The Drosophila melanogaster importin α3 Locus Encodes an Essential Gene Required for the Development of Both Larval and Adult Tissues
D. Adam Mason, Endre Máthé, Robert J. Fleming, David S. Goldfarb

Abstract

The nuclear transport of classical nuclear localization signal (cNLS)-containing proteins is mediated by the cNLS receptor importin α. The conventional importin α gene family in metazoan animals is composed of three clades that are conserved between flies and mammals and are referred to here as α1, α2, and α3. In contrast, plants and fungi contain only α1 genes. In this study we report that Drosophila importin α3 is required for the development of both larval and adult tissues. Importin α3 mutant flies die around the transition from first to second instar larvae, and homozygous importin α3 mutant eyes are defective. The transition to second instar larvae was rescued with importin α1, α2, or α3 transgenes, indicating that Importin α3 is normally required at this stage for an activity shared by all three importin α's. In contrast, an α3-specific biochemical activity(s) of Importin α3 is probably required for development to adults and photoreceptor cell development, since only an importin α3 transgene rescued these processes. These results are consistent with the view that the importin α's have both overlapping and distinct functions and that their role in animal development involves the spatial and temporal control of their expression.

MOST proteins targeted to the nucleus contain nuclear localization signals (NLSs) that are recognized by soluble receptors called karyopherins (Macara 2001; Bednenkoet al. 2003; Weis 2003). Proteins containing classical NLSs (cNLSs) are imported bound to the importin α/β1 heterodimer. Importin α serves as an adapter that links cNLS cargo to the karyopherin importin β1, which ferries the complex through the nuclear pore complex (Macara 2001; Bednenkoet al. 2003; Weis 2003).

The genomes of metazoan organisms encode multiple importin α genes. For example, the human genome encodes six importin α's (Köhleret al. 2002; Cabot and Prather 2003). Phylogenetic analyses of the importin α gene family revealed that most importin α's belong to one of three evolutionarily conserved clades, designated by our nomenclature as conventional α1's, α2's, and α3's (Köhler et al. 1997, 1999; Maliket al. 1997; Máthéet al. 2000; Masonet al. 2002). Conventional importin α's from plants and fungi are all α1's. In contrast, metazoan animals, with the exception of Caenorhabditis elegans (Geles and Adam 2001), contain representatives from each of the three groups. Parsimony arguments suggest that metazoan α2 and α3 genes arose from α1 progenitors in ancestral single-cell eukaryotic lineages.

Vertebrate importin α's show distinct tissue- and cell-type-specific expression patterns (Prieveet al. 1996; Köhler et al. 1997, 2002; Tsujiet al. 1997; Nachuryet al. 1998; Kameiet al. 1999), and human importin α paralogs are differentially regulated in quiescent and proliferating cultured cells and tissue differentiation models (Köhleret al. 2002). In vitro binding assays and permeabilized cell transport assays indicate that α1's, α2's, and α3's have both overlapping and distinct sets of transport cargoes (Miyamotoet al. 1997; Nadleret al. 1997; Sekimotoet al. 1997; Prieveet al. 1998; Köhler et al. 1999, 2001; Welchet al. 1999; Kumaret al. 2000; Nemergut and Macara 2000; Talcott and Moore 2000; Jianget al. 2001; Guillemainet al. 2002; Melénet al. 2003). For example, a vertebrate α3 has unique specificity for RCC1 (Köhleret al. 1999; Nemergut and Macara 2000; Talcott and Moore 2000), Ran BP3 (Welchet al. 1999), interferon regulatory factor 3(Kumaret al. 2000), and adenoviral E1A (Köhleret al. 2001). An α2 selectively bound the glucose transporter GLUT2 (Guillemainet al. 2002) and an α1 specifically transported STAT1 and STAT2 transcription factors (Sekimotoet al. 1997; Melénet al. 2003). Importantly, the preference of importin α's for NLS cargo can be altered when two different substrates are presented together in permeabilized cell transport assays (Kohleret al. 1999). This latter finding underscores the complexity of the functional interactions between importin α's and different NLS cargo and indicates that in vivo studies are needed to unravel the physiological roles of individual importin α's.

In vivo studies are consistent with the notion that the different importin α's play distinct roles in animal development. The RNAi-mediated disruption of an α3 paralog, but not an α2, had a severe effect on the development of porcine embryos (Cabot and Prather 2003). Likewise, RNAi-mediated reductions in the expression of different importin α's caused distinct developmental defects in C. elegans (Geles and Adam 2001; Askjaeret al. 2002; Geleset al. 2002). Specifically, the C. elegans α3 paralog ima-3 is required for meiosis in the developing female germline (Geles and Adam 2001), while the nonconventional ima-2 is required for mitosis (Geleset al. 2002). Proper spindle formation requires ima-2 but not ima-3 (Askjaeret al. 2002).

The Drosophila genome encodes four importin α's (Küssel and Frasch 1995; Töröket al. 1995; Dockendorffet al. 1999; Adamset al. 2000; Máthéet al. 2000; Giarrèet al. 2002; Masonet al. 2002), three of which contain conserved Importin β1 (Görlichet al. 1996; Weiset al. 1996) and cNLS-binding domains (Contiet al. 1998; Dockendorffet al. 1999). The fourth predicted Drosophila importin α, cg14708 (Adamset al. 2000), is extremely divergent, has a weakly conserved IBB domain, and is missing the conserved tryptophan-asparagine array that is crucial for binding to cNLS cargo (Contiet al. 1998).

The three conventional Drosophila importin α paralogs have different developmental stage- and cell-type-specific expression patterns (Küssel and Frasch 1995; Töröket al. 1995; Dockendorffet al. 1999; Máthéet al. 2000; Fanget al. 2001; Giarrèet al. 2002). In addition, α1 and α2, but not α3, accumulate in the nucleus at the onset of mitosis (Küssel and Frasch 1995; Töröket al. 1995; Máthéet al. 2000; Giarrèet al. 2002). Null α2 mutations result in defects in gametogenesis that cause incompletely penetrant male sterility and complete female sterility (Giarrèet al. 2002; Gorjánáczet al. 2002; Masonet al. 2002). The α2 activity essential for female fertility appears to be unique to α2 since it cannot be replaced by the ectopic expression of α1 or α3 transgenes. In contrast, male sterility was rescued to a similar extent by the expression of α1, α2, and α3 (Masonet al. 2002).

Drosophila Importin α3 has been identified as a binding partner of germ cell-less (Dockendorffet al. 1999), DNA polymerase α (Máthé et al. 2000), and heat-shock factor (HSF; Fanget al. 2001). Importin α3 mRNA and protein were not detected in early embryos, coincident with the restriction of HSF to the cytoplasm (Fanget al. 2001). Finally, defects in α3 nuclear export correlate with specific cell fate transformations in mechano-sensory organs observed in hypomorphic mutations in the importin α recycling factor Dcas (Tekotteet al. 2002). In this study we describe the developmental defects associated with severe mutations in α3 and conclude that α3 is required for the development of both larval and adult tissues. Transgene rescue studies demonstrate that the requirement for α3 in larval development can be partially replaced by ectopic expression of α1 or α2. In contrast, only ectopic α3 expression can support development to adults. In the eye, α1, but not α2, can partially replace α3 in at least some cell types, but α3 appears to be uniquely required for the proper differentiation of photoreceptor cells.

MATERIALS AND METHODS

Genetic stocks and markers: Flies were kept on standard cornmeal-dextrose media and grown at 25° unless indicated otherwise. The importin α31/TM6C stock is described in Máthé et al. (2000). The FRT82B, α317-7/TM3 {Kr-GFP}, Sb1 and FRT82B, α3w73/TM3 {Kr-GFP}, Sb1 stocks were created and provided by Tory Herman and Larry Zipursky [University of California (UCLA), Los Angeles]. The α2D14/y+ CyO stock was created by Bernard Mechler (Department of Developmental Genetics, DKFZ, Heidelberg, Germany) and provided by István Kiss (Hungarian Academy of Sciences, Szeged, Hungary) (Töröket al. 1995; Giarrèet al. 2002; Gorjánáczet al. 2002). The Gal4pnos-VP16 stock (Rørth 1998) was provided by Pernille Rørth (EMBL, Heidelberg, Germany). The (1) Gal4tubP (P{w + mC = tubP-GAL4}LL7)/TM3,Sb1; (2) Gal4Act5C (P{w[+mC] = Act5CGAL4} 17bFO1)/TM6B, Tb1; (3) Gal4arm (P{w[+m W.hs] = GAL4-armS}4a P{w[+mW.hs] + GAL4-arm.S}4b)/TM3, Sb1; (4) Sb1/TM3 P{w + mC = ActGFP}JMR2, Ser1; (5) Df(3R)GB104/ TM3, Sb1; (6) Df(3R) by 416/TM3, Sb1; (7) Df(3R)by62/TM1; (8) TM3, Sb1, P{ry[+t7.2] =Δ2-3}99B/Df(3R)C7, ry[506]; and (9) Gal4eye, UASt FLP/CyO; FRT82B, GMR-hid, l(3)CL-R1/TM2 (Stowers and Schwarz 1999) stocks were obtained from the Bloomington Drosophila Stock Center at Indiana University.

PCR of importin α31: Genomic DNA was prepped from single flies of the indicated genotypes and used for PCR. PCR conditions were: 2 μm primers; 1.5 mm MgCl2; 2.5 units Taq DNA polymerase; and 2 mm dATP, dCTP, dGTP, and dTTP (annealing temperature is 62°, 30 cycles). The 5′ P-element primer, PF-2 or primer 1 in Figure 1, had the sequence CGAC GGGACCACCTTATGTTAT (Eggertet al. 1998), and the 3′ primer, α3 3′NsacII or primer 2 in Figure 1, had the sequence CGCACGCCGCGGCCTTTGCCAGCTTCTTCAGG. The resulting band that appears only when importin α31 is present was sequenced and shown to correspond to a P-element insertion ∼780 bp from the ATG of α3 (see also Máthé et al. 2000).

Expression constructs and germline transformations: UASp Importin α transgenes were created by cloning α1 (Masonet al. 2002), α2, or α3 PCR fragments containing a 5′ Cavener consensus sequence (AAAATG; Cavener 1987) and the ∼1.5-kb coding region into KpnI and NotI sites in the UASp P-element transformation vector (Rörth 1998). In contrast to transgenes used previously (Masonet al. 2002), the α2 and α3 transgenes do not contain any α2- or α3-specific 5′ or 3′ untranslated region (UTR) sequences. The UASp α1 transgene contains 1 nucleotide of α1 3′UTR. Transgenic UASp α1, α2, and α3 lines were created using standard germline transformation procedures (Spradling 1986). The UASp α1, α2, and α3 inserts used in this study were all located on the second chromosome.

Larval cuticle preps: Importin α3 alleles were balanced with a green fluorescent protein (GFP)-tagged TM3 chromosome and the appropriate crosses were set up in egg-laying cups. Eggs were laid on apple juice plates overnight. Resulting first instar larvae were examined with a UV dissecting microscope and nonfluorescent larvae were removed to a plate containing standard cornmeal-dextrose media. After ∼24–48 hr dead larvae were isolated and larval cuticles were prepared as previously described (Stern and Sucena 2000), except that 4% paraformaldehyde was used as the fixative. Cuticles were observed by differential interference contrast (DIC) imaging with a Leica TCS NT microscope equipped with UV, Ar, Kr/Ar, and He/Ne lasers. Digital images were processed using Adobe PhotoShop (Adobe Systems, San Jose, CA).

Northern and Western blots: Total RNA and protein were isolated from Drosophila tissues with Tri-Reagent LS (Molecular Research Center, Cincinnati; Chomczynski 1993) following the recommended protocols. Protein isolated from larvae of the indicated stage and genotype was analyzed by Western blot with rabbit anti-Importin α2 (Töröket al. 1995) provided by Istvan Török (DKFZ, Heidelberg, Germany), rabbit anti-α3 (Máthéet al. 2000), or a mouse anti-α-tubulin antibody (Amersham Biosciences, Piscataway, NJ). Blots were developed using alkaline phosophatase-tagged goat anti-rabbit secondary antibodies. The Fermentas Prestained Protein Ladder, an ∼10- to 180-kD size marker (Fermentas, Hanover, MD; Figure 3, A and C), or the GIBCO BRL (Gaithersburg, MD) benchmark size marker (Life Technologies, Grand Island, NY; Figure 3B) were used as size markers. RNA isolated from heterozygous and homozygous α3D93 and α3D165 mutant first instar larvae was analyzed by Northern blot with an α3 full-length random prime 32P-labeled probe (not shown).

Scanning electron microscopy of adult eyes: Flies of the indicated genotypes were dehydrated in a graded ethanol series and stored in 100% ethanol. Flies were critical point dried, coated with gold, and examined by scanning electron microscopy with a LEO 982 FESEM microscope. Digital images were processed using Adobe PhotoShop (Adobe Systems).

Eye sectioning: Fly eyes of the indicated genotypes were embedded in Durcapan resin according to standard procedures (Wolff 2000), sectioned to 1 μm thickness, stained with toluidine blue, and observed by DIC imaging. Digital images were processed using Adobe PhotoShop (Adobe Systems).

Immunofluorescence: Ovaries were dissected from females of the indicated genotypes, fixed in 1× PBS, 4% paraformaldehyde, and blocked in PBS-saponin (1× PBS, 0.1% saponin, and 1% normal goat serum). Ovaries were incubated with a mouse anti-Kelch antibody (Xue and Cooley 1993) diluted 1:1 (Gorjánáczet al. 2002) in PBS-saponin, followed by a goat anti-rabbit FITC-labeled secondary antibody diluted 1:300 in PBS-saponin. DNA was stained with 4′,6-diamidino-2-phenylindole (DAPI) in PBS. Samples were examined by confocal microscopy and digital images were processed using Adobe PhotoShop (Adobe Systems). The anti-Kelch antibody was developed by L. Cooley and provided by the Developmental Studies Hybridoma Bank (Iowa City, IA).

Crosses: Recombination of importin α31: Gal4arm/TM3, Sb1 or Gal4Act5C/TM6B, Tb1 males were crossed to importin α31/TM6C, Sb1, Tb1 virgin females. Gal4arm31 or Gal4Act5C/α31 virgin female offspring were collected and mated to TM3, Sb1/TM6B, Tb1 males. Resulting male offspring were selected for a dark red eye. These males were utilized to make stocks and the presence of the P element in α31 was verified by PCR (see above).

To allow importin α31 to recombine with a “wild-type” third chromosome, w1118 males were crossed to α31/TM6C, Sb1, Tb1 females. Importin α31/+ virgin female offspring were collected and mated to TM3, Sb1/TM6B, Tb1 males. Recombinant α31/ TM6B, Tb1 males were mated individually to virgin females from the original α31/TM6C, Sb1,Tb1 stock. Crosses were incubated at room temperature and each vial was examined for the presence of non-Tb1 pupal offspring. If non-Tb1 pupae were observed the adult offspring from this cross were analyzed. Four recombinant α31 chromosomes were viable over the original α31 chromosome [designated α31(R1), α31(R2), α31(R3), and α31(R4)]. Stocks were made for the α31(R1) and α31(R2) chromosomes, and the presence of the P element was verified by PCR (Figure 1).

Analysis of importin α31 viability: Importin α31/TM6C, Sb1, Tb1 females were crossed to: (a) α31/TM6C, Sb1, Tb1; (b) α31(R1)/TM3, Sb1; (c) α31(R2)/TM6C, Sb1,Tb1; (d) Df(3R)by416/TM3, Sb1; (e) Df(3R)GB104/TM3, Sb1; (f) Df(3R)by62/TM1; or (g) Gal4arm, α 31/TM6B, Tb1 males. Likewise, α31(R1)/TM3, Sb1 flies were crossed to (a) α31(R1)/TM3, Sb1; (b) Df(3R)by416/TM3, Sb1; (c) Df(3R)GB104/TM3, Sb1; or (d) Df(3R)by62/TM1 flies. Finally, α31(R2)/TM6C, Sb1, Tb1 flies were crossed to (a) α31(R2)/TM6C, Sb1,Tb1 or (b) Df(3R)GB104/TM3, Sb1. For all crosses offspring were analyzed for the presence of the appropriate markers on balancers and viability indices were calculated by dividing the number of observed offspring by the number of expected offspring if all nonhomozygous balancer genotypes were equally viable (Table 1). Offspring inheriting two copies of the same balancer (e.g., TM3, Sb1/TM3, Sb1) were assumed to be completely lethal. Offspring inheriting two different balancers (e.g., TM3, Sb1/TM6C, Sb1, Tb1) were often lethal. However, in crosses in which these offspring were observed the viability was calculated assuming them to be fully viable.

Creating deletions in importin α3: Importin α31(R1)/TM3, Sb1 virgin females were crossed to TM3, Sb1, P{ry[+t7.2] =Δ2-3}99B/Df(3R)C7, ry[506]. “Jump start” male offspring of the genotype α31(R1)/TM3, Sb1, P{ry[+t7.2] =Δ2-3}99B were collected and mated to TM3, Sb1/TM6B, Tb1 virgin females. Resulting white-eyed, non-ebony, TM3, Sb1 or TM6B, Tb1 male offspring were selected and mated individually to TM3, Sb1/ TM6B, Tb1 virgin females. After ∼5 days of mating 10–20 males were pooled together for a genomic DNA extraction. Approximately 3 μl of this genomic DNA was then used in a PCR reaction (60° annealing temperature, 1.5 mm MgCl2) with the α3 3′NSacII primer, primer 2 in Figure 1 (sequence above), and the α3 5′ prom 2 primer, primer 3 in Figure 1 (α3 5′ prom 2 sequence, CCAGTTCATTGCTGTTGCTCC). Small deletions in α3 were detected by the presence of a smaller PCR product. If a pool of DNA was shown to contain an α3 deletion, DNA was extracted separately from offspring of each of the 10–20 males. This DNA was utilized in the PCR reaction as described above, enabling the identification of the specific line that contained the deletion. The shortened PCR products were gel purified with QiaQuick columns (QIAGEN, Valencia, CA) and sequenced.

Analysis of importin α3 mutant alleles: Importin α3D93/TM3, Sb1 flies were crossed to: (a) α3D93/TM3, Sb1; (b) FRT82B, α317-7/ TM3 {GFP}, Sb1; (c) FRT82B, α3w73/TM3 {GFP}, Sb1; (d) Df(3R)by416/TM3, Sb1; (e) Df(3R)GB104/TM3, Sb1; (f) Df(3R) by62/TM1; or (g) α31/TM6C, Sb1, Tb1 flies. Likewise, α3D165/TM3, Sb1 or TM6B, Tb1 or TM3 {GFP}, Ser1 flies were crossed to: (a) α3D165/TM3, Sb1; (b) FRT82B, α317-7/TM3 {GFP}, Sb1; (c) FRT82B, α3w73/TM3 {GFP}, Sb1; (d) Df(3R)by416/TM3, Sb1; (e) Df(3R)GB104/TM3, Sb1; (f) Df(3R)by62/TM1; or (g) α3D93/TM6B, Tb1. Finally, Df(3R)GB104/TM3, Sb1 flies were crossed to (a) FRT82B, α317-7/TM3 {GFP}, Sb1 or (b) FRT82B, α3w73/TM3 {GFP}, Sb1 flies. For all crosses offspring were analyzed for the presence of the appropriate markers on balancers and viability indices were calculated as described above (Table 2).

To determine the approximate stage of lethality the importin α3 mutant alleles and deficiency chromosomes were balanced with TM3 {GFP} chromosomes and the appropriate crosses were repeated in egg-laying cups. Embryos were allowed to hatch and early first instar larvae were sorted by fluorescence. Nonfluorescent first instar larvae were collected on a cornmeal agar plate and their development was observed at 25°.

Rescue of importin α3D93/α3D93 lethality: Male flies of the genotype UASp importin α1, α2 or α3/UASp α1, α2, or α3; FRT82B, α3D93/TM3 {GFP}, Ser1 (males carrying UASp α1 had the original α3D93 chromosome instead of FRT82B, α3D93) were crossed to virgin females of the genotype Gal4tubP, α3D93/TM3 {GFP}, Ser1. Progeny were scored at the onset of pupariation for fluorescence when observed through the side of the vial with a UV dissecting microscope. The resulting adult progeny were scored for the presence of the Ser1 marker on the TM3{GFP}, Ser1 chromosome. Viability indices were calculated by dividing the number of observed offspring by the number of expected offspring if all nonhomozygous balancer genotypes were viable to the indicated stage (Table 3). Due to the partial penetrance of Ser1, all non-Ser1 flies were assayed for fluorescence with a UV dissecting microscope to conclusively determine their genotype. As a negative control, α3D93/TM3{GFP}, Ser1 males were mated to Gal4tubP, α3D93/TM3{GFP}, Ser1 females.

The approximate stage of lethality for UASp importin α1, α2, or α33D93/Gal4tubP, α3D93 offspring was determined as previously described. FRT82B, α3D93/Gal4tubP, α3D93 and α3D93/ Gal4tubP, α3D93 offspring served as negative controls.

Rescue of importin α3D93/α317-7 lethality: Male flies of the genotype UASp importin α1, α2, or α3/CyO; Gal4tubP, α3D93/TM3 {GFP}, Ser1 were crossed to females of the genotype FRT82B, α317-7/TM3 {GFP}, Sb1. Progeny were scored at the onset of pupariation as previously described. It was assumed that all nonfluorescent pupae had inherited the UASp importin α transgene and not CyO. Adult offspring were scored for the presence of the CyO, TM3 {GFP}, Ser1, and TM3 {GFP}, Sb1 balancers using the appropriate markers. Viability indices were calculated as previously described, except we assumed that all CyO; Gal4tubP, α3D93/ FRT82B, α317-7 offspring died as first/second instar larvae and, therefore, zero offspring of this genotype were expected at later stages. The viability of adult progeny was calculated for (a) UASp α1, α2 or α3; Gal4tubP, α3D93/ FRT82B, α317-7 experimental flies and (b) UASp α1, α2, or α3; Gal4tubP, α3D93/TM3 {GFP}, Sb1 positive control flies (Table 4). As a negative control Gal4tubP, α3D93/TM3 {GFP}, Ser1 males were crossed to FRT82B, α317-7/TM3{GFP}, Sb1 females and the viability of Gal4tubP, α3D93/FRT82B, α317-7 offspring was calculated at puparium, pharate adult, and adult stages (Table 4). Due to the partial penetrance of Ser1, all non-Ser1 flies were assayed for fluorescence with a UV dissecting microscope to conclusively determine their genotype. The approximate stage of lethality for UASp α1, α2, or α3;FRT82B, α317-7/Gal4tubP, α3D93 off-spring was determined as previously described. FRT82B, α317-7/Gal4tubP, α3D93 and FRT82B, α317-7/α3D93 offspring served as negative controls.

Generating homozygous importin α3D93 eyes using EGUF/hid: UASt FLP, Gal4eye/CyO; FRT82B, GMR-hid, l(3)CL-R1/TM6B females were crossed to: (a) FRT82B, α3+, 87E P-lacW [w+]/FRT82B, importin α3+, 87E P-lacW [w+] or (b) FRT82B, α3D93/ TM6B flies. Adult offspring of the genotype (a) UASt FLP, Gal4eye/+; FRT82B, GMR-hid, l(3)CL-R1/FRT82B, α3+, 87E P-lacW [w+] or (b)UASt FLP, Gal4eye/+; FRT82B, GMR-hid, l(3)CL-R1/FRT82B, α3D93 were selected for and their eyes were analyzed by scanning electron microscopy (SEM) and tangential sectioning.

Rescue of homozygous importin α3D93 eyes: UASt FLP, Gal4eye/ CyO; FRT82B, GMR-hid, l(3)CL-R1/TM6B females were crossed to UASp importin α1, α2, or α3/CyO; FRT82B, α3D93/TM6B males. Adult offspring of the genotype UASt FLP, Gal4eye/UASp α1, α2, or α3; FRT82B, GMR-hid, l(3)CL-R1/FRT82B, α3D93 were selected for and their eyes were analyzed by SEM and tangential sectioning.

RESULTS

Phenotypes associated with importin α31 can be removed by recombination: A Drosophila importin α3 hypomorphic mutation, α31, was reported to greatly reduce viability, and all surviving females were sterile (Máthéet al. 2000). The α31 allele is associated with a P-element insertion (P-lacW [w+]) located ∼780 bp upstream of the start codon (Máthéet al. 2000). To determine if UASp α1, α2, or α3 transgenes could rescue the homozygous α31 phenotypes a Gal4arm driver was recombined onto the same chromosome as α31. Unexpectedly, the Gal4arm, α31 chromosome was homozygous viable (Table 1) and homozygous females were fertile (not shown). The presence and correct position of the α31 P element was confirmed by PCR (Figure 1). Therefore, the low viability and female sterility of α31 flies is likely not due to the hypomorphic α3 mutation. Alternatively, it is possible that the chromosome containing the Gal4arm driver carries a suppressor of the α31 mutation.

View this table:
TABLE 1

Viability of importin α.31 chromosomes

To distinguish between these possibilities we repeated the recombination experiment using an independent third chromosome from a “normal” w1118 stock. This analysis showed that 4 out of 74 recombinant P-lacW [w+]-containing chromosomes [importin α31(R1), α31(R2), α31(R3), and α31(R4)] supported good viability over the original α31 chromosome (Table 1; not shown). The presence and position of the P-element insert was confirmed by PCR for two of the recombinant chromosomes [α31(R1) and α31(R2); Figure 1]. Flies homozygous for α31(R1) were viable (Table 1) and the females were fertile (not shown). These results confirm that the original P-lacW[w+] insertion could not have been solely responsible for the reported phenotypes. For unknown reasons flies homozygous for α31(R2) were homozygous lethal despite the fact that they were viable over the original α31 allele (Table 1).

We also examined the viability of the original and recombinant importin α31 alleles over various deficiencies. Flies carrying the original α31 allele or the recombinant α31(R1) were viable and female progeny were fertile over Df(3R)by416, breakpoints 085D10–12;085E01–03; Df(3R)GB104, breakpoints 085D12;085E10; and Df(3R) by62, breakpoints 085D11–14; 085F06 (Table 1; Máthéet al. 2000; not shown). Further experiments demonstrated that all three of these deficiencies uncover α3 (Table 2; Figure 3; not shown). Taken together, these results indicate that a second site mutation(s) on the original α31 chromosome either caused or contributed strongly to the published phenotypes (Máthéet al. 2000). The discovery that the α31 allele is viable and female fertile over deficiencies and loss-of-function α3 alleles (see below) suggests that the second-site mutation is the major contributor to these phenotypes. Although α31 flies were demonstrably hypomorphic for α3 protein expression (Máthéet al. 2000), the reduced α3 levels are apparently not deleterious to the organism.

Figure 1.

—PCR of importin α31 chromosomes and creation of deletion mutants in importin α3. (A) PCR to detect the P element in importin α31. Importin α3 and cg8273 coding regions are shown in white, the noncoding region is shown in gray, the P-element insertion in α31 is indicated by the black triangle, and primers used for PCR are indicated with arrows. Nucleotide numbers are indicated relative to the start of the α3 coding region (ATG =+1). To verify the presence of the P element in α31, DNA was isolated from single flies of the indicated genotypes and analyzed by PCR using a P-element primer (primer 1) and a primer in α3 (primer 2). Lane 1, 1-kb DNA ladder; lane 2, w1118; lane 3, α31/TM6C; lane 4, Gal4Act5C, α31/TM6B; lane 5, Gal4arm, α31/TM6B; lane 6, Gal4arm, α31/Gal4arm, α31; lane 7, Gal4arm/TM6B; lane 8, α31(R1)/TM6B; lane 9, α31(R1)/α31(R1); lane 10, α31(R2)/ TM6C. DNA size markers are shown in kilobases. (B) PCR to detect P-element excision-induced deletions in importin α3. DNA isolated from P-element-excised flies was analyzed by PCR using primers that flank the P-element insertion site (primers 2 and 3 in A). Two lines, importin α3D93 and α3D165, carrying small deletions in the α3 locus were identified (lanes containing bands <1.4 kb). DNA size markers are shown in kilobases. (C) Diagram of importin α3 alleles. Importin α3 and cg8273 coding regions are shown in white, the noncoding region is shown in gray, and deleted regions in α3D93 and α3D165 are indicated by black boxes. The α317-7 allele is also used in this study and contains a stop codon in place of amino acid W132 (T. Herman and L. Zipursky, personal communication). Nucleotide numbers are indicated relative to the start of the α3 coding region (ATG =+1).

P-element excision-induced alleles of importin α3: In search of more severe importin α3 mutations, a P-element excision mutagenesis was used to create small deletions in the α3 coding sequence. The P element in the clean α31(R1) stock was mobilized and offspring were selected for the loss of the P element (loss of white+). A PCR assay using primers flanking the P-element insertion site was used to screen for imprecise P-element excision events (Figure 1). Two small deletions in the α3 gene (α3D93 and α3D165) were identified (Figure 1). Sequencing of the shortened PCR products revealed that the α3D93 deletion removes 897 bp from the 5′ region of α3, including the coding sequence for the first 20 amino acids. The α3D165 deletion removes 619 bp but no coding sequence (Figure 1). Because the original α31 P element was inserted in the 5′ region between α3 and the convergently transcribed predicted open reading frame cg8273 (Adamset al. 2000), only ∼125 nucleotides remain upstream of one of the predicted start sites of cg8273 in α3D93 and α3D165 (Figure 1). Thus, it is possible that the expression of cg8273 will be affected by α3D93 and α3D165 deletions.

View this table:
TABLE 2

Viability of importin α3 mutants

The importin α3D93 and α3D165 mutations were both homozygous lethal and lethal over each other (Table 2). In addition, α3D93 was lethal over Df(3R)by416, Df(3R) GB104, and Df(3R)by62 (Table 2). Importin α3D93/α31 flies were completely viable (Table 2) and the females were fertile (not shown), confirming our conclusion that the α31 allele does not yield severe phenotypes. In conclusion, this strategy produced recessive lethal α3 mutants, thereby demonstrating that α3 is an essential gene in flies.

Importin α3 deletion mutants do not develop past larval stages: To determine when importin α3 mutant flies die, α3D93 and α3D165 chromosomes were balanced with a TM3 chromosome marked by GFP (TM3{GFP}, Ser1). In this fashion homozygous mutant offspring could be distinguished from heterozygotes by GFP fluorescence. Homozygous α3D93 and α3D165 offspring completed embryogenesis and formed normal-appearing first instar larvae. Most homozygous mutant α3D93 and α3D165 first instar larvae were able to duplicate their mouth hooks in preparation for molting (Figure 2, B and C), but most died before completing ecdysis. Of the offspring that did begin ecdysis most died while still attached or immediately adjacent to their first instar cuticles. Rarely first instar cuticles were found that were not associated with dead larvae. These larvae presumably died as very early second instar larvae, since we never observed any crawling homozygous mutant second instar larvae. Offspring carrying α3D93 or α3D165 deletions over Df(3R)by416 or Df(3R)GB104 also died around the first instar molt (not shown). We conclude that α3 serves an essential role in larval development and that the majority of α3 mutant offspring die before or during ecdysis of the first instar larval molt.

importin α3D93 and α3D165 larvae express little if any full-length Importin α3 protein: Since the phenotypes of importin α3D93 and α3D165 were no more severe over deficiencies than when homozygous, it is likely that both mutations are either null or strongly hypomorphic. This conclusion was supported by Northern blot analysis showing that α3D93/α3D93 and α3D165/α3D165 first instar larvae contained very little or no α3 mRNA (not shown, see materials and methods). This was expected since the two deletions removed significant portions of the α3 5′UTR (Máthéet al. 2000; Figure 1C). Immunoblot analysis was used to investigate whether the first-second instar arrest phenotypes of homozygous α3D93 and α3D165 mutants are due to the complete or partial absence of α3 protein. Total protein was isolated from first instar larvae and examined by immunoblotting with an antiserum against the C-terminal domain of α3 (Máthéet al. 2000). As shown in Figure 3, homozygous α3D93, α3D165, and α3D93/Df(3R)GB104 first instar larvae contained very little or no full-length α3 protein. Curiously, a faster-migrating anti-α3 cross-reactive band appeared in both wild-type and mutant larvae (Figure 3A; see below). The identity of this band is currently unknown.

Rescue of importin α3D93 larval lethality: If the developmental defects of importin α3D93 and α3D165 mutants are due to the lack of α3 protein, the defects should be rescued by an α3 transgene. A Gal4tubP driver was used to express a UASp α3 transgene in a homozygous α3D93 background. As shown in Table 3, the α3 transgene rescued many α3D93/α3D93 and α3D93/α3D165 offspring to the pigmented pharate adult stage and α3D93/Df(3R)GB104 offspring to pupal stages, but none to full adulthood. Some rescued larvae completed the second and third instar molts to form morphologically normal pupae before dying, and a few became well-developed pigmented pharate adults that never eclosed (Table 3). When expressed with a Gal4Act5c driver the α3 transgene rescued α3D93/α3D93 offspring to wandering third instar larvae. These partially rescued larvae sclerotized their cuticles but did not extend their spiracles and died before becoming puparia. Full rescue may require the expression of the α3 transgene in the correct tissues at the correct time and at appropriate levels. The fact that Gal4tubP and Gal4Act5c drivers rescued to different degrees supports this possibility. Another concern is the likelihood that the α3D93 and α3D165 deletions affected not only the expression of both α3's but also the divergently transcribed cg8273 (Figure 1). The partial rescue by the α3 transgene does, however, demonstrate that the death of α3D93/α3D93 larvae around the first molt is due to defects in α3 and not in cg8273.

An objective of this study is to determine if importin α1's, α2's, and α3's have distinct and/or overlapping functions. Previously, using the Gal4/UAS expression system (Brand and Perrimon 1993), we showed that α1, α2, and α3 transgenes all rescued the partial male sterility of α2 null flies, but only α2 transgenes rescued the sterility of α2 null females (Masonet al. 2002). Thus the role of α2 in gametogenesis appears not to be redundant with α1 and α3 in females but is redundant in males. A similar approach was taken to determine if α1 and α2 transgenes could rescue the death of α3 mutant offspring.

For these rescue experiments to be meaningful it is important that the transgenes be expressed at reasonable levels in mutant first instar larvae. Extracts from homozygous importin α3D93 first instar larvae expressing UASp α1, α2, or α3 transgenes were examined by Western blot using antibodies directed against α2 (Töröket al. 1995), α3 (Máthéet al. 2000), or α-tubulin (Figure 3B). As shown in Figure 3B, α2 and α3 were both expressed at high levels in first instar larvae carrying the UASp α2 or α3 transgenes, respectively. A slower-migrating anti-α3 cross-reactive band in mutant first instar larvae expressing UASp α1 (* in Figure 3B) is consistent with results observed when UASp α1 was expressed in α2 mutant ovaries with the Gal4pnos-VP16 driver (Masonetal. 2002). Since α1 is predicted to be ∼60 kD and α3 is predicted to be ∼56.6 kD, it is likely that this band represents a cross-reaction of α1 with the anti-α3 anti-serum. We conclude that all three transgenes are expressed at high levels in first instar larvae.

Figure 2.

—Stage of lethality for importin α3 mutant larvae. Larval cuticles were prepared from (A) first instar larvae of the genotype importin α3D93/TM3 {GFP} or dead larvae of the genotypes (B) α3D93/α3D93, (C) α3D165/α3D165, and (D) α317-7/ α3D93. Note the duplicated mouth hooks in B–D.

The expression of either the importin α1 or α2 transgene delayed the death of homozygous α3D93 offspring. Many offspring expressing UASp α1 with the Gal4tubP or Gal4Act5c drivers completed the first instar molt before dying as late-stage second instar larvae, although some larvae appeared to survive to early third instar stages (not shown). Similarly, many α3D93/α3D93 offspring expressing UASp α2 also survived the first instar larval molt before dying as late second or early third instar larvae. In some trials a few of these animals developed to puparia (not shown). We note that α3D93/α3D93 offspring expressing α1 or α2 were not able to reach pupal stages as efficiently as mutant offspring expressing α3 (Table 3). We conclude that α1 and α2 can, at least partially, replace the function(s) of α3 during larval development.

Nonsense mutation alleles of importin α3: The results described above suggest that α3 is important for developmental events during or after the first larval molt. There are two caveats to this conclusion. First, the α3D93 and α3D165 deletions may affect the expression of cg8273, a divergently transcribed gene whose possible role in development is not known. Second, it is not certain whether the α3D93 and α3D165 alleles are null or hypomorphic. If hypomorphic, it is possible that α3 is required for even earlier stages of development. To address these issues we used two nonsense alleles, α317-7 containing a stop codon after amino acid (aa) 131 in the second ARM repeat and α3w73 containing a stop codon after aa 158 within the third ARM repeat (kindly provided by T. Herman and L. Zipursky). Since the nonsense mutations were isolated in a screen using heavy mutagenesis they may carry multiple mutations (T. Herman and L. Zipursky, personal communications). However, by working with independently derived nonsense mutations over the α3D93 deletion chromosome we can alleviate the possible contribution of recessive mutations in cg8273 or other loci to the phenotype. Both α317-7 and α3w73 were completely lethal over α3D93, α3D165, and Df(3R)GB104 (Table 2; not shown). Importin α3w73/α3D93 offspring died as mid to late second instar larvae, indicating that the α3w73 allele may not be null. In contrast, many α317-7/α3D93 and α317-7/Df(3R)GB104 larvae died at the first/second instar molt with duplicated mouth hooks (Figure 2D; not shown), although some died as early second instar larvae. More α3D93/α317-7, like α3D93/ α3D165, offspring appeared to complete ecdysis and died as early second instar larvae compared to α3D93/α3D93 and α3D165/α3D165 mutants. This is possibly due to subtle genetic background differences. Generally, then, α317-7, α3D93, and α3D165 alleles all cause death at or soon after the first larval molt. Therefore, the larval deaths of α3D93 and α3D165 mutant flies reflect defects in α3 expression that are independent of cg8273.

Figure 3.

—Importin α3 protein levels in importin α3 mutants and Importin α transgene expression. (A) Protein isolated from first instar larvae of the indicated genotype was examined by Western blot with anti-Importin α3 or anti-α-tubulin antibodies. (B) Protein isolated from first instar larvae of the indicated genotype was examined by Western blot with anti-α2, anti-α3, or anti-α-tubulin antibodies. (C) Protein isolated from first or third instar larvae of the indicated genotype was examined by Western blot with anti-α3 or anti-α-tubulin antibodies. The estimated sizes of the markers are indicated next to the arrows; the thick arrowhead indicates the size of the full-length α3 protein, and the thin arrowhead represents a smaller anti-α3-cross-reactive band (A–C).*, the slower migrating anti-α3-cross-reactive band that appears only when α1 transgenes are expressed (B); **, a second slower-migrating anti-α3-cross-reactive band that appears only when an α2 transgene is expressed (B and C).

As expected, levels of Importin α3 protein in first instar larvae carrying α317-7 over α3D93 or Df(3R)GB104 were extremely low or undetectable (Figure 3C). The faster-migrating cross-reactive band described above is also apparent in extracts from α317-7/Df(3R)GB104 flies (Figure 3C). This finding rules out the possibility that this band is an N-terminal truncation expressed from the α3D93 chromosome. The band is unlikely to be a degradation product of maternal α3 since the faster-migrating band is also present in extracts from α3D93/ α3D93 and α317-7/α3D93 larvae that were rescued to third instar by an α2 transgene (see below). The identity of this faster-migrating band is a mystery, since it appears in extracts from flies (potentially) expressing native, N-terminally truncated (α3D93 and α3D165), and C-terminally truncated (α317-7) α3 proteins, each of which has a different predicted mass. The simplest explanation is that the band is a spurious cross-reactive species that is unrelated to α3.

Importantly, expression of UASp importin α3 with Gal4tubP was able to rescue α317-7/α3D93 and α3w73/α3D93 offspring to fully viable fertile adults (Table 4; not shown). Thus, it is likely that the inability to rescue α3D93/α3D93 flies to adulthood with an α3 transgene is due to disruption of cg8273 expression. In contrast, UASp α1 and α2 transgenes were both unable to rescue α317-7/α3D93 flies to adulthood (Table 4). However, the α2 transgene was able to rescue some offspring to abnormal pharate adults, although most died at earlier stages (Table 4). Dissecting these partially rescued pharate adults from the pupal cases revealed that some adult structures were at least partially formed, including wings, tergites, sternites, thorax, and bristles. Expression of the α1 transgene rescued less well, as these offspring survived at best to late second or early third instar larvae. It is worth noting that heterozygous α3D93 flies expressing UASp α1 have a lower than expected viability (Table 4). Thus, it is possible that the overexpression of α1 causes a partial-dominant lethal phenotype. On the basis of these rescue experiments we conclude that α3 serves a mostly redundant function during larval development. However, since α1 and α2 transgenes do not rescue to adult stages it is likely that the role of α3 in the development of some adult tissues cannot be replaced by α1 or α2.

View this table:
TABLE 3

Rescue of importin α3D93/α3D93 lethality

View this table:
TABLE 4

Rescue of importin α3D93/α317-7 lethality

Analysis of importin α3D93 mutant eyes: The observation that importin α3 does not appear to play an exclusively paralog-specific role in larval development led us to examine the effects of the loss of α3 on the development of adult tissues. To address this issue we created an FRT82B, α3D93 chromosome that can be used to create clones of homozygous α3D93 cells in an otherwise heterozygous fly using the FLP/FRT recombinase system (Xu and Harrison 1994). We subsequently generated eyes that were homozygous for α3D93 using a stock that expresses the FLP recombinase in the eye and contains a FRT82B, GMR-hid chromosome (Stowers and Schwarz 1999). The GMR-hid element drives the expression of the apoptosis-inducing gene hid under the control of the eye-specific GMR enhancer. Consequently, any eye cell that contains one or two copies of GMR-hid undergoes apoptosis during early pupal stages. Therefore, only cells that have lost this chromosome through mitotic recombination survive to form adult eyes (Stowers and Schwarz 1999; Figure 4A). Consistent with previous results, we observed that FRT82B, α3+/FRT82B, GMR-hid flies had well-formed eyes when FLP recombinase was expressed in the eye (Stowers and Schwarz 1999; Figure 4B), although the photoreceptor patterning observed in eye sections appears to be slightly defective (Figure 5A).

Figure 4.

—SEM of importin α3 mutant and rescued eyes. Flies of the genotypes (A) Gal4eye, UASt FLP; FRT82B, GMR-hid/TM6B; (B) Gal4eye, UASt FLP; FRT82B, GMR-hid/FRT82B, importin α3+; (C) Gal4eye, UASt FLP; FRT82B, GMR-hid/FRT82B, α3D93; (D) Gal4eye, UASt FLP/UASp α1; FRT82B, GMR-hid/FRT82B, α3D93; (E) Gal4eye, UASt FLP/UASp α2; FRT82B, GMR-hid/FRT82B, α3D93; and (F) Gal4eye, UASt FLP/UASp α3; FRT82B, GMR-hid/FRT82B, α3D93 were critical-point dried and examined by scanning electron microscopy.

Eyes homozygous for the importin α3D93 allele were highly defective (Figure 4C). Under the light microscope these eyes appeared “glassy.” The α3D93 mutation did not cause a general cell-lethal phenotype in the eye, since ommatidial-like structures were visible (Figure 4C). However, the ommatidia were disorganized and did not fully develop, and interommatidial bristles were often missing (Figure 4C). Examination of tangential sections of homozygous α3D93 eyes demonstrated that the ommatidia were defective, since the photoreceptor cell rhabdomeres were not visible and the ommatidia were severely misshapen (Figure 5B).

Figure 5.

—Tangential sections through importin α3 mutant and rescued eyes. Eyes dissected from flies of the genotypes (A) Gal4eye, UASt FLP; FRT82B, GMR-hid/FRT82B, α3+; (B) Gal4eye, UASt FLP; FRT82B, GMR-hid/FRT82B, importin α3D93; (C) Gal4eye, UASt FLP/UASp α1; FRT82B, GMR-hid/FRT82B, α3D93; (D) Gal4eye, UASt FLP/UASp α2;FRT82B, GMR-hid/FRT82B, α3D93; (E) Gal4eye, UASt FLP/UASp α3;FRT82B, GMR-hid/FRT82B, α3D93; and (F) α31(R1)/α3D93 were embedded in resin, sectioned, and toluidine blue stained. Note the absence of photoreceptor rhabdomeres in B–D.

To test whether the defective eye phenotype is truly the result of the lack of Importin α3 activity in the eye, UASp α3 was expressed in α3D93 mutant eyes using the Gal4eye driver. The α3 transgene was able to partially rescue the defect in ommatidia formation (Figure 4F), demonstrating that the glassy-eye phenotype is indeed due to the lack of α3. However, expression of the α3 transgene did not completely rescue the phenotype, as most interommatidial bristles were missing (Figure 4F). Tangential sectioning of these rescued eyes revealed that the photoreceptor cell rhabdomeres were visible, demonstrating that their loss was indeed due to a disruption in α3. However, the photoreceptors in the rescued eyes were not wild type in appearance. They appeared disorganized and misshapen and most ommatidia had an incorrect number of photoreceptors (Figure 5E). This partial rescue may be due to problems with the Gal4eye expression pattern or may be the result of the disruption of the neighboring gene as previously discussed.

To examine the specificity of Importin α3 function in the eye, UASp α1 and α2 transgenes were expressed. Eyes mutant for α3, but ectopically expressing α1, appeared to be at least partially rescued (Figure 4D). Eyes rescued with α1 still appeared partially glassy by light microscopy, but when examined by SEM it was clear that they had more well-developed ommatidia than when the transgene was not expressed (Figure 4D). Tangential sections of these eyes revealed that α1-rescued ommatidia were still largely defective, since no photoreceptor cell rhabdomeres were observed (Figure 5C). Expression of UASp α2 did not appear to affect the phenotype, since homozygous α3D93 eyes expressing α2 looked identical to those not expressing the transgene (Figure 4E) and tangential sections demonstrated that photoreceptor cell rhabdomeres were not present in these ommatidia (Figure 5D). These data suggest that α1, but not α2, is able to partially replace α3 in the eye.

We have also observed that a null mutation in importin α2, α2D14 (Töröket al. 1995; Giarrèet al. 2002; Gorjánáczet al. 2002), was able to enhance the eye defect observed in homozygous α3D93 eyes. Eyes that were homozygous for the α3D93 mutation and heterozygous for the α2D14 allele lacked almost all ommatidial structures. These eyes appeared to be thin sheets of tissue with very little differentiation (not shown). However, for unknown reasons this enhancement was not rescued by the expression of an α2 transgene. Specifically, flies expressing UASp α2 in eyes that were homozygous for α3D93 and heterozygous for α2D14 appeared identical to those that were not expressing the transgene (not shown).

Finally, we looked at whether a hypomorphic importin α3 condition caused slight defects in eye development. To examine this we looked at tangential sections of eyes from α31(R1)/α3D93 flies. This analysis demonstrated that these ommatidia had the correct number, shape, and patterning of photoreceptor rhabdomeres (Figure 5F). We conclude that a hypomorphic α3 condition does not cause defects in eye development.

Importin α3 does not function in ring canal formation: Gorjánácz et al. (2002) have recently demonstrated that Importin α2 is required in the female germline to correctly form ring canals. In homozygous α2D14 ovaries, Kelch (Xue and Cooley 1993) is not targeted to the ring canal correctly, correlating with a defect in transfer of maternal material from the nurse cells to the developing oocyte (Gorjánáczet al. 2002). Expression of a UASp α2 transgene is able to rescue the Kelch mislocalization phenotype observed in α2 null females (Gorjánáczet al. 2002; Figure 6D). To examine the ability of α1 or α3 to function in ring canal formation we examined the localization of the Kelch protein in ovaries from homozygous α2D14 females expressing UASp α1, α2, or α3 transgenes. Consistent with previous observations that α2 mutant females expressing α1 or α3 are sterile (Masonet al. 2002), we found that expression of α1 or α3 did not rescue the mislocalization of Kelch (Figure 6, C and E). In rare cases some accumulation of weak Kelch fluorescence was observed in mutant ovaries expressing α3 (Figure 6E, arrow). It is not known whether this signal represents poorly formed ring canals or, more likely, is a staining artifact. We conclude that α1 and α3 do not function to properly target Kelch to ring canals in the same manner as α2 does.

DISCUSSION

Nuclear transport is facilitated largely by members of the karyopherin gene family. These importins and exportins bind nuclear import or export signal-bearing proteins and ferry them across the nuclear pore complex. Importin α's are adaptors that link many cNLS-containing cargoes to the karyopherin importin β1 (Macara 2001; Bednenkoet al. 2003; Weis 2003). The conventional importin α gene family is composed of three clades, α1's, α2's, and α3's, although fungi and plants contain only α1 genes. With the exception of C. elegans, invertebrate and vertebrate animal genomes encode at least one importin α from each clade (Köhler et al. 1997, 1999; Maliket al. 1997; Máthéet al. 2000; Masonet al. 2002). For example, humans contain three α1's, one α2, and two α3's. There is ample in vitro evidence that conventional importin α's mediate the import of cNLS-containing cargoes as well as paralog-specific NLS cargoes (Miyamotoet al. 1997; Nadleret al. 1997; Sekimotoet al. 1997; Prieveet al. 1998; Köhler et al. 1999, 2001; Welchet al. 1999; Kumaret al. 2000; Nemergut and Macara 2000; Talcott and Moore 2000; Jianget al. 2001; Guillemainet al. 2002; Melénet al. 2003). The in vivo analysis of the importin α gene family in metazoan animals is complicated by the individual paralogs' poorly defined NLS-cargo-binding repertoires, their differing cell type- and tissue-specific expression patterns and levels, and the likelihood that some may perform non-transport-related activities (Matsusakaet al. 1998; Tabbet al. 2000; Grusset al. 2001; Nachuryet al. 2001; Wieseet al. 2001; Askjaeret al. 2002).

A previous study concluded that a hypomorphic mutation in importin α3 was partially lethal with all surviving females being sterile (Máthéet al. 2000). However, these phenotypes did not cosegregate with the P element in the α31 allele, which itself caused no phenotypes. Therefore, even though the α31 allele is hypomorphic, the reported phenotypes were most likely due to a second-site mutation(s). To determine whether more severe mutations in α3 cause phenotypes we generated new 5′ deletion alleles (α3D93 and α3D165) and studied the effects of nonsense mutation alleles (α317-7 and α3w73) provided by T. Herman and L. Zipursky (UCLA). The fact that homozygous α3D93 and α3D165 flies and flies containing α3D93, α3D165, and α317-7 alleles over α3 deficiencies die at approximately the same stage is consistent with all three alleles being null. However, since we could not rule out by Western blot the possibility that these mutants retained low residual levels of α3 protein, they could be severe hypomorphs rather than nulls.

The analysis of these alleles demonstrates that Drosophila Importin α3 is required for the development of both larval and adult tissues. Importin α3 mutant flies die around the first larval molt, and α3 mutant eyes are severely defective and lack photoreceptor cells. The α3 mutant phenotypes are dramatically different from those of α2 mutant flies. Specifically, α2 is required for gametogenesis and apparently not for somatic development (Giarrèet al. 2002; Gorjánáczet al. 2002; Masonet al. 2002). The loss of α2 causes sterility, total in females and partial in males. Interestingly, α1, α2, and α3 transgenes all rescued the male sterility defect to the same degree, but only α2 transgenes rescued the female sterility defect. These results are consistent with α2 playing a paralog-specific role in oogenesis that cannot be performed by either α1or α3 (Masonet al. 2002). Normally, α2 mRNA is expressed in a number of larval tissues and imaginal discs (Töröket al. 1995), but, since α2 null flies develop normally to adulthood—the germline not withstanding—its role in somatic development must be either unimportant or redundant with at least one of the other paralogs.

Figure 6.

—Kelch staining of ovaries from importin α2 null females rescued with importin α transgenes. Ovaries dissected from females of the genotypes (A) w1118; (B) importin α2D14/ α2D14; Gal4pnos-VP16/TM6B; (C) α2D14, UASp α1/ α2D14; Gal4pnos-VP16; (D) α2D14, UASp α22D14; Gal4pnos-VP16; and (E) α2D14, UASp α3/α2D14; Gal4pnos-VP16 were stained with an anti-Kelch antibody (green) and DAPI (blue). Note the Kelch-stained ring canals (arrowheads) in A and D and their absence in B, C, and E. Rarely, some very faint fluorescent anti-Kelch-stained structures are observed in α2D14, UASp α32D14; Gal4pnos-VP16 ovaries (arrows in E).

Transgene rescue studies support the conclusion that Importin α3, like α2, serves both paralog-specific and redundant roles during development. On the basis of our criteria, paralog-specific roles for α3 are those that can be rescued by only α3 transgenes. Redundant functions are those that could be rescued by an α3 and α1 and/or α2 transgenes. The most likely redundant function is the housekeeping transport of cNLS cargoes. This class of phenotype probably arises when only the mutated importin α type, in this case α3, is normally expressed at sufficient levels in the relevant tissue or when a high level of general importin α activity is required.

Drosophila importin α3 mutant offspring complete embryogenesis and hatch to first instar larvae without any apparent defects. The majority of α3 mutant offspring die during the first instar molt. At the end of each larval stage ecdysone pulses signal significant changes in gene expression that are necessary for the generation of second instar larvae (Riddiford 1993; Bender 1995; Thummel 1996; Kozlova and Thummel 2000). It is possible that α3 plays a role in facilitating this transition, perhaps by mediating the nuclear transport of signaling proteins or transcription factors that specify the transition. In this regard, homozygous α3D93 mutants containing a Gal4Act5C-expressed α3 transgene successfully completed first and second instar molts only to die near the transition from wandering third instar larvae to puparium. This is consistent with the notion that α3 is required for both the transition from first to second instar larva and the transition from larva to puparium. Thus, α3 may play a general role in developmental transitions.

Importin α3 is likely required during the first molt for a redundant function. First, since some α3 mutant larvae reach the second instar, there may be enough endogenous α1 and α2 present to partially replace the loss of α3 during the first larval molt. Importin α1 in particular is well expressed in larval tissues (Giarrèet al. 2002). Alternatively, the requirement for α3 during the first molt may be important but nonessential, whether or not α1or α2 is present. The most convincing argument that a redundant function of α3 is required during this developmental transition is the finding that the expression of α1 and α2 transgenes rescued α3 mutant flies to later stages. In conclusion, it is likely that the cause of this phenotype is due to the preferential expression of redundant α3 activity(s) in one or more larval tissues, rather than of an α3-specific activity. Here, the preferential use of α3 to perform a redundant importin α function during larval development is analogous to the role of α2 in spermatogenesis.

The interpretation of the rescue experiments is complicated by differences in the capacity of the various transgenes to rescue the similar defects of homozygous importin α3D93 vs. α3D93/α317-7 flies. For example, only α3 transgenes rescued α3D933D93 flies to pupal stages, whereas both α2 and α3, but not α1, transgenes rescued α3D93/α317-7 flies to pupal stages. Importin α2-rescued α3D93/α317-7 progeny do not properly complete pupation and, consequently, never eclose. Only α3 transgenes are capable of rescuing the latter stages of development through eclosion. We trust the α3D93/α317-7 results more because these flies would not suffer the effects of deleterious recessive alleles potentially present in homozygous α3D933D93 offspring. Therefore, focusing on α3D93/α317-7 results, we conclude that an activity of α3 that is essential for pupation is at least partially redundant with α2. These functional results are consistent with phylogenetic analyses showing that α3's are more closely related to α2's than to α1's (Köhleret al. 1999; Masonet al. 2002; not shown). However, the fact that only an α3 transgene is able to rescue α3D93/α317-7 progeny to adults suggests that α3 does serve an α3-specific function in the development of some adult tissues.

Importin α3 has both redundant and paralog-specific roles in eye development. Homozygous α3D93 eyes appear glassy and lack photoreceptor cell rhabdomeres in adult ommatidia. These phenotypes can be mostly rescued by the expression of α3 transgenes. Only an α3 transgene was able to partially rescue the photoreceptor cell defect, indicating that α3 likely serves a paralog-specific function in the differentiation of these cells. A recent study demonstrated that a dominant-negative Importin β1 protein expressed in the eye caused defects in development of photoreceptor cells (Kumaret al. 2001), suggesting that α3 and Importin β1 may work together to perform a nuclear transport function essential for the development of the eye. We cannot rule out the possibility that α3 is important for eye development only under the EGUF/hid experimental conditions (Stowers and Schwarz 1999).

Expression of importin α1 improved the overall morphology of α3 mutant eyes, but these eyes still lacked recognizable photoreceptor cell rhabdomeres. Importin α1 may rescue the differentiation of nonphotoreceptor accessory cells, like pigment and cone cells, or may improve photoreceptor development enough to allow more efficient specification of accessory cell lineages. Importin α2 expression has little if any effect on the development of α3 mutant eyes. Curiously, an α2 null mutation enhanced the α3 glassy eye phenotype, suggesting that endogenous α2 and α3 may function together during eye development. However, this enhancement could not be rescued by the expression of an α2 transgene. Flies homozygous for the null α2 allele have morphologically wild-type eyes, so α2 does not appear to be required for eye development when α3 is present (Giarrèet al. 2002; Gorjánáczet al. 2002; Masonet al. 2002). Interestingly, α1 was better than α2 at rescuing eye development, but the opposite was true for pupation, where α2 was better than α1. Rather than being contradictory, we believe these results underscore just how complex the physiology of the importin α gene family is likely to be.

We cannot rule out the possibility that the differing capacity of UASp importin α1, α2, or α3 transgenes to rescue α2 and α3 mutant phenotypes is due to differences in transgenic protein expression levels, protein stability, or post-translational modifications. However, we do note that all three transgenes appear by Western blot analyses to be well expressed (Masonet al. 2002; Figure 3). In addition, a UASp α2, but not α3, transgene fully rescued phenotypes associated with the loss of α2 (Figure 6 and not shown), while the same α3 transgene, but not the α2, fully rescued phenotypes caused by the loss of α3 (Table 4). These results strongly suggest that α2 and α3 differ in their ability to perform cellular functions in vivo.

Previously, in vitro studies showed that vertebrate importin α3's specifically transported presumably essential cellular proteins, such as RCC1 and RanBP3 (Köhleret al. 1999; Welchet al. 1999; Nemergut and Macara 2000; Talcott and Moore 2000). Our finding that embryos and larvae do not appear to require an α3-specific activity was, therefore, surprising. It is possible that α3 protein or mRNA may be stored maternally at a low level and maintained until larval stages. However, Fang et al. (2001) did not observe any α3 mRNA or protein in 0- to 2-hr embryos, suggesting that α3 is not stored maternally. Small amounts of α3 protein were observed in 0- to 2-hr embryos by Máthé et al. (2000) and the source of this discrepancy is currently unclear. Therefore, residual α3 activity may be present in mutant embryos and α1- and α2-rescued mutant larvae to perform all α3 functions necessary for cell survival. Alternatively, α3-specific nuclear transport of RCC1 and RanBP3 observed in vitro may not be specific for α3 in vivo or these α3-specific functions may not be conserved from vertebrates to flies. Finally, the nuclear import of α3-specific essential cellular proteins may also be imported by a redundant nuclear targeting pathway. This appears to be the case for the import of RCC1 in vertebrate cells. RCC1 import is mediated by two distinct pathways, only one of which requires α3 (Nemergut and Macara 2000).

The analyses of importin α2 and α3 mutant phenotypes demonstrate that α2 is essential only for gametogenesis, while α3 appears to serve a more widespread developmental role (Giarrèet al. 2002; Gorjánáczet al. 2002; Masonet al. 2002; this study). These observations are consistent with the defects in somatic tissues associated with RNAi-mediated disruption of α3 paralogs in C. elegans and porcine embryos (Geleset al. 2002; Cabot and Prather 2003) and suggest that the α2's are the most derived of the three importin α types. In addition, rescue experiments with UASp α1, α2, and α3 transgenes suggest that these differential developmental roles are due, at least partly, to distinct α2 and α3 biochemical activities (Masonet al. 2002; this study). Much work remains to be done to determine the precise cellular processes and molecular mechanisms that yield the observed mutant phenotypes. In conclusion, these studies lay the groundwork for future in vivo and in vitro studies of the importin α gene family.

Acknowledgments

We thank T. Herman and L. Zipursky for the importin α317-7 and α3w73 nonsense alleles; I. Kiss for the α2D14 allele; the Bloomington Stock Center for fly stocks; I. Török and B. Mechler for the anti-α2 antibodies; P. Rørth for the Gal4pnos-VP16 stock and the UASp vector; N. Shulga for assistance with confocal microscopy; A. Jonth for assistance isolating viable recombinant α31 chromosomes; H. Jasper for assistance embedding eyes; R. Angerer for the use of the UV dissecting scope; K. Bentley for sectioning embedded eyes; B. McIntyre for SEM; and T. Herman, T. Schwarz, and members of the Goldfarb and Fleming labs for helpful discussions. The monoclonal anti-Kelch antibody developed by L. Cooley was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. This work was supported by grants from the March of Dimes (1-FY01-313) to D. S. Goldfarb, from the National Science Foundation (IBN-0234751) to R. Fleming, and from the National Science Foundation (CTS-6571042) to B. McIntyre.

Footnotes

  • Communicating editor: K. Anderson

  • Received July 25, 2003.
  • Accepted August 21, 2003.

LITERATURE CITED

View Abstract