Nup154, a New Drosophila Gene Essential for Male and Female Gametogenesis Is Related to the Nup155 Vertebrate Nucleoporin Gene

Drosophila stocks were raised at 25°C on a standard sucrose-corn-meal-yeast medium. The l(2)01501 strain was provided by Dr. U. Schaefer (Max-Planck-Institut für biophysikalische Chemie, Gottingen, Germany).

Standard techniques were used for RNA and DNA manipulations, genomic and cDNA library screening (Chomczynski and Sacchi, 1987; Sambrook et al., 1989). Subclones were generated by insertion of DNA restriction fragments, or PCR-amplified products, into appropriate site(s) of the pGem3 vector.

A genomic library constructed by partial Sau3A digestion of Oregon R genomic DNA and ligation into the BamHI site of the EMBL3 vector (Frischauf et al., 1983) was provided by Dr. V. Pirrotta (Department of Zoology, University of Geneva, Geneva, Switzerland). An embryonic cDNA library in λgt10 (Poole et al., 1985) was provided by Dr. T. Kornberg (Department of Biochemistry and Biophysics, University of California, San Francisco, CA). Nucleotide sequencing was performed using synthetic oligonucleotide primers and the Thermo Sequenase–radiolabeled terminator cycle sequencing kit (Amersham International, Little Chalfont, UK) according to manufacturer's instructions.

In situ hybridizations with digoxigenin-labeled (Boehringer Mannheim GmbH, Mannheim, Germany) probes on whole-mount ovaries were performed as described (Tautz and Pfeifle, 1989), with minor modifications.

Eggs were collected at 24°C on small agar plates. After two 20-min pre-collections, eggs were collected at different intervals to obtain meiotic stages and first mitotic divisions. Eggs were dechorionated in 50% bleach, washed in distilled water, and then devitellinized as described (Warn and Warn, 1986), except for a final fixation step with acetone at −20°C for 5 min.

Cuticle preparation was performed as described (Schupbach and Wieschaus, 1986). Eggs were mounted in Hoyers/lactic acid 1:1 and cleared at 65°C.

To generate anti-Nup154 antibodies, a Nup154 cDNA fragment coding for amino acids 253–645 of the predicted Nup154 protein was cloned in the bacterial expression plasmid pQE-31, fused in frame with its 6× his tag coding region (QIAExpress; QIAGEN Inc., Valencia, CA). The recombinant protein was expressed in the Escherichia coli bacterial strain M15(pREP4) and affinity purified over a NiNTA column according to manufacturer's instructions. Polyclonal antibodies directed against the fusion protein were produced in rabbit and affinity purified using the antigen immobilized on nitrocellulose filters. For Western blot analyses the proteins (35 μg/lane) were separated on 10% SDS–polyacrylamide gels (Laemmli, 1970) using a mini-PROTEAN II gel apparatus (Bio-Rad Laboratories, Hercules, CA). Proteins were transferred onto nitrocellulose filters in transfer buffer (25 mM Tris [Sigma Chemical Co., Poole, UK], 192 mM Glycine [Sigma Chemical Co.], and 20% methanol) at a constant voltage (100 V) for 1 h at 4°C using a miniature trans-blot module (Bio-Rad Laboratories). Membranes were blocked for 16 h in blocking buffer (10 mM Tris-HCl, pH 8, 150 mM NaCl, 5% nonfat dry milk, 0.05% Triton X-100). Anti-Nup154 antibodies were used at 1:100 dilution in blocking buffer for 2 h at room temperature. After washing (three times for 10 min in 10 mM Tris-HCl, pH 8, 150 mM NaCl, 0.05% Triton X-100) immunoblots were incubated with HRP-conjugated goat anti–rabbit antibodies (Bio-Rad Laboratories) (1:16,000 in 10 mM Tris-HCl, pH 8, 150 mM NaCl, 0.05% Triton X-100, 1 h at room temperature), washed as above, and then incubated in ECL+Plus (Amersham International) detection reagents according to vendor instructions.

Fixation and antibody staining of hand-dissected ovaries were carried out as previously described (Bopp et al., 1993). Primary anti-Nup154 antibodies were used at 1:20 dilution and detected using FITC-conjugated sheep anti–rabbit IgG secondary antibodies (Boehringer Mannheim GmbH) at 1:100 dilution. mAb414 (Berkeley Antibody Co., Richmond, CA) was used at 1:10,000 dilution and detected using rhodamine-conjugated goat anti–mouse IgG secondary antibodies (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) at 1:200 dilution. DNA staining was performed by 5-min incubation in PBS containing 1 μg/ml DAPI (4′,6-diamidino-2-phenylindole), followed by extensive washes in PBS. Stained egg chambers were mounted in Aquamount and analyzed with conventional epifluorescence or with a Zeiss laser confocal microscope attached to a Zeiss Axiofot microscope (Carl Zeiss, Oberkochen, Germany). The images were processed in Photoshop 3.05 (Adobe Systems Inc., Mountain View, CA).

Embryo staining with Hoechst 33258 and antibodies against β-tubulin was carried out as previously described (Riparbelli and Callaini, 1996).

Testes from larvae, pupae, and adult flies were dissected in PBS, placed in a small drop (0.5 μl) of 5% glycerol in PBS on a glass slide and covered with a coverslip. The slides were frozen on a copper bar precooled in liquid nitrogen or directly in liquid nitrogen, and after removal of the coverslip with a razor blade, were immersed in methanol for 10 min at −20°C and in acetone for 5 min at −20°C. γ-Tubulin and cyclin B were detected using 1:100 dilutions of Rbcs1 or Rb271 antibodies (provided by W.G.F. Whitfield, University of Dundee, Dundee, UK), respectively. Incubations with both antibodies were carried out overnight at 4°C. In double-staining experiments, β-tubulin was detected by subsequent incubation of the samples with anti–β-tubulin mAb (Boehringer Mannheim Corp., Indianapolis, IN), used at 1:200 dilution, for 1 h at room temperature. Nuclear envelope was identified with an mAb raised against human fibroblast vimentin (Biocine, Siena, Italy) that was previously shown to recognize nuclear membrane in early Drosophila embryos (Riparbelli and Callaini, 1992). The appropriate secondary antibodies, either goat anti–mouse or goat anti–rabbit IgG (Cappel Laboratories, Malvern, PA) coupled with FITC or TRITC, were used at 1:600 dilution and incubated for 1 h at room temperature. DNA was stained by 3–4 min incubation in PBS containing 1 μg/ ml Hoechst 33258. Samples were mounted on glass microscope slides in 90% glycerol containing 2.5% n-propyl gallate (Giloh and Sedat, 1982). Fluorescence observations were made on a Leitz Aristoplan microscope equipped with FITC, TRITC, and UV filters.

Testes from pupae and adult flies were fixed in trialdehyde solution (Kalt and Tandler, 1971) for 2 h at room temperature or overnight at 4°C. After rinsing in 0.1 M cacodylate buffer, pH 7.2, samples were postfixed in 1% osmium tetroxide for 2–3 h, bulk stained in 1% uranyl acetate in distilled water, dehydrated in a graded series of ethanol, treated with propylene oxide, embedded in Epon-Araldite mixture, and then polymerized at 60°C for 48 h. Random and serial sections were cut using an LKB Nova ultramicrotome and a diamond knife. Sections were collected on copper grids, stained with uranyl acetate and lead citrate, and then observed and photographed with a Philips CM10 electron microscope (Philips Electron Optics, Mahwah, NJ).

The cDNA sequence data reported in this paper are available from GenBank/EMBL/DDBJ under accession number Y17111.

In a large screen for P element-induced mutations in region 32 of the second chromosome, several female-sterile and semi-sterile lines were isolated. Complementation tests indicated that two of them identified the same gene and were allelic to l(2)01501, a lethal P insertion in 32D1 band (Karpen and Spradling, 1992). We called the mutation tulipano (tlp) and the alleles we isolated tlp1 and tlp2, because of the peculiar truncated aspect of the mutant eggs. We later renamed the gene Nup154 and maintained the original names to indicate the mutation and the alleles isolated. Remobilization of the P element from l(2)01501 lead to the isolation of viable and fertile lines, providing evidence that the lethal mutation and the failure to complement tlp1 and tlp2 was indeed due to the presence of the transposon. The two alleles fully complemented other female-sterile mutations mapping in the region, including hold up, abnormal oocyte, and wavoid like (Sandler, 1977).

In homozygous condition, both tlp1 and tlp2 not only exhibited female semi-sterility but also induced complete male sterility.

Genomic regions flanking the P element insertion site were obtained by plasmid rescue from both tlp1 and tlp2. A small DNA fragment immediately adjacent to the transposon was used to screen a wild-type genomic library. A fragment, spanning the P insertion site was, in turn, used to isolate cDNAs of different lengths from a 0–3 h embryonic cDNA library. In Fig. 1 the sequence of a cDNA, reconstructed from overlaps of three partial cDNA clones, is reported, together with its putative translation product. The predicted protein is composed of 1,365 amino acid residues, having a calculated mol wt of 154 kD. It contains 32 cysteine residues, 12 potential N-linked glycosylation sites and several potential phosphorylation sites for protein kinase C, casein, and tyrosine kinases. Computer searches identified no hydrophobic stretches at the NH₂ terminus (Nielsen et al., 1997), suggesting that the protein has no signal peptides. Two small regions enriched in proline are present in the middle of the protein. Computer Blast analysis (Altschul et al., 1990; Gish and States, 1993) of the putative translation product revealed a striking similarity all along the rat Nup155 nucleoporin (32.4% identity in 1,358 residues), (Radu et al., 1993). Yeast Nup170, and Nup157, reported as structurally related to rat Nup155 (Aitchison et al., 1995b), also show similarities with the Drosophila protein, mostly in two blocks, in the NH₂-terminal (32.5% identity in a 191-residue region) and the COOH-terminal (21.5% identity in a 395-residue region) halves, while the central region, as well as the first and last 150–amino acid regions, are not conserved. The Drosophila Nup154 protein, like rat Nup155 and yeast Nup170 and Nup157 nucleoporins, did not have the XFXFG, FG, or GLFG repeats present in the most numerous nucleoporin sub-family (Davis, 1995; Fabre et al., 1995).

Nucleoporins are known to be involved in both protein and mRNA transport across nuclear membranes (Gerace, 1992). We therefore investigated if polyA⁺ RNA accumulated in nurse cell nuclei of tlp1 ovaries by in situ hybridization, using as probe a population of antisense DNA molecules produced by reverse transcription of polyA⁺ RNA extracted from wild-type ovaries. The results (Fig. 2) indicated that the global level of RNA transport in the mutant (Fig. 2,b) is comparable with that of wild type (Fig. 2,a), at least in the tested tissue and within the resolution limits of the technique. Because of the presence of pseudo-tumorous egg chambers in tlp mutants (see below and Fig. 6 b), we chose as a marker for protein nuclear import in the ovary the Sex lethal (Sxl) protein, which undergoes a relocalization process from cytoplasm to nucleus in wild-type germaria (Bopp et al., 1993). In tlp1 the nuclear import of Sxl occurs normally (data not shown). From these data we obviously cannot exclude that trafficking of other proteins, as well as of specific RNA molecules could be impaired in tlp mutants.

Polyclonal antibodies directed against a recombinant fragment of the Nup154 protein were obtained and tested in Western blot analyses. A protein of the expected molecular weight was detected in both 0–6 h embryo (Fig. 3,a, lanes 1 and 2, and b, lane 1) and ovarian (Fig. 3,b, lane 2) extracts. To investigate Nup154 localization in developing egg chambers, immunofluorescence experiments were performed. In the wild type, the protein was present in the germarium and throughout egg chamber development, both in germ line and soma, around the nuclear periphery (Fig. 3,c). In stage 10 egg chambers it was also present in a network inside the nuclei (Fig. 3,e), where its distribution seemed to overlap with that of chromatin (Fig. 3,f). This was even more clear in the oocyte nucleus, where chromosomes were condensed in a central karyosome (Fig 3,d) and anti-Nup154 antibodies identified a very bright central dot (Fig. 3 c). The intranuclear localization of Nup154 protein was also observed in spermatocytes and other cell types (data not shown).

Since very few data are available on nucleoporin localization in Drosophila and intranuclear localization has not been reported for rat Nup155 and yeast Nup170 and Nup157, we then looked at the localization of other nuclear envelope proteins in wild-type egg chambers, using antibodies directed against lamin or repeat-containing nucleoporins (mAb414) (Fig. 3). While anti-lamin antibodies clearly stained only the nuclear periphery (Fig. 3,h), the mAb414 against repeat-containing nucleoporins also stained the nuclear interior (Fig. 3 g), resembling the staining pattern seen with anti-Nup154 antibodies. This last finding is not surprising since intranuclear staining with mAb414 has been previously reported in yeast cells (Aris and Blobel, 1989).

Nup154 protein was present in our mutant alleles, indicating that tlp1 and tlp2 were not null, as also suggested by the hypomorphic phenotype of the mutations. However, in both tlp1 and tlp2 homozygous egg chambers showing morphological alterations (see below) the protein mainly accumulated in the cytoplasm of nurse and follicle cells and was only barely detected inside the nuclei (Fig. 3, i and j). Residual protein was still localized to the nuclear envelopes. Moreover, the level of anti-Nup154 staining was reduced in tlp2 homozygous flies (Fig. 3 i).

We investigated whether mislocalization of Nup154 in mutant alleles could influence the localization of the nucleoporins recognized by mAb414. The results obtained (Fig. 3,k) showed a prominent decrease in the nuclear mAb414 staining intensity and a concomitant increase in the cytoplasmic signal, compared with the wild type (Fig. 3,g). On the contrary, anti-lamin antibodies showed no gross differences in the overall antigen localization between wild type (Fig. 3,h) and mutant alleles (Fig. 3 l).

tlp1 and tlp2 alleles affected oogenesis to different extents: tlp2 homozygous females were almost completely sterile and produced very few eggs compared with tlp1. The majority of eggs produced by tlp1 homozygous females (No. of eggs observed = 135) had fused, reduced or no chorionic dorsal appendages, and could be grouped in different phenotypic classes. The most striking group was represented by cup-shaped eggs (41%; Fig. 4,c); 42% showed a reduced A/P axis (Fig. 4,b), while in 13% the dorsal appendages were abnormal but length along the A/P axis was not grossly affected (not shown). The remaining 4% were similar to wild type (Fig. 4 a).

Despite their abnormal features, about half of the eggs (44 out of 87 scored) produced by tlp1 homozygous mothers were fertilized and resumed meiosis as they were laid. Fig. 4, d–i, shows fertilized eggs during the metaphase of the second meiotic division. The tlp1 egg was irregular in shape and smaller (Fig. 4,e) than the wild-type egg (Fig. 4,d). Nevertheless, the spindle apparatus (Fig. 4,g) and chromosome organization (Fig. 4,i) in the mutant egg looked very similar to those observed in wild type (Fig. 4, f and h) (Riparbelli and Callaini, 1996).

We examined egg chamber development in homozygous tlp1 and tlp2 as well as in heterozygous tlp2/l(2)01501 females (Fig. 5), which showed the strongest ovarian defects. In the latter genetic combination, ovarioles appeared as strings of pearls in which egg chambers degenerated before entering vitellogenic stages (Fig. 5,c). tlp2 homozygous egg chambers reached later developmental stages (Fig. 5, b and e) but showed abnormal degrees of nurse cell chromatin condensation compared with wild type (Fig. 5, a and d). In the most extreme cases, nuclei retained the five blocks or even the polytene chromosome morphology (Fig. 5 e, and enlarged insets) characteristic of early developmental stages. Different levels of chromatin disorganization were observed even in nurse cell nuclei belonging to the same egg chamber. The failure to decondense nurse cell chromatin was barely detectable in tlp1 homozygous egg chambers.

Other abnormal features, including altered oocyte growth rate and follicle cell movements, were observed in both tlp1 and tlp2 homozygous egg chambers, but especially in tlp1, which exhibited the weakest ovarian phenotype and allowed a greater number of cysts to reach more mature stages (Fig. 6). In a small fraction of egg chambers, abnormal nurse cell number (Fig. 6, c and d) or bicaudal-like phenotypes (not shown) were observed.

In tlp1 and tlp2, male gametogenesis was affected in a more severe manner than oogenesis, as indicated by the fully male sterility conferred by both alleles. The spermatogenesis phenotypes of the two alleles were very similar and what follows applied to both. Testes from mutant pupae and adults were two to three times smaller than those from wild type. The most abundant germ cell stage observed in tlp testes from both pupae (n = 74) and adults (n = 63) were cysts of primary spermatocytes during the growth phase (Fig. 7,b). However meiotic figures were never observed. Two classes of accumulating cysts were easily distinguished by co-staining tlp testes (n = 134) with antibodies against β-tubulin (Fig. 7,c) and Hoechst dye (Fig. 7,d): cysts with microtubules forming elongated bundles, tapered at the extremities, which clearly contained only two nuclei, and round cysts containing spermatocytes (Fig. 7, b and c). The ratio between these two classes ranged from 1:1 to 2:1, respectively, in each testis. Several criteria indicated that the elongated cysts contained only the two somatic cyst cells and did not contain germ cells (see Discussion).

Although 16 spermatocytes were present in most of tlp cysts containing germ cells, in ∼10% (n = 27) of the scored cysts (n = 291) lower numbers of primary spermatocytes, ranging from 2 to 8 were found (Fig. 7, e and f). These incomplete cysts always contained the pair of somatic cyst cells.

To define the meiotic arrest point in tlp testes, we analyzed centrosome behavior, spindle formation, chromosome condensation, nuclear envelope breakdown, and cyclin B localization (Figs. 8 and 9). In both wild-type (Fig. 8, e and f) and tlp (Fig. 8, a and b) testes, centrioles duplicated at the beginning of the spermatocyte growth phase and were associated with two small microtubule asters. At the end of prophase, centrioles elongated at least threefold in tlp spermatocytes (Fig. 8,d), and became associated with large asters (Fig. 8,c), as in wild type (Fig. 8, g and h). The interphase microtubule network completed reorganization at the end of prophase and broke down in the mutant, also as in wild type (Cenci et al., 1994). However meiotic spindles did not form in testes from tlp homozygous males. Whereas in wild-type testes a pair of centrosomes was inherited by each secondary spermatocyte at the end of the first meiotic division (Fig. 8, k and l), in tlp testes cells often contained four aster-like structures (Fig. 8,i) closely apposed to condensed chromatin masses (Fig. 8 j).

Although meiotic divisions did not appear to occur in tlp spermatocytes, some aspects of spermiogenesis seemed to continue. In older tlp cysts, the germ cell microtubule cytoskeleton appeared to have undergone spatial reorganization as if germ cells were attempting to perform postmeiotic differentiation. Microtubules elongated and arranged in circular bundles that formed a dense shell below the plasma membrane of round germ cells (Fig. 8,m) containing one or two chromatin masses (Fig. 8 n). Small anucleate cytoplasmic fragments similarly enveloped by round microtubule shells were also present. In addition, examination of old cysts by transmission electron microscopy revealed that irregular axonemes were sometimes present (not shown), confirming that differentiation program and meiosis had been uncoupled in the mutant.

An antibody previously demonstrated to recognize components of the nuclear membrane in Drosophila syncytial embryos (Riparbelli and Callaini, 1992) strongly stained the nuclear envelope of young primary spermatocytes in both tlp (Fig. 9,a) and wild type (Fig. 9,e). During late prophase the nuclear staining became weak (Fig. 9,c), suggesting that the nuclear envelope normally broke down in tlp. In wild-type testes, cyclin B was highly expressed during prophase but underwent degradation starting from metaphase, and it was no longer visible at anaphase (Fig. 9, i and j). High levels of cyclin B were always detected in the mutant during prophase of the first meiotic division (Fig. 9 h) and at later stages of differentiation (not shown), indicating that cyclin B did not undergo normal degradation. We conclude from all these data that in mutant germ cells the meiotic block takes place at the transition between prophase and metaphase.

Ultrastructural analyses of mutant spermatocytes showed that the nuclear membrane formed small evaginations, that presumably detached in the surrounding cytoplasm as distinct vesicles (Fig. 10,a). This peculiarity was never observed in wild-type spermatocytes (Fig. 10,b) and was reminiscent of the herniations described in nuclear envelopes of some nucleoporin yeast mutants (Wente and Blobel, 1993; Bastos et al., 1995). Examination of serial sections suggested that the nuclear envelope bulged out near nuclear pores (Fig. 10, c and d). When the nuclear envelope did not show surface blobs, the pore complexes of tlp spermatocytes (Fig. 10,e) looked very similar to those observed in wild type (Fig. 10 f).

In an attempt to isolate female-sterile mutations in region 32 of the second chromosome of D. melanogaster (Gigliotti et al., 1993; Malva et al., 1994), we identified the first Drosophila mutants in a gene showing similarity with known nucleoporins. Mutation of the Nup154 gene of Drosophila affects the spatial organization of the NPCs, resulting in nuclear membrane structural defects similar to those described for some yeast nucleoporin gene mutants (Wente and Blobel, 1993; Bastos et al., 1995). The various nucleoporins are found in distinct parts of the NPC, from the cytoplasmically disposed filaments to the nuclear basket (Davis, 1995) and are involved in protein and RNA trafficking between nucleus and cytoplasm (Gerace, 1992; Fabre et al., 1995; Gorlich and Mattaj, 1996; Laemmli and Tjian, 1996; Nigg, 1997). Some proteins associated with the NPC have been demonstrated to be localized also to the nuclear interior (Cordes, 1993; Powers et al., 1995; Zimowska et al., 1997), suggesting that they could have additional functional role(s), which remain to be understood. One hypothesis predicts that the components of the NPC, together with the peripheral nuclear lamina, could be important players in the maintenance and the alteration of the three-dimensional genome structure during cell life (Blobel, 1985; Sukegawa and Blobel, 1993). The data reported in this paper suggest an intranuclear function also for the Drosophila Nup154 protein. Intranuclear localization has not been reported for rat Nup155 and yeast Nup170 and Nup157. Our observation could have been facilitated by the peculiar features of nurse cells. Because of the size and DNA content of their polytenic nuclei, nurse cells can be easily analyzed in more details than the majority of other cells, such as rat cells in culture or yeast cells or even the somatic follicle cells in the Drosophila ovary, where it is more difficult to look at Nup154 protein behavior. Alternatively, Nup154 of Drosophila has acquired, in Drosophila, new and more specialized functions.

The most interesting aspect of the tlp ovarian phenotype in our hypomorphic alleles is the failure to decondense nurse cell chromosomes. Several genes are required to regulate nurse cell chromatin structure, including cup, otu, and fs(2)B (King et al., 1957; King and Storto, 1988; Keyes and Spradling, 1997). Cup is transiently associated with the nuclear membrane and has been proposed to indirectly influence chromatin structure by acting in a pathway that affects the nuclear membrane (Keyes and Spradling, 1997). By virtue of its dual nuclear localization, Nup154 could be an essential component of this pathway. In this hypothesis, the altered chromatin structure observed in tlp ovaries could be directly related to the mislocalization of Nup154. On the other hand, tlp mutations could have a more generalized impact on egg chamber physiology and development, by impairing the nucleo-cytoplasmic traffic of essential factors, including those required for chromatin organization. In both cases, the failure of Nup154 to attain a correct localization could produce alterations in nuclear pore assembly and structure and this interpretation is supported by the observed mislocalization of mAb414 immunoreactive proteins in tlp mutants. It has been demonstrated that yeast Nup170, to which Nup154 is structurally related, is required for normal stoichiometry of FG nucleoporins within the NPC (Kenna et al., 1996). Our results could suggest that the Nup154 Drosophila protein also plays a role in localizing specific FG nucleoporins within the NPC. However, a striking discrepancy can be found between mAb414 immunolocalization in Drosophila tlp mutants and in yeast nup170 deletion mutants, where no evident differences in staining intensities or overall localizations of mAb414 could be detected compared with the wild type (Kenna et al., 1996). This discrepancy could indicate that Drosophila Nup154 is not the functional homologue of yeast Nup170. This hypothesis is supported by the observation that Drosophila Nup154 is essential for viability while yeast Nup170 is dispensable (Aitchison et al., 1995a; Kenna et al., 1996) Furthermore, the obvious nuclear envelope ultrastructural abnormalities observed in tlp mutants are absent in nup170Δ mutants (Aitchison et al., 1995a).

The most peculiar tlp testis phenotype is the occurrence of cysts that do not contain any developed germ cells but only the somatic cyst cell pair. The identity of these cells has been proven by several structural and molecular criteria: (a) the presence of large microtubule bundles; (b) the absence of any γ-tubulin spots; (c) the persistence of the nuclear envelope throughout development; and (d) the lack of cyclin B in the cytoplasm.

The presence of cysts containing only the two somatic cyst cells suggests that during gametogenesis cell division is independently regulated in somatic and germ line stem cells. This is consistent with the observation that cyst cells can form in testes of males in which pole cells have been ablated by UV (Aboim, 1945). However, it has been recently demonstrated that, when germ cells are missing, resulting from genetic ablation of pole cells, cyst cells do not withdraw from the cell cycle, suggesting that signals from germ cells are necessary to regulate their proliferation (Gonczy and DiNardo, 1996). These authors conclude that they cannot discriminate whether these signals arise from germ line stem cells or their proliferating daughters. Our data suggest that the appropriate signals are not produced in each single cyst in a direct cross talk between the germ line and the somatic components of the cyst itself, but probably represent an early event occurring before cyst formation.

Despite the high number (500) of male germ line cysts analyzed in pupae and adults, we never found spermatocytes completing meiotic division. This observation suggests that meiosis is prevented in those germ cells that escaped the early block induced by the mutation, and successfully underwent the four spermatogonial divisions. Comparison of centrosome behavior, nuclear envelope breakdown, and cyclin B localization in tlp and wild-type germ cells indicated that meiotic prophase is normal in tlp mutants. However, cyclin B is not abruptly degraded at metaphase but persists in arrested tlp spermatocytes until they degenerate or differentiate irregular axonemes. The simplest interpretation of our results is that the transition to metaphase does not occur in tlp mutants. Alternatively, we should postulate that impairment of Nup154 function leads to cyclin B stabilization and this in turn would explain the observed block in meiotic progression. This hypothesis could be supported by the finding that mutations in the yeast cse1 gene, coding for the export factor for kariopherin α, prevent cell cycle–dependent degradation of B-type cyclins (Irniger et al., 1995). Molecular components of the nuclear import and export machinery, including Cse1 and Nup154, could therefore be part of a regulatory mechanism controlling cell transit through mitosis and/or meiosis.

The multiplicity of phenotypes observed in tlp mutants could be attributed to pleiotropic effects as well as to the hypomorphic nature of our alleles. In any case, Nup154 is required not only for female and male germ line growth and development, as reported in this paper, but also in other tissues and developmental stages, as demonstrated by the existence of lethal alleles and the observation of complex embryonic phenotypes (Kiger, A., and M. Fuller, manuscript in preparation).

This notwithstanding, impairment of Nup154 function affects male versus female gametogenesis to different extents. In addition, Nup154 mutations have different consequences for germ line versus soma differentiation in males. These findings argue that distinct regulatory mechanisms, differentially involving Nup154 function, underlie these processes. Several scenarios can be envisaged. On one hand, the residual activity of Nup154 protein in the mutants could be sufficient to go through some processes but not other ones. On the other hand, Nup154 could interact with sex- and tissue-specific proteins and this interaction could establish sex- and tissue-specific functions. One candidate for this interaction in the germ line could be the product of the germ cell–less gene, which acts early in the germ cell specification pathway, is a component of the NPC and requires for its function the presence of unidentified factors (Jongens et al., 1994). Finally, the data obtained so far cannot exclude the possibility that the Nup154 gene is differentially spliced or gives rise to transcripts and translation products of slightly different lengths in different sex and/or tissues. These minor differences could have relevant consequences on protein function and protein– protein interactions. Clarification of the nature of the molecular mechanism(s) underlying the mutant phenotypes in our alleles may shed light on this point.

The studies presented here indicate that Nup154 is an essential protein involved in nuclear envelope structure. In addition, its dual localization, both at the nuclear envelope and in the nuclear interior, suggests multiple roles in nuclear functions. Its future study will be important for progress in understanding how nuclear pore components regulate nuclear and cellular physiology, also in relation to specialized developmental processes such as oogenesis and spermatogenesis.

We are grateful to Dr. W.G.F. Whitfield for generously providing antibodies against γ-tubulin and cyclin B, and to Dr. D. Bopp who provided anti-Sxl antibodies. We thank D. Rotoli for the skilled help in using the confocal microscope, and A. Bellopede for technical assistance. We thank M. Fuller, A. Kiger, P. Bazzicalupo, and U. Di Porzio for helpful discussion and critical reading of the manuscript.