Drosophila rae1 is required for male meiosis and spermatogenesis

The discovery of genes underlying regulation and execution of meiotic events is essential to gain insights into the mechanisms leading to faulty chromosome segregation which contributes to infertility and birth defects in many species including humans. Drosophila melanogaster is a particularly amenable organism to study the regulation of meiotic chromosome behavior during gametogenesis. Spermatogenesis in Drosophila, and insects in general, has been the favored material for cytological studies because of the large size of male meiocytes. An additional advantage of using D. melanogaster is the availability of large collections of mutations disrupting the process. Such mutations have been particularly informative for understanding how entry into meiosis is controlled and how its regulation is coordinated with other complex processes of cellular morphogenesis and differentiation that ultimately lead a functional sperm.

In Drosophila, each spermatogonium undergoes four mitotic divisions to generate 16 interconnected primary spermatocytes, which then enter into a long ‘growth period’ (∼90 hours) and become committed to meiosis. The exit from the mitotic cell cycle and the entry into meiosis, is accompanied by a radical change in the gene expression landscape of the primary spermatocyte (Chintapalli et al., 2007). Subsets of newly transcribed genes are required for the meiotic G2/M transition, the meiotic division execution, or to provide regulatory or structural proteins that will be used postmeiotically for spermatid differentiation. Insights into how specific proteins permit the coordination of premeiotic, meiotic and postmeiotic events have been obtained through the study of two classes of genes that differ in their influence on spermatid differentiation. In ‘meiotic arrest’ class, which includes can, mia and aly, mutant primary spermatocytes do not cross the G2/M border, meiotic divisions do not occur, and unreduced spermatids fail to differentiate, resulting in complete sterility. In the ‘twine’ class, which includes twine and Dmcdc2, mutant spermatocytes skip one or both male meiotic divisions, the unreduced spermatids still undergo some aspects of differentiation, but fail in the final steps of sperm individualization. Interestingly studies of the meiotic arrest and twine class genes have identified several proteins that are also involved in mitotic cell cycle, and are either testis-specific paralogs or testis-enriched protein isoforms produced by alternative splicing (reviewed by White-Cooper, 2010). For example, twine, encodes a testis-specific Cdc25 paralog, which regulates the activity of the Cdc2-Cyclin B complex and the entry into meiosis (Courtot et al., 1992; White-Cooper et al., 1993), similar to the mitotic role played by its paralog, string.

Here we report the characterization of a male sterile recessive mutation, Z5584 (Wakimoto et al., 2004) and provide additional insight into the regulation of male meiosis and its relationship to regulation of mitosis. We show that Z5584 results in pleiotropic defects which include disruptions of nuclear integrity in primary spermatocytes, meiotic chromosome missegregation and multiple postmeiotic defects. Molecular analyses show that Z5584 is a point mutation in the Drosophila rae1 gene. Ectopic expression of a rae1-RNAi construct in the germline resulted in male meiotic defects and expression in somatic cells indicated an effect on mitosis. Previous studies have shown that Schizosaccharomyces pombe and mouse homologues are essential genes involved in mRNA export and cell cycle progression in mitotic cells (Brown et al., 1995; Whalen et al., 1997; Babu et al., 2003). In D. melanogaster tissue culture cells, RAE1 mainly localizes to the nuclear envelope and is required for the progression through the G1 phase of the mitotic cell cycle but not for RNA export (Sitterlin, 2004). Moreover, Tian et al. recently showed that the D. melanogaster rae1 is an essential gene that regulates neural development as a component of a E3 ubiquitin ligase complex (Tian et al., 2011). Taken together, these new findings and our in vivo studies broaden the role of RAE1 in cell regulation and development. In particular, we show that Drosophila rae1 has a male germ line meiotic function that can be separated by mutation from its somatic function.

The ms(2)Z5584 strain was identified by Wakimoto et al. (Wakimoto et al., 2004) as one of over 1000 lines from the Zuker Collection (Koundakjian et al., 2004) that carried an EMS-treated homozygous viable, recessive male sterile chromosome 2. It was originally classified as defective in meiosis I or II based on the presence of large spermatid nuclei.

Since the ms(2)Z5584 chromosome was recovered from a heavily mutagenized strain, we used meiotic recombination to map the male sterile phenotype and also obtain derivatives in which the bulk of the chromosome was replaced with unmutagenized sequences. We compared recombinant derivatives which replaced either the left arm or right arm of chromosome 2. One such chromosome named Rec#57 (al⁻ dp⁻ b⁻ pr⁻ cn⁺ c⁻ ms(2) Z5584⁻ px⁺ bw⁻ sp⁺) resulted in male sterility when homozygous and resulted in the same cytological defects as observed with the original chromosome (data not shown). Conversely, male homozygotes for the reciprocal recombinant (al⁺ dp⁺ b⁺ pr⁺ cn⁻ c⁺ ms(2)Z5584⁺ px⁻ bw⁺ sp⁻) were fertile and exhibited cytologically normal spermatogenesis.

To assess the severity of the ms(2)Z5584 mutation, we compared the phenotypes of males homozygous for the mutated second chromosome and males hemizygous using the smallest available deletion identified as uncovering ms(2)Z5584 mutation (see below: mapping results). Both genotypes exhibited complete sterility and coincident meiotic phenotypes. As a whole, these observations suggest that ms(2)Z5584 second chromosome bears a single severe mutation of a gene acting during, and essential for, spermatogenesis.

To determine more precisely which stages of spermatogenesis were affected by the mutation, we first examined testis preparations of wild-type (OreR) and ms(2)Z5584 homozygous males using phase contrast microscopy. This assay of live, gently squashed tissue permitted a survey of the progression of spermatogenesis. Monitoring the structure and relative sizes of premeiotic, meiotic and postmeiotic nuclei and the spermatid mitochondrial derivative (Nebenkern) provides indicators of normal meiosis and cytokinesis, and the production of motile sperms indicate normal postmeiotic differentiation. As shown in Fig. 1, ms(2)Z5584 males show prominent defects at multiple stages of spermatogenesis. The earliest defect observed was abnormal nuclear herniations in primary spermatocytes (Fig. 1A,A′). In addition, in contrast to wild-type postmeiotic round spermatids which have uniformly sized nuclei adjacent to similarly sized Nebenkern, the ms(2)Z5584 spermatids showed wide variations of nuclear and Nebenkern sizes within and among cysts (Fig. 1B,B′). Defects in spermatid differentiation were also observed. Although the mutant produced spermatids with elongated tails, large, round spermatid nuclei failed transition to the needle shaped thin nuclei typical of wild-type spermatids (Fig. 1C,C′). Differentiation arrested prior to spermatid individualization so that the seminal vesicle, which is located at the base of the testis and normally contains motile sperms, was completely lacking mature sperm.

For a higher resolution characterization of the ms(2)Z5584 defects and to determine consequences on meiosis I versus II, we used fixed testis preparations and immunostaining to examine the distribution of several nuclear, chromosomal and cytoskeletal components at different stages of spermatogenesis. As shown in Fig. 2A, the aberrant nuclear herniations in ms(2)Z5584 young apolar spermatocytes at S3 stage [staging as referred to by Cenci et al. (Cenci et al., 1994)] contained chromatin, as revealed by DAPI staining and a distinct lamin signal detected by anti-laminDm0 immunostaining. Hence, these nuclear extensions likely contain an organized lamin network but may result from abnormal chromatin organization or distribution. As previously described by Cenci and colleagues, growth of the primary spermatocyte from early to late stages is accompanied by substantial increase in nuclear size and chromatin movement from the central positions in the nucleus to the nuclear periphery (Cenci et al., 1994). In primary spermatocytes at early prometaphase (M1a stage), the four chromosome pairs occupy distinct nuclear territories, and are detected as DAPI bright chromatin clumps (Fig. 2B). In ms(2)Z5584 mutants, chromosome pairs appear distinct and occupy separate territories in primary spermatocytes. However, in greater than 90% of prometaphase cells, we observed aberrant condensation, with chromatin clumps showing varying levels of condensed or diffuse DAPI staining. There is also evidence of smaller than normal and variably sized DAPI-positive fragments, suggestive of chromatin fragmentation (Fig. 2C).

To further explore the state of chromatin condensation, we immunostained wild-type and mutant testes preparations with an antibody raised against histone H3 phosphorylated at serine 10 (H3S10ph), a marker of meiotic-entry-associated chromosome condensation. As shown in Fig. 2D, the H3S10ph signal is distributed largely throughout the metaphase chromosomes in wild-type primary spermatocytes at M3 stage. In the mutant, H3S10ph is reduced to a small region of the chromatin.

To determine whether the mutation affected meiotic chromosome behavior and spindle organization, we investigated the meiosis progression in the mutant, analyzing stages from metaphase I to telophase I. To recognize the different stages in the mutant, we took advantage of antibodies specific for gamma-tubulin or the centriolar protein SPD-2 (Giansanti et al., 2008), as centrosome markers. In ms(2)Z5584 cells, all meiotic cysts had at least one abnormal metaphase I or anaphase I figure. Specifically, mutant cysts differed from wild type (Fig. 3A) with metaphases I cells exhibiting poorly organized chromosome configurations and undercondensed chromatin fragments lying outside of the metaphase plates (Fig. 3B), as well as abnormal meiotic spindles with centrosomes mislocated from their typical polar location (Fig. 3C). Anaphases I cells in ms(2)Z5584 mutants also showed several defects, including lagging chromosomes (Fig. 4A′), chromatin bridges (Fig. 4A″) and unbalanced chromosome segregation to the opposite poles of the dividing cell (Fig. 4A″,A‴).

Chromosome segregation defects were accompanied by anomalies in the spindle structure. Relative to those of wild type, central spindles in the mutant were markedly reduced in microtubule density and were often displaced from the spindle axis (Fig. 4B′). Misoriented microtubules were also visible in telophase spindles (Fig. 4C). Telophases I in the mutant also exhibited abnormal actomyosin ring assembly which is a key component of cytokinesis apparatus together with central spindle. We used an actin-specific Rhodamine–phalloidin fluorescent staining to detect the actomyosin ring, which appears as a thin band at the midzone in wild-type early telophase, then as a ring-shaped structure between the late telophase nuclei (Fig. 4D). In the mutant, the actomyosin ring showed variable behavior, being properly formed and constricted in some cells (Fig. 4D′, arrow) and unconstricted (Fig. 4D′, arrowhead) or absent (not shown) in other cells. Abnormalities in chromosome segregation also showed variability at cytological level. Meiosis I cysts showed occasional cells with spindles that were devoid of chromatin (Fig. 4E′). Strikingly, we did not detect meiosis II figures in ms(2)Z5584 mutant testes.

These findings prompted us to undertake a statistical analysis of the occurrence of cysts containing 16 primary spermatocytes, 32 secondary spermatocytes or 64 spermatids in wild-type and ms(2)Z5584 homozygotes. As expected, wild-type cysts containing 16, 32 or 64 nuclei were observed but cysts with intermediate numbers were not observed. In contrast, ms(2)Z5584 testes had cysts with 16 nuclei and cysts with a number of nuclei ranging between 16 and 32; no cysts with 32 or 64 nuclei were observed (Fig. 4F). These data indicate that mutant meiocytes routinely fail prior to the completion of Meiosis I and do not progress to Meiosis II.

Despite the multiple meiotic alterations described above, phase-contrast microscopy of ms(2)Z5584 mutant showed that some aspects of cellular differentiation proceed (Fig. 1B′,C′).

We used immunocytological analysis to monitor centriolar, nuclear, and actin distributions in later stages of spermatogenesis and found frequent abnormalities in differentiation. These defects included two mislocalized centrioles instead of a single normally positioned centriole in onion stage spermatids (Fig. 5A,A′), improper nuclear shaping and cyst polarization in elongating sperm bundles, and failure of the assembly of the actin-rich individualization complexes (Fig. 5B,B′), preventing the complete differentiation and release of mature gametes from the testis.

In conclusion, the ms(2)Z5584 mutation resulted in pleiotropic effects evident throughout spermatogenesis. The earliest visible defect observed was nuclear protrusions in growing primary spermatocyte and abnormalities in prometaphase chromosome condensation. Meiosis I chromosome segregation and spindle morphologies were also defective and the likely consequence of these may be a failure to complete meiosis I. One or more of the premeiotic or meiosis I defects may lead to the multiple postmeiotic anomalies.

To identify the ms(2)Z5584 gene, we first mapped the mutation using meiotic recombination to the distal right arm of chromosome 2, then tested a series of 2R deficiency chromosomes (see Materials and Methods) for their ability to complement the ms(2)Z5584 male sterility. Two overlapping deletions, Df(2R)ED3923 and Df(2R)BSC424, refined the localization to within the 57F6 cytogenetic interval and a 40.592 kb region defined by the proximal breakpoints of the two deficiencies. The most recent release of the Drosophila Genome Project (McQuilton et al., 2012) annotated 17 putative genes within this region, of which 11 had insertions that disrupted the gene. We tested these insertions in combination with ms(2)Z5584 and found that all complemented ms(2)Z5584 resulting in fertile males and suggesting that the corresponding genes were unlikely candidate genes. Four of the remaining genes, namely CG30284, rae1, CG15676 and NC2alpha, were selected as genes of higher interest on the basis of previously reported expression in the testis and/or a predicted molecular function consistent with the ms(2)Z5584 phenotype. Sequencing these four genes in the ms(2)Z5584 strain identified a point mutation in the open reading frame of only the rae1 gene. This mutation resulted in a G/C to A/T transition which was predicted to result in a glycine (G) to aspartic acid (D) substitution, at position 129 in the predicted 346 amino acid RAE1 protein.

Previous studies have shown that RAE1 is a WD40-repeat-containing protein with orthologs present in a wide variety of eukaryotes, including yeast, protozoans, plants and animals (Brown et al., 1995; Murphy et al., 1996; Bharathi et al., 1997; Pritchard et al., 1999; Sabri and Visa, 2000; Sitterlin, 2004; Lee et al., 2009). Amino acid sequence alignment of RAE1 orthologs using CLUSTALW shows that the Drosophila protein has extensive sequence identity with other family members (supplementary material Fig. S1). The functional importance of glycine 129 is suggested by its evolutionary conservation in RAE1 proteins from yeast to humans, and its location within a highly conserved 12 amino acid domain in the third putative WD40 repeat (supplementary material Fig. S1). Taken together, these studies warrant the designation of ms(2)Z5584 as rae1^Z5584.

rae1^Z5584 is the first and thus far only reported Drosophila rae1 mutation with a documented role in meiosis and spermatogenesis. The availability of strains to conditionally express double stranded rae1 RNA (RNAi) in the testis allowed us to test whether RNAi-induced phenotypes were similar to those observed in rae1^Z5584 homozygotes. To this end, rae1 dsRNA was expressed under the control of Gal4 promoter whose expression is regulated by a germline-specific Gal4 driver, BamG4UASDicer2 (see Materials and Methods). This resulted in male sterility with testes defects identical to those observed in rae1^Z5584 homozygotes (Fig. 6). In particular, nuclear herniations in primary spermatocytes were clearly apparent (Fig. 6A) while meiotic division defects both in vivo (Fig. 6B) and in fixed preparations (Fig. 6D,E) strongly resembled those of rae1^Z5584 homozygotes (compare irregular metaphase plates in Fig. 6E,E′ and those in Fig. 3B; compare lagging chromosomes in Fig. 6D and those in Fig. 4A′–A‴). These results are consistent with our gene assignment and our conclusion of rae1^Z5584 requirement for normal nuclear organization in primary spermatocytes and completion of Meiosis I. In D. melanogaster, a Rae1 homolog, CG12782, exists and maps on the second chromosome right arm. We tested two deficiencies that uncovered CG12782, each in trans-heterozygous condition with rae1^Z5584, but we did not detect an effect on male fertility. This does not exclude the possibility of interactions between the two homolog genes, that might be revealed in more sensitive assays. We did note that the highly conserved RAE Gly 129 which our mutation shows is critical for male meiosis, is not present in the CG12782 predicted protein.

To assess the localization of RAE1 in meiotic cells, we took advantage of a D. melanogaster transgenic strain carrying a UAS-GFP-rae1 transgene (Tian et al., 2011). To drive the transgene expression in the testis, we combined it with the germline-specific BamG4UASDicer2 and the ubiquitous tubulinGal4 drivers. Both the drivers gave the same GFP–RAE1 distribution pattern in meiotic cells, but the Bam Gal4 driver did not yield a detectable GFP signal in pre and postmeiotic cells. Therefore, the ensuing description refers to results obtained with the constitutive tubulinGal4 driver. In intact testes, the fusion protein GFP signal was apparent in all the cellular stages from spermatogonia to sperms. In spermatogonia (not shown) and in young primary spermatocytes (S1 to S4 stages) (Cenci et al., 1994) GFP–RAE1 appeared as a distinct rim at the nuclear envelope (Fig. 7A), similar to that observed in D. melanogaster SL2 culture cells transfected with a HA-tagged RAE1 (Sitterlin, 2004). In S1–S4 spermatocytes, the chromatin appeared devoid of GFP–RAE1. As the primary spermatocytes progressed toward late prophase stages, GFP–RAE1 localized with the three chromatin clumps (Fig. 7B, arrows) and cytoplasmic GFP–RAE1 was also apparent (Fig. 7B). In anaphase I and telophase I, the GFP–RAE1 coincided with segregating chromatin and with the DAPI-positive mitochondria that localized between the two daughter nuclei (Fig. 7C). In postmeiotic onion stage spermatids, GFP–RAE1 appeared predominantly associated with Nebenkern (Fig. 7D), with little or no nuclear signal detected. In mature sperms, GFP–RAE1 appeared associated with both nuclei and tails (Fig. 7E).

To assess if the GFP–RAE1 protein could substitute for the wild-type RAE1 we challenged the chimeric construct to rescue the rae1^Z5584 homozygous male sterility (see supplementary material Fig. S2 for details). We found that the rae1 transgene was unable to restore the male fertility. However, cytological analysis of the GFP-rae1-expressing, mutant males revealed partial rescue of at least some of the meiotic defects. In particular, young primary spermatocytes appeared normal and lacked the characteristic nuclear herniations that were commonly observed in the mutant (supplementary material Fig. S3A). Moreover, chromosome segregation appeared normal at metaphase I (data not shown) and onion stage spermatids exhibited the 1∶1 nuclear to Nebenkern ratio typical of wild-type cells, although some nuclei and Nebenkerns appeared to have irregular shapes (supplementary material Fig. S3B,C). Spermatid elongation was evident but cysts showed nuclear misalignment typical of the rae1^Z5584. However, nuclear morphology appeared to be intermediate between the round uncondensed spermatid nuclei of the mutant and the highly condensed sperm head of normal sperm (supplementary material Fig. S3D).

To determine the extent to which GFP–RAE1 supported the completion of the meiotic divisions, we asked whether males that were homozygous for rae1^Z5584 and expressing the GFP–RAE1 transgene had cysts containing 16, 32, 64 or an intermediate number of germ cells. We performed nuclear counts within intact cysts of two males. In contrast to rae1^Z5584 homozygotes (Fig. 4F), the transgenic males showed some 32 and 64 cell cysts, indicating that GFP–RAE1 is capable of supporting the progression and completion of meiosis I and II, although variably so. The apparent absence of motile sperm in these males show expression of the GFP–RAE1 is not able to rescue postmeiotic differentiation.

To investigate a possible role of Drosophila rae1 in mitosis, we asked if the rae1^Z5584 mutation affected the mitotic parameters of neuroblasts in the central nervous system (CNS) in third instar larvae. We determined a mitotic index for this tissue for rae1⁺ control (Ore-R) larvae (see Materials and Methods) and compared this to indices for rae1^Z5584 homozygotes and hemizygotes. We also counted the number of anaphase configurations for each genotype. As summarized in Table 1, rae1^Z5584 hemizygotes showed a statistically significant higher mitotic index and lower frequency of anaphases relative to the control. rae1^Z5584 homozygotes show instead only a decrease of anaphase figures. In this context it should be noted that neither the rae1^Z5584 homozygotes nor the hemizygotes showed reduced viability.

As a complementary approach, we expressed rae1 RNAi under the control of the neuroblast-specific driver wor-GAL4 (Cabernard and Doe, 2009), and examined the consequences. The mitotic index in the CNS of third instar larvae expressing the rae1 RNAi was higher than controls (Table 1) but the frequency of anaphase configurations was not statistically different from controls. However, the majority of anaphase figures (70%) appeared defective in rae1 interfered larval brains (Table 1). The defects included lagging chromosomes, chromatin bridges and uneven chromosome segregation (not shown). Interestingly, in colchicine-treated, Rae1-depleted larval brains, metaphase chromosomes appeared undercondensed (Fig. 8). Thus, reduction of RAE1 induces a detectable effect on the mitotic cell cycle and mitotic chromosome morphology. The developmental consequences of these defects are not known, but we note that we did not detect an effect on viability.

Our discovery that a male sterile mutation, ms(2)Z5584, resides in the D. melanogaster ortholog of the rae1 gene has implications for understanding the function of the RAE1 protein. The protein is a WD40-repeat β propeller protein, with four evolutionarily conserved WD repeats. A wide variety of functions have been identified for the members of the WD40-repeat protein superfamily, leading to the proposal the common property is the ability to assemble multiple proteins in large macromolecules (Neer et al., 1994). rae1 is evolutionarily highly conserved from yeast to mammals (Reddy et al., 2008), and it was first identified in the fission yeast Schizosaccharomyces pombe as a nucleoporin involved in poly(A)⁺ mRNA export and cell cycle progression (Brown et al., 1995; Whalen et al., 1997). The interaction between RAE1 and the nucleoporin NUP98 was demonstrated in human cells (Pritchard et al., 1999) and aberrant poly(A)⁺ mRNA accumulation in the nucleus was documented in rae1 (gle2) mutants in Saccharomyces cerevisiae (Murphy et al., 1996). However, subsequent studies revealed pleiotropic effects of rae1 in various organisms, indicating a broader function for the protein in mitosis. For example, in human cells, RAE1 interacts with the mitotic checkpoint protein BUB1 (Wang et al., 2001) and with the mitotic apparatus protein NUMA (Wong et al., 2006), suggesting a role for rae1 as mitotic regulator. Moreover, a direct binding of RAE1 to securin, blocking its degradation and thus regulating the metaphase to anaphase transition was reported (Jeganathan et al., 2005; Jeganathan et al., 2006).

In D. melanogaster, rae1 was investigated in SL2 culture cells by dsRNA interference by Sitterlin who showed that this resulted in cell cycle arrest at G1/S transition, but neither mitotic anomalies nor mRNA accumulation in the nucleus were observed (Sitterlin, 2004). Very recently, Tian et al. clearly demonstrated that in Drosophila, RAE1 is involved in the regulation of neural development (Tian et al., 2011). Using a combination of genetic and biochemical analyses, these authors elegantly showed that RAE1 positively regulates the abundance of the E3 ubiquitin ligase Highwire, which in turn restrains the growth of synapses at the neuromuscular junctions (Tian et al., 2011).

Our cytological and molecular characterization of the ms(2)Z5584 mutation provides the first evidence of an in vivo role of rae1 in the meiotic cell cycle. The specific mutation we discovered is a glycine to aspartic acid substitution. This glycine is an invariant amino acid within a 12 amino acid domain which is highly conserved in RAE1 hortologs in a wide range of species from yeast to humans. By mobilizing a transposable P-element in D. melanogaster, Tian et al. obtained two rae1 imprecise excision alleles which were homozygous lethal at the second instar larval stage (Tian et al., 2011). In the light of these results and the lethality of rae1 null mice (Babu et al., 2003), and given the well-established role of rae1 in mitosis in many species, it is rather surprising that rae1^Z5584 homozygous flies are fully viable with no effect on the mitotic index of larval neuroblasts. Hemizygous individuals, however, even though fully viable, show an increased mitotic index in the CNS. We also observed a reduction in the number of anaphase configuration which is indicative of a mitotic delay. In this context, it is notable that rae1 haploinsufficiency produces mitotic anomalies in mammals (Babu et al., 2003). In accordance with the results in other organisms, the CNS-specific expression of rae1 dsRNA produced mitotic anomalies in neuroblast cells. Because the main effect of decreased RAE1 in dividing cells is cell cycle arrest at the G1/S transition, the apparent absence of mitotic anomalies in D. melanogaster SL2 culture cells following rae1 dsRNA interference (Sitterlin, 2004) might have been due to the fact that successfully interfered cells arrest prior to the G1/M transition so do not enter mitosis. Taken together, studies using Drosophila and those using other organisms provide a consistent view of RAE1 as a conserved mitotic cell cycle regulator (Brown et al., 1995; Whalen et al., 1997; Wang et al., 2001; Wong et al., 2006; Sitterlin, 2004; Jeganathan et al., 2006). Our studies of rae1^Z5584 are the first to suggest a meiotic role. We do not yet know if the meiotic role of RAE1 is limited to Drosophila males or if it is also important for Drosophila females and meiosis in other organisms.

The multiple defects we observed in ms(2)Z5584 mutant males affected both early meiotic stages and late spermiogenesis. The earliest defects we observed in mutant primary spermatocytes are nuclear envelope herniations, which resemble those found in gle-1 (the rae1 ortholog) mutants in S. cerevisiae (Murphy et al., 1996). This observation is consistent with a localization of RAE1 to the nuclear envelope in wild-type spermatogonia and primary spermatocytes as suggested by our studies of the GFP–RAE1 fusion protein and with the pattern of HA-tagged RAE1 in D. melanogaster SL2 culture cells (Sitterlin, 2004). Primary spermatocytes lacking RAE1 showed nuclear herniations. Notably the nuclear lamin appeared continuous even in these herniations. Most consistently we observed that the mutant showed defective meiotic chromatin condensation which accompanied by the nearly total loss of the chromosome condensation marker, histone H3 serine 10 phosphorylation. We may thus hypothesize that RAE1 is necessary at nuclear envelope in primary spermatocytes to direct meiotic chromatin organization and condensation, whose alteration may explain the ensuing meiotic anomalies. In this context it is notable that the defective chromosome condensation is also a distinctive trait of rae1-interfered somatic cells. Interestingly, mutations in dTopors, the Drosophila homolog of the mammalian dual ubiquitin/small ubiquitin-related modifier (SUMO) ligase Topors result in male sterility and spermatocyte nuclear membrane blebbing similar to ms(2)Z5584 mutants. Remarkably, dTopors mutants also exhibit meiosis I defects, such as chromosome missegregation and anaphase I bridges (Matsui et al., 2011). These results suggest that rae1 and dTopors genes may be involved in the same pathway. A demonstration of a genetic and physical interaction of RAE1 and dTOPORS in male germ cells would be consistent with results of Tian et al. showing that RAE1 physically interacts with the E3 ubiquitin ligase Highwire in the CNS (Tian et al., 2011). The analysis of GFP–tagged RAE1 distribution in wild-type meiosis I cells indicated co-localization of RAE1 with chromatin starting from late prophase primary spermatocytes until ana/telophase I nuclei, thus suggesting that the role of RAE1 in chromatin condensation may be more direct. In meiosis, bivalents need to reach a correct condensation state during late G2/prophase I to ensure a faithful chromosome partitioning to daughter cells, through the resolution of homologous associations, necessary to prevent chromatin bridges (Hartl et al., 2008). Hence, defective chromosome condensation could explain the lagging chromosomes and anaphase bridges, culminating in an uneven chromosome segregation to the opposite cell poles of the first meiotic division. Concomitantly, we detected also several defects affecting meiotic spindle and centrosomes. Metaphase I figures are reminiscent of the abnormal, multipolar spindles associated with multiple pericentrin foci, which cause defects of mitotic chromosome alignment and segregation in rae1 dsRNA-treated HeLa cells (Blower et al., 2005). The involvement of RAE1 in mitotic spindle assembly has been previously proposed by the identification of RAE1 association with NuMA (nuclear mitotic apparatus) (Wong et al., 2006). This complex associates with dynein at spindle poles, and is crucial for spindle bipolarity in HeLa cells, where the interaction between RAE1 and the cohesin subunit SMC1 was also reported, which is required for proper spindle formation (Wong et al., 2006; Nakano et al., 2010; Wong, 2010). Moreover, a disorganization of tubulin and actin networks was also documented in S. pombe rae1-1 mutant cells (Brown et al., 1995). Furthermore, in ms(2)Z5584 mutant males we also observed ana/telophase I central spindle anomalies. The mitochondria are uniformly distributed within the cytoplasm of late prophase I cells and associate with the central spindle at anaphase (Fuller, 1993; Cenci et al., 1994). Our cytological preparations were particularly helpful to reveal the behavior and distribution of mitochondria by DAPI staining. In this context it is remarkable that in wild type, GFP-tagged RAE1 localized with mitochondria distributed between ana/telophase I daughter nuclei where the central spindle is assembled. ms(2)Z5584 male meiosis thus showed both karyokinesis and cytokinesis defects that account for the presence of spermatids at onion stage with nuclei and Nebenkern of variable size. ms(2)Z5584 non-reduced spermatid exhibited two mispositioned centrioles as opposite to wild-type spermatids where a single centriole, juxtaposed to the nucleus, will become the basal body. This latter will then play a central role in nuclear shaping and cyst polarization (Fuller, 1993). Consistently with this scenario, in ms(2)Z5584 late spermatids the nuclear transition from a rounded to a needle-like shape is impaired, and cyst polarization does not occur. Nevertheless, in ms(2)Z5584 mutants, unreduced spermatids that exit from the defective meiotic divisions proceed to accomplish some differentiation including axoneme assembly and elongation. This behavior is quite similar to that of the Twine Class of male sterile mutants. The genes of this class, such as Dmcdc2 and twine itself, have been shown to preside the entry into meiosis, regulating the activity of Cdc2–Cyclin B complex, underlying M-phase progression. Their null mutants skip both meiotic divisions; nonetheless they initiate and enact some steps of differentiation (Alphey et al., 1992; Courtot et al., 1992; White-Cooper et al., 1993; Maines and Wasserman, 1999; Sigrist et al., 1995). ms(2)Z5584 mutants enter first meiotic division, as testified by the presence of, albeit irregular, metaphase I and anaphase I figures. This indicates that rae1 acts downstream of the meiotic entry. In rae1^Z5584 mutants, secondary spermatocytes are formed with unbalanced chromosome distribution and even devoid of chromosomes. Nevertheless, mutant secondary spermatocytes fail to execute the second meiotic division. This is rather surprising because of the lack of an efficacious spindle checkpoint in Drosophila spermatocytes (Basu et al., 1999; Rebollo et al., 2004), and the behavior of fusolo and solofuso mutants, in which secondary spermatocytes, although devoid of chromatin, execute a second meiotic division (Bucciarelli et al., 2003). This would imply that RAE1 plays a role in the second meiotic division execution too, as suggested also by the colocalization of RAE1 with the ana/telophase I chromatin and with mitochondria at the central spindle in wild type, that indicates an accurate partitioning between the two secondary spermatocytes.

The persistence of GFP-tagged RAE1 in wild-type postmeiotic cells may reflect a role for RAE1 in spermatid differentiation, consistent with the observed defects of rae1^Z5584 homozygotes at these stages. Interestingly, we noted a difference between the fully round spermatid nuclear phenotype in the mutant and the partially condensed spermatid nuclei induced with GFP–RAE1 expression. This may reflect partial rescue and a direct role of RAE1 in spermatid chromatin condensation. We do not know if the inability to achieve complete condensation is due to abnormal function of the chimeric protein or to its abnormal regulation since the construct did not use the rae1 promoter. Regardless, the failure of spermatid with abnormal chromatin condensation to initiate individualization is not surprising since this is a common mutant phenotype (Wakimoto et al., 2004) and it accounts for the inability of the GFP-rae1 transgene to restore fertility to rae1^Z558 mutants.

The documented interaction of RAE1 with a E3 ubiquitin ligase (Tian et al., 2011) may account for an indirect role of RAE1 on chromatin dynamics via the ubiquitin pathway. This interaction may also account for the pleiotropic defects seen in ms(2)Z5584 spermatogenesis. The proposed involvement of RAE1 with nucleoporins has been strengthened in recent years, with its inclusion in a wide variety of molecular networks, embracing processes that also include chromosome condensation, kinetochore assembly and spindle formation during mitosis (Nakano et al., 2010). One possibility is that there is a novel and unexpected meiosis-specific function of RAE1 as a nucleoporin. We can hypothesize a model in which NPC proteins such as RAE1 are required at nuclear pores in meiotic prophase for docking chromosomes to the nuclear envelope and promoting proper chromosome condensation, as essential prerequisite for a correct chromosome segregation. Accordingly, rae1 loss of function would cause both the chromosome condensation and segregation defects which are distinctive of ms(2)Z5584 mutant 1st meiotic division. Moreover, the ms(2)Z5584 mutant meiotic spindle organization and behavior is also damaged, in agreement with the reported RAE1 role in mitosis (Wong et al., 2006; Nakano et al., 2010; Wong, 2010). The severe postmeiotic defects of ms(2)Z5584 may be the results of the lack of an efficacious spindle checkpoint in Drosophila spermatocytes, able to arrest the system in case of meiotic anomalies (Basu et al., 1999; Rebollo et al., 2004), and of a possible role for RAE1 in postmeiotic spermatid differentiation stages. This is exemplified by the pleiotropic effects of Drosophila meiotic mutants like, for example, asunder (Anderson et al., 2009) and larp (Ichihara et al., 2007).

In conclusion, our data strongly indicate an unanticipated involvement of RAE1 in meiosis and a requirement for RAE1 function for postmeiotic differentiation of the male germ line in Drosophila. The possibility of a conserved function for RAE1 in meiosis and male fertility provides new avenues for studying how the protein acts at the cellular and mechanistic levels in two different types of cell division.

The ms(2)Z5584 line was provided by Koundakjian et al. (Koundakjian et al., 2004). It was identified from the Zuker collection in a screen for ethyl methanesulfonate (EMS)-induced recessive male sterile mutations and classified as defective in male meiosis by Wakimoto et al. (Wakimoto et al., 2004). The mutation was induced on a second chromosome marked with cn and bw, kept in stock with the balancer CyO, Cy cn². We used the following GAL4 driver stocks: BamG4UASDicer (kindly provided by M. Fuller, Stanford University); and w; wor-Gal4,UAS-cherry::Zeus,w+/CyO;Dr/TM6B,Tb,Hu,e (kindly provided by C. Doe, University of Oregon); and tubulinGal4/TM6B;Sb. The UAS-GFP-rae1/TM6B;Sb line was kindly provided by Chunlai Wu (Neuroscience Center, Louisiana State University, Health Sciences Center). The rae1 RNA-interfered line was obtained from the VDRC (Dietzl et al., 2007). All other stocks were obtained from the BDSC at Indiana University.

Flies were raised on standard Drosophila medium at 25°C, except for crosses involving BamG4UASDicer which required growth at 29°C. For assays of male fertility, crosses generally involved one to three males and three to five females in a single cross. Those that yielded no progeny in two separate trials were classified sterile.

The ms(2)Z5584 male sterility was mapped by meiotic recombination relative to the recessive al, dp, b, pr, c, px and sp markers and tests involving 100 recombinant chromosomes indicated a map position of ∼96.2 m.u. on the right arm of chromosome 2. Two recombinant chromosomes were maintained in stocks over the CyO, or T(2;3)Cy; Tb balancers for use in subsequent experiments. One chromosome was al⁻ dp⁻ b⁻ pr⁻ cn⁺ c⁻ px⁺ bw⁻ sp⁺, resulted in sterility and was denoted Rec#57 ms(2)Z5584. A reciprocally marked chromosome al⁺ dp⁺ b⁺ pr⁺ cn⁻ c⁺ s⁺ px⁻ bw⁺ sp⁻ resulted in fertility. These chromosomes were tested in homozygotes to verify that the observed spermatogenesis defects were due to solely to the region to which the male sterility mapped. Mapping with deficiency chromosomes allowed us to refine the localization of ms(2)Z5584 to the 57F6 cytogenetic interval, within the 40.592 kb region defined by the proximal breakpoints of Df(2R)ED3923 and Df(2R)BSC424. Complementation crosses were also performed between ms(2)Z5584/CyO and strains that contained insertions proposed to affect the following genes located in this region: CG10082, CG10320, pirk, CG42362, CG42379/CG42380/CG42381/CG9865, CG30263, CG10306 and Tbp. In all of cases, the insertion mutation complemented the ms(2)Z5584 sterility.

Testes from ms(2)Z5584 mutant males or control males (OreR) were dissected within 1 day of eclosion in testis isolation buffer, TIB (183 mM KCl, 47 mM NaCl, 10 mM Tris pH 6.8). For observations of unfixed samples, testes were placed in 10 µl of buffer on a slide, overlaid with a coverslip, and observed using phase contrast optics on an Zeiss Axiophot microscope. To preserve whole cysts, testes were placed on a slide under a siliconized coverslip, the slide frozen in liquid nitrogen, then the coverslip removed with a razor blade, before the tissue was dehydrated in cold ethanol, then fixed for 7 minutes in 2% paraformaldehyde in 1×phosphate-buffered saline (PBS). Fixation was followed by a 5-minute PBS rinse, 10-minutes staining with 2 µg/ml DAPI (4,6-diamidino-2-phenylindole; Boehringer, Mannheim) in 2× SSC (0.3 M NaCl, 0.03 M sodium citrate) and mounting in anti-fade medium (DABCO, Sigma), to count the number of nuclei/cyst. Nuclear numbers from a minimum of 100 cysts of each genotype were compared statistically using a χ²-test.

For immunofluorescence analyses, slides of gently squashed, fixed testis were prepared. For lamin, α-tubulin, Spd2 or γ-tubulin immunostaining, testes were fixed with cold methanol for 7 minutes and permeabilized in 1× PBS containing 0.1% Tween 20 for 10 minutes. The best fixation procedure for immunostaining with anti-H3 phosphorylated at Ser10 was 3.7% paraformaldehyde in 1× PBS, followed by treatment with 45% acetic acid for 30 seconds and the squashing in 60% acetic acid as described by Giansanti et al. (Giansanti et al., 1999). Preparation of meiotic chromosomes from adult testes was performed by fixing with 45% acetic acid for 1 minute, and gently squashing under a coverslip, followed by freezing in liquid nitrogen. In all testis preparations, slides were incubated in a moist chamber for 1 hour at room temperature with the following antibody dilutions in 1× PBS: 1∶100 mouse anti-LaminDm0 monoclonal (isolated by P. A. Fisher and obtained from Developmental Studies Hybridoma Bank, Iowa City, IA), 1∶50 mouse anti-α-tubulin monoclonal (Sigma-Aldrich, St Louis, MO), 1∶80 rabbit anti-DSpd2 [kindly provided by S. Bonaccorsi, University of Roma ‘Sapienza’ (Giansanti et al., 2008)], 1∶1000 GTU-88 mouse anti-γ-tubulin monoclonal (Sigma-Aldrich, St Louis, MO) and 1∶100 mouse anti-H3 phosphorylated at Ser10 (provided by S. Bonaccorsi, University of Roma ‘Sapienza’). To detect primary antibodies, slides were incubated for 1 hour at room temperature in a moist chamber with the secondary antibodies, either Alexa-Fluor-488-conjugated goat anti-mouse IgG or Alexa-Fluor-594-conjugated goat anti-rabbit IgG (Molecular Probes, Eugene, OR), diluted according to the supplier's instructions. To visualize actin staining by phalloidin fluorescence, tissue was squashed on slides, frozen in liquid nitrogen, dehydrated in cold ethanol for 10 minutes, fixed in 3.7% paraformaldehyde and 1× PBS for 7 minutes, rehydrated in 1× PBS, incubated with blocking solution (1× PBS containing 0.1% Tween 20 and 3% BSA) for 30 minutes, then incubated for 2 hours at 37°C with 100 nM Rhodamine–phalloidin (Cytoskeleton, Inc., Denver, CO). Nuclei were stained with DAPI as described above. Testis preparations from a minimum of 10 control and 10 mutant individuals were examined for each assay.

The distribution pattern of GFP-tagged RAE1 in wild-type male meiotic cells was analyzed in the Sb⁺ male progeny of the crosses between UAS-rae1-GFP/TM6B,Sb individuals and BamG4UASDicer homozygous flies; and between UAS-rae1-GFP/TM6B,Sb individuals and tubulinG4/TM6B;Sb flies. Testis preparations from Sb⁺ males were obtained as described above.

Fluorescence and phase contrast images were observed with a Zeiss Axiophot fluorescence microscope equipped with a 100-W mercury light source and filter combinations suitable for the different fluorochromes (Chroma Technology Corp., Rockingham, VT, USA). Images were captured with a CCD camera (Series 200; Photometrics, Tucson, AZ) using IPLab software (Signal Analytics Corp., Vienna, VA) and processed using Adobe Photoshop (Adobe Systems, Mountain View, CA). Confocal images were captured by Zeiss LSM710 confocal microscope.

To analyze mitotic parameters, the central nervous system of third instar larvae were dissected and prepared essentially as described by Gatti and Pimpinelli (Gatti and Pimpinelli, 1983) with DAPI staining to detect chromatin and with or without colchicine treatment as described in the results. Ore-R larvae were used as the control. Larvae hemizygous for ms(2)Z5584 were selected as Tb+ offspring from a cross of Rec#57 ms(2)Z5584/T(2;3)Cy; Tb females and Df(2R)ED3923/T(2;3)Cy Tb males. To analyze mitotic parameters of RNA-interfered strains, larvae bearing the UAS-RNAi rae1 in combination with the Gal4 driver wor-Gal4 were selected. Mitotic index was defined as the number of mitotic cells per optical field. Among the mitotic cells, we also classified phases as prophase/metaphase and anaphase. The optical field was the circular area defined by a 100× Zeiss objective/1.30 Plan-NEOFLUAR, using 10× oculars and the Optovar set at 1.25. Every field occupied by brain tissue on the slide was scored and at least three slides were examined for each genotype. Frequencies of mitotic figures for mutant and control genotypes were compared statistically using χ²-test.

To identify the ms(2)Z5584 molecular lesion, genomic DNA was extracted from 30 adult males, and polymerase chain reaction (PCR) and primers (supplementary material Table S1) were used to amplify coding regions and 5′ upstream and 3′ downstream regions of specific genes (CG30284, rae1, CG15676 and NC2alpha). PCR fragments were sequenced by Eurofins MWG Operon Facility (Germany) and sequences from the ms(2)Z5584 chromosome were compared to those we obtained from control strains [ms(2)Z3162 and ms(2)Z1395 from the Zuker Collection for cn bw sequences and OreR for wild type] and to the y; cn bw sp stain, available from the Drosophila Genome Project (Release 5.3, http://flybase.org). The WD40-repeats of the predicted RAE1 protein were identified by SMART (Simple Modular Architecture Research Tool) motif analysis tool (http://smart.embl-heidelberg.de). The predicted RAE1 protein sequences were aligned using CLUSTALW (1.81) (http://www.genome.jp/tools/clustalw) and the following sequences for Drosophila melanogaster (no. AAF46745), Mus musculus (no. AAH59051), Homo sapiens (no. AAC28126), Caenorhabditis elegans (no. Q93454), Schizosaccharomyces pombe (no. CAA16856) and Saccharomyces cerevisiae (no. AAB64662).

Transgenic males homozygous for the ms(2)Z5584 mutation and containing a third chromosome harboring both tubulin-G4 and UAS-GFP-rae1 were obtained (supplementary material Fig. S2). Fertility of these males were tested in matings to virgin y w females. These males were also used for cytological studies to compare features of spermatogenesis with those of wild-type and ms(2)Z5584 mutant males and to monitor localization of GFP–Rae1 protein. In total, we tested 17 rae1^Z5584 homozygous males for fertility, half of which, in theory, should have been transgenic (supplementary material Fig. S2). None of these males proved fertile. Eight of these males were investigated at confocal microscope for the cytological phenotype and two of them were GFP positive.

We are deeply indebted to Chris Doe, University of Oregon, the laboratory of Margaret Fuller, Stanford University, and Chunlai Wu, Louisiana State University, for generously providing fly strains. We thank Maurizio Gatti, Silvia Bonaccorsi (University Sapienza, Rome) and members of their laboratories for valuable discussions and Luca Proietti De Santis (University of Tuscia) for his valuable advice on confocal microscopy. Confocal analysis was performed at the Major Research Equipment Center (CGA) of the University of Tuscia.