During spermatogenesis, cells coordinate differentiation with the meiotic cell cycle to generate functional gametes. We identified a novel gene, which we named off-schedule (ofs), as being essential for this coordinated control. During the meiotic G2 phase, Drosophila ofs mutant germ cells do not reach their proper size and fail to execute meiosis or significant differentiation. The accumulation of four cell cycle regulators-Cyclin A, Boule, Twine and Roughex-is altered in these mutants,indicating that ofs reveals a novel branch of the pathway controlling meiosis and differentiation. Ofs is homologous to eukaryotic translation initiation factor eIF4G. The level of ofs expression in spermatocytes is much higher than for the known eIF4G ortholog (known as eIF-4G or eIF4G),suggesting that Ofs substitutes for this protein. Consistent with this, assays for association with mRNA cap complexes, as well as RNA-interference and phenotypic-rescue experiments, demonstrate that Ofs has eIF4G activity. Based on these studies, we speculate that spermatocytes monitor G2 growth as one means to coordinate the initiation of meiotic division and differentiation.
INTRODUCTION
In spermatogenesis, progenitor cells must execute the meiotic divisions in coordination with acquiring the specialized morphology and functionality of sperm. This conserved process is particularly amenable to analysis in Drosophila. The fly testis is a blind-ended tube organized as an assembly line for spermatogenesis. Germline stem cells at the blind end give rise to gonialblasts, which divide mitotically four times with incomplete cytokinesis to produce a cyst of 16 interconnected spermatogonia. These cells exit the mitotic cycle and enter meiosis as spermatocytes, exhibiting an extended G2 phase characterized by a significant increase in cell mass and robust transcription. At the end of G2, the spermatocytes undergo the meiotic divisions and begin the conversion from round spermatids to specialized spermatozoa (Fuller,1993).
Ten `spermatocyte arrest' genes are required for both meiosis and differentiation and are sorted into two classes according to their molecular targets and specific role in promoting transcription(Ayyar et al., 2003; Hiller et al., 2004; Hiller et al., 2001; Jiang and White-Cooper, 2003; Lin et al., 1996; Perezgasga et al., 2004). The always early (aly) class affects the transcription of meiotic genes such boule, twine and cyclin B, as well as that of differentiation genes such as fuzzy onions (fzo) and don juan (White-Cooper et al.,1998). Notably, these mutations do not effect transcription of other spermatocyte genes, such as pelota, cyclin A and roughex. The Aly class proteins are thought to alter chromatin structure to permit the high levels of transcription necessary in spermatocytes (Ayyar et al.,2003; Jiang and White-Cooper,2003; Perezgasga et al.,2004; White-Cooper et al.,2000). The cannonball (can) class affects boule and twine expression post-transcriptionally only and has no effect on cyclin B. The post-transcriptional effects must be indirect, because all can class loci encode testis-specific components of the general transcriptional machinery (Hiller et al.,2004; Hiller et al.,2001). Together, the spermatocyte arrest genes reveal how a diverse set of genes is selectively transcribed in spermatocytes.
The transcriptional regulatory pathway does not address the timing of meiotic entry and differentiation, however. Although transcripts necessary for these processes accumulate in early spermatocytes, the corresponding proteins do not appear until much later(White-Cooper et al., 1998). Because there is little, if any, transcription after the G2-M transition in flies (Olivieri and Olivieri, 1965), spermatocytes must delay meiotic division until all the necessary transcripts have accumulated. A similar dilemma exists during the mitotic cycle in yeast. For cells to maintain the same average size over several divisions, control points act during the gap phases and allow cell cycle progression only when the cell has reached a threshold size, with G1 predominating in budding yeast and G2 in fission yeast (Rupes, 2002). Cell growth rates also feed back on mitotic cell cycle progression in Drosophila cells (Stocker and Hafen, 2000). Less is known about how growth might affect the specialized meiotic cell cycle.
Our identification and characterization of off-schedule(ofs) provides evidence that cell growth is linked to the coordination of meiosis and differentiation. Spermatocytes in ofsmutant males failed to execute the G2-M transition of meiosis or substantive post-meiotic differentiation and had a significant cell size defect. The Off-schedule protein was found to resemble the eukaryotic initiation factor 4G (eIF4G), which is a member of the eIF4F translation initiation complex and bridges mature mRNAs and the ribosome(Prevot et al., 2003). The eIF4G activity of Ofs was apparent in its ability to associate with mRNA caps and to functionally replace canonical eIF4G (also known as eIF-4G-FlyBase) in cell culture. Because translation is primarily regulated at initiation, eIF4G is instrumental in determining the translational capacity of a cell and thus its ability to accumulate mass. Thus, the ofs mutant phenotype suggests that sufficient cell mass must accumulate before spermatocytes execute meiosis and differentiation.
MATERIALS AND METHODS
Fly stocks
Testes from males homozygous for a P element-generated deletion, ofs3572D-175ca, contained few sperm tails. Complementation against EMS-induced male sterile mutants identified the second allele, ofsZ3-3283 (Wakimoto et al., 2004). rux9(roughex9) is a presumed null(Gönczy et al., 1994). Mosaic analysis with repressible cell marker (MARCM) clones were generated in y w hsFLP Tub-GAL4 UAS-GFP:myc-nls; +; FRT82B ofsZ3-3283/FRT82BTub-GAL80 males [MARCM stocks from Ken Irvine(Rutgers, Piscataway, NJ) and Gary Struhl (Columbia University, New York, NY)](Lee and Luo, 1999). Clones were induced in wandering third instar larvae, and flies were aged 1-3 days at 25°C and an additional 6 days at 29°C. For lacZ-marked clones, we used FRT82B ms(3)1011 (an insertion at effete, which results in lacZ expression throughout the male germline)(Castrillon et al., 1993). These clones were induced in pupae or newly eclosed adult males; flies were aged at 25°C. twine-lacZ has been described previously(White-Cooper et al.,1998).
Microscopy
For phase microscopy of unfixed samples, testes were dissected in Drosophila ringers solution, placed in 15 μl on a slide and overlaid with a coverslip. BrdU pulse labeling was for 20-30 minutes at room temperature, as was previously described(Wallenfang et al., 2006). Frozen split squashes were as described in Li et al.(Li et al., 2004).β-galactosidase-activity stains on testes from newly eclosed males homozygous for the twine-lacZ insertion were as in Gönczy et al.(Gönczy et al., 1992). Immunofluorescence on whole mounts was as described in Terry et al.(Terry et al., 2006). Primary antibody incubations were overnight at 4°C unless noted: 1:50-100 mouseα-Cyclin A-A19, 1:1000 rabbit α-Boule (R383)(Cheng et al., 1998), 1:4000 rabbit α-Aly (White-Cooper et al.,1998), 1:700 rabbit α-Fzo(Hales and Fuller, 1997), 1:20 mouse α-Cyclin B-F2, 1:4000 rabbit α-lamin-Dm0(Smith and Fisher, 1989),1:1000 rabbit α-GFP (1 hour, room temperature; Molecular Probes), 1:1000 rabbit α-Rux (Avedisov et al.,2000), 1:500 rabbit α-eIF4G (p180)(Zapata et al., 1994) and 1:10 mouse α-BrdU (90 minutes, 25°C; Becton-Dickinson). Images were captured using a Zeiss microscope with Apotome illumination.
In situ hybridizations
Probes were synthesized from cDNAs AT30049 (ofs; Fig. 5A) and RE34257(eIF4G) using the Dig RNA labeling kit (Roche Boehringer Mannheim),treated with DNase I, hydrolyzed in carbonate buffer and precipitated. Testes were fixed in 4% formaldehyde in PBS for 15 minutes on ice, then in 4%formaldehyde in PBS with 0.1% Tween-20 and 0.1% Na deoxycholate for 15 minutes at room temperature, treated with Proteinase K and fixed a third time in 4%formaldehyde in PBS with 0.1% Tween-20 for 30 minutes at room temperature. Hybridization was overnight at 65°C in 50% formamide, 5×SSC, 100μg/ml salmon sperm DNA, 50 μg/ml heparin and 0.1% Tween-20, adjusted to pH 4.5 with citric acid. An α-Digoxygenin-AP secondary antibody (Roche)was used at 1:2000, either overnight at 4°C or for 1 hour at room temperature, and reactions were developed with NBT/BCIP.
Cloning
The 3572D-175ca deletion was generated by male recombination using insertion EP(3)3572 (Preston and Engels,1996). Complementation between 3572D-175ca and other deletions that we generated, including 3572D-197e (which complemented 3572D-175ca),defined a roughly 2 kb relevant region. This comprised the 5′ UTR for two nested genes, CG10192 and CG33111, and included 45 bases of the CG10192 coding region (Fig. 5A). CG10192 and CG33111 coding regions were then sequenced from ofsZ3-3283 and two isogenic controls (Z3-0105 and Z3-5364). Only one base change was found: a C to T change (Gln798>Stop) in CG10192.
cDNA mapping and 5′ and 3′ RACE showed that CG10192 and CG33111 share an alternatively spliced 5′ UTR(Fig. 5A, transcripts A-D) but have distinct ORFs and 3′ UTRs. A 117 bp region encoding no recognizable protein domains is alternatively spliced out of the first exon of CG10192(Fig. 5A,B). Reverse transcriptase (RT)-PCR analysis from testis RNA verified that this was the only alternative splicing in the CG10192 coding region. Details of transcript mapping are available upon request. The rescue construct contained 328 bp of 5′ UTR common to all splice variants, the full-length coding region and 298 bp of the 3′ UTR, assembled by RT-PCR and cloned into an hs83 promoter P element vector (Horabin and Schedl, 1993).
Cell culture
We plated 4.5 million S2R+ cells in six-well plates in Insectagro(Mediatech) supplemented with 7.5% fetal bovine serum and penicillin/streptomycin. Transfections were carried out the following day(Effectene; Qiagen). For the cap-association experiments, cells were harvested 2 days post-transfection and lysed in 50 mM Tris-Cl (pH 7.5), 150 mM NaCl, 1%Tween-20, 0.3% Igepal with a protease inhibitor cocktail (Sigma), 1 mM NaF, 1 mM activated sodium orthovanadate and 1 mM PMSF on ice for 15 minutes. GTP(100 mM) and 1 mM DTT were added for 10 minutes on ice, and then lysates were incubated with pre-blocked 7-methyl-GTP Sepharose 4B or Glutathione Sepharose 4B (GE, Amersham) overnight at 4°C. Beads were washed twice in lysis buffer then in PBS and resuspended in 1×SDS-PAGE sample buffer, boiled and analyzed by western blot.
dsRNA target regions were: DSRed, first ∼600 bp of the pDSRed2 coding region (Clontech); eIF4G, first ∼650 bp of the coding region(Herold et al., 2001); ofs, HFA14207 (GenomeRNAi Database, www.dkfz.de/signaling2/rnai/);and eIF4A (eIF-4A), HFA03526. Target regions were amplified from cDNA using primers with 5′ T7 promoter sequences appended, and dsRNAs were synthesized (MEGAscript T7 kit. Ambion). dsRNA knockdown was performed as described (Clemens et al.,2000). Double knockdowns used twice as much total dsRNA.
Growth and rescue experiment overview: day 1, plated 4.5×106 cells per well and dsRNA treated; day 2, transfected as above; days 4 and 7, split cells 1:3; days 2, 3, 4, 6, 8 and 10, counted cells (hemacytometer). On day 8, cells were harvested, washed, fixed overnight in cold 70% ethanol and stained with propidium iodide solution. They were then sorted on a Becton-Dickinson FACSort machine running CellQuest Pro software. Cell cycle profiles were determined on FlowJo software using the Watson algorithm. All constructs used the Actin5C promoter, with eGFP kindly provided by Sara Cherry (University of Pennsylvania, Philadelphia, PA), and GFP-ofs and GFP-eIF4G both generated using the Drosophila Gateway system and the pENTR/D-TOPO kit (Invitrogen)(Huynh and Zieler, 1999).
Western blotting
Each 1 ml of S2R+ cells from the growth experiment was lysed in 50 μl of 50 mM Tris (pH 7.5), 150 mM NaCl, 1% NP40, 10% glycerol, 1.5 mM EGTA on ice,combined with 2×SDS-PAGE sample buffer and boiled. Approximately 1-5μl of lysate was loaded, adjusting for differences in cell numbers. Blots were blocked in 5% milk in TBS +0.1% Triton X-100, then cut to probe the same filter with experimental and loading control antibodies: 1:5000 mouseα-GFP (Clontech), 1:100,000 rabbit α-eIF4G (p180)(Zapata et al., 1994),1:10,000 mouse α-alpha-tubulin (B512; Sigma), 1:5000 rabbitα-eIF4E (eIF-4E), 1:5000 mouse α-FLAG (F3165; Sigma), diluted in 2% milk (GFP antibody) or 5% milk (all others). HRP-coupled secondary antibodies (1:2000; Vector) were used for ECL detection (Amersham).
RESULTS
off-schedule is required for meiotic progression and spermatid differentiation
In wild-type flies, elongated sperm tails fill the majority of the testis tube. In the course of screening for defects in spermatogenesis, we noticed a severe reduction in sperm tails in the small deletion mutant 3572D-175ca (data not shown). Complementation analysis against EMS-induced male sterile mutants(Wakimoto et al., 2004)uncovered a second, stronger allele, Z3-3283, mutations in which produced no sperm tails (Fig. 1B). Z3-3283 testes were instead filled with cells resembling primary spermatocytes, having large nuclei and prominent nucleoli (Fig. 1D,E, red nuclei). In wild type, spermatid differentiation was first evident with the appearance of a spherical, phase-dark mitochondrial derivative (Fig. 1D, purple). In the mutants, mitochondria instead formed a dissociated mass next to the nucleus (Fig. 1E, blue). These observations suggested that mutant spermatocytes did not execute the meiotic G2-M transition or significant spermatid differentiation. The phenotype was recessive (Fig. 1A), and no stronger with Z3-3283 placed over a deficiency(Fig. 1C). Thus, the Z3-3283 allele was probably a strong hypomorph or null for the gene we tentatively called off-schedule (ofs), and was used for all subsequent analyses.
Mutant spermatocytes appeared smaller than wild type(Fig. 1D,E, compare red nuclei). To confirm this, we induced both homozygous-mutant and homozygous-wild-type clones in heterozygous testes, and compared the cross-sectional area of young spermatocytes of each genotype. Mutant cells were identified by the loss of a lacZ-expressing chromosome(lacZ-), whereas homozygous wild-type cells were identified by higher lacZ expression (roughly 2× higher)compared with heterozygotes (homozygous wild-type cells were termed 2×lacZ+). We outlined the cell periphery, revealed by anti-phospho-Tyrosine, and found that the cross-sectional area of lacZ-ofs cells was significantly smaller than age-matched 2×lacZ+ wild-type cells induced in the same testis. For example, the mean±s.e.m. was 192±5.2μm2 versus 291±6.1 μm2 (n=5 cells of each genotype; Student's t-test for matched pairs, P<0.001). Similar results were obtained in comparisons in two other testes. To rule out any contribution from second-site mutations, we also compared spermatocyte size in ofs/Df(3R)mbc-R1 testes versus heterozygous siblings. Young ofs/Df spermatocytes, defined as those first accumulating Boule protein (see below), were smaller than heterozygous sibling cells (mean±s.e.m.=314±15 μm2 versus 448±14 μm2, respectively; n=25 cells, five testes for each genotype; P=3.3×10-8; see Fig. S1 in the supplementary material). Because decreased size could reflect a difference in ploidy, we assayed for pre-meiotic S phase in ofs mutants by BrdU incorporation. All germ cells within a cyst cycle synchronously, allowing us to distinguish pre-meiotic S phase from the mitotic S phases by counting the number of cells in each BrdU-positive (BrdU+) cyst. The fraction of cysts per testis that were undergoing pre-meiotic S phase was similar in ofs compared to heterozygotes (0.4 versus 0.6, n=21 and 15 testes, respectively; P=0.42). Thus, the ofs spermatocytes progressed through pre-meiotic S phase normally, and were indeed in G2 of their meiotic cell cycle. The smaller size of ofsspermatocytes must therefore reflect a defect in mass accumulation.
ofs mutants fail to undergo meiosis or differentiation.(A-C) Phase images, apical-tip oriented to the left. (A) Heterozygous testes appear wild-type, with elongating spermatids filling most of the testis. (B) Mutant testes lack elongating spermatids. (C) ofs/Df(3R)mbc-R1 is indistinguishable from ofs/ofs (B). (D,E) High-magnification, with diagrams of the represented stages. Color code: primary spermatocyte nucleus, red;nucleolus, black; haploid spermatid nucleus, orange; fused mitochondrial derivative, purple; dissociated mitochondria, blue; cell outlines (where discernable), gray. (D) Heterozygous testes have examples of mature primary spermatocytes (left) and onion-stage spermatids (right). (E) The most prevalent cells found in ofs testes resemble primary spermatocytes.(F,G) Whole-mount immunofluorescent images of wild-type (F, arrowhead,2×lacZ+) and ofs (G, arrow, lacZ-) clones in a heterozygous background. Flies were aged 4 days after clone induction, which allowed clones to progress midway through the meiotic G2 of wild-type spermatocytes.(F′,G′) Brackets are the same size in each panel,highlighting the cell size difference between the two clones. The entire ofs cell outline (PTyr, phospho-Tyrosine) is approximately the same size as the nucleus of the control cell.
ofs mutants fail to undergo meiosis or differentiation.(A-C) Phase images, apical-tip oriented to the left. (A) Heterozygous testes appear wild-type, with elongating spermatids filling most of the testis. (B) Mutant testes lack elongating spermatids. (C) ofs/Df(3R)mbc-R1 is indistinguishable from ofs/ofs (B). (D,E) High-magnification, with diagrams of the represented stages. Color code: primary spermatocyte nucleus, red;nucleolus, black; haploid spermatid nucleus, orange; fused mitochondrial derivative, purple; dissociated mitochondria, blue; cell outlines (where discernable), gray. (D) Heterozygous testes have examples of mature primary spermatocytes (left) and onion-stage spermatids (right). (E) The most prevalent cells found in ofs testes resemble primary spermatocytes.(F,G) Whole-mount immunofluorescent images of wild-type (F, arrowhead,2×lacZ+) and ofs (G, arrow, lacZ-) clones in a heterozygous background. Flies were aged 4 days after clone induction, which allowed clones to progress midway through the meiotic G2 of wild-type spermatocytes.(F′,G′) Brackets are the same size in each panel,highlighting the cell size difference between the two clones. The entire ofs cell outline (PTyr, phospho-Tyrosine) is approximately the same size as the nucleus of the control cell.
off-schedule identifies a distinct pathway controlling spermatocyte development
The lack of meiosis and of any significant differentiation in ofsmutants was strongly reminiscent of the phenotype of spermatocyte-arrest mutants (Lin et al., 1996). ofs did not control Aly accumulation, which had the same spermatocyte nuclear localization in ofs mutants as in wild type(Fig. 2A,B, red). Furthermore,Cyclin A accumulation in ofs mutants revealed a phenotype distinct from that of the spermatocyte-arrest class. Normally, Cyclin A accumulated in the cytoplasm of G2 spermatocytes(Fig. 2A, green). As spermatocytes near the G2-M transition, Cyclin A translocates into the nucleus, where it is visible only briefly before it is degradation at metaphase (Eberhart et al.,1996; Lin et al.,1996). In spermatocyte-arrest mutants, Cyclin A accumulates normally, but then persists (Lin et al.,1996). By contrast, in ofs mutants, Cyclin A shifted to the nuclei of spermatocytes and was degraded(Fig. 2B, green, arrows; also see Fig. 3B, gold arrow). In addition, the timing of both nuclear translocation and eventual degradation dramatically differed from wild type. First, nuclear accumulation appeared more proximal to the testis tip than would normally occur(Fig. 2B). To establish whether this implied precocious movement of Cyclin A to the nucleus in mutant cells,we analyzed ofs mutant clones using cell position along the testis as a measure of developmental time. Indeed, Cyclin A was visible in the nucleus of mutant cells (Fig. 3A,arrows) that were much more proximally located relative to heterozygous cells,in which the translocation only occurred at meiosis I(Fig. 3A, arrowhead). Second,degradation of Cyclin A appeared delayed, because each ofs mutant testis had several cysts with nuclear Cyclin A, whereas there was, on average,less than one cyst in wild type (Fig. 2, green) (Lin et al.,1996). Mosaic analysis confirmed this, because Cyclin A persisted much longer in ofs mutant cells than in heterozygous control cysts(Fig. 3B, white arrows versus outline), although, eventually, it was degraded in mutant cells(Fig. 3B, gold arrow).
Whereas Cyclin A translocation was precocious, Boule accumulation was delayed. Boule is one target of the spermatocyte-arrest pathway and accumulates late in G2 (Cheng et al., 1998; White-Cooper et al., 1998). In ofs/+ heterozygotes, germ cells approaching meiotic entry expressed both proteins(Fig. 3C, arrow). In ofs mutants, however, Cyclin A was degraded before Boule was first detected (Fig. 3D, line). Because Cyclin A persisted longer in ofs mutants, the fact that Cyclin A and Boule did not overlap suggested that the onset of Boule accumulation was significantly delayed. This was confirmed by scoring for Boule accumulation in homozygous mutant (lacZ-ofs) and homozygous wild-type (2×lacZ+)clones at 2, 3, 6 and 8 days after clone induction. Whereas wild-type 2×lacZ+ clones accumulated Boule by day 3, we did not detect Boule in ofs clones until day 6(Table 1).
Boule accumulation is delayed in ofs mutant clones
. | . | ofs (lacZ−), # cysts . | . | Wild type (2×lacZ+), # cysts . | . | ||
---|---|---|---|---|---|---|---|
Age of clones (days) . | # testes with clones . | Bol+ . | Bol− . | Bol+ . | Bol− . | ||
2 | 5 | - | 7 | - | 2 | ||
3 | 9 | - | 17 | 4 | 9 | ||
6 | 9 | 2 | 32 | 13 | 18 | ||
8 | 1 | - | 3 | 2 | 2 |
. | . | ofs (lacZ−), # cysts . | . | Wild type (2×lacZ+), # cysts . | . | ||
---|---|---|---|---|---|---|---|
Age of clones (days) . | # testes with clones . | Bol+ . | Bol− . | Bol+ . | Bol− . | ||
2 | 5 | - | 7 | - | 2 | ||
3 | 9 | - | 17 | 4 | 9 | ||
6 | 9 | 2 | 32 | 13 | 18 | ||
8 | 1 | - | 3 | 2 | 2 |
Flies were aged for 2, 3, 6 and 8 days after clone induction. ofsclones lack the lacZ marker, whereas wild-type clones contain two copies of lacZ. Boule (Bol) accumulation was assessed by immunofluorescence. Both types of clones are equally as likely to be induced. The observed difference in the number of ofs versus wild-type clones is probably a reflection of our ability to reliably detect two copies of lacZ versus the one copy in the background.
ofs identifies a distinct pathway controlling spermatocyte development. (A,B) Immunofluorescent images of frozen split squashes. Aly staining (red) is indistinguishable between heterozygotes (A)and homozygotes (B). (A) Cyclin A (green) is predominantly cytoplasmic in heterozygous spermatocytes. (B) In the mutants, Cyclin A accumulates in nuclei(arrows) relatively close to the apical tip.
ofs identifies a distinct pathway controlling spermatocyte development. (A,B) Immunofluorescent images of frozen split squashes. Aly staining (red) is indistinguishable between heterozygotes (A)and homozygotes (B). (A) Cyclin A (green) is predominantly cytoplasmic in heterozygous spermatocytes. (B) In the mutants, Cyclin A accumulates in nuclei(arrows) relatively close to the apical tip.
As a proposed translational regulator, Boule is necessary for the accumulation of Twine, the Cdc25 phosphatase that activates Cdc2 to initiate the G2-M transition (Alphey et al., 1992). The delay in Boule accumulation should therefore cause a delay in Twine accumulation. Indeed, in ofs mutant testes,accumulation of a Twine translational reporter(White-Cooper et al., 1998)was observed in most mutant testes (15 out of 25), but at low levels and only in cells towards the base of the testis(Fig. 3F). Taken together, our data show that several key meiotic regulators exhibited advanced or delayed patterns in ofs mutants; hence the name off-schedule.
off-schedule spermatocytes overexpress Roughex, which causes aberrant Cyclin A localization
The early cytoplasmic-to-nuclear shift of Cyclin A in ofsspermatocytes either reflected precocious execution of a normal event associated with the G2-M transition or a different mechanism entirely. Normally, Cyclin A translocation is coincident with chromosomal condensation from a crescent-shaped to being more compact(Lin et al., 1996)(Fig. 3A, inset ii,lower cells). However, ofs mutant cells with nuclear Cyclin A exhibited chromosomal organization that more closely resembled early-stage heterozygous spermatocytes (Fig. 3A, insets). Therefore, the nuclear import of Cyclin A in ofs mutants was dissociated from other aspects of the G2-M transition.
Similar precocious nuclear import had been observed upon overexpression of Roughex (Rux), a cyclin-dependent kinase inhibitor(Avedisov et al., 2000; Gönczy et al., 1994; Thomas et al., 1997). Indeed,we found that ofs mutant spermatocytes exhibited abnormal accumulation of Rux protein, and that Rux colocalized with nuclear Cyclin A(Fig. 4B, arrow). Furthermore,removal of rux activity prevented Cyclin A from accumulating in the nuclei of mutant spermatocytes (Fig. 4D, arrow). Thus, in ofs mutants, excess Rux accumulated,leading to aberrant, early nuclear Cyclin A localization(Fig. 4A). In contrast to this aberrant program, the normal cytoplasmic-to-nuclear transition of Cyclin A in wild-type cells was not associated with Rux accumulation, and still occurred when rux was absent (Fig. 4C). In summary, the changes in Cyclin A subcellular distribution observed in ofs mutants do not reflect precocious timing for the normal mechanism, but rather the engagement of a different mechanism entirely.
Examining the block to meiosis in off-schedulespermatocytes
We considered whether the changes in protein accumulation that we observed for Cyclin A, Twine and Rux were sufficient to explain the block the G2-M transition in ofs mutants. Because Twine accumulation was delayed in ofs mutants, we drove Twine accumulation in early spermatocytes using the transcriptional and translational regulatory sequences of β2-tubulin to try to rescue meiotic entry(Maines and Wasserman, 1999). However, the ofs mutant phenotype was not noticeably suppressed, even after also removing rux inhibition of Cyclin A/cyclin-dependent kinase (Cdk) activity (data not shown). This suggests that other effectors of the G2-M transition are also affected in ofs mutants(perhaps Cyclin B, see Discussion).
off-schedule encodes the predominant eIF4G-like protein in spermatocytes
To clone ofs, we defined the portion of the ofs3572D-175ca deletion crucial for its effects on spermatogenesis and identified a nonsense mutation in ofsZ3-3283 (Fig. 5A; see Materials and methods). Both affected the CG10192 locus. We conducted transcript mapping, and assembled a full-length cDNA for transgenic rescue, using a promoter that expresses well in spermatocytes (see Materials and methods). ofs homozygotes inheriting the transgene were fully rescued to fertility (data not shown).
ofs encodes a 2072-residue protein with striking homology in the C terminus to the eukaryotic initiation factor 4 gamma (eIF4G). The N terminus did not exhibit any significant homology to other proteins. The bona fide eIF4G ortholog in D. melanogaster is CG10811 (also known as eIF-4G or eIF4G) (Hernandez et al.,1998; Zapata et al.,1994), which is 1666 amino acids in length and has higher overall similarity to eIF4G proteins from other species than does Ofs(Fig. 5B). Ofs and the Drosophila eIF4G protein have strong conservation in their C termini. The premature stop in ofsZ3-3283 is upstream of the conserved regions, consistent with its strong loss-of-function phenotype.
ofs mutants exhibit altered timing of Cyclin A, Boule and Twine accumulation. (A-D) Whole-mount immunofluorescence. (A,B) ofsclones (GFP+, green) in a heterozygous background. (A) ofsclones precociously exhibit nuclear Cyclin A (CycA, arrows, red), whereas heterozygous spermatocytes are only beginning to accumulate Cyclin A, and nuclear entry occurs only in cysts entering the G2-M transition(arrowhead). (Inset i) In early spermatocytes, the degree of chromatin condensation in ofs clones (outlined, upper right) is similar to heterozygous cells (left). (Inset ii) By contrast, older mutant cysts retain less condensed chromatin (outlined, upper right), whereas heterozygotes begin to condense chromatin as they approach meiosis (left-most and lower cells). (B) Midsection of the testis (tip is up) showing persistence of nuclear Cyclin A in clones (white arrows) past the meiotic region and its eventual degradation (yellow arrow). The position of the most advanced Cyclin A-containing heterozygous spermatocytes (located in a different focal plane) is outlined (merged panel). (C,D) In contrast to heterozygotes (C, arrow), in ofs/Df(3R)mbc-R1 mutants,Cyclin A (green) does not overlap with Boule (red) expression (D, line). Notice that ofs/Df has a slightly different (perhaps stronger) Cyclin A phenotype in which all spermatocyte staining is nuclear. (E,F) twine-lacZ reporter, X-Gal activity staining. (E) In heterozygotes (boxed area magnified on right) expression first appears faintly in nuclei of mature primary spermatocytes (arrow), becomes stronger during the meiotic divisions and is strongest in spermatids (arrowhead). Asterisk labels nonspecific, endogenous enzymatic activity. (F) In ofsmutants, expression is first detected relatively more distal in the testis tube (arrow) and resembles the mature primary spermatocyte signal.
ofs mutants exhibit altered timing of Cyclin A, Boule and Twine accumulation. (A-D) Whole-mount immunofluorescence. (A,B) ofsclones (GFP+, green) in a heterozygous background. (A) ofsclones precociously exhibit nuclear Cyclin A (CycA, arrows, red), whereas heterozygous spermatocytes are only beginning to accumulate Cyclin A, and nuclear entry occurs only in cysts entering the G2-M transition(arrowhead). (Inset i) In early spermatocytes, the degree of chromatin condensation in ofs clones (outlined, upper right) is similar to heterozygous cells (left). (Inset ii) By contrast, older mutant cysts retain less condensed chromatin (outlined, upper right), whereas heterozygotes begin to condense chromatin as they approach meiosis (left-most and lower cells). (B) Midsection of the testis (tip is up) showing persistence of nuclear Cyclin A in clones (white arrows) past the meiotic region and its eventual degradation (yellow arrow). The position of the most advanced Cyclin A-containing heterozygous spermatocytes (located in a different focal plane) is outlined (merged panel). (C,D) In contrast to heterozygotes (C, arrow), in ofs/Df(3R)mbc-R1 mutants,Cyclin A (green) does not overlap with Boule (red) expression (D, line). Notice that ofs/Df has a slightly different (perhaps stronger) Cyclin A phenotype in which all spermatocyte staining is nuclear. (E,F) twine-lacZ reporter, X-Gal activity staining. (E) In heterozygotes (boxed area magnified on right) expression first appears faintly in nuclei of mature primary spermatocytes (arrow), becomes stronger during the meiotic divisions and is strongest in spermatids (arrowhead). Asterisk labels nonspecific, endogenous enzymatic activity. (F) In ofsmutants, expression is first detected relatively more distal in the testis tube (arrow) and resembles the mature primary spermatocyte signal.
ofs mRNA accumulates in spermatocytes and at post-meiotic stages(Fig. 5A,C). By contrast, eIF4G showed variable, low-level expression near the testis tip and less signal thereafter (Fig. 5D). eIF4G protein was detectable at low levels in early-stage germ cells, diminishing in spermatocytes, and at high levels in somatic tissues, such as the testis sheath (data not shown). Taken together with the cell size differences noted earlier, we propose that Ofs protein serves as the main eIF4G of the meiotic male germ line. As such, Ofs should associate with mRNA cap complexes, as does canonical eIF4G, and should functionally substitute for eIF4G in cultured cells.
Off-schedule is pulled down by 7-methyl-GTP Sepharose from S2R+ cell extracts
In cap-dependent translation, the eIF4F complex, made up of eIF4G, eIF4A and eIF4E, acts as a bridge between the mRNA cap and the 40S ribosomal subunit. eIF4E binds the modified cap structure of mature mRNAs, whereas eIF4A is an RNA helicase that unwinds secondary structure(Prevot et al., 2003). To establish whether Ofs could associate with mRNA caps, we incubated extracts of S2R+ cells expressing epitope-tagged versions of Ofs or eIF4G with 7-methyl-GTP Sepharose and identified bound proteins by SDS-PAGE western analysis. Both Ofs and eIF4G associated with 7-methyl-GTP Sepharose but not with control Sepharose beads (Fig. 6). Thus, by inference, Ofs can associate with mature mRNAs in cultured cells.
ofs spermatocytes activate an abnormal Cyclin A pattern due overexpression of Rux. Whole-mount immunofluorescent images of siblings from a single cross. (A,B) Rux+ progeny show nuclear Cyclin A (CycA, green) co-localized with Rux (red) accumulation in ofsmutants (B, arrow), but not in their heterozygous siblings (A, arrowhead).(C) Siblings heterozygous for ofs still show Cyclin A nuclear translocation in the absence of rux function (arrowhead). (D) ofs mutants inheriting the rux9 mutation have only cytoplasmic localization of Cyclin A (arrow).
ofs spermatocytes activate an abnormal Cyclin A pattern due overexpression of Rux. Whole-mount immunofluorescent images of siblings from a single cross. (A,B) Rux+ progeny show nuclear Cyclin A (CycA, green) co-localized with Rux (red) accumulation in ofsmutants (B, arrow), but not in their heterozygous siblings (A, arrowhead).(C) Siblings heterozygous for ofs still show Cyclin A nuclear translocation in the absence of rux function (arrowhead). (D) ofs mutants inheriting the rux9 mutation have only cytoplasmic localization of Cyclin A (arrow).
Off-schedule has an eIF4G-like function in S2R+ cells
To address the ability of Ofs to functionally replace eIF4G, we depleted cultured cells of eIF4G or Ofs, singly and in combination, using double-stranded RNA (dsRNA). Depletion of eIF4G alone caused a statistically significant decrease in cell number (Fig. 7A, orange triangles). Although depletion of Ofs alone had no effect (Fig. 7A, green Xs), the combined depletion of Ofs and eIF4G exhibited a more dramatic effect than eIF4G alone, approaching the effect of eIF4A (also known as eIF-4A-FlyBase)knockdown (Fig. 7A, purple asterisks and gray circles). Thus, Ofs is partially redundant with eIF4G for S2R+ cell proliferation and/or viability. Western blots showed specific,extensive reduction in the levels of eIF4G and of transiently transfected GFP-Ofs by day 4 (Fig. 7B).
To examine what effect depletion of these proteins had on the cell cycle,we analyzed DNA content by FACS on day 8. A larger proportion of cells in the eIF4G alone and eIF4G+Ofs depletion conditions had G1 DNA content,indicating that the G1-S transition was delayed(Fig. 7D). This is consistent with the fact that the G1-S transition is thought to be the major control point for growth sensing in mitotic cells of the fly(Stocker and Hafen, 2000). If a G1 delay was the only effect, we would have expected to see a similar, compensatory decrease in both S and G2. Instead, most of the decrease occurred in S phase. Thus, G2 had also slowed, but to a lesser degree than G1. Similar results have been reported previously (Bjorklund et al.,2006). As with the growth curves, the Ofs single depletion had no effect on the cell cycle profile.
We next assessed the ability of GFP-Ofs to rescue the growth defect. Cells were treated with dsRNA on day 1, transfected with a rescue construct on day 2, and GFP expressers (transfected) and non-expressers (untransfected) were counted for each well from days 3 through to 10. Whereas GFP alone could not rescue, cells expressing either GFP-Ofs or GFP-eIF4G now exhibited growth curves and cell cycle profiles indistinguishable from controls(Fig. 7C,E, respectively). Either protein rescued eIF4G deficiency with equal effectiveness. Because dsRNA treatment probably reduced the effectiveness of the cognate GFP-tagged protein (Fig. 7B), rescue was likely to be underestimated.
DISCUSSION
Spermatocytes must execute meiosis and morphogenesis in the correct sequence to form functional sperm. In flies, many of the key genes necessary for this to occur are transcribed at once via activity of the spermatocyte-arrest genes. These transcripts are then regulated post-transcriptionally to ensure the proper timing of their target processes. The ofs mutant phenotype suggests that spermatocytes monitor growth during G2 as one way to ensure that execution of the meiotic divisions is timed coordinately with spermatid differentiation.
ofs encodes the main eIF4G in spermatocytes. (A) Map of the genomic sequence of a portion of the right arm of Chromosome 3 indicating breakpoints of informative deletions (dark gray) and the Z3-3283 point mutation relative to transcripts (red, oriented 5′ to 3′)and coding regions (blue) for both CG10192 (Ofs) and CG33111. Also shown: in situ probes (light gray), insertion sites for nearby transposable elements (red circles) and flanking genes (green). (B) Predicted Ofs protein and Drosophila eIF4G. Locations are noted for the alternate splice (red), the premature stop in Z3-3283 (asterisk), predicted eIF4E-binding sites (green) and homologous regions (gray). Conserved domains are 27% identical and 43% similar, and include the predicted eIF4A-binding domains. (C-D′) In situ hybridizations to wild-type (wt) testes. Tips are marked with asterisks. (C) Antisense probe for ofs (see panel A) showed expression in spermatocytes. (D) Antisense probe to eIF4G showed low expression at the testis tip, and was undetectable thereafter. By contrast, somatic tissue stained strongly (arrows).(C′,D′) Sense probe controls.
ofs encodes the main eIF4G in spermatocytes. (A) Map of the genomic sequence of a portion of the right arm of Chromosome 3 indicating breakpoints of informative deletions (dark gray) and the Z3-3283 point mutation relative to transcripts (red, oriented 5′ to 3′)and coding regions (blue) for both CG10192 (Ofs) and CG33111. Also shown: in situ probes (light gray), insertion sites for nearby transposable elements (red circles) and flanking genes (green). (B) Predicted Ofs protein and Drosophila eIF4G. Locations are noted for the alternate splice (red), the premature stop in Z3-3283 (asterisk), predicted eIF4E-binding sites (green) and homologous regions (gray). Conserved domains are 27% identical and 43% similar, and include the predicted eIF4A-binding domains. (C-D′) In situ hybridizations to wild-type (wt) testes. Tips are marked with asterisks. (C) Antisense probe for ofs (see panel A) showed expression in spermatocytes. (D) Antisense probe to eIF4G showed low expression at the testis tip, and was undetectable thereafter. By contrast, somatic tissue stained strongly (arrows).(C′,D′) Sense probe controls.
Off-schedule, translation initiation and growth control
Alignment among eIF4G sequences suggests that Ofs would be part of the eIF4F complex with eIF4A and eIF4E, and our demonstration of its association with 7-methyl GTP Sepharose strongly supports this. Although we did not test binding of Ofs directly to eIF4A, alignment of human and fly eIF4G proteins shows conservation of three out of four sets of amino acids necessary to bind eIF4A (Imataka and Sonenberg,1997). Of 12 crucial residues, ten were identical in Ofs, one was a conservative (L>I) change, and the twelfth diverged in Drosophila eIF4G as well (data not shown). With regard to eIF4E binding, the putative binding site in Ofs has an arginine substituted for the usual hydrophobic residue (Mader et al.,1995). However, a similar substitution is tolerated in Drosophila eIF4E binding protein 1(Miron et al., 2001), and Baker and Fuller present evidence for interaction with DrosophilaeIF4E1 (Baker and Fuller,2007). Taken together, it is quite likely that Ofs participates in cap-dependent translation initiation.
eIF4G (CG10811) and Ofs (CG10192) appear to be the only two eIF4G proteins encoded in the fly genome. One other candidate, l(2)01424, is more related to the proposed translational inhibitor, NAT1/p97 (Rpn1)/DAP5, than to eIF4G proteins (Takahashi et al.,2005). Although the novel N-terminus of Ofs raised the possibility that it would play a role distinct from eIF4G, our data suggest that Ofs can act as the only eIF4G in cultured cells. Whether these two proteins always act redundantly in vivo cannot be assessed without mutations in eIF4G. Nevertheless, eIF4G, at its endogenous level, cannot substitute for Ofs in spermatocytes. Perhaps this is simply due to a relatively lower level of eIF4G compared with Ofs. Alternatively, Ofs might uniquely aid in the translation of a special class of mRNAs, specific to spermatocyte development. Perhaps sequences in its novel N-terminus assist in such a role. Although further experiments are needed to distinguish between these possibilities, one reason for a distinction between spermatocytes and other cells might be in their respective mode of growth control. In cultured eIF4G-deficient mitotic cells,the cell cycle effect we observed was on G1, whereas the defect in spermatocyte progression was in G2. Although the G1-S transition is the major control point for growth sensing in mitotic cells of the fly (Stocker and Hafen,2000), G2 might make more sense as the control point for meiosis, because it is during this phase of the cycle that spermatocytes need to prepare not just for division, but for differentiation. Furthermore,spermatocytes might commit to the meiotic cycle, versus returning to the mitotic cycle, during G2, as is the case for the yeast Saccharomyces cerevisiae (Simchen et al., 1972). Perhaps expressing a unique eIF4G (Ofs) in spermatocytes helps serve this role. Given the functional role for ofs as presented here and in the accompanying paper(Baker and Fuller, 2007), we propose that ofs henceforth be known as eIF4G2 (also known as eIF-4G2-FlyBase).
Ofs associates with an mRNA cap analog. (A,B) Western blots of S2R+ cells transfected with Flag-Ofs or Flag-eIF4G (A) or with GFP-Ofs (B)and incubated with either 7-m-GTP-Sepharose (7-methyl-GTP Sepharose) beads or glutathione-Sepharose beads (control). Relevant bands are indicated with an arrow (Ofs constructs) or arrowhead (Flag-eIF4G). Asterisks indicate cross-reacting bands. Input lanes represent 2.5% of total lysate; pulldown lanes represent 100% of protein recovered from the beads. (A) Both Flag-Ofs and Flag-eIF4G are pulled down along with eIF4E by 7-m-GTP-Sepharose beads but not by control beads. (B) Binding is not due to the Flag epitope, because GFP-tagged Ofs also associated with 7-m-GTP-Sepharose beads. All conditions were repeated at least three times with comparable results.
Ofs associates with an mRNA cap analog. (A,B) Western blots of S2R+ cells transfected with Flag-Ofs or Flag-eIF4G (A) or with GFP-Ofs (B)and incubated with either 7-m-GTP-Sepharose (7-methyl-GTP Sepharose) beads or glutathione-Sepharose beads (control). Relevant bands are indicated with an arrow (Ofs constructs) or arrowhead (Flag-eIF4G). Asterisks indicate cross-reacting bands. Input lanes represent 2.5% of total lysate; pulldown lanes represent 100% of protein recovered from the beads. (A) Both Flag-Ofs and Flag-eIF4G are pulled down along with eIF4E by 7-m-GTP-Sepharose beads but not by control beads. (B) Binding is not due to the Flag epitope, because GFP-tagged Ofs also associated with 7-m-GTP-Sepharose beads. All conditions were repeated at least three times with comparable results.
Is sufficient growth during G2 essential for meiotic progression?
Because ofs (eIF4G2) encodes the predominant eIF4G in spermatocytes, one might expect that mutant cells would exhibit decreased translation of many mRNAs. Just as we found a striking delay in Boule accumulation, other proteins would be expected to be similarly affected. Such a global deficit could account for the delayed development of these cells, and would be predicted to influence cell size, because the translational capacity of a cell predicts its ability to accumulate mass(Stocker and Hafen, 2000). Indeed, one of the earliest phenotypes in eIF4G2 spermatocytes was their small size. Yet, we also found that Aly accumulation appeared normal and that Rux protein appeared to accumulate to an excess degree in early spermatocytes. These data demonstrate that some mRNAs are not affected by the translational deficit, and raise an alternative scenario wherein spermatocytes actively monitor their size. If they do not achieve proper growth, a checkpoint is induced to prohibit meiosis and differentiation. Because meiosis involves two cell divisions with little intervening interphase, size monitoring would be especially important before these cells commit to divide.
Circumstantial support for a growth checkpoint includes the accumulation of the Cdk inhibitor Rux, which leads to aberrant behavior of Cyclin A. In this model, the postulated checkpoint causes the striking delay in the accumulation of Boule, which, in turn, explains the delay in Twine accumulation. Eventually, Boule does accumulate to reasonable levels, perhaps as cells leak through the checkpoint, just as eventually occurs in mitotic checkpoints(Hartwell and Weinert, 1989). However, by then, Cyclin A has been degraded, and without it, the eventual accumulation of Twine cannot trigger meiosis, so the checkpoint has succeeded.
To establish that a checkpoint exists, one would need to identify the sensor, which detects the problem, and effectors, which execute inhibitory functions until the cell resolves the problem. We do not have a candidate for the sensor that detects growth at this time, nor for effectors controlling differentiation. However, we can speculate that Rux is one effector regulating the meiosis branch, where it could serve to inhibit Cyclin A-driven Cdc2 kinase activity (Avedisov et al.,2000). Rux is not the only effector regulating meiosis, however. Previous work showed that directly increasing the level of Rux only blocked entry into the second meiotic division(Gönczy et al., 1994). Consequently, the accumulation of Rux that we observe in eIF4G2mutants cannot fully explain the absence of the first meiotic division or the defect in differentiation. As would be typical for cell cycle regulation,several effectors must be activated at once to completely block the G2-M transition.
The existence of other effectors could explain why forcing early Twine accumulation failed to restore meiotic entry to eIF4G2 mutants in a rux background. Alternatively, there might be additional positive factors necessary for G2-M transition that have not accumulated in eIF4G2 spermatocytes. Consistent with this, prior work driving expression of another Cdc25, string (stg), in early spermatocytes directed a normal rather than a precocious G2-M transition (Sigrist et al.,1995). Thus, advancing Cdc25 activity is insufficient to trigger a precocious G2-M even in the absence of a growth defect. Perhaps early spermatocytes have not had enough time to accumulate an essential component, such as Cyclin B, for the meiotic divisions. We found that eIF4G2 mutant clones exhibited Cyclin B levels comparable to neighboring heterozygous cells (data not shown). However, there is a peak in Cyclin B accumulation just prior to meiosis I(White-Cooper et al., 1998),and Baker and Fuller describe a deficit of this Cyclin B peak in eIF4G2 mutants (Baker and Fuller,2007). Thus, Cyclin B remains a candidate factor.
Ofs has an eIF4G-like function in S2R+ cells. (A) Growth curves representing the average counts of GFP-(untransfected) cells from 10-13 wells with error bars representing s.e.m. (B) Western blot against eIF4G and GFP-Ofs assessing the degree of knockdown at days 4 and 7. The major eIF4G band is indicated with an arrow. We variably saw a second band migrating slightly faster (asterisk); its identity is unclear. (C) Matched growth curves for both GFP-(untransfected) and GFP+ (transfected) cells from wells transfected with GFP alone, GFP-Ofs or GFP-eIF4G. Curves were normalized for transfection efficiency at day 3. Key in A applies. Expression of GFP-Ofs or GFP-eIF4G completely rescues the defects caused by eIF4G or eIF4G+Ofs knockdown, but not by eIF4A knockdown. Each curve represents average counts from 3-6 wells, with error bars indicating s.e.m. (D) Cell cycle profiles for the cells represented in A showing a shift towards G1with either eIF4G or eIF4G+Ofs knockdown. eIF4A knockdown caused an increase in the percentage of cells with sub-G1 DNA content, normally indicative of apoptotic cells (Oancea et al., 2006). (E) Summary of cell cycle profiling showing rescue by GFP-Ofs and GFP-eIF4G, but not by GFP alone. Gray bars (GFP-) and green bars (GFP+) are paired for each experimental condition. Within each set,from left to right: sub-G1, G1, S and G2. Notice the relative shifts in the G1 and G2 bars:depletion of eIF4G-like proteins caused an increase in the proportion of G1 cells (compare G1 with G2 bars), and this trend was reversed by transfection of either GFP-Ofs or GFP-eIF4G.
Ofs has an eIF4G-like function in S2R+ cells. (A) Growth curves representing the average counts of GFP-(untransfected) cells from 10-13 wells with error bars representing s.e.m. (B) Western blot against eIF4G and GFP-Ofs assessing the degree of knockdown at days 4 and 7. The major eIF4G band is indicated with an arrow. We variably saw a second band migrating slightly faster (asterisk); its identity is unclear. (C) Matched growth curves for both GFP-(untransfected) and GFP+ (transfected) cells from wells transfected with GFP alone, GFP-Ofs or GFP-eIF4G. Curves were normalized for transfection efficiency at day 3. Key in A applies. Expression of GFP-Ofs or GFP-eIF4G completely rescues the defects caused by eIF4G or eIF4G+Ofs knockdown, but not by eIF4A knockdown. Each curve represents average counts from 3-6 wells, with error bars indicating s.e.m. (D) Cell cycle profiles for the cells represented in A showing a shift towards G1with either eIF4G or eIF4G+Ofs knockdown. eIF4A knockdown caused an increase in the percentage of cells with sub-G1 DNA content, normally indicative of apoptotic cells (Oancea et al., 2006). (E) Summary of cell cycle profiling showing rescue by GFP-Ofs and GFP-eIF4G, but not by GFP alone. Gray bars (GFP-) and green bars (GFP+) are paired for each experimental condition. Within each set,from left to right: sub-G1, G1, S and G2. Notice the relative shifts in the G1 and G2 bars:depletion of eIF4G-like proteins caused an increase in the proportion of G1 cells (compare G1 with G2 bars), and this trend was reversed by transfection of either GFP-Ofs or GFP-eIF4G.
Whether a growth checkpoint exists or not, mass accumulation could be used to time the G2-M transition by coupling rate-limiting cell cycle proteins to the translational capacity of the cell. In the budding yeast, S. cerevisiae, cyclin CLN3 (also known as YHC3) contains an upstream open reading frame in the 5′ UTR that slows its translation in G1 under poor growth conditions(Polymenis and Schmidt, 1997). Similarly, during G2 in the fission yeast, Schizosaccharomyces pombe, accumulation of CDC25 is disproportionately affected by defects in translation (Daga and Jimenez,1999). Perhaps the translation of Boule, along with a few other meiotic cell cycle regulators, is disproportionately affected when translation is compromised in spermatocytes. Although this should be investigated, this simpler model does not explain the aberrant accumulation of Rux and the nuclear sequestration of Cyclin A that we observed.
The role of growth control in spermatid differentiation
The defects in differentiation in eIF4G2 mutants are not secondary to the meiotic block, because several cell cycle mutants fail to divide but still undergo substantial post-meiotic differentiation(Alphey et al., 1992; Eberhart et al., 1996; Eberhart and Wasserman, 1995; Sigrist et al., 1995). Several spermatid differentiation genes, such as don juan and fuzzy onions, are transcribed in primary spermatocytes under the control of spermatocyte arrest genes (White-Cooper et al., 1998). Translational control delays the accumulation of their protein products. This delay is functionally relevant, because precocious don juan accumulation leads to sterility(Hempel et al., 2006). In principle, then, the lack of significant differentiation in eIF4G2mutants could simply be due to a more pronounced translational delay for key differentiation genes. Alternatively, the block in differentiation might reflect a direct effect of the proposed growth checkpoint. Consistent with either model, the accumulation of the mitochondrial fusion protein Fuzzy onions is delayed, although we did not time this precisely (data not shown). We expect that other differentiation targets will also be abnormally delayed in eIF4G2 mutants.
Possible conservation of coupling translation initiation with meiotic progression
There are striking parallels to the role of eIF4G2 during spermatogenesis in other organisms. For instance, there are also two major isoforms of eIF4G in Caenorhabditis elegans, encoded by ifg-1. When the longest isoform was depleted from the germ line, oocytes arrested in meiosis I(B. D. Keiper, personal communication). The requirement for ifg-1 in spermatogenesis has not yet been examined. However, one of the five isoforms of eIF4E in the worm, IFE-1, is clearly essential for spermatogenesis. RNA interference against ife-1 results in delayed meiotic progression,and in defective sperm, in both hermaphrodites and males(Amiri et al., 2001). Furthermore, mouse testes carrying the Y chromosome deletion Spy(also known as Eif2s3y-Mouse Genome Informatics) have a meiotic arrest phenotype due to a lack of EIF2 (also known as EIF2S2-Mouse Genome Informatics) function (Mazeyrat et al.,2001). Taken together, these examples suggest that translational control, and therefore possibly growth control, is a common theme for meiotic cycle cells.
Acknowledgements
We thank Jean Maines for unpublished work leading to our discovery of ofs; Catherine Baker, Minx Fuller and Brett Keiper for sharing unpublished data; Brooke Middlebrooks for generating transgenic fly lines;Paul Fisher, Ken Irvine, Gary Struhl, Julie Brill, Jamila Horabin, Jose Sierra, Helen White-Cooper, Barbara Thomas, Terence Murphy and Sara Cherry for essential reagents; Bruce Edgar for helpful insights on growth control; Edmund Koundakjian and Charles Zuker for their valuable mutant collection; Eric Sebzda for providing FACS expertise; and Colin Palmer, Tim Kelliher and Kate Rochlin for technical assistance. This work was funded by a NICHD grant T32HD07516 (T.M.F.-D.) and R01 grants from the NIH (GM60804, S.D. and HD60990,S.A.W.).