNA cleavage/polyade-nylation complex are overexpressed or exhibit unique isoformsin the testis or spermatogenic cells relative to their somaticcounterparts, including CstF-64 and PAP [21–23]. Theseunique expression patterns suggested a potential role for thesefactors in germ cell-specific alternative polyadenylation.However, the machinery responsible for noncanonical poly(A)site selection during spermatogenesis remains uncertain.

Although the majority of somatic transcripts utilizecanonical poly(A) sites, noncanonical site utilization alsooccurs in somatic cells [9]. It was recently found that the CFImcomplex composed of NUDT21/CPSF6, in combination withthe CPSF subunit FIP1L1 and PAP, is sufficient to directalternative, A(A/U)UAAA-independent poly(A) site selectionand polyadenylation of somatic mRNAs [23]. Further, theCFIm complex is present along the entirety of activetranscription units, consistent with a cotranscriptional mecha-nism of poly(A) site selection [14]. These findings suggestedthat CFIm is a primary determinant of noncanonical poly(A)site selection. In the present study, we demonstrate that CFImsubunits are expressed at very high levels throughout spermdevelopment from germ cell-enriched, truncated mRNAsformed by alternative polyadenylation. The implications ofthese results for the regulation of cell-specific polyadenylationduring spermatogenesis are discussed.

MATERIALS AND METHODS

Tissues and Cells

Testis and somatic tissues were obtained from adult CD-1 male mice(Charles River, Wilmington, MA) following approved guidelines from theUniversity of Massachusetts Institutional Animal Care and Use Committee.Enriched adult spermatogenic cells were prepared by successive proteasedigestions of decapsulated mouse testes (0.5 mg/ml collagenase and 0.005%trypsin at 338C) as previously described [3]. Individual mouse spermatogenicstages were purified from adult and immature testes using unit gravitysedimentation on gradients of BSA, as previously described [24, 25].

RNA Isolation and Analysis

Total RNA was isolated from mouse tissues and cells using TRI reagent(Sigma, St. Louis, MO) according to the manufacturer’s directions. Humantissue total RNAs were purchased from Clontech (Mountain View, CA). Theintegrity of RNA was visually assessed by the ratio of 28S and 18S ribosomalRNA after ethidium bromide staining of formaldehyde gels and spectropho-tometric analysis.

For Northern blot analysis, total RNAs from mouse (20 lg) and human (4lg) were separated on 10% formaldehyde agarose gels and transferred to GeneScreen Plus membranes (NEN Life Science, Boston, MA). Gene-specificprobes were labeled with [a-32P] dCTP by random priming using the Prime-a-Gene Labeling system (Promega, Madison, WI). Membranes were prehybrid-ized at 558C for 4 h followed by hybridization with labeled probe for 24 h.Prehybridization and hybridization solutions included 53 SSC (0.15 M sodium

chloride plus 0.015 M sodium citrate), 50% formamide, 12.5 mM sodiumphosphate (pH 6.6), 1% SDS, and 103 Denhardt solution (1% bovine serumalbumin, Ficoll 400 [Sigma], 1% polyvinylpyrrolidine [Sigma]. Afterhybridization, membranes were washed three times at room temperature in13 SSC, 0.1% SDS then washed three times for 30 min each in 0.13 SSC,0.1% SDS at 558C. Bound probe was visualized by autoradiography at�808C.

RT-PCR and 3’RACE

Full-length cDNA sequences were generated using a combination ofoverlapping 30-RACE and RT-PCR. Complementary DNAs were generatedfrom mouse testis total RNA by RT reaction as previously described [26].Sequences extending into the 50 and 30UTRs were generated using gene-specific and nested primers with the Gene Racer kit (Invitrogen, Carlsbad, CA).Refer to Figures 2A and 4A for specific primer sequences used to generatemouse testis sequences for Nudt21 and Cpsf6. Other primer sequences areavailable upon request. PCR products were gel purified, subcloned into pGEM-T easy vector (Promega, Madison, WI), and sequenced in both directions.

Western Blots

Proteins were isolated from tissues and cells using whole-cell extractionbuffer (20 mM HEPES [pH 7.9], 0.4 M NaCl, 1 mM EDTA, 10 mM NaF, 25%glycerol, 10 mM dithiothreitol, 1 mM PMSF, 13 protease inhibitor cocktail,and 0.5% NP-40). Nuclear extracts were prepared by high-salt extraction ofisolated nuclei [27]. Protein concentration was determined by Coomassieprotein assay (Pierce, Woburn, MA). Proteins were separated on 12% SDS-PAGE gels and transferred to Immobilon-P PVDF membranes (Millipore,Billerica, MA). After blocking with 1% nonfat milk for 1 h at roomtemperature, the blots were incubated with antibody raised against NUDT21/CPSF5 (1:1000; #BC001403; ProteinTech Group, Chicago, IL) or CPSF6(1:1000; Abnova, Taipei City, Taiwan) overnight at 48C. The next day, blotswere washed three times with PBS-Tween-20 (0.05%) and incubated withrespective secondary antibody for 1 h, and bound antibody was visualized withWestern Lightning (Perkin Elmer, Wellesley, MA).

Immunolocalization

Bouin fixed mouse testes were subjected to a deparaffinization, rehydration,and Proteinase K antigen retrieval process. Sections were blocked in 2.5%normal goat serum for 1 h at room temperature followed by three 10-minwashes with PBS. Mouse anti-CPSF6 antibody (1:100) was applied to testistissue sections and incubated overnight at room temperature. Negative controltestis sections were incubated in parallel with PBS alone. Sections were thenwashed three times (10 min each) and then incubated with a goat anti-mouseCy3 secondary antibody (1:1000) for 1 h at room temperature followed by threeadditional 10-min washes with PBS. Mounting media was added to the sectionsprior to visualization under epifluorescence.

Chromatin Immunoprecipitation

Isolated adult mouse germ cells were cross-linked in 1% formaldehyde for10 min at 378C. Cross-linking was stopped by adding glycine (125 mM), andthe cells were collected and lysed in SDS buffer (1% SDS, 10 mM EDTA, 50mM Tris-HCl [pH 8.1], 1mM PMSF and 13 protease inhibitor cocktail mix).Chromatin was sonicated to an average length of ;600 bp, and the anti-NUDT21 antibody was added to the lysate. After overnight incubation at 48C,immune complexes were collected with protein A-agarose (Upstate, Charlottes-ville, VA). DNA was extracted and used as a template for the PCR assay ofgenomic sequences contained within the 30UTRs for mouse Srebf2_v1 (sense,ATGTCTCACCCTCAGAATCTCTCAT; antisense, CCTCTGCCAGGCTG-GGTTACAGGCA), Srebf2 encoding the SREBF2 precursor protein (sense,GGTCTCCAGCATGACTTGTTTGC; antisense, CTCTGCTGCCTGC-CCTGAAA), Nr6a1 (sense, GAGTGAGCAAAGACACCCTGAT; antisense,CACTCCTTCACCGTACTTGTCTT). Mouse Gabra6 genomic sequences(sense, TTCCAGATTTCCTCACAGCC; antisense, TGCAGATACC-CAGCCTCC) were assayed as a negative control.

RESULTS

Abundant Amounts of a Novel Nudt21 Transcript in MouseMale Germ Cells

Based on recent evidence for the importance of CFImcomplexes in noncanonical poly(A) site selection in somatic

CFIm SUBUNITS ARE ENRICHED IN MALE GERM CELLS 473

cells [14], we examined the expression of its subunit genes inmouse testis and male germ cells. Northern blot analysis usinga cDNA probe to the coding region of Nudt21 (product A; Fig.1A) detected an abundant 1.1-kb transcript in mouse testis andenriched adult germ cells that either was not observed or wasbarely detectable in mouse kidney, liver, brain, and heart (Fig.1B). This male germ cell-enriched Nudt21 transcript was muchshorter than the somatic form amplified from mouse brain(;4.4 kb; GenBank Accession No. BC090834). A larger;4.5-kb transcript was detected in several mouse somatictissues only after long film exposure (data not shown),indicating that Nudt21 mRNA is preferentially elevated inmouse testis and male germ cells. A faint ;2.2-kb band wasalso variably detected in multiple tissues (Fig. 1B).

We used products generated by RT-PCR and 30-RACE (A–D; Fig. 1A) to determine the nucleotide sequences encompass-ing the coding region, 30UTR, and a portion of the 50UTR ofthe mouse germ cell Nudt21 RNA. Because the A fragmentused as a Northern probe contained both somatic Nudt21 exon1 and exon 2 sequences, we also probed mouse total RNA withRT-PCR product B, which encompasses only exon 1 (Fig. 1A).This confirmed that the spermatogenic Nudt21 RNA containedsomatic exon 1 sequences (Fig. 1C). The overall length of thecDNA (1078 bp) was very similar to that of the male germ cellNudt21 transcript (1.1 kb; Fig. 2A). The predicted translationproduct for germ cell NUDT21 is 100% similar to that for

mouse brain (Accession No. BC090834) and to a previouslyreported mouse testis cDNA that lacks a complete 30UTRsequence (Accession No. AK160147; Fig. 2B). Mouse testisNUDT21 also is highly similar to a predicted human NUDT21isoform from retina (Accession No. CAD97606) having only asingle divergent amino acid residue (hNUDT21–1; Fig. 2B).An alternatively spliced human NUDT21 isoform isolated fromskin (Accession No. CAI46057) also has been identified that,except for the final 8 residues, lacks the entire C-terminal halfof mouse NUDT21 and human NUDT21–1 (hNUDT21–2; Fig.2B).

The mouse somatic and germ cell Nudt21 cDNAs differ inthe 30UTR, which is much shorter in the germ cell form (Fig.1A). This structural difference was confirmed in Northern blotsusing a probe specific to the somatic Nudt21 30UTR (probe E;Fig. 1A). The 1.1-kb Nudt21 transcript was not detected inmouse testis, but a less abundant, ;4.5-kb transcript wasidentified in brain and other tissues after a long exposure (Fig.1D). This larger transcript is highly similar in size to thepreviously reported mouse brain Nudt21 cDNA (see above).These data suggest that the germ cell-enriched Nudt21 RNA isgenerated by alternative polyadenylation, although more rapidturnover of the longer transcript form in mouse testis also maycontribute to its reduced relative abundance. The usage of testis-specific transcription start sites as well cannot be ruled out.

FIG. 1. A novel Nudt21 RNA is highly enriched in mouse testis and male germ cells. A) Schematic of male germ cell and somatic Nudt21 transcripts formouse, showing that structural differences occur in their respective 30UTRs. DNA products generated by RT-PCR (A, B, C, E) and 3’RACE (D) from mousetestis or kidney total RNA are shown below the transcripts. Arrows indicate the position of the exon 1/intron 1 junction and translation initiation sites. B)Northern blot of total RNA isolated from mouse tissues and enriched adult male germ cells using RT-PCR product A as a probe (see A). Lanes: mouse testis(T), enriched adult male germ cells (gc), kidney (K), liver (L), brain (B), and heart (H). Ethidium bromide staining of RNA loading is shown below each lane.C) Northern analysis of mouse testis and kidney total RNA using a probe specific to Nudt21 exon 1 (product B, see A). Ethidium bromide staining ofribosomal RNAs is shown below. D) Hybridization of total RNA from mouse tissues with RT-PCR product E specific to the 30UTR of mouse somaticNudt21 mRNA (see A). The blot used in the right panel of B was reprobed for somatic-specific 30UTR sequences and detected a ;4.5-kb mRNA in brainand other tissues, but not the 1.1-kb testis-enriched transcript.

474 SARTINI ET AL.

A Male Germ Cell-Enriched CPSF6 RNA Containing a

Truncated 30UTR

We next examined the transcript abundance for the otherCFIm subunit, Cpsf6, in several mouse tissues. 30-RACE wasused to generate a 2-kb fragment (product F; Fig. 3A) forNorthern blotting. Similar to Nudt21, RNAs for Cpsf6 wereelevated in mouse testis and male germ cells relative tosomatic tissues (Fig. 3B). A predominant 2.1-kb RNA wasdetected in testis and enriched germ cell total RNA (Fig. 3B),with minor species of ;4 and 1.5 kb variably detected. The2.1-kb transcript was shorter than the previously reportedCpsf6 somatic cDNA amplified from brain tissue (;3 kb;GenBank Accession No. BC068133) and mammary tumor (2.5kb; Accession No. BC031189). Cpsf6 transcripts weregenerally of low abundance in total RNA of most somatictissues, with faint bands detected in mouse kidney (;2.5–3 kb)and heart (;1.8 kb). A very weak, ;4-kb band was alsodetected in multiple tissues (Fig. 3B). A 126-bp probe to thefirst exon of Cpsf6 was generated by RT-PCR of mouse testistotal RNA (probe G; Fig. 3A). Northern analysis confirmedthat the major 2.1-kb Cpsf6 transcript contained exon 1sequences (Fig. 3C). Poly(A) mRNA was used for this

experiment due to a high level of nonspecific hybridizationto ribosomal RNA by this probe.

RACE/RT-PCR products F, G, H, and I (Fig. 3A) werecharacterized to obtain mouse testis Cpsf6 cDNA sequences(Fig. 4A). The overall length of this composite cDNA (2145kb) was very close to that of the major mouse germ cell-enriched transcript. Comparison of the mouse testis Cpsf6sequence to several mouse somatic cDNA sequences inGenbank (e.g., Accession Nos. BC068133 and BC031189)confirmed protein sequence identity and apparent lack ofalternative splicing (data not shown). As for Nudt21, the mousegerm cell and somatic Cpsf6 cDNAs differed in the length oftheir 30UTRs (Fig. 3A). Consistent with this, a probe specific tothe 30UTR of the somatic Cpsf6 transcript (product J; Fig. 3A)did not display detectable hybridization to mouse testis RNA(data not shown).

The mouse testis CPSF6 coding region showed differenceswith human somatic CPSF6 (Fig. 4B). Two alternativelyspliced human kidney Cpsf6 cDNAs have been reported: onepossessing additional amino acid residues not in the predictedmouse protein (hCPSF6–1; Accession No. AAH00714) andanother (hCPSF6–2; Accession No. AAH05000) that lacks acontiguous stretch of internal residues present in mouse CPSF6and hCPSF6–1 (Fig. 4B).

FIG. 2. Nucleotide and coding sequences for mouse testis Nudt21. A) Composite cDNA sequence for mouse testis Nudt21. Primers used to amplifyvarious RT-PCR and 3’RACE products are underlined. The translational initiation and stop codons (bold) as well as putative canonical and noncanonicalpolyadenylation signals (dashed boxes) and upstream CFIm binding sites (solid line boxes) in the 30UTR are indicated. B) Predicted coding sequences forNUDT21 from mouse testis and somatic human NUDT21 isoforms (hNUDT21–1, 2). See text for GenBank Accession numbers. Asterisks indicate identitywith the predicted mouse testis amino acid sequence, while residues absent in the human splice variant hNUDT21–2 are shown with dashes.

CFIm SUBUNITS ARE ENRICHED IN MALE GERM CELLS 475

Cleavage/Polyadenylation Sequences Within the Germ CellCFIm Transcripts

The preceding results showed that both Nudt21 and Cpsf6RNAs in mouse spermatogenic cells were generated byalternative 30-end formation. Analysis of their 30UTRs revealedthe presence of both canonical and noncanonical polyadenyl-ation sequences upstream of their respective poly(A) additionsites (Figs. 2A and 4A). Further, both 30UTRs contain CFImbinding sequences (UGUAN) upstream of these sites (Figs. 2Aand 4A). This suggested that 3’-end formation for both Nudt21and Cpsf6 RNAs may be subject to auto-regulation via CFImcomplexes in male germ cells.

CFIm Subunit Proteins Are Highly Expressed in MouseTestis and Spermatogenic Cells

Western blotting was used to determine the expressionlevels of CFIm proteins in mouse tissues and germ cells.Similar to their transcripts, NUDT21 and CPSF6 proteins werehighly elevated in extracts of mouse testis when compared tothose of kidney, a representative somatic tissue (Fig. 5A). Theapparent sizes of the testis proteins were highly similar to thosepredicted from the Nudt21 and Cpsf6 testis cDNAs (;25 and68 kDa, respectively). Further, both CFIm subunits are presentin the nuclei of enriched adult mouse germ cells, consistentwith a role for CFIm in pre-mRNA cleavage/polyadenylationevents in male germ cells (Fig. 5B). Thus, both the mRNAs

and proteins for NUDT21 and CPSF6 are highly enriched inmouse spermatogenic cells.

CFIm Subunit mRNAs and Proteins Are DifferentiallyRegulated During Spermatogenesis

Northern blot analysis of isolated germ cell populationsdemonstrated that Nudt21 and Cpsf6 mRNA levels increasedduring meiotic and spermatid stages of spermatogenesis (Fig.6A). Both mRNAs were lower in spermatogonia, increased inpachytene spermatocytes, and remained elevated in the roundspermatid population. Thus, CFIm subunit gene expression isdevelopmentally up-regulated in meiosis and early haploidstages of spermatogenesis.

In contrast to their mRNAs, the levels of CFIm subunitproteins were fairly constant during different stages ofspermatogenesis, with no large increases in meiotic cellsrelative to spermatogonia (Fig. 6B). Thus, CFIm subunit geneand protein expression show discoordinate regulation duringmouse spermatogenesis. Immunohistofluorescence furtherconfirmed that CPSF6 protein was present in spermatogonia,spermatocytes, and spermatids, and appeared particularlyintense in early elongating spermatids (Stage XI) in fixedmouse testis sections (Fig. 7). CPSF6 antibody staining was notpresent in later elongating spermatids (Stages VI-VII).Immunostaining with the NUDT21 antibody was non-specificand not informative (data not shown).

FIG. 3. Analysis of testis-enriched Cpsf6 transcripts in the mouse. A) Schematic illustration of Cpsf6 transcripts in mouse testis/male germ cells andsomatic tissues. Products generated by RT-PCR (G, H, I, J) and 3’RACE (F) to identify and characterize the germ cell and somatic RNAs are shown beloweach one. Arrows indicate the junction between exons 1 and 2. B) Northern blots detected abundant amounts of a 2.1-kb Cpsf6 RNA in mouse testisrelative to somatic tissues. The blot was probed using the 3’RACE product F (see A). Lanes: mouse testis (T), liver (L), kidney (K), heart (H), lung (Lu), andenriched adult spermatogenic cells (gc). All samples were hybridized together on the same blot and exposed to film following washing for the same lengthof time. Ethidium bromide staining of RNA loading is shown below. C) Northern analysis of poly(A) RNA from mouse testis using a probe to Cpsf6 exon 1sequences (RT-PCR product G; see A).

476 SARTINI ET AL.

CFIm Subunit Gene Expression in Human Testis

We also addressed whether Nudt21 and Cpsf6 geneexpression is elevated in the human testis. Northern blots forNudt21 revealed a common ;2.2-kb transcript in all humantissues (Fig. 8). However, a Nudt21 mRNA that was similar insize to that for mouse testis (;1.1 kb) was detected within thehuman testis, although its abundance was not highly elevatedrelative to the common transcript form, as occurs in the mouse.Cpsf6 transcripts were greatly elevated in human testis relativeto somatic tissues. Multiple Cpsf6 RNAs were detected, with

major bands of 1.5 and ;4 kb and a less abundant ;2.1-kbspecies (Fig. 8). Thus, shorter transcript forms for CFImsubunits also are uniquely generated in human testis.

CFIm Binding Sites Are Present Within the 30UTRs ofNumerous Spermatogenic Cell mRNAs

Elevated expression of CFIm subunit proteins in male germcells suggested their potential involvement in A(A/U)UAAA-independent poly(A) site selection during spermatogenesis. Wetherefore examined the 30UTR sequences of numerous male

FIG. 4. Nucleotide and predicted coding sequences for mouse testis Cpsf6 cDNA. A) Composite sequence for mouse testis Cpsf6 cDNA. Primers usedfor RT-PCR and 3’RACE are underlined. The initiator and stop codons are shown in bold lettering, and CFIm binding sites and putative poly(A) signals areindicated by solid and dashed boxes, respectively. B) Predicted coding sequences for Cpsf6 cDNAs from mouse testis (testis CPSF6) and human somatictissues (hCPSF6–1, 2). Asterisks indicate identity with the mouse testis amino acid residues, while missing sequences are shown by dashes.

CFIm SUBUNITS ARE ENRICHED IN MALE GERM CELLS 477

germ cell transcripts harboring noncanonical poly(A) signalsfor the presence of CFIm binding sites. In all cases, UGUANsites were identified either upstream or downstream of thesesignals [28–42] (Table 1). Thus, CFIm binding sites are acommon feature of mRNAs undergoing alternative, A(A/U)UAAA-independent polyadenylation during spermatogene-sis.

We also examined whether CFIm complexes are present ongenes that give rise to alternatively polyadenylated mRNAsduring spermatogenesis using chromatin immunoprecipitation(ChIP) assays for NUDT21. The mouse Nr6a1 and Srebf2genes generate predominant, shorter transcripts by alternativepolyadenylation during later stages of spermatogenesis [26,

43]. The 30UTRs for mouse germ cell Nr6a1 and rat andhuman Srebf2_v1 transcripts contain A(A/U)UAAA-indepen-dent poly(A) signals (1), whereas the poly(A) signal for mousegerm cell Srebf2_v1 mRNA is undetermined [3]. ChIP analysisof adult mouse spermatogenic cells using NUDT21 antibodies

FIG. 6. Distinct developmental variation of CFIm subunit mRNAs andproteins during male germ cell development. A) Total RNA from purifiedtype A spermatogonia (lane A), type B spermatogonia (lane B), prepubertalpachytene spermatocytes (lane PP), pachytene spermatocytes (lane PS),round spermatids (lane RS), and residual bodies (RB) were hybridized withNudt21 probe A and Cpsf6 probe F. The same blot was probed in eachcase. Ethidium bromide staining of RNA loading is shown below. B)NUDT21 and CPSF6 protein concentrations are similar in mitotic,meiotic, and spermiogenic germ cell populations. B, B type spermatogo-nia; PS, pachytene spermatocytes; RS, round spermatids; gc, enrichedadult mouse male germ cells.

TABLE 1. CFIm binding sites (UGUA) within male germ cell-enriched transcripts having noncanonical polyadenylation signals.

Gene 30UTR Sequencea Reference

Bzw1 UGUAGAAUGG...50 bp...AAUUUUGAAACAACAUCCUCAGUAAAGCAAACAGGAGUUGUA [24]B4galT1 UGUAUUUUUA...45 bp...GGAAAAAAAAUAAU [25]Crem UGUACAGUUGCUUUUGAAUGCAAUACAAAUAUAU [4]

UAAAUAGAACCAUGUUGGUUGCUCUUUGUAc-abl UGUACCUGCACCUUUGAUGCUUACAAACUGCCCCGAGA [26]Cyp51 GCUGCAGCUUGGCAGAGAAUGAAGCUUUGACACAGCUUUCAUACUGUACUG [27]Ccna2 UGUACAGUGGGGAGUAGGAAAAGAUUUUAUAAAA [28]

UGUAAAAAUUGAUGUUUUCUAUAAANr6a1 UGUAUGACC..70BP..AUAGCCACAGACUUCUAAGUAAAGAAG [5]His.H3 UGUAUAUCGAUAAGCUUUUAUAAA [29]Klf4 GCAAUACACACGUAAAGAUCACCUUGUA [30]Papp5 UGUAAGAAAAAUGACCAUUUUAUAAA [31]PhLP UGUAUUGUCUCUUGCAAAUAUGAA [32]Ppp3r2 UGUAACAACUAGCAAA [33]Pou3f1 UUUAAAUCGCUGUA [34]Srebf2_v1 UGUAAAUACGAUUGUUU..260bp..UUACAAAAGUGAGCAAAAGCAAACCUUCCCUUUAAA [23]Sry UGUAAAGAAUUCAGACUUUCCAAUAUAA [35]Translin UGUAGUU..85BP..AAACAAA [36]

ACAUUUAAAAUGUAUAGUUUAGGGUAGUCAUGUATpi2 UGUAAAAGAACCGGUAAUGAAA [37]Zfp35 UGUAGUAUAAAAAAGGCUAUGAGAAA [38]

a Bold text indicates the polyadenylation sequences, underlined text indicates the CFIm binding sites (UGUA), and italic text indicates thepolyadenylation signal hexamer sequences.

FIG. 5. CFIm subunit proteins are elevated in mouse testis and malegerm cells. A) Western analysis of NUDT21 (left panel) and CPSF6 (rightpanel) proteins in whole testis (T) and kidney (K). Equivalency of proteinloading was confirmed by Ponceau-S staining (not shown). B) NUDT21(left) and CPSF6 (right) proteins in nuclear extracts of enriched adultmouse spermatogenic cells (gc).

478 SARTINI ET AL.

detected abundant levels of both Nr6a1 and Srebf2 genomicsequences, whereas those for Gabra6, a gene not transcribed inmouse germ cells, were undetectable (Fig. 9A). Thus, CFImsubunits are resident on genes that are alternatively polyade-nylated at noncanonical poly(A) sites within spermatogenic cellchromatin.

Interestingly, sequences spanning the distal 30UTR for theSrebf2 mRNA encoding the SREBF2 precursor protein werealso abundant in adult mouse germ cell chromatin precipitatedby NUDT21 antibody (Fig. 9B). Further, the ratios of therespective proximal Srebf2_v1 and distal Srebf2 3 0-UTRgenomic sequences were very similar in NUDT21-cross-linkedchromatin from mouse germ cells and kidney (Fig. 9B).Previous work has shown that CFIm complexes are presentthroughout the entirety of a transcribed gene [14], presumablyas part of a large transcription/RNA processing complex.NUDT21 ChIP data therefore suggest that elevated accumu-lation of Srebf2_v1 transcripts in adult male germ cells is due topreferential pre-mRNA processing and/or mRNA stabilization,not enhanced early transcriptional termination. Similar circum-stances have been reported for the formation of alternativelypolyadenylated mouse Dhfr transcripts [44].

DISCUSSION

The discovery of a large number of male germ cell-enrichedtranscripts possessing noncanonical poly(A) signals indicatedthat A(A/U)UAAA-independent polyadenylation was an im-portant regulatory mechanism during spermatogenesis [2, 21].In support of this, alternative polyadenylation controls theformation of novel transcription factor isoforms such asSrebf_v1 in male germ cells [3]. The basis for alternative,noncanonical poly(A) site selection remains generally unclear.Recent studies have indicated that CFIm complexes perform animportant role in this process in somatic cells. In particular,NUDT21 and CPSF6 protein subunits promote A(A/U)UAAA-independent polyadenylation in conjunction with CPSF [14]. Inaddition, down-regulation of Nudt21 expression in HeLa cellsled to changes in alternative poly(A) site utilization for severalsomatic mRNAs [45]. As shown here, male germ cells contain

elevated amounts of both NUDT21 and CPSF6 proteins, andgerm cell-enriched mRNAs utilizing noncanonical poly(A)signals contain consensus CFIm binding sites within their30UTRs. This suggests that CFIm may be important for A(A/U)UAAA-independent polyadenylation site selection in malegerm cells.

The male germ cell mRNAs for Nudt21 and Cpsf6 areshorter than the previously described somatic isoforms,reflecting the utilization of proximal alternative polyadenyla-tion sites in each case. Sequence analysis did not detectpotential canonical or noncanonical polyadenylation signals

FIG. 8. CFIm transcripts are enriched in human testis. Northern blots oftotal RNA from different human tissues were probed for Nudt21 (upperpanel; probe A) and Cpsf6 (lower panel; probe F) mRNAs. Ethidiumbromide staining is shown below for each blot. Lanes: human liver (L),pancreas (P), heart (H), lung (Lu), muscle (M), kidney (K), spleen (S), testis(T), and small intestine (SI).

FIG. 7. Immunolocalization of CPSF6 inmultiple germ cell stages of fixed adultmouse testis sections. A–C) Anti-CPSF6antibody. D) No primary antibody. Cellslabeled for CPSF6 are indicated by arrows:Spg, spermatogonium; Spc, spermatocyte;Spt, spermatid; Rspt, round spermatid.Roman numerals V, VI-VII, and XI indicateseminiferous tubule stages containing dis-tinct mouse spermatogenic cell types.Original magnification is 360 (C) and 340(A, B, D).

CFIm SUBUNITS ARE ENRICHED IN MALE GERM CELLS 479

further upstream of the respective sites for either germ cellmRNA. This suggests that during spermatogenesis, the firstsuitable polyadenylation signal for these transcripts is utilized.Further, the presence of CFIm consensus binding sitesupstream of the Nudt21 and Cpsf6 mRNA poly(A) sitessuggests that alternative polyadenylation during spermatogen-esis is driven in part by autoregulation of cleavage/polyade-nylation factor mRNA 30-end formation.

Nudt21 and Cpsf6 mRNAs were substantially elevated inmeiotic and spermatid stages of spermatogenesis compared toearlier spermatogonial stages, indicating that formation of thesetranscripts is developmentally regulated during spermatogen-esis. In contrast to this, the levels of CFIm subunit proteinsremained relatively constant in spermatogonia, pachytenespermatocytes, and round spermatids. Elevated ratios ofmRNA to protein abundance in meiotic and postmeioticspermatogenic cells have been reported for a large number ofmale germ cell gene products [46], including the transcriptionfactors SP1 and TBP [47, 48]. For CFIm mRNAs, we feel thislikely reflects inefficient translation, which is generallyobserved for mRNAs expressed in later spermatogenic stages[46]. Altered protein turnover may also contribute todiscoordinate regulation in these cell types.

The significance of cell-specific alternative polyadenylationfor spermatogenesis is not fully understood. Alternative 30-endformation has been hypothesized to regulate protein codingcapacity, localization, translational efficiency, and stability[49]. For example, alternative transcripts can give rise to germcell-specific proteins having unique structures and apparentfunctions, as for SREBF2_v1 and CREMs [3, 50]. Theseregulatory mechanisms may also provide a means forcontrolling the expression of specific genes in unique cell- orstage-dependent patterns [51].

As for Nudt21 and Cpsf6 mRNAs, alternative polyadenyl-ation in male germ cells tends to give rise to smaller transcriptswith shortened 30UTRs via usage of proximal, frequentlynoncanonical poly(A) signals [8]. Sequences within 30UTRshave been implicated in the regulation of both mRNA turnover[52–54] and translation [55] in spermatogenic cells. MessengerRNAs bearing short 30UTRs may exhibit increased mRNAstability relative to their longer counterparts because of theabsence of destabilization elements, thus contributing to theirelevated accumulation during spermatogenesis. In turn,increased mRNA stability may facilitate continued proteinsynthesis during later stages of spermiogenesis followingtermination of gene transcription, as suggested for Pgk2transcripts [54]. Further, it was recently shown that a shorter,testis-specific Bzw1 mRNA containing a truncated 30UTR istranslated with reduced efficiency relative to its longer

isoforms [28]. Thus, selection of proximal, noncanonicalpoly(A) signals in male germ cells may lead to alternativemRNAs that are in some cases more stable and/or more poorlytranslated.

Elevated expression and novel isoforms of pre-mRNAcleavage/polyadenylation trans-acting factors both occur dur-ing spermatogenesis. Mouse male germ cells express very highamounts of Cstf2 as well as sCstf2, a novel isoform of thecleavage stimulation factor that has been implicated in germcell-specific polyadenylation [21]. Germ cell-enriched expres-sion of a testis-specific form of PAP [23], as well as other 30-end processing factors [8], has also been reported. In contrast toour results, the latter study [8] reported that Cpsf6 geneexpression was down-regulated as mouse spermatogenesisproceeds based on analysis of microarray results. Thisdivergent finding may reflect lack of detection of the truncatedgerm cell-enriched Cpsf6 mRNA within the original arraychips, which can be sensitive to alternative poly(A) site usage[56]. Further, our findings show that human testis also uniquelyexpresses shorter mRNAs for both NUDT21 and CPSF6,suggesting that the general mechanisms controlling theirformation are at least partially conserved in humans. Onedifference between the mouse and human was that these shortermRNAs were not highly elevated relative to the longertranscript forms in human testis, although both CPSF6 mRNAswere highly elevated relative to those in somatic tissues. Thismay reflect distinct requirements for elevated mRNA abun-dance between these two species during spermatogenesis.

The significance of novel 30-end processing factor expres-sion for alternative polyadenylation in male germ cells remainsto be clarified. Presumably, this reflects collaboration amongdifferent components of the cleavage/polyadenylation complexwith cis-elements in germ cell pre-mRNAs. It was recentlyfound that standard 3 0-end processing sites occur lessfrequently for transcripts expressed in meiotic and spermio-genic germ cells [8], which may favor less stringent recognitionof polyadenylation signals. Increased concentrations of cleav-age/polyadenylation factors in combination with this cisorganization may foster the utilization of noncanonicalpolyadenylation sites in later germ cell stages, which otherwisetend to be less efficiently processed [57]. Because pre-mRNAcleavage and transcription are coupled [10], scanning for thefirst suitable noncanonical polyadenylation site may lead topreferential generation of truncated 30UTRs in male germ cells.Consistent with this, proximal poly(A) signals tend to benoncanonical in contrast to distal signals [57].

The present findings suggest that elevated levels of CFImcomplexes coupled with the presence of neighboring UGUANtarget sequences may promote the usage of proximal and less

FIG. 9. ChIP analysis of NUDT21 in adultmouse germ cells and kidney. Formalde-hyde-cross-linked chromatin was precipi-tated with antibodies to NUDT21 (Anti-NUDT21) or using beads alone withoutprimary antibody (No Ab Ctl). A) Followingreversal of crosslinks, DNA precipitatedfrom adult male germ cells was assayed formouse Srebf2, Nr6a1, and Gabra6 genomicsequences by PCR. Specific bands generat-ed by the respective PCR reactions areshown in each case. Aliquots of input DNA(Input) were assayed as positive controls. B)NUDT21 ChIP analysis of proximal(Srebf2_v1) and distal (Srebf2) 30UTRs forthe mouse Srebf2 gene in mouse sper-matogenic cell (Germ cells) and kidneychromatin.

480 SARTINI ET AL.

favorable A(A/U)UAAA-independent poly(A) signals. There-fore, NUDT21 and CPSF6 may be integral participants in malegerm cell-specific alternative polyadenylation. In this context,it was recently found that the interaction of components of theRNA splicing and 30-end processing complex with upstreamU-rich elements promotes 30-end formation at noncanonicalpoly(A) sites [15, 58]. Interestingly, CFIm interacts with thesplicing factor U2AF2 [59] as well as with the CPSF subunitFIP1L1 [14], both of which are present at these upstream U-rich elements [14, 15]. Therefore, it is possible that CFIm andfactors interacting with adjacent upstream U-rich elementbinding complex(es) functionally interact to promote usage ofnoncanonical poly(A) sites during spermatogenesis.

Finally, UGUA sequences within the 30UTR of RINGOtranscripts were recently shown to be part of translationalrepressor elements bound by Pumilio proteins in developingXenopus oocytes [60]. Stage-dependent translational repressionis a prominent mechanism of regulated gene expression duringspermiogenesis [46]. These observations thus raise theinteresting possibility that UGUA sequences present inspermatogenic transcripts have dual functions in certaininstances, regulating both alternative 30-end cleavage/polyad-enylation of pre-mRNAs as well as stage-dependent mRNAtranslation.

ACKNOWLEDGMENT

The authors would like to thank Mr. George Gagnon for his excellenttechnical assistance.

REFERENCES

1. Eddy EM. Male germ cell gene expression. Recent Prog Horm Res 2002;57:103–128.

2. MacDonald CC, Redondo JL. Reexamining the polyadenylation signal:were we wrong about AAUAAA? Mol Cell Endocrinol 2002; 190:1–8.

3. Wang H, Sartini BL, Millette CF, Kilpatrick DL. A developmental switchin transcription factor isoforms during spermatogenesis controlled byalternative messenger RNA 30-end formation. Biol Reprod 2006; 75:318–323.

4. Foulkes NS, Schlotter F, Pevet P, Sassone-Corsi P. Pituitary hormone FSHdirects the CREM functional switch during spermatogenesis. Nature 1993;362:264–267.

5. Zhang YL, Akmal KM, Tsuruta JK, Shang Q, Hirose T, Jetten AM, KimKH, O’Brien DA. Expression of germ cell nuclear factor (GCNF/RTR)during spermatogenesis. Mol Reprod Dev 1998; 50:93–102.

6. McMahon KW, Hirsch BA, MacDonald CC. Differences in polyadenyl-ation site choice between somatic and male germ cells. BMC Mol Biol2006; 7:35.

7. Zhang H, Lee JY, Tian B. Biased alternative polyadenylation in humantissues. Genome Biol 2005; 6:R100.

8. Liu D, Brockman JM, Dass B, Hutchins LN, Singh P, McCarrey JR,MacDonald CC, Graber JH. Systematic variation in mRNA 30-processingsignals during mouse spermatogenesis. Nucleic Acids Res 2007; 35:234–246.

9. Zhao J, Hyman L, Moore C. Formation of mRNA 30-ends in eukaryotes:mechanism, regulation, and interrelationships with other steps in mRNAsynthesis. Microbiol Mol Biol Rev 1999; 63:405–445.

10. Proudfoot N. New perspectives on connecting messenger RNA 30-endformation to transcription. Curr Opin Cell Biol 2004; 16:272–278.

11. Salisbury J, Hutchison KW, Graber JH. A multispecies comparison of themetazoan 30-processing downstream elements and the CstF-64 RNArecognition motif. BMC Genomics 2006; 7:55.

12. Colgan DF, Manley JL. Mechanism and regulation of mRNA polyade-nylation. Genes Dev 1997; 11:2755–2766.

13. Hu J, Lutz CS, Wilusz J, Tian B. Bioinformatic identification of candidatecis-regulatory elements involved in human mRNA polyadenylation. RNA2005; 11:1485–1493.

14. Danckwardt S, Kaufmann I, Gentzel M, Foerstner KU, Gantzert AS,Gehring NH, Neu-Yilik G, Bork P, Keller W, Wilm M, Hentze MW,Kulozik AE. Splicing factors stimulate polyadenylation via USEs at non-canonical 30-end formation signals. EMBO J 2007; 26:2658–2669.

15. Kaufmann I, Martin G, Friedlein A, Langen H, Keller W. Human Fip1 is a

subunit of CPSF that binds to U-rich RNA elements and stimulatespoly(A) polymerase. EMBO J 2004; 23:616–626.

16. Brown KM, Gilmartin GM. A mechanism for the regulation of pre-mRNA30-processing by human cleavage factor Im. Mol Cell 2003; 12:1467–1476.

17. Ruegsegger U, Blank D, Keller W. Human pre-mRNA cleavage factor Imis related to spliceosomal SR proteins and can be reconstituted in vitrofrom recombinant subunits. Mol Cell 1998; 1:243–253.

18. Ruegsegger U, Beyer K, Keller W. Purification and characterization ofhuman cleavage factor Im involved in the 30-end processing of messengerRNA precursors. J Biol Chem 1996; 271:6107–6113.

19. Venkataraman K, Brown KM, Gilmartin GM. Analysis of a noncanonicalpoly(A) site reveals a tripartite mechanism for vertebrate poly(A) siterecognition. Genes Dev 2005; 19:1315–1327.

20. Dettwiler S, Aringhieri C, Cardinale S, Keller W, Barabino SM. Distinctsequence motifs within the 68-kDa subunit of cleavage factor Im mediateRNA binding, protein-protein interactions, and subcellular localization. JBiol Chem 2004; 279:35788–35797.

21. Wallace AM, Dass B, Ravnik SE, Tonk V, Jenkins NA, Gilbert DJ,Copeland NG, MacDonald CC. Two distinct forms of the 64,000 Mrprotein of the cleavage stimulation factor are expressed in mouse malegerm cells. Proc Natl Acad Sci U S A 1999; 96:6763–6768.

22. Dass B, Attaya EN, Michelle Wallace A, MacDonald CC. Overexpressionof the CstF-64 and CPSF-160 polyadenylation protein messenger RNAs inmouse male germ cells. Biol Reprod 2001; 64:1722–1729.

23. Kashiwabara S, Zhuang T, Yamagata K, Noguchi J, Fukamizu A, Baba T.Identification of a novel isoform of poly(A) polymerase, TPAP,specifically present in the cytoplasm of spermatogenic cells. Dev Biol2000; 228:106–115.

24. Bellve AR, Millette CF, Bhatnagar YM, O’Brien DA. Dissociation of themouse testis and characterization of isolated spermatogenic cells. JHistochem Cytochem 1977; 25:480–494.

25. Bellve AR, Cavicchia JC, Millette CF, O’Brien DA, Bhatnagar YM, DymM. Spermatogenic cells of the prepuberal mouse. J Cell Biol 1977; 74:68–85.

26. Wang H, Liu F, Millette CF, Kilpatrick DL. Expression of a novel, sterol-insensitive form of sterol regulatory element binding protein 2 (SREBP2)in male germ cells suggests important cell- and stage-specific functions forSREBP targets during spermatogenesis. Mol Cell Biol 2002; 22:8478–8490.

27. Liu F, Tokeson J, Persengiev SP, Ebert K, Kilpatrick DL. Novel repeatelements direct rat proenkephalin transcription during spermatogenesis. JBiol Chem 1997; 272:5056–5062.

28. Yu M, Sha H, Gao Y, Zeng H, Zhu M, Gao X. Alternative 30UTRpolyadenylation of Bzw1 transcripts display differential translationefficiency and tissue-specific expression. Biochem Biophys Res Commun2006; 345:479–485.

29. Shaper NL, Wright WW, Shaper JH. Murine B1,4-galactosyltransferase:both the amounts and structure of the mRNA are regulated duringspermatogenesis. Proc Natl Acad Sci U S A 1990; 87:791–795.

30. Meijer D, Hermans A, von Lindern M, van Agthoven T, de Klein A,Mackenbach P, Grootegoed A, Talarico D, Della Valle G, Grosveld G.Molecular characterization of the testis specific c-abl mRNA in mouse.EMBO J 1987; 6:4041–4048.

31. Noshiro M, Aoyama Y, Kawamoto T, Gotoh O, Horiuchi T, Yoshida Y.Structural and evolutionary studies on sterol 14-demethylase p450(CYP51), the most conserved p450 monooxygenase. I. Structural analysesof the gene and multiple sizes of mRNA. J Biochem 1997; 122:1114–1121.

32. Ravnik SE, Wolgemuth DJ. Regulation of meiosis during mammalianspermatogenesis: the A-type cyclins and their associated cycle dependentkinases are differentially expressed in the germ-cell lineage. Dev Biol1999; 207:408–418.

33. Moss S, Ferry R, Groudine M. An alternative pathway of histone mRNA30-end formation in mouse round spermatids. Nucleic Acids Res 1994; 22:3160–3166.

34. Godmann M, Kromberg I, Mayer J, Behr R. The mouse Kruppel-likeFactor 4 (Klf4) gene: four functional polyadenylation sites which are usedin a cell-specific manner as revealed by testicular transcript analysis andmultiple processed pseudogenes. Gene 2005; 361:149–156.

35. Muramatsu T, Giri P, Higuchi S, Kincaid RL. Molecular cloning of acalmodulin-dependent phosphatase from murine testis identification of adevelopmentally expressed nonneural isoenzyme. Proc Natl Acad SciU S A 1992; 89:529–533.

36. Lopez P, Yaman R, Lopez-Fernandez LA, Vidal F, Puel D, Clertant P,Cuzin F, Razzoulzadegan M. A novel germ line-specific gene of thephosducin-like protein (PhLP) family. J Biol Chem 2003; 278:1751–1757.

CFIm SUBUNITS ARE ENRICHED IN MALE GERM CELLS 481

37. Ueki K, Muramatsu T, Kincaid RL. Structure and expression of twoisoforms of the murine calmodulin-dependent protein phosphataseregulatory subunit (calcineurin B). Biophys Res Commun 1992; 187:537–543.

38. Hara Y, Rovescalli AC, Kim Y, Nirenberg M. Structure and evolution offour POU domain genes expressed in mouse brain. Proc Natl Acad Sci U S A1992; 89:3280–3284.

39. Daneau I, Houde A, Ethier JF, Lussier JG, Silversides DW. Bovine SRYgene locus: cloning and testicular expression. Biol Reprod 1995; 52:591–599.

40. Gu W, Wu X, Meng XH, Morales C, el-Alfy M, Hecht NB. The RNA-andDNA-binding protein TB-RBP is spatially and developmentally regulatedduring spermatogenesis. Mol Reprod Dev 1998; 49:219–228.

41. Russell D, Kim K. Expression of triosephosphate isomerase transcripts inrat testis: evidence for retinol regulation and a novel germ cell transcript.Biol Reprod 1996; 55:11–18.

42. Cunliffe V, Koopman P, McLaren A, Trowsdale J. A mouse zinc fingergene which is transiently expressed during spermatogenesis. EMBO J1990; 9:197–205.

43. Yang G, Zhang YL, Buchold GM, Jetten AM, O’Brien DA. Analysis ofgerm cell nuclear factor transcripts and protein expression duringspermatogenesis. Biol Reprod 2003; 68:1620–1630.

44. Frayne EG, Leys EJ, Crouse GF, Hook AG, Kellems RE. Transcription ofthe mouse dihydrofolate reductase gene proceeds unabated through sevenpolyadenylation sites and terminates near a region of repeated DNA. MolCell Biol 1984; 4:2921–2924.

45. Kubo T, Wada T, Yamaguchi Y, Shimizu A, Handa H. Knock-down of 25kDa subunit of cleavage factor Im in Hela cells alters alternativepolyadenylation within 30-UTRs. Nucleic Acids Res 2006; 34:6264–6271.

46. Kleene KC. Patterns of translational regulation in the mammalian testis.Mol Reprod Dev 1996; 43:268–281.

47. Persengiev SP, Raval PJ, Rabinovitch S, Millette CF, Kilpatrick DL.Transcription factor Sp1 is expressed by three different developmentallyregulated messenger ribonucleic acids in mouse spermatogenic cells.Endocrinology 1996; 137:638–646.

48. Persengiev SP, Robert S, Kilpatrick DL. Transcription of the TATAbinding protein gene is highly up-regulated during spermatogenesis. MolEndocrinol 1996; 10:742–747.

49. Edwalds Gilbert G, Veraldi KL, Milcarek C. Alternative poly(A) siteselection in complex transcription units: means to an end? Nucleic AcidsRes 1997; 25:2547–2561.

50. Foulkes NS, Mellstrom B, Benusiglio E, Sassone-Corsi P. Developmentalswitch of CREM function during spermatogenesis: from antagonist toactivator. Nature 1992; 355:80–84.

51. Eddy EM. Regulation of gene expression during spermatogenesis. SeminCell Dev Biol 1998; 9:451–457.

52. Salehi-Ashtiani K, Goldberg E. Posttranscriptional regulation of primateLdhc mRNA by its AUUUA-like elements. Mol Endocrinol 1995; 9:1782–1790.

53. Zhang H, Barnoski BL, Sol-Church K, Stabley DL, Martin-Deleon PA.Murine Spam1 mRNA: involvement of AU-rich elements in the 30UTRand antisense RNA in its tight post-transcriptional regulation inspermatids. Mol Reprod Dev 2006; 73:247–255.

54. Xu M, Hecht NB. Polypyrimidine tract binding protein 2 stabilizesphosphoglycerate kinase 2 mRNA in murine male germ cells by binding toits 30-UTR. Biol Reprod 2007; 76:1025–1033.

55. Hawthorne SK, Busanelli RR, Kleene KC. The 50-UTR and 30-UTR of thesperm mitochondria-associated cysteine-rich protein mRNA regulatetranslation in spermatids by multiple mechanisms in transgenic mice.Dev Biol 2006; 297:118–126.

56. D’Mello V, Lee JY, MacDonald CC, Tian B. Alternative mRNApolyadenylation can potentially affect detection of gene expression byAffymetrix genechip arrays. Appl Bioinformatics 2006; 5:249–253.

57. Beaudoing E, Freier S, Wyatt JR, Claverie JM, Gautheret D. Patterns ofvariant polyadenylation signal usage in human genes. Genome Res 2000;10:1001–1010.

58. Danckwardt S, Gehring NH, Neu-Yilik G, Hundsdoerfer P, Pforsich M,Frede U, Hentze MW, Kulozik AE. The prothrombin 3’end formationsignal reveals a unique architecture that is sensitive to thrombophilic gain-of-function mutations. Blood 2004; 104:428–435.

59. Millevoi S, Loulergue C, Dettwiler S, Karaa SZ, Keller W, Antoniou M,Vagner S. An interaction between U2AF 65 and CF I(m) links the splicingand 30-end processing machineries. EMBO J 2006; 25:4854–4864.

60. Padmanabhan K, Richter JD. Regulated Pumilio-2 binding controlsRINGO/Spy mRNA translation and CPEB activation. Genes Dev 2006;20:199–209.

482 SARTINI ET AL.