Drosophila melanogaster Importin α1 and α3 Can Replace Importin α2 During Spermatogenesis but Not Oogenesis
D. Adam Mason, Robert J. Fleming, David S. Goldfarb

Abstract

Importin α’s mediate the nuclear transport of many classical nuclear localization signal (cNLS)-containing proteins. Multicellular animals contain multiple importin α genes, most of which fall into three conventional phylogenetic clades, here designated α1, α2, and α3. Using degenerate PCR we cloned Drosophila melanogaster importin α1, α2, and α3 genes, demonstrating that the complete conventional importin α gene family arose prior to the split between invertebrates and vertebrates. We have begun to analyze the genetic interactions among conventional importin α genes by studying their capacity to rescue the male and female sterility of importin α2 null flies. The sterility of α2 null males was rescued to similar extents by importin α1, α2, and α3 transgenes, suggesting that all three conventional importin α’s are capable of performing the important role of importin α2 during spermatogenesis. In contrast, sterility of α2 null females was rescued only by importin α2 transgenes, suggesting that it plays a paralog-specific role in oogenesis. Female infertility was also rescued by a mutant importin α2 transgene lacking a site that is normally phosphorylated in ovaries. These rescue experiments suggest that male and female gametogenesis have distinct requirements for importin α2.

THE nuclear targeting of proteins is mediated by a number of nuclear localization signal (NLS)-specific receptors called importins or karyopherins (Mattaj and Englmeier 1998; Weis 1998; Görlich and Kutay 1999; Nakielny and Dreyfuss 1999). Importin β-family members bind NLS cargo in the cytoplasm and act as chaperones to facilitate their translocation across the nuclear envelope through serial interactions with nucleoporins arrayed along the central channel of the nuclear pore complex (Routet al. 2000; Rabut and Ellenburg 2001). Importin β1 is unusual in that it functions in conjunction with importin α to mediate the import of “classical” NLS (cNLS) cargo (Nakielny and Dreyfuss 1999), which includes a large variety of nuclear proteins (Michaud and Goldfarb 1993).

Importin α’s are composed of 10 tandem armadillo (Arm) repeats, bracketed by shorter N- and C-terminal domains (Peiferet al. 1994; Contiet al. 1998). Interestingly, importin β-family members contain large domains composed of tandem HEAT motifs, which are related to Arm repeats by a degenerate ∼40- to 45-amino-acid consensus sequence (Maliket al. 1997; Andradeet al. 2001). Arm and HEAT repeats fold into structurally related superhelical rods that serve as selective scaffolds for binding proteins (Huberet al. 1997; Contiet al. 1998). Importin α’s bind cNLS cargo via their Arm domains (Contiet al. 1998) and importin β through N-terminal importin β-binding (IBB) sequences (Görlichet al. 1996; Weiset al. 1996). After cNLS-importin α/β1 ternary complexes enter and dissociate in the nucleus, importin α and β1 are independently recycled back to the cytoplasm (Koeppet al. 1996; Percipalleet al. 1997). Export of importin α is mediated by another importin β-family member, an exportin called Cse1p in yeast and CAS in higher eukaryotes (Kutayet al. 1997; Hood and Silver 1998), which binds within the tenth Arm repeat of importin α (Heroldet al. 1998).

In contrast to the single importin α gene of Saccharomyces cerevisiae (SRP1), vertebrates contain as many as eight conventional importin α genes. Phylogenetic analysis has revealed that the majority of importin α genes belong to one of three conserved clades (Köhler et al. 1997, 1999; Maliket al. 1997), referred to here as α1, α2, and α3. All conventional yeast and plant importin α genes are α1 paralogs. In contrast, metazoan animals typically contain representatives of all three clades. The functional basis for the conservation of multiple importin α genes in animals is not known. However, their occurrence only in metazoan animals suggests they may be involved in tissue differentiation and development. Consistent with this notion, importin α1’s, α2’s, and α3’s show distinct tissue and cell type-specific expression patterns (Prieveet al. 1996; Köhleret al. 1997; Tsujiet al. 1997; Nachuryet al. 1998; Kameiet al. 1999).

In vitro evidence suggests that all conventional importin α’s bind a broad class of cNLS sequences, but they do so with different affinities (Prieve et al. 1996, 1998; Nadleret al. 1997; Miyamotoet al. 1997). In permeabilized cell nuclear import assays, representative importin α paralogs show overlapping preferences for different cNLS sequences (Köhleret al. 1999). In some cases, specific importin α’s show strong preferences for NLS cargo. For example, only an importin α3 mediates the in vitro import of RCC1 (Köhleret al. 1999) and Ran BP3 (Welchet al. 1999), and only an importin α1 imports the Stat 1 transcription factor (Sekimotoet al. 1997). In addition, the preference of importin α’s for certain NLS cargo is significantly increased when two different substrates are presented together in the import assay (Köhleret al. 1999). This latter finding underscores the complexity of the functional interactions betweenimportin α’s and different NLS cargo and indicates that in vivo studies are needed to unravel the physiological roles of the importin α1, α2, and α3 genes. In this study we exploit the finding that Drosophila importin α2 is required for male and female fertility to examine functional interactions among the conserved Drosophila importin α genes.

MATERIALS AND METHODS

Cloning of Drosophila importin α genes and phylogenetic analysis: Drosophila importin α genes were cloned using degenerate oligonucleotide-mediated PCR. The 5′ oligonucleotide was made to the conserved amino acid sequence AWALT-NIA found in the third helix of arm repeat 2. To facilitate cloning, an EcoRI site was added, generating the sequence 5′ ATCGCGAATTCGC/TITGGGCIT/CTIACIAAT/CATT/C/AGC 3′ (I represents inosine). The 3′ oligonucleotide was made to the conserved VGNIVTG sequence in the third helix of arm repeat 6. With the addition of a BamHI site, the nucleotide sequence of the 3′ primer is 5′ GACGTAGGATCCCCIGT TACT/G/AATG/ATTICCIA 3′. The primers were used to PCR amplify importin α gene fragments [PCR conditions: 1.5 mm MgCl2, 20 mm Tris-Cl pH 8.4, 50 mm KCl, 0.2 μg SalI-digested Drosophila genomic DNA, 2 μm 5′ and 3′ primers, 200 μm dNTPs, and 1 unit of Taq DNA polymerase (Bethesda Research Laboratories, Gaithersburg, MD) amplified at 94° for 20 sec, 43° for 20 sec, ramp to 55° for 2 min, 55° for 30 sec, 72° for 30 sec with 50 cycles] from genomic Drosophila DNA. Genomic DNA was digested with SalI to inhibit importin α2 amplification since PCR with undigested DNA yielded only importin α2 products. Two distinct PCR products were subcloned into the pGAD 424 vector (CLONTECH, Palo Alto, CA) and sequenced to reveal that they encoded novel importin α1 and α3 genes. The two importin α gene fragments were Digoxigenin labeled by PCR (Boehringer Mannheim, Indianapolis) and used to probe an ovary λgt11 cDNA library (Zinnet al. 1988) to obtain full-length coding regions (α1, accession no. AAC26055; α3, accession no. AAC26056). A CLUSTAL V (Higginset al. 1992) alignment of the new Drosophila importin α genes to the importin α-like genes found in GenBank was used to construct a phylogenetic tree by the neighborjoining method (Saitou and Nei 1987).

Genetic stocks and markers: Flies were kept on standard cornmeal-dextrose media and grown at 25°. The importin α2 gene was previously referred to as the oho31 gene and encodes the Pendulin protein (Küssel and Frasch 1995; Töröket al. 1995). The clean importin α2D14/y+ CyO and importin α2D3/y+ CyO stocks were provided by Dr. Istvan Kiss (Hungarian Academy of Sciences, Szeged, Hungary; see Töröket al. 1995). Deficiency stocks Df(3L)kto2/TM6B, Tb[1], Df(3R)by416/TM3, Sb[1], Gal4Hsp70/TM6B, and w1118 stocks were obtained from the Bloomington Stock Center. The Gal4nanos-VP16 stock (Rørth 1998) was a gift from Pernille Rørth (EMBL, Heidelberg, Germany).

Fertility assays and testes squashes: Male fertility was assayed by crossing individual males to five w1118 virgin females. Female fertility was assayed by mating individual females to three w1118 males. Flies were allowed to mate for ∼10 days before being discarded. To examine sperm motility, importin α2D14 or importin α2 D3 homozygous and heterozygous testes were dissected from 4-day-old males in cold testis dissecting buffer (183 mm KCl, 47 mm NaCl, 10 mm Tris-HCl pH 6.8), gently ruptured under a coverslip, and visualized by dark field microscopy.

Transmission electron microscopy of testes: Testes were dissected in cold testis dissecting buffer from importin α2D14/ importin α2D14, importin α2D14/y+CyO, and w1118 0- to 4-day-old males and prepared for electron microscopy as previously described (Tokuyasuet al. 1977). Briefly, samples were fixed in 2.5% glutaraldehyde, postfixed in 1.0% osmium tetroxide, infiltrated in Spurr epoxy resin, and embedded. These samples were then cut into ultra-thin sections, stained with uranyl acetate and lead citrate, and subsequently examined with a Hitachi 7100 transmission electron microscope.

RNA and protein isolation: Total RNA and proteins were isolated from Drosophila tissues with Tri-Reagent LS (Molecular Research Center, Cincinnati; Chomczynski 1993) following the recommended protocols. RNA was quantified by determining the A260, and protein concentration was determined using the Bio-Rad (Richmond, CA) DC protein assay.

Northern and Western blots: To determine the endogenous importin α1, α2, and α3 mRNA expression patterns, 20 μg of RNA isolated from adult males, adult females, dissected testes, and dissected ovaries was separated on a 1% agarose, 6% formaldehyde gel; transferred to a nylon membrane; and probed with 32P-random prime-labeled importin α1, α2, or α3 probes [GIBCO (Grand Island, NY)/Bethesda Research Laboratories random primers DNA labeling system]. Bands were visualized by phosphoimaging. To examine protein expression patterns, ∼10 μg of protein isolated from adult males, adult females, dissected testes, and dissected ovaries was separated on an 8% PAGE gel; transferred to a polyvinylidene fluoride membrane; and blotted with rabbit anti-importin α2 (Töröket al. 1995) provided by Istvan Török (DKFZ, Heidelberg, Germany) or rabbit anti-importin α3 (Máthéet al. 2000) provided by Endre Máthé (University of Cambridge, Cambridge, United Kingdom). Blots were developed using alkaline phosophatase-tagged goat anti-rabbit secondary antibodies. To quantify transgene expression levels RNA and protein were isolated from dissected ovaries from mated w1118 or homozygous importin α2D14 females expressing UASp importin α1, α2, α3, 2×α3, or α2S56A transgenes (see next section). Ovarian RNA (15 μg) was tested in a Northern blot using 32P-random prime-labeled K10 3′ untranslated region (UTR) and RP49 probes as described above. Bands were visualized and quantified by phosphoimaging. To examine protein levels 10 μg of protein was tested in a Western blot using rabbit anti-importin α2, rabbit anti-importin α3, or mouse anti-α-tubulin antibodies (Amersham, Arlington Heights, IL) as described above.

Immunofluorescence of testes: Testes were dissected from wild-type or homozygous importin α2D14 males in 1× PBS; fixed in 1× PBS, 4% paraformaldehyde; and blocked in PBS-saponin (1× PBS, 0.2% saponin, and 0.3% normal goat serum). Testes were then incubated with rabbit anti-importin α2 (Töröket al. 1995) diluted 1:50 in PBS-saponin, followed by a goat anti-rabbit FITC-labeled secondary antibody diluted 1:300 in PBS-saponin. DNA was stained with 10 μm Hoechst in PBS. Confocal microscopy was performed on a Leica TCS NT microscope equipped with UV, Ar, Kr/Ar, and He/Ne lasers, and digital images were processed using Adobe PhotoShop (Adobe Systems, San Jose, CA).

Expression constructs and germline transformations: Importin α transgenic flies were created by cloning importin α1, α2, and α3 PCR fragments corresponding to the 1.5-kb coding region of each gene into EcoRI and NotI sites in the pUASt P-element transformation vector (Brand and Perrimon 1993) or into KpnI and NotI sites in the UASp P-element transformation vector (Rørth 1998). UASp α3 contains an additional 22 nucleotides of its 5′ UTR. The UASp importin α2 S56A transgene was created by PCR amplifying an importin α2 3′ AvaII and NotI fragment containing a single point mutation that changed the TCG codon for serine-56 to a GCG alanine codon. This PCR fragment was then ligated with a 5′ KpnI and AvaII importin α2 PCR fragment into KpnI and NotI sites in UASp. The pUASt and UASp importin α2 and α3 transgenes contain an additional 7 and 42 nucleotides of their 3′ UTRs, respectively, and UASp importin α3 contains 22 nucleotides of its 5′ UTR.

Transgenic UASt and UASp importin α1, α2, α3, and UASp importin α2 S56A lines were created using standard germline transformation procedures (Spradling 1986). The UASt and UASp importin α1 and UASp importin α2 inserts used in this study were located on the second chromosome, while the UASt importin α2, UASt importin α3, UASp importin α3, and UASp importin α2 S56A inserts map to the third chromosome.

Crosses:To examine genetic interactions between the importin α2D14 allele and deficiencies that uncover the α1 or α3 genes, females of the genotype importin α2D14/CyO were crossed to (1) importin α1Df(3L)kto2/TM6B or (2) importin α3Df(3R) by416/TM3. Male offspring without balancers are heterozygous for importin α2 and importin α1 or α3. CyO males without TM3 or TM6B balancers serve as heterozygous importin α1 or α3 flies in a wild-type importin α2 background. To test for rescue of importin α2D14 male sterility following ectopic importin α1, α2, or α3 expression, females of the genotype importin α2D14/CyO; Gal4Hsp70/TM6B were crossed to (1) UASt importin α1, importin α2D14/CyO; (2) importin α2D14/CyO; UASt importin α2/UASt importin α2; or (3) importin α2D14/CyO; UASt importin α3/UASt importin α3. Male offspring that are heterozygous for the importin α2D14 allele and have the Gal4Hsp70 driver and the UASt importin α transgenes were used as the positive control for fertility (e.g., importin α2D14/CyO; UASt importin α2/Gal4Hsp70); homozygous importin α2D14 males that have the Gal4Hsp70 driver and the UASt importin α transgenes served as the experimental group (e.g., importin α2D14/importin α2D14; UASt importin α2/Gal4Hsp70); homozygous mutant males that contain the UASt importin α transgene but inherited the TM6B balancer instead of Gal4Hsp70 were utilized as the negative control (e.g., importin α2D14/importin α2D14; UASt importin α2/TM6B). Male flies were collected 0-18 hr after eclosion, heat shocked for 2 hr in a 37° air incubator, and then assayed for fertility by mating individually to five w1118 virgin females at 25°. To test for rescue of importin α2D14 female sterility following ectopic importin α1, α2, α3, or α2 S56A expression, females of the genotype importin α2D14/CyO; Gal4nanos-VP16/TM6B were crossed to (1) UASp importin α1, importin α2D14/CyO; (2) UASp importin α2, importin α2D14/CyO; (3) importin α2D14/CyO; UASp importin α3/ UASp importin α3; or (4) importin α2D14/CyO; UASp importin α2 S56A/TM6B. Female offspring that are heterozygous for the importin α2D14 allele and have the Gal4nanos-VP16 driver and the UASp importin α transgenes were used as the positive control for fertility (e.g., importin α2D14, UASp importin α2/CyO; Gal4nanos-VP16/+); homozygous importin α2D14 females that have the Gal4nanos-VP16 driver and the UASp importin α transgenes served as the experimental group (e.g., importin α2D14/importin α2D14, UASp importin α2; Gal4nanos-VP16/+); homozygous mutant females that contain the UASp importin α transgene but inherited the TM6B balancer instead of Gal4nanos-VP16 were utilized as the negative control (e.g., importin α2D14, UASp importin α2/importin α2D14; TM6B/+). To express two copies of importin α3 in an α2 mutant background importin α2D14/CyO;/UASp importin α3, Gal4nanos-VP16/TM6B females were crossed to importin α2D14/CyO; UASp importin α3/UASp importin α3 and female offspring of the genotype importin α2D14/importin α2D14; UASp importin α3, Gal4nanos-VP16/UASp importin α3 were collected. Females of the described genotypes were assayed for fertility as described above or used as a source for RNA and protein isolation from ovaries.

RESULTS

Phylogenetic structure of the Drosophila importin α gene family: Phylogenetic analysis of known importin α gene sequences indicates that metazoan animals contain at least one representative of each of three conserved importin α clades, designated here as conventional α1, α2, and α3 genes (Figure 1; Maliket al. 1997; Köhler et al. 1997, 1999). For example, the current version of the human genome encodes four importin α1, one importin α2, and three importin α3 genes (Venteret al. 2001). Of these three clades, only importin α1 genes have been found in fungi and plant genomes. Nonconventional importin α genes, which contain recognizable IBB domains and conserved cNLS-binding sites, occur in plants, invertebrates, and protozoa. These divergent importin α genes probably arose from the same progenitors that gave rise to extant conventional genes. At the time this project was initiated, Pendulin, an importin α2, was the only known Drosophila importin α (Küssel and Frasch 1995; Töröket al. 1995).

Prior to the completion of the Drosophila genome sequence, we attempted to clone by PCR the entire Drosophila importin α gene family using degenerate oligonucleotide primers designed against sequences that were well conserved among known importin α genes. Successfully amplified PCR products were then used to screen a Drosophila ovary cDNA library for full-length importin α cDNAs. In addition to recloning the Drosophila importin α2, this strategy yielded novel importin α1 and α3 genes. The new importin α genes encode Arm domain proteins with consensus importin β and cNLS binding domains. These results demonstrate that the phylogenetic structure of the importin α gene family is conserved between invertebrates and vertebrates (see also Köhleret al. 1999). The Drosophila importin α3 gene was recently recloned by two hybrid screens on the basis of its interactions with DNA polymerase α (Máthéet al. 2000) and germ cell-less (Dockendorffet al. 1999). In addition, the published Drosophila genome sequence (Adamset al. 2000) contains the importin α1, α2, and α3 genes, as well as a fourth nonconventional importin α gene [protein identification (PID) no. g7295403] that may or may not function in cNLS transport.

Figure 1.

—A phylogeny of the importin α’s. Alignments of importin α protein sequences were used to construct a phylogeny (materials and methods). Shown here is the consensus tree of those importin α sequences whose Arm domains are demonstrably homologous (see text). Bootstrap values are represented as a percentage of 1000 replications. Species, protein identification (PID) numbers, and proposed α1, α2, and α3 nomenclature for conventional genes are indicated. Vertebrate genes are shown in brown, invertebrates in red, plants in green, and fungi in blue.

The phylogeny shown in Figure 1 does not contain all importin α-like sequences found in GenBank. Because importin α’s contain tandem Arm repeats, sequence alignments cannot be used in constructing a rigorous phylogeny unless it can be determined that the order of individual repeats has been conserved (Maliket al. 1997). If, for example, shuffling, substitution, or deletion of repeats within an Arm domain had occurred, then the alignment of this altered sequence with other sequences would effectively be comparing nonhomologous sequences. All putative importin α sequences included in the phylogeny shown in Figure 1, including the nonconventional genes, contain virtually the same order of individual Arm repeats, although Arm repeat one is naturally extremely divergent. The Caenorhabditis elegans genome contains three putative importin α genes, only two of which pass this homology test and are included on the phylogeny. The Arm repeats of the other C. elegans importin α-like gene are dissimilar to other known importin α-like sequences and cannot, therefore, be aligned with confidence, even though it is a functional importin α (Geles and Adam 2001). This is basically the reason that β-catenins, plakoglobins, the flagellar protein PF16, and other non-importin α Arm domain proteins that share relatively high sequence similarity with importin α’s are also left off this tree. The nonconventional Oryza sativa (rice) and Arabidopsis thaliana importin α sequences are included in the phylogeny because 6 of their 10 Arm repeats are homologous with those of bona fide conventional genes. Interestingly, the nonconventional O. sativa and A. thaliana importin α genes define a distinct clade that arose at least ∼75 million years ago, prior to the division between monocotyledonous (sativa) and dicotyledonous (thaliana) angiosperms.

On the basis of the robust distribution of most known importin α genes into three distinct clades (Figure 1), we propose that the α1, α2, α3 nomenclature be adopted for naming conventional importin α genes. This approach clarifies nonsystematic schemes (e.g., Qip1, RCH1, etc.) and has the advantage over the previously proposed “S, P, Q” system (Tsujiet al. 1997) in that the numerical identifier is incorporated directly into the importin α name (e.g., importin α1). This scheme is not meant to replace the original names given by their discoverers. Rather, as we learn more about the physiological role(s) of individual genes it will be helpful for comparative purposes to know that human Qip1, for example, is an importin α3.

View this table:
TABLE 1

Fertility of importin α-deficient flies

Drosophila importin α2 is required for male and female fertility: Two early studies reported that flies homozygous for a null allele of the importin α2 gene (oho-31) die during development while exhibiting an overgrowth of hematopoietic tissues (Küssel and Frasch 1995; Töröket al. 1995). However, these phenotypes were subsequently found to be due to a second site mutation (I. Kiss, personal communication). Bona fide importin α2 null alleles, importin α2D14 and importin α2D3, generated by imprecise P-element mobilization (Töröket al. 1995) and subsequent recombination to remove the oho31 mutation, were kindly provided to us by I. Kiss (Hungarian Academy of Sciences, Szeged, Hungary). These deletions removed 1.7 kb of 5′ sequence in importin α2D14 and 1 kb of 5′ sequence in importin α2D3 (Töröket al. 1995). The large amount of coding sequence removed and the absence of importin α2 mRNA (not shown) or protein (Töröket al. 1995) in homozygous importin α2D14 flies demonstrate that these mutations are indeed protein null alleles. Consistent with the unpublished observation that the oho31 phenotype was caused by a second site mutation (I. Kiss, personal communication), we do not observe any larval lethality or overgrowth of hematopoietic tissues in flies homozygous for these alleles, despite the fact that Importin α2 protein is absent. Because the importin α2D14 and importin α2D3 alleles are indistinguishable phenotypically (not shown), we used only the importin α2D14 allele in our subsequent experiments.

View this table:
TABLE 2

Rescue of importin α2 null male and female sterility

Males homozygous for importin α2D14 were mostly sterile but were otherwise morphologically wild type. Male fertility was quantified by mating individual male flies with five wild-type (w1118) virgin females. In this experiment, all 20 heterozygous importin α2D14 males tested were fertile, and all 20 homozygous importin α2D14 males were sterile (Table 1). However, in rare cases, homozygous importin α2D14 males produced a few offspring when mated with wild-type females (Table 2). Therefore, importin α2 is important but not essential for male fertility. Likewise, homozygous importin α2D14 adult females appear morphologically wild type but are sterile (G. Adam and I. Kiss, unpublished results). We independently quantified female fertility by individually mating 20 homozygous importin α2D142-/-) or 20 heterozygous importin α2D142+/-) females with three wild-type males. All α2+/- females were fertile and all α2-/- mutant females were sterile (Table 1). In addition, we have repeatedly mated α2-/- females with wild-type and α2+/- males during the course of our studies and have never observed any fertile females. Therefore, importin α2 appears to be essential for female fertility but only important for male fertility.

Importin α2 is required for normal gametogenesis in males and females: Because α2-/- males and females develop and behave indistinguishably from their heterozygous α2+/- siblings, we focused on the role of importin α2 in gametogenesis. Intact testes isolated from α2-/- males appeared normal under the light microscope and contained characteristically large numbers of elongated sperm bundles. However, squashed testes from 4- to 5-day-old α2-/- males released few, if any, motile sperm (data not shown). In contrast, swarms of motile sperm were released from squashed testes isolated from wild-type and α2+/- males. In addition, no sperm were observed in the seminal receptacle or spermathecum of wild-type females mated to α2-/- males (data not shown). Therefore, it is likely that the infertility of males lacking importin α2 is due to a defect during spermatogenesis that results in defective sperm. The occasional fertility of α2-/- males is consistent with the occurrence of small numbers of motile sperm in some testes. An analysis of mutant testes by transmission electron microscopy (TEM) indicated that α2-/- males display a defect in spermatogenesis upon individualization. Prior to individualization, the sperm bundles of homozygous α2-/- flies were indistinguishable from bundles in heterozygous testes (not shown). Following individualization, most sperm bundles from heterozygous testes contained close to the full complement of 64 properly individualized spermatids (Figure 2A). In contrast, in α2-/- testes a large number of spermatids failed to individualize and remained syncytial (Figure 2, B and C). This phenotype is not likely to be caused by a defect in the individualization machinery since normal-appearing individualization complexes, visualized with rhodamine-phallodin staining (Fabrizioet al. 1998), were observed in α2-/- testes (not shown). We were curious about the axonemal structure of α2-/- sperm tails, partly because mutants in a Chlamydomonas importin α-like flagellar protein, PF16, are paralyzed, and detergent-treated mutant axonemes lack central C1 microtubules. PF16 is conserved between Giardia lamblia and mice, but appears to be absent from the Drosophila genome (Adamset al. 2000). Thus it is possible that an importin α replaces the function of PF16 in Drosophila flagella. However, the organization of the sperm axonemes and the structure of the major and minor mitochondrial derivatives were normal in α2-/- testes (Figure 2). We conclude that Drosophila Importin α2 probably does not serve a comparable function to PF16 in flagella.

In our hands, the eggs laid by α2-/- females showed gross morphological defects that corroborate earlier observations (G. Adam and I. Kiss, personal communication). Specifically, eggs laid by α2-/- mothers were smaller and appeared deflated as if they were deficient in internal contents (not shown). In addition, dorsal appendages were missing, fused together, or otherwise severely deformed, and the chorions appeared thinner than in eggs laid by wild-type females (not shown). These results suggest that the complete sterility of α2-/- mutant female flies is due to a severe defect in oogenesis.

Figure 2.

—Ultrastructure of importin α2+/- and α2-/- testes. (A) Electron microscopic images of mature sperm bundles of α2+/- males showing uniformly packed and individualized (arrowheads) spermatids; (B and C) α2-/- testes showing both individualized (arrowheads) and unindividualized syncytial spermatids (arrows). (D) A detail of individualized α2+/- sperm tails and (E) α2-/- sperm tails. The axoneme shown in E lacks a membrane (compare to D) and has not, therefore, completed individualization.

Importin α1, α2, and α3 are expressed in testes and ovaries: The results described above indicate that importin α2 is required for normal gametogenesis. Northern blot analysis was used to begin to assess the particular roles of the three conventional importin α genes in gametogenesis. As shown in Figure 3A, importin α1, α2, and α3 mRNAs were all detected in testes, ovaries, and in adult male and female flies. Therefore, as is the case with the three characterized C. elegans importin α genes (Geles and Adam 2001), the three conventional Drosophila Importin α genes are all expressed in the germline. Importin α3 mRNA levels were about equally expressed in ovaries and testes. In contrast, importin α1 mRNA levels were slightly elevated in the testes compared to ovaries, and importin α2 mRNA levels were much higher in the testes than ovaries. Thus importin α2 and, to a lesser extent, importin α1 mRNAs are selectively enriched in testes.

Figure 3.

—Analysis of importin α gene expression. (A) A total of 20 μg of total RNA isolated from adult wild-type males, adult wild-type females, dissected wild-type testes (Test.), and dissected wild-type ovaries (Ova.) was hybridized with importin α1, α2, or α3-specific probes (materials and methods). Importin α1, α2, and α3 mRNA migrate at ∼2.5 kb. (B) A total of 10 μg of total protein isolated from adult wild-type males, adult wild-type females, dissected wild-type testes (Test.), and dissected wild-type ovaries (Ova.) was examined on a Western blot with rabbit anti-Importin α2 or α3 antibodies (materials and methods).

Western blot analyses of Importin α2 and α3 protein levels are shown in Figure 3B. Importin α1 protein levels could not be examined because antibodies are not currently available. The concentration of Importin α3 protein is about the same in testes and ovaries, but somewhat higher in adult females than in males (see also Máthéet al. 2000). This result is consistent with the Northern blot analysis (Figure 3A), which shows similar importin α3 mRNA levels in adult and germ cell extracts. Thus, Importin α3 protein levels appear to be uniformly expressed in somatic and germline tissues in proportion to its mRNA levels. The Importin α2 case is complex. Here, Importin α2 protein is much more highly expressed in ovaries than in total female flies. In males, Importin α2 protein is present at low levels in testes, at about the concentration found in whole adult females, but is undetectable in adult males (see below). In general, then, Importin α2 protein appears to be enriched in ovaries and testes relative to whole flies. However, the very high level of importin α2 mRNA in the testes (Figure 3B) is out of proportion to the relatively low level of Importin α2 protein (see also Töröket al. 1995). These results suggest the possibility that translation of importin α2 mRNA is negatively regulated in the testes.

Figure 4.

—Selective localization of Importin α2 in testes. Testes dissected from wild-type (A-I) or α2-/- (J-L) males were stained for DNA with Hoechst (A, D, G, and J) and Importin α2 (B, E, H, and K) with anti-Importin α2 antibodies (materials and methods). Overlays are shown in C, F, I, and L. Confocal images were taken of the apical tip of wild-type (A-F) and α2-/- (J-L) testes and elongated sperm bundles (G-I) from wild-type testes. Bars, ∼100 μm.

To further examine Importin α2 expression, the protein was localized by immunofluorescence in wild-type and α2-/- testes. These data clearly show that in wild-type testes Importin α2 is highly expressed in immature germ cells located at the apical tip (Figure 4, A-C), where it is predominantly cytoplasmic (Figure 4, D-F), and is undetectable in more mature elongated sperm bundles (Figure 4, G-I). At the apical tip of testes premeitoic stem cells and spermatogonial cells exhibit stronger Hoechst fluorescence than postmeiotic spermatocytes (Gönczyet al. 1997). On the basis of this criterion, we conclude that Importin α2 is preferentially localized to stem and spermatogonial cells. The observed fluorescence is specific for Importin α2 since it is absent in α2-/- testes (Figure 4, J-L). These results suggest that Importin α2 functions during the earliest stages of sperm development.

Rescue of α2-/- male sterility with importin α transgenes: The coexpression of importin α1, α2, and α3 in the testes raises the possibility that the occurrence of some fertile α2-/- males could be due to the capacity of endogenous Importin α1 and/or α3 to partially perform the function(s) of the lost Importin α2. If true, then the ectopic expression of importin α1 and α3 transgenes in the testes might raise their levels enough to more efficiently rescue the α2-/- sterility phenotype. Alternatively, if Importin α2 plays a paralog-specific role that is important but not essential in spermatogenesis, then importin α1 or α3 transgenes would not rescue male sterility. The capacity of each of the three importin α genes to rescue male sterility was determined using the UASt transgene system (Brand and Perrimon 1993) with a Gal4Hsp70 driver to ectopically express importin α1, α2, or α3 transgenes (materials and methods). As shown in Table 2, α2-/- males containing UASt importin α1, α2, or α3 transgenes and the Gal4Hsp70 driver were significantly more fertile than sibling control males lacking either the transgenes or the driver. These results indicate that the activity(s) of importin α2 that is important for spermatogenesis can be replaced by elevating the expression levels of importin α1 or α3. Furthermore, testes squashes from 4- to 5-day-old flies demonstrated that each of the three transgenes partially rescued the sperm motility defect of α2-/- males (data not shown). Specifically, testes from rescued flies contained significantly greater numbers of motile sperm than those from α2-/- testes. In this experiment, flies received a single 2-hr heat-shock treatment. Repeated heat-shock treatments throughout development increased the rescue efficiency of the transgenes: 100% fertility (six of six males) for importin α1, 89% fertility (eight of nine males) for importin α2, and 78% fertility (seven of nine males) for importin α3. We conclude that the requirement for importin α2 in spermatogenesis is not due to a unique biochemical property of Importin α2 that is completely lacking from Importin α1 and α3.

Rescue of the α2-/- female sterility with importin α transgenes: We next investigated the capacity of UASp importin α1, α2, and α3 transgenes to rescue the sterility defect in α2-/- female flies. The Gal4nanos-VP16 driver was used to express UASp importin α transgenes in the female germline (Rørth 1998). The UASp importin α2 transgene efficiently rescued the α2-/- sterility defect to the level found in α2+/- females (Table 2). Eggs laid by importin α2-transgene-expressing α2-/- females were morphologically wild type (not shown). Control sibling α2-/- flies were sterile and laid defective eggs. As in males, the capacity of an importin α2 transgene to efficiently rescue the sterility of α2-/- females is strong evidence that the phenotype is caused solely by a mutation in the importin α2 gene and not to a second, unlinked mutation. Interestingly, UASp importin α1 and α3 transgenes did not rescue the sterility of α2-/- females (Table 2). Also, the characteristic deflated morphology of eggs laid by α2-/- flies was unaffected by the ectopic expression of either importin α1 or α3 transgenes (data not shown).

Figure 5.

Importin α1, α2, and α3 transgene expression. (A) A total of 15 μg of total RNA isolated from dissected ovaries of the indicated genotype (materials and methods) was hybridized with a K10 3′ UTR probe that is common to all UASp-derived transgenes and a RP49 probe as an internal loading control. Transgene expression levels were determined by normalizing the K10 3′ UTR band to the RP49 band by phosphoimaging and are indicated relative to the expression level of the rescuing importin α2 transgene. (B) A total of 10 μg of protein isolated from dissected ovaries of the indicated genotype was examined by Western blot with anti-Drosophila Importin α2, α3, or anti-α-tubulin antibodies (materials and methods). * indicates the slower migrating anti-Importin α3-crossreactive band that appears only when α1 is overexpressed. The Importin α2 S56A protein was made by changing the serine at position 56 of α2 to alanine.

To control for transgene expression levels, importin α1, α2, and α3 transgene mRNA levels were determined by Northern blot using a K10 3′ UTR probe that is common to all UASp-expressed mRNAs (Rørth 1998). Ribosomal protein RP49 mRNA levels were used as loading controls. Importin α transgene expression levels were quantified and normalized to RP49 levels by phosphoimaging. As shown in Figure 5A, the importin α1 transgene was expressed ∼2.3 times higher than the rescuing importin α2 transgene. Initially, we found that single importin α3 transgenes were expressed at levels slightly less than the importin α2 transgene. Therefore, flies containing two importin α3 transgenes were created. The combined expression of both transgenes raised importin α3 mRNA levels in the ovary to 1.4 times that of the importin α2 transgene. The α2-/- females that expressed two copies of importin α3 were still completely sterile (0/20 fertility) and laid morphologically defective eggs that never hatched. These results argue that the failure of importin α1 and α3 transgenes to rescue the importin α2D14 female sterility was not due to poor transcription of the transgenes, which, in both cases, exceeded that of a rescuing importin α2 transgene.

The levels of importin α proteins in dissected ovaries from transgenic flies were determined by Western blot using available anti-Importin α2 and anti-Importin α3 antisera. As shown in Figure 5B, ovarian Importin α2 migrates as a doublet that is absent from α2-/- ovaries and is mostly replenished by expression of a UASp importin α2 transgene. In flies expressing two copies of the importin α3 transgene, the level of Importin α3 protein increased approximately five times over wild-type levels. The level of Importin α1 could not be directly determined since anti-Importin α1 antibodies are not currently available. However, because the untranslated sequences of the importin α1 and α2 transgenes differ by only 7 nucleotides (materials and methods), it is unlikely that they are differentially translated. In addition, a slower migrating anti-Importin α3-crossreactive band always appeared in extracts from ovaries expressing a UASp importin α1 transgene (Figure 5B). Because Drosophila Importin α1 is 3300 daltons larger than Importin α3 it is possible that this band is Importin α1. Taken together, these results indicate that importin α1 and α3 transgenes are both highly expressed in the ovary. We conclude that Importin α2 has a unique role(s) in oogenesis that cannot be performed in vivo by either Importin α1 or α3.

Curiously, it appears that Importin α3 protein levels in the ovary varied in inverse proportion to Importin α2 protein levels. For example, Importin α3 levels were higher in the ovaries of α2-/- flies than in wild-type flies (Figure 5B). Importin α3 levels decreased in α2-/- ovaries in flies expressing an importin α2 transgene (Figure 5B), and the decrease was proportional to the level of transgene expression (not shown). Thus, it is possible that the expression of Importin α3 is influenced by the level of Importin α2.

The conserved Importin α2 phosphorylation site is not required for fertility: Importin α2 isolated from ovaries and preblastoderm embryos is partially phosphorylated and migrates as a doublet on Western blots (Figures 3B and 5B; Küssel and Frasch 1995; Töröket al. 1995; Máthéet al. 2000). Consistent with this, the upper band disappears upon phosphatase treatment (Töröket al. 1995). The S. cerevisiae importin α1 homolog Srp1p is phosphorylated at serine-67 just downstream of the IBB domain. However, mutation of this site resulted in no discernable phenotypes (Azumaet al. 1997). This serine is conserved, but not necessarily phosphorylated, in most conventional importin α’s (Azumaet al. 1997), including the three Drosophila importin α’s. Török et al. (1995) speculated that Drosophila Importin α2 was phosphorylated at this site (serine-56) due to its consensus cdc2 phosphorylation sequence. We tested their hypothesis by introducing a mutant importin α2 transgene containing alanine instead of serine at position 56 (S56A). As shown in Figure 5B, ovarian Importin α2 S56A appeared as a single band that comigrated with the lower of the two wild-type Importin α2 bands. We conclude that serine-56 in Importin α2 is probably phosphorylated; however, we cannot rule out the possibility that serine-56 is required for the phosphorylation of some other residue. Interestingly, an importin α2 S56A transgene efficiently rescued α2-/- female sterility. Specifically, 90% (18/20) of α2-/- females expressing an importin α2 S56A transgene were fertile, compared with 0% (0/20) of α2-/- females containing the transgene but not the Gal4nanos-VP16 driver. In the same cross, 90% (18/20) of α2+/- females expressing importin α2 S56A were fertile. These results demonstrate that, as in yeast, the phosphorylation of Importin α2 is not required for an essential in vivo function. Therefore, the paralog-specific role of Importin α2 in female gametogenesis is not likely due to its phosphorylation.

Effect of importin α gene dosage on fertility: The finding that all three importin α transgenes rescued α2-/- male sterility is consistent with the notion that a threshold concentration of some combination of the three importin α’s normally perform an important function that any single paralog is capable of performing when ectopically expressed. If true, then the partial reduction in the concentration of any two importin α’s might also cause sterility. This was tested by simultaneously reducing the gene dosage from two to one for two different importin α genes. Because discrete mutations in either importin α1 or α3 genes were not available, we employed chromosomal deficiencies in combination with the importin α2D14 allele. Males and females heterozygous for both importin α2D14 and deficiencies that uncovered either the importin α1 (Df (3L) kto2) or the importin α3 (Df (3R) by416) genes (analysis not shown) were tested for fertility. Male flies heterozygous for importin α1 and α2 were significantly less fertile than males heterozygous for only the importin α2D14 allele or the importin α1 deficiency alone (Table 1). Importin α1 and α2 double heterozygotes also produced greatly reduced numbers of motile sperm (not shown). In contrast, males heterozygous for importin α2 and α3 genes exhibited wild-type levels of fertility (Table 1). Females heterozygous for the importin α2D14 mutation and the importin α1 or α3 deficiencies were fully fertile (Table 1) and laid morphologically wild-type eggs. Thus, spermatogenesis appears to be more sensitive than oogenesis to overall importin α levels.

DISCUSSION

Animals contain both conserved (conventional) and organism-specific (nonconventional) importin α genes (Figure 1). Little is known about the individual physiological roles of the various animal importin α’s or their functional interplay in vivo. Conventional importin α genes fall into three clades designated α1, α2, and α3 (Figure 1; see also Maliket al. 1997; Köhler et al. 1997, 1999). Here we report the cloning and phylogenetic analysis of the first complete set of importin α1, α2, and α3 genes from an invertebrate, D. melanogaster, thereby demonstrating that the conventional importin α gene family arose in animals prior to the split between invertebrates and vertebrates. On the basis of parsimony arguments and the fact that importin α2 and α3 genes are more similar to one another than to α1 genes (Figure 1), the phylogeny supports, but does not prove, that importin α2 and α3 genes arose from an α1 progenitor(s). Less likely is the alternative hypothesis that importin α2 and α3 progenitors were lost from fungal and plant lineages (Aravindet al. 2000). The C. elegans genome, which contains a single conventional importin α3 (IMA3) and two divergent worm-specific genes (Geles and Adam 2001), shows that the conventional importin α gene family is not strictly conserved in all animals. However, it is possible that the two nonconventional worm importin α genes (IMA1 and -2) derive from progenitor α1 and α2 genes. Because IMA1 and IMA2 are restricted to the germline, they may have evolved as rapidly as have many sex determination genes (de Bono and Hodgkin 1996; Hansen and Pilgrim 1999).

The phylogeny of the conventional importin α gene family (Figure 1) supports the adoption of a numbering scheme—importin α1, α2, and α3—to designate to which conserved clades a particular importin α gene belongs. Additional levels of gene duplication and (presumably) functional specialization have in many species given rise to additional conventional isoforms. For example, Schizosaccharomyces pombe contains two importin α1 isoforms, and humans contain a total of eight importin α1, α2, and α3 genes. However, because it is not always clear how, or if, isoforms within individual clades are related among species, the current phylogenetic analysis does not support a numbering system for individual isoforms (e.g., importin α1-a, α1-b, α1-c, etc.).

The presence in the Drosophila genome of a single representative of each conventional clade simplifies the analysis of conventional animal importin α genes and argues that flies are the most suitable genetic system for investigating their functional relationships in vivo. Previous genetic studies in Drosophila addressed the in vivo functions of importin α2 and α3 genes (Küssel and Frasch 1995; Töröket al. 1995; Máthéet al. 2000). The initial description of overgrowth of hematopoietic organs in homozygous importin α2 mutant flies (Küssel and Frasch 1995; Töröket al. 1995) proved to be incorrect (G. Adam and I. Kiss, personal communication). Similarly, we found that the mutant phenotypes ascribed to the insertion of a P element upstream of importin α3 (Máthéet al. 2000) could be recombined away from the P element (A. Mason, unpublished observations).

In our study, bona fide importin α2 null alleles provided by I. Kiss were used to investigate functional interactions among the conventional Drosophila importin α genes during gametogenesis. Interestingly, the only phenotypes associated with homozygous null importin α2 mutations are male and female sterility. However, we cannot rule out the possibility that maternally derived Importin α2 also serves a role in early embryonic development, since oogenesis is disrupted in females lacking Importin α2. In gametogenesis, we have found that importin α1 and α3 transgenes can replace importin α2 during spermatogenesis but not during oogenesis. These rescue experiments suggest that Importin α2 has distinct roles during male and female gametogenesis.

Importin α2-/- males exhibit defects in sperm individualization, produce very few motile sperm, and most flies are completely sterile. Sperm individualization appears to function as a quality checkpoint to weed out abnormal spermatids (Fuller 1993; Fabrizioet al. 1998). In fact, mature sperm bundles from wild-type testes often contain fewer than the full complement of 64 spermatids (Tokuyasuet al. 1972), suggesting that the removal of defective spermatids is a normal process. Also, it has been observed that spermatids with aberrant nuclear morphologies fail to individualize, presumably due to the inability of individualization complexes to assemble at the head (Tokuyasuet al. 1977; Fabrizioet al. 1998).

An intriguing example of the relationship between nuclear transport, individualization, and chromosome condensation is the segregation distorter meiotic drive system (Tokuyasuet al. 1977). In males, segregation distorter (SD) chromosomes are preferentially passed on to offspring at the expense of chromosomes that contain SD+ (Sandleret al. 1959). Transmission electron microscopy (TEM) analysis of SD/SD+ sperm bundles revealed that SD+ spermatids remain in a syncytium after SD sperm are individualized (Tokuyasuet al. 1972). It has further been demonstrated that the failure of SD+ spermatids to become individualized is associated with defects in chromatin condensation and nuclear morphology. Specifically, when SD/SD+ postelongated sperm bundles are examined by TEM, SD nuclei stain much more darkly than SD+ nuclei in the same bundle, presumably due to differences in chromatin condensation (Tokuyasuet al. 1977). The primary meiotic drive element on SD chromosomes is a dominant neomorphic mutation known as Sd. Recently, Sd was shown to encode a truncated form of the Ran GTPase-activating protein (RanGAP; Merrillet al. 1999) that is partly mislocalized to the nucleus (Kusanoet al. 2001). Normally, RanGAP is restricted to the cytoplasm and is required to keep cytoplasmic Ran GTP levels low, while the strict nuclear localization of the Ran guanine nucleotide exchange factor (RanGEF) keeps nuclear Ran GTP levels high (Mattaj and Englmeier 1998). The resulting Ran GTP gradient provides the energy source for nuclear transport reactions. Therefore, the introduction of RanGAP activity into the nucleus in SD cells should affect nuclear trafficking pathways (Mattaj and Englmeier 1998). Nuclear transport defects were, in fact, observed in the salivary glands of flies expressing the truncated RanGAP (Kusanoet al. 2001); however, it has not been determined if a nuclear transport defect is the primary cause of the SD phenotype. Nonetheless, it is possible that the chromosome condensation defect in SD sperm is a consequence of defects in nuclear transport. Similarly, the effect of the importin α2D14 mutation, which also likely causes a general defect in nuclear transport, on sperm individualization could be due to faulty chromosome condensation. However, TEM analysis of nuclei in α2-/- postelongated sperm bundles exhibited the staining intensity characteristic of normally condensed chromatin (data not shown). Therefore, the individualization defect observed in α2-/- flies is unlikely due to a gross defect in chromosome condensation.

Although the α2-/- spermatids that fail to individualize lack obvious morphological defects, they probably do have subtle defects that preclude their proper association with individualization complexes. In a percentage of mutant spermatids the defect appears to be subtle enough to escape the individualization checkpoint. Therefore, the severity of the spermatogenesis defect in α2-/- flies varies among spermatids within the same sperm bundle. In fact, in a small percentage of mutant flies enough motile sperm properly develop to produce small numbers of progeny.

Importin α1, α2, and α3 transgenes all rescued the sterility defect of importin α2-/- males, demonstrating for the first time in that all three conventional importin α genes can perform a common physiologically relevant function. These results raise the question of why males missing Importin α2 are almost completely sterile if Importin α1 and α3 can perform the missing function. One possible explanation for the sterility of α2-/- males is that a minimum level of combined importin α1, α2, and α3 expression in the testes is required for efficient spermatogenesis. The partial penetrance of α2-/- male sterility suggests that endogenous Importin α1 and α3 levels in the testes are sufficient in some α2-/- spermatids to perform the in vivo function(s) of the lost Importin α2. Although all three importin α genes are expressed in the testes, α2 mRNA is expressed much more highly than either α1 or α3 mRNAs. At first glance, its loss would be expected to greatly deplete total importin α protein levels, resulting in a significant decrease in general importin α function (but see below). In mouse, representatives of each of the three conventional α1, -2, and -3 clades are all more highly expressed in testes compared to other tissues (Tsujiet al. 1997; Nachuryet al. 1998), suggesting that high importin α levels may also be important during mammalian spermatogenesis. Finally, males heterozygous for both importin α1 and α2 are mostly sterile, as are males heterozygous for both the original oho31-containing importin α2 null mutations and mutations in Ketel, the Drosophila importin β1 homolog (Erdélyiet al. 1997). These observations suggest that spermatogenesis is sensitive to a decrease in the capacity of cNLS-cargo import in the testes.

It may seem odd that a partial defect in a cellular housekeeping process like nuclear transport would cause defects only in spermatogenesis. However, importin α2 is preferentially expressed in the testes compared to other adult tissues (Figure 3), so its loss would be expected to disproportionately affect spermatogenesis. Also, recessive male-sterile mutations are observed 10-15% as often as recessive lethal mutations, and the number of genes that can be mutated to cause male sterility has been estimated at between 500 and 1750 (Fuller 1993). Many of these are caused by nonnull mutations in genes required for general metabolism (Fuller 1993). Although formally null, α2-/- males are “functionally hypomorphic” (nonnull) for general importin α function because importin α1 and α3 are coexpressed in the testes and, presumably, can perform the same functions.

Alternatively, it might not be the combined concentration of all three importin α’s that is important for spermatogenesis. Although all three importin α’s are almost certainly capable of mediating general cNLS import, different paralogs are known to exhibit in vitro preferences for various cNLS cargoes (Miyamotoet al. 1997; Nadleret al. 1997; Prieveet al. 1998; Köhleret al. 1999). Drosophila Importin α2 could have a preference over α1 and α3 for certain cargo whose import is required for spermatogenesis. Importin α1 and α3 might still bind weakly to this cargo(s) in the absence of α2, but not strongly enough to fully supplant α2 function. In this case, when Importin α1 and α3 are expressed at high levels they can more efficiently bind the α2 specific cargo(s) and rescue sterility more consistently. The relatively low level of Importin α2 protein in testes is consistent with the hypothesis that Importin α2 serves a partially distinct specialized activity in the testes. Assuming the Western blot data (Figure 3B) reflect the true relative levels of Importin α2 and α3, the loss of α2 should not greatly reduce total importin α protein levels in testes (we are currently unable to assess Importin α1 levels). Following this line of reasoning it makes more sense that Importin α2 is required in the testes at low levels to perform an important function for which it has a higher specific activity than either α1 or α3. Finally, although the concentration of Importin α2 protein is low in adult males and in whole testes, the protein is selectively expressed in stem cells and primary and secondary spermatogonial cells (Figure 4). Therefore, the requirement for Importin α2 during spermatogenesis could derive from a necessity for a high level of cNLS transport in these specific testis cell types.

Perhaps more interesting than the redundant role(s) that the conventional Drosophila importin α genes play in spermatogenesis is the unique requirement for Importin α2 during oogenesis. Importin α2-/- females lay severely deformed eggs and are completely sterile. Strikingly, the importin α2-/- female sterility phenotype was completely rescued by expression of importin α2 but not α1 or α3, even when the mRNA levels in the ovary from importin α1 and α3 transgenes were higher than mRNA levels from the α2 transgene. Both Northern and Western blot analyses confirmed that all three transgenes were highly expressed in the ovary. In addition, all three transgenes are expressed by the same Gal4nanos-VP16 driver. Therefore, it is unlikely that importin α1 and α3 transgenes are expressed in a different pattern from the rescuing importin α2 transgene. We conclude that Importin α1 and α3 lack a specific biochemical activity(s) that α2 alone is capable of performing in the ovaries. It is formally possible that Importin α1 and α3 could replace Importin α2 in the female germline were they expressed at higher levels than has been achieved in these experiments. However, complete sterility is observed in females that are extensively overexpressing Importin α1 or α3, indicating that the capacity of Importin α1 or α3 to replace Importin α2 must be exceedingly low.

The unique requirement for Importin α2 during oogenesis might be due to a requirement for the nuclear transport of α2-specific NLS cargo. Thus, female infertility could be a pleiotropic consequence of the mislocalization of a nuclear protein(s) that is required for proper oogenesis. Importin α paralog-specific nuclear transport functions have been suggested by the strong preferences shown by importins α1 and α3 for certain transport cargoes, using binding and permeabilized cell import assays (Sekimotoet al. 1997; Köhleret al. 1999; Welchet al. 1999). Although importin α2-specific NLS cargoes have yet to be identified, not enough potential binding partners have been screened to rule out their existence.

Alternatively, Importin α2 may play an essential role in the ovary in a process(es) that is distinct from its role in nuclear transport. Indeed, there is growing experimental evidence that importin α’s can perform import-independent functions. For example, mutations in one of the two S. pombe importin α1’s, Cut15, affect mitotic chromosome condensation without reducing the efficiency of cNLS-cargo import (Matsusakaet al. 1998). Similarly, Srp1p, the importin α gene from S. cerevisiae, has been genetically linked to proteasome-mediated protein degradation (Tabbet al. 2000). Further experiments are needed to determine if nuclear transport-independent and/or NLS-cargo-specific transport functions explain the unique role that Importin α2 plays during oogenesis.

Importin α2 is partially phosphorylated on serine-56 in ovaries and preblastoderm embryos but not in other stages of development (Küssel and Frasch 1995; Töröket al. 1995; Figures 3B and 5B). In contrast, Importin α3 appears not to be phosphorylated at any developmental stage or adult tissues (Máthéet al. 2000) (Figures 3B and 5B). Because of these observations, we tested the hypothesis that phosphorylation of Importin α2 is necessary for its unique role in oogenesis. In fact, female fertility was successfully rescued by an importin α2 transgene that contained a mutation that prevented its in vivo phosphorylation (importin α2 S56A). The function of importin α phosphorylation, which is conserved from yeast to Drosophila, and probably to mammals, remains a mystery. In Drosophila, it has been suggested that phosphorylation of Importin α2 may be related to the observation that in precellularized embryos the protein shifts between the nucleus and cytoplasm in a cell cycle-dependent fashion (Küssel and Frasch 1995; Töröket al. 1995; Máthéet al. 2000). Specifically, Importin α2 protein is cytoplasmic during interphase and nuclear during mitosis (Küssel and Frasch 1995; Töröket al. 1995). Although it is clear that phosphorylation is not essential for Importin α2 function in vivo, it may still play a role in regulating the protein’s subcellular localization.

The fact that extant importin α2 and α3 genes occur only in the genomes of metazoan animals suggests that importin α2 and α3 genes evolved to function in cellular process(es) that are conserved among invertebrate and vertebrate lineages. Several observations suggest that multiple animal importin α genes may have arisen during the evolution of gametogenesis. Importin α2 is required predominantly for gametogenesis in Drosophila, and importin α1, -2, and -3 genes are exceptionally highly expressed in mouse testes (Tsujiet al. 1997; Nachuryet al. 1998). In addition, nonconventional C. elegans IMA1 and IMA2 importin α genes are expressed exclusively in the gonads (Geles and Adam 2000). The sole conventional C. elegans importin α, IMA3, is also expressed in the gonads and is required for gametogenesis. Therefore, there is at least a consistent relationship between the expression patterns and functions of multiple importin α genes and gametogenesis in animals. If true, then the occurrence of a nonconventional importin α clade in monocotyledonous and dicotyledonous plants raises the possibility that a distinct lineage of importin α genes arose to function in plant gametogenesis.

In conclusion, these experiments lay a foundation for future studies on the roles and interplay among conventional animal importin α’s. It will be interesting to identify the specific molecular function(s) of Importin α2 in oogenesis and to learn if the functions of all three conventional Drosophila importin α genes are conserved in vertebrates. Future studies on the importin α gene family in Drosophila will also depend on the isolation and characterization of importin α1 and α3 mutations.

Acknowledgments

We thank I. Kiss for providing the importin α2D14 and importin α2D3 strains; I. Török and E. Máthé for antibodies; M. Frasch for importin α2 cDNA; P. Rørth for the UASp plasmids and the Gal4nanos-VP16 stock; the Bloomington Stock Center for Df(3L)kto2, Df(3R)by416, Gal4Hsp70, and w1118 stocks; K. Jensen for electron microscopy; N. Shulga for confocal microscopy; and members of the Goldfarb and Fleming labs for helpful discussions. This work was supported by grants from the National Institutes of Health (GM40362) and American Cancer Society (BE-104C) to D. S. Goldfarb and National Science Foundation (IBN-9727951) to R. Fleming.

Footnotes

  • Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AAC26055 and AAC26056.

  • Communicating editor: K. V. Anderson

  • Received August 27, 2001.
  • Accepted December 3, 2001.

LITERATURE CITED

View Abstract