Domains of Importin-[alpha] 2 required for ring canal assembly during Drosophila oogenesis

Journal of

www.elsevier.com/locate/yjsbi

Journal of Structural Biology 154 (2006) 27–41

StructuralBiology

Domains of Importin-a2 required for ring canal assemblyduring Drosophila oogenesis

Matyas Gorjanacz a,c,1, Istvan Torok c, Istvan Pomozi b, Gy}oz}o Garab b,Tamas Szlanka a,c, Istvan Kiss a, Bernard M. Mechler c,*

a Institute of Genetics, Biological Research Center of the Hungarian Academy of Sciences, Szeged, Hungaryb Institute of Plant Biology, Biological Research Center of the Hungarian Academy of Sciences, Szeged, Hungary

c Department of Developmental Genetics, Deutsches Krebsforschungszentrum, Heidelberg, Germany

Received 20 May 2005; received in revised form 10 November 2005; accepted 7 December 2005Available online 17 January 2006

Abstract

Null-mutation in Drosophila importin-a2, such as the deficiency imp-a2D14, causes recessive female sterility with the formation ofdumpless eggs. In imp-a2D14 the transfer of nurse cell components to the oocyte is interrupted and the Kelch protein, an oligomeric ringcanal actin organizer, is normally produced but fails to associate with the ring canals resulting in their occlusion. To define domainsregulating Kelch deposition on ring canals we performed site-directed mutagenesis on protein binding domains and putativephosphorylation sites of Imp-a2. Phenotypic analysis of the mutant transgenes in imp-a2D14 revealed that mutations affecting theImp-b binding-domain, the dimerization domain, and specific serine residues of putative phosphorylation sites led to a normal or nearlynormal oogenesis but arrested early embryonic development, whereas mutations in the nuclear localization signal (NLS) and CAS/expor-tin binding domains resulted in ring canal occlusion and a drastic nuclear accumulation of the mutant proteins. Deletion of the Imp-bbinding domain also gave rise to a nuclear localization of the mutant protein, which partially retained its function in ring canal assembly.Thus, we propose that mutations in NLS and CAS binding domains affect the deposition of Kelch onto the ring canals and prevent theassociation of Imp-a2 with a negative regulator of Kelch function.� 2006 Elsevier Inc. All rights reserved.

Keywords: Importin-a; Drosophila; Ring canal assembly; Kelch; Fluorescence anisotropy imaging

1. Introduction

The nuclear envelope divides eukaryotic cells into nucle-ar and cytoplasm compartments, disjoining transcriptionfrom translation. This compartmentalization requiresactive nucleo-cytoplasm transport to import componentsof the nuclear machinery synthesized in the cytoplasmand, reciprocally, to export components from the nucleusinto the cytoplasm. The transport process is mediated bysoluble receptors designated as importins and exportins,also called karyopherins, which carry cargoes into and

1047-8477/$ - see front matter � 2006 Elsevier Inc. All rights reserved.

doi:10.1016/j.jsb.2005.12.007

* Corresponding author. Fax: +49 6221 424552.E-mail address: [email protected] (B.M. Mechler).

1 Present address: European Molecular Biology Laboratory, Directors’Research, Heidelberg, Germany.

out of the nucleus (Chook and Blobel, 2001). The mecha-nism of nuclear transport has been the subject of intensescrutiny and includes essentially three major players, name-ly Importin-a (Imp-a), Importin-b (Imp-b), and Ran (Con-ti and Izaurralde, 2001; Gama-Carvalho and Carmo-Fonseca, 2001; Gorlich and Kutay, 1999; Jans et al.,2000; Mattaj and Englmeier, 1998; Stewart et al., 2001;Weis, 2003). In the cytoplasm, the Imp-a protein recogniz-es the nuclear localization signal (NLS) peptide of the car-go protein, binds to Imp-b, and the whole complex istranslocated into the nucleus (Weis, 2003). Followingnuclear import the NLS-cargo-Imp-a/b ternary complexbecomes dissociated by nuclear RanGTP, which binds toImp-b. Imp-a is recycled back to the cytoplasm throughinteraction with CAS/Cse1p exportin, which is also associ-ated with RanGTP (Hood and Silver, 1998; Kutay et al.,

mailto:[email protected]

28 M. Gorjanacz et al. / Journal of Structural Biology 154 (2006) 27–41

1997). Imp-a/Cas/RanGTP complexes then become disso-ciated in the cytoplasm when RanGTP is converted toRanGDP by RanGAP. The steep RanGTP/RanGDP con-centration gradient between the nucleus and cytoplasmorchestrates the nuclear import and export of macromole-cules (Kuersten et al., 2001).

Imp-a exerts other functions unrelated with nucleo-cyto-plasm transport. During mitosis it plays a critical role inspindle assembly and centriole formation through a RanG-TP-induced release of the microtubule-associated proteinsTPX2 and NuMA from Importins (Gruss et al., 2001;Nachury et al., 2001; Wiese et al., 2001). Furthermore,Imp-a is also required following mitosis for nuclear enve-lope assembly at the surface of chromatin (Hachet et al.,2004).

Imp-a is an adaptor protein that binds proteins contain-ing NLS motif and Imp-b. It can be divided into threestructural units: a central NLS-binding (NLSB) domainwith 10 Armadillo (Arm) repeats, a positively charged N-terminal (IBB) domain that binds to either Importin-b orits own NLSB domain, and a C-terminal CAS/Cse1p-bind-ing (CASB) domain (Chook and Blobel, 2001; Matsuuraand Stewart, 2004). The ARM repeats stack to form aright-handed superhelix onto which the NLS peptidesbecome associated at two sites on the concave groove. EachARM repeat consists of three helices, H1, H2, and H3,with H3 orientated towards the concave side of Imp-a(Andrade et al., 2001; Coates, 2003). All the residues inter-acting with the NLS motif of the cargo substrate are locat-ed on the H3 helices of Imp-a (Conti et al., 1998; Fonteset al., 2000; Kobe, 1999).

Phylogenetic studies indicate that the importin-a genes ofhigher eukaryotes can be classified in three conservedclades, designated as a1, a2, and a3 (Kohler et al., 1997,1999; Malik et al., 1997). Yeast and plant imp-a genesbelong to the a1 clade whereas metazoan animals typicallycontain members of all three clades. Drosophila melanogas-

ter owns one gene of each subfamily (Giarre et al., 2002;Goldfarb et al., 2004; Mason et al., 2002).

Most intriguing is the crucial role of Drosophila Imp-a2protein during oogenesis (Gorjanacz et al., 2002). A loss-of-function mutation in imp-a2, designated as imp-a2D14

or D14 (Torok et al., 1995), leads to female sterility charac-terized by the occlusion of the ring canals (Gorjanacz et al.,2002), that link the nurse cells to the oocyte and providethe transcriptionally inactive oocyte with a ‘‘nutrientstream’’ of RNA and proteins produced in the nurse cells(Robinson and Cooley, 1997b; Cooley, 1998). This processcan be divided into two phases. A selective transfer, result-ing more particularly in the deposition of embryonic deter-minants, characterizes the first phase of egg chamberdevelopment up to stage 10. Then, between stages 10 and12, the dumping of the cytoplasmic content of the nursecells into the oocyte takes place and leaves behind thenuclei of the apoptotic nurse cells (Mahajan-Miklos andCooley, 1994). In D14, the occlusion of the ring canals pre-vents dumping due to a retention of the Kelch protein in

the cytoplasm (Gorjanacz et al., 2002), which usuallybecomes associated with the ring canals in their final stageof assembly (Robinson et al., 1994; Robinson and Cooley,1997a).

To learn more about the function of Imp-a2 in Drosoph-

ila we mutagenized known domains of imp-a2, includingthe IBB domain, the dimerization (DIM) domain, theNLS-binding (NLSB) domain, the putative CAS-binding(CASB) domain, and a series of putative phosphorylationserine residues. The mutant transgenes were studied in aD14 null-mutant background and we determined whetherthese constructs blocked oogenesis, or exerted no effecton oogenesis but caused embryonic arrest. We also exam-ined the intracellular location of the mutant proteins. Thisapproach allowed us to identify the domains of Imp-a2necessary for ring canal assembly and those critical forearly embryonic development. Finally, by using differentialpolarization laser scanning confocal microscopy(DP-LSM) we determined the anisotropic moleculararchitecture of actin filaments in wild-type, as well as innull-mutant D14 and kelch ring canals.

2. Materials and methods

2.1. Fly stocks and P-element-mediated transformation

Flies were maintained under standard culturing condi-tions at 25 �C on standard cornmeal–yeast–agar medium.

Wild-type imp-a2 K9 cDNA (Torok et al., 1995) andin vitro mutagenized imp-a2 cDNAs were subcloned intothe pUASp2 P-element mediated transformation vector(see below). Wild-type and mutant constructs were micro-injected along with the D2–3 transposase helper plasmidinto w1118 syncytial blastoderm embryos according to stan-dard techniques. The following P-element insertions:UAS::imp-a2+, UAS::S37A, UAS:: S56A, UAS::S98A,UAS::3xSA, UAS::DIBB, UAS::NLSB�, UAS::LNLSB�,UAS::SNLSB�, UAS::DIM�, and UAS::CASB� werecrossed with y w; Sp/ y+ CyO; Sb/TM6,Ubx flies and foreach construct three to five independent third chromosom-al insertions were introduced into D14 genetic background.yw; imp-a2D14/y+ CyO; UAS/TM6, Ubx stocks were estab-lished. To express UAS transgenes in ovaries we used ananos::Gal4VP16 germ line-specific Gal4 enhancer trap line(Van Doren et al., 1998). In rescue experiments we crossedyw; imp-a2D14/y+ CyO; UAS/TM6,Ubx with yw; imp-a2D14/y+ CyO; nanos::Gal4VP16/TM6,Ubx flies and yw;imp-a2D14/imp-a2D14; UAS/nanos::Gal4VP16 females wereinvestigated. The other stock used in this study werekelchDE1 (Xue and Cooley, 1993), dcasts1 and dcasts28

(Tekotte et al., 2002) kindly provided by Ilan Davis.

2.2. In vitro mutagenesis of importin-a2 gene

For in vitro mutagenesis of imp-a2 gene we cloned imp-a2 K9 cDNA (Torok et al., 1995) into the EcoRI siteof pBluescript-II-SK(+) vector. Mutant imp-a2 cDNA

M. Gorjanacz et al. / Journal of Structural Biology 154 (2006) 27–41 29

constructs were generated by using the PCR-based Quick-Change Site-Directed Mutagenesis Kit (Stratagene),checked that they produced the appropriate size polypep-tide by in vitro transcription–translation using TNT-Cou-pled Reticulocyte Lysate System (Promega), andsequenced (T7 Sequenase Version 2.0 DNA SequencingKit, Amersham Pharmacia Biotech) to confirm the absenceof PCR-induced errors. The constructs were then sub-cloned into the NotI and KpnI cloning sites of pUASp2

P-element mediated transformation vector (kindly provid-ed by P. Rorth), a modified pUASp (Rorth, 1998).

The following PCR primers (bold letters indicate theintroduced nucleotide substitutions) were used to generateS37A, 5 0-CATCGAGCTGCGCAAGGCCAAAAAGGAG-3 0; S56A, 5 0-GAGGATCTAACGGCGCCGCTCAAAGAG-3 0; S98A, 5 0-CAAGATGCTCGCTCGGGAACGCAATC-3 0. To generate the 3xSA construct all three primerswere used in three successive PCR amplifications.

To generate NLSB�, LNLSB�, and SNLSB� constructswe used the following five primers: W135 and N139, 5 0-GAGGCCGCTGCGGCGCTTACCGCCATCGCCTCTG-3 0;W177 and N181, 5 0-GCAGGCAGTCGCGGCTCTGGGCGCCATTGCCGGCG-3 0; W220 and N224, 5 0-CAACATCGTCGCGCTGATGTCCGCCCTGTGCCG-3 0; W346

and N350, 5 0-GGAGGCTGCCGCGACGGTCAGCGCCATCACAGCAGC-3 0; W388 and N392, 5 0-GAGGCTGCCGCGGCGGTGACAGCCACCACGACATC-3 0. In DIM�

construct the following four primers substitute the residuesR227, 5 0-CAACCTGTGCGCAAACAAGAATCCATCG-3 0; Y266, 5 0-GGGCTTTGTCCGCCGTCACGGACGACG-3 0; R304, 5 0-GCCCGCCCTGGCCAGCGTTGGCAAC-3 0; and Y475, 5 0-CGAGGAGGTCGCCAAGAAGGCCTACGCC-3 0. The CASB� construct was generated withthe primer 5 0-GGAGATGGGCGCTGCAGCCGCGGCGGCAACTCTGCAGC-3 0. The DIBB interstitial deletionwas made by taking advantage of the SphI restriction endo-nuclease that cleaves a DNA fragment removing codons25–90.

2.3. Micro-filament association assay

Micro-filament association assay was performed asdescribed (Gorjanacz et al., 2002). Soluble nuclear proteinfractions from NLSB�, CASB�, and DIBB ovaries in aD14 background were made in buffer containing 25 mMHepes, pH 7.6, 100 mM KCl, 12.5 mM MgCl2, 0.1 mMEDTA, 400 mM K-glutamate, and 20% glycerol (Kam-akaka et al., 1991) and centrifuged at 200000g for 2 h at4 �C.

2.4. Western blot analysis

Immuno-blotting was performed as described (Toroket al., 1995) using 10% SDS–polyacrylamide gels. Rabbitpolyclonal anti-Imp-a2 antibodies (Torok et al., 1995) in1:1000 dilution, rabbit polyclonal anti-P40 antibodies(Torok et al., 1999) in 1:15000 dilution, and anti-actin

hybridoma supernatant from the Developmental StudiesHybridoma Bank (Iowa City, IA) in 1:10 dilution wereused.

2.5. Immuno-histochemistry and microscopy

Immuno-staining of Drosophila ovaries (Gorjanaczet al., 2002) and embryos (Giarre et al., 2002) was per-formed by using the following antibodies: anti-Imp-a2antibody (1:50), anti-Filamin antibody (1:1000; Sokol andCooley, 1999), mouse monoclonal anti-a-tubulin DM1Aantibody (1:300; Sigma–Aldrich GmbH, Germany), anti-Kelch 1B hybridoma supernatant (1:1; Xue and Cooley,1993) kindly provided by L Cooley, and anti-Hts-RChybridoma supernatant (1:1; Robinson et al., 1994) fromDevelopmental Studies Hybridoma Bank (Iowa City, IA).F-actin was detected with rhodamine-conjugated Phalloi-din or Alexa 488 Fluor Phalloidin (Molecular Probes).For DNA staining we used TO-PRO3 (1:1000; MolecularProbes) or DAPI. All Cy2-, Cy3-, and Cy5-coupled second-ary antibodies were used in 1:300 (Jackson ImmunoRe-search Laboratories). The samples were examined with aZeiss LSM-410 confocal laser scanning microscope (CarlZeiss, Jena, Germany) or a Leica DMR fluorescence micro-scope (Leica GmbH, Wetzlar, Germany) equipped with acooled charge-coupled Spot camera (Visitron Systems,Puchheim, Germany).

For chorion examination, eggs were washed in 0.1%Nonidet P40, mounted in Hoyer’s medium/lactic acid mix(7:3), and investigated under a Leitz microscope (Leica).

2.6. Differential polarization laser scanning microscopy

In general, differential polarization (DP) quantities canbe calculated by measuring the intensity differences (DI)between two orthogonal polarization states and the aver-age intensity (Ia), which is related to the total intensity.The anisotropy of polarized fluorescence emission wasdetermined with respect to a reference plane is

I jj�I?I jjþ2I?

, whereIi is the fluorescence intensity parallel to and I^ is perpen-dicular to the reference. The anisotropy can be determinedfrom DI and Ia as 2DI

6Ia�DI � DI3Ia

.The design and construction of the DP-LSM was based

on a Zeiss LSM-410 laser scanning confocal invertedmicroscope with an external argon-ion laser. It wasequipped with a DP attachment for the analysis of thepolarization state of the fluorescence emission with theaid of a linear polarizer, a photo-elastic modulator(PEM-90, HINDS Instruments), and a digital lock-inamplifier (DPSD, KFKI, Budapest). The DPSD card wasconnected to a microcomputer (IBM compatible PentiumIII PC) via ISA bus, yielding DI value for each pixel (thedwell time was approximately 200 ls, the modulator wasset at 520 nm). Ia, the mean value of the fluorescence emis-sion, was obtained from the 100 kHz modulated signal ofthe photo-multiplier by using an analogue low pass electri-cal circuit and an analogue-digital converter connected to a


PCI bus of the microcomputer. Similar polarization stateanalyzer was used earlier in a polarization modulation laserscanning microscope (Gupta and Kornfield, 1994; Guptaet al., 1994). To avoid photo-selection and contributionfrom polarized fluorescence emission due to polarized exci-tation, an optical depolarizer was placed in the laser beamin front of the beam expander. The measuring program,written in C environment, controlled the DPSD andADC cards, and performed on-line calculations yieldingDI and Ia. These values, introduced as external signals tothe video-board of LSM-410, were used for DP-imaging;in this particular case for imaging the anisotropy distribu-tion of the fluorescence emission dipoles.

To determine the preferential orientation of the emissiondipoles and their magnitude r, two DP images were record-ed on the same sample, r0,90 and r+45,�45, i.e., anisotropyvalues obtained for the ‘vertical minus horizontal’ and the(+)45� minus (�)45� directions, respectively. The calcula-tion was performed on each pixel of the image pair. Theaccuracy of the measurement is ±0.05 for the magnitudeand ±3� for the direction of the anisotropy. These valueswere largely invariant on the pixel position in the ring,i.e., displaying no significant selectivity for the orientationof the emission dipoles with respect to the laboratory co-ordinate system.

3. Results

3.1. Generation of mutant Imp-a2 proteins

To determine the role of specific Imp-a2 domains in ringcanal assembly, we constructed a series of imp-a2 trans-genes either carrying nucleotide changes producing alaninesubstitutions at critical amino acid residues, or containingan interstitial deletion removing a particular domain ofImp-a2 (Fig. 1A). The mutated cDNA constructs wereplaced under a UAS promoter in a pUASp2 P-element vec-tor and used to transform flies. Previously, we showed thatthe synthesis of wild-type Imp-a2 protein from aUAS::imp-a2+ transgene driven by a nanos::Gal4VP16 driv-er can fully rescue the development of mutated D14 ani-mals, which produce no Imp-a2 proteins, due to aninterstitial deletion removing the promoter region and alarge portion of the gene (Gorjanacz et al., 2002; Toroket al., 1995). Therefore, to study the requirement of the dif-ferent Imp-a2 domains we expressed the mutant imp-a2transgenes in a D14 background by using the same nanos::-

Gal4VP16 driver.Structural studies of Importins-a point out the residues

critical for interaction with mono-partite and bipartiteNLS sequences. The NLS-binding residues are distributedin two domains (Conti and Kuriyan, 2000; Conti et al.,1998; Fontes et al., 2000; Kobe, 1999; Matsuura and Stew-art, 2004). The major, or large, NLS-binding site (LNLSB)covers ARM repeats 2–4 and interacts with mono-partiteNLS motifs or the larger basic segment of the bipartiteNLS motif. The minor, or small, NLSB domain (SNLSB)

covers ARM repeats 7–8 and accommodates the smallerbasic segment of the bipartite NLS motif. In both NLS-binding domains, pairs of conserved WXXXN motifs bindto each NLS. The conserved N residues form hydrogenbonds with the NLS basic residues whereas the conservedW residues form hydrophobic interactions with long ali-phatic portions of the NLS side chain. By site-directedmutagenesis we modified all conserved W and N residuesof the NLSB domain and made three transgenes. In thetransgene NLSB� all conserved W residues at positions135, 177, 220, 346, and 388 and all conserved N residuesat position 139, 181, 224, 350, and 392 were substitutedby A. In the transgene LNLSB�, only the residues affectingthe major NLS-binding site including the residues at posi-tions 135, 139, 177, 181, 220, and 224 were modified,whereas the residues at positions 346, 350, 388, and 392,that cover the minor NLS-binding site, were substitutedby A in the transgene SNLSB�.

Structural analysis shows that the Y283 and R321 resi-dues of yeast Imp-a are involved in the binding of thebipartite NLS (Leung et al., 2003). These residues corre-spond to Y266 and R304 of Drosophila Imp-a2, which inter-act with the linker sequence connecting the two basicclusters of the bipartite NLS. However, these two residuesas well as R277 and Y475 form an interface possibly leadingto Imp-a dimerization (Conti et al., 1998). In the dimeriza-tion defective transgene, or DIM�, the residues R227, Y266,R304, and Y475 were substituted by A.

In Imp-a the interaction to the CAS nuclear export factoris essentially mediated by a row of conserved amino acid res-idues located in its C-terminal domain (Herold et al., 1998).Consequently we substituted with six A the residues GLD-KLE at positions 459–464, similarly to the substitutionsmade in the human Rch1 protein that eliminate binding toCAS without affecting its binding to Imp-b or NLS peptides(Herold et al., 1998). The Drosophila imp-a2 transgene defec-tive for CAS binding was designated as CASB�.

Since Imp-a2 can be phosphorylated at S residues dur-ing oogenesis and early embryogenesis (Torok et al.,1995), we substituted by A three S residues at positions37, 56, and 98, respectively, that correspond to potentialphosphorylation sites and generated four different trans-genes. Three transgenes, designated as S37A, S56A, andS98A, contain a single A substitution at one of the S residueswhereas in the fourth transgene, named 3xSA, all three Sresidues were changed into A residues.

Finally, we generated a transgene in which the IBB-do-main of Imp-a2 was deleted between residues 25 and 90.This transgene was named DIBB.

3.2. In vivo analysis of imp-a2 mutants

To test the function of the modified Imp-a2 proteins, weestablished three to five independent transgenic lines foreach transgene. Expression of the UAS::imp-a2 transgeneswas driven by nanos::GAL4VP16 in D14 null-mutant back-ground. We selected lines in which the relative amount of

Fig. 1. Construction, expression, and phenotypic analysis of imp-a2 transgenes. (A) Diagram of the Imp-a2 protein containing an N-terminalImp-b-binding (IBB) domain, ten Armadillo (ARM1-10) repeats, and a short C-terminal domain and map of the introduced mutations in the transgenes.Positions of IBB, LNLSB, SNLSB, and CASB domains are indicated above the diagram. Positions of the amino acid residues substituted by A areindicated for each construct. Rates of egg lying (number of laid eggs/day/female) and percentages of egg hatching (%) were analyzed for wild-type, D14

mutant and transgenic lines in a D14 background (ND, not determined). (B) Immunoblot of protein extracts from (lane 1) wild-type, (lane 2) D14, and(lanes 3–13) imp-a2 transgenic ovaries. The blot was bisected at about 45 kDa, and the upper portion was probed with anti-Imp-a2 antibodies and thebottom portion with anti-P40 antibodies, as a loading control.


Imp-a2 produced in ovaries was similar to that made inwild-type. Western blot analysis of ovarian extractsrevealed that, with the exception of DIBB, all other modi-fied Imp-a2 proteins were synthesized in approximatelythe same amount as in wild-type (Fig. 1B). In the case ofDIBB, we found that all lines produced a reduced level ofDIBB by comparison to wild-type, suggesting that theinterstitial deletion might either alter the rate of transcrip-tion–translation, or induce protein instability.

We previously showed that Imp-a2 can be resolved in atleast two species by SDS–PAGE and demonstrated that the

slower migrating polypeptide corresponds to a phosphory-lated form (Torok et al., 1995). As shown in Fig. 1B, theS56A mutation significantly reduced, albeit not completely,the amount of the slower migrating form indicating thatS56 is a target of phosphorylation. Interestingly, the DIBBprotein migrated as a single species in SDS–PAGE, sug-gesting that the phosphorylation of an additional residuedistinct from S56 might account for the residual amountof the slowly migrating Imp-a2.

We then investigated the ability of the transgenes to res-cue D14 development (Fig. 1A). As previously described


(Gorjanacz et al., 2002), one normal copy of imp-a2 cDNAcan fully restore fertility in D14 females. We found that88% of the eggs laid by these females hatched and devel-oped to adults. Based on the degree of rescue we classifiedthe imp-a2 mutants into two groups. In the first group thatincludes the S37A, S56A, S98A, 3xSA, DIBB, and DIM� lines,the egg chambers developed normally and the size of theeggs was similar to that of wild-type eggs. However,embryo development was either completely blocked inDIBB and DIM� or strongly affected in the S37A, S56A,S98A, and 3xSA mutations.

DIBB females laid nearly the same number of eggs aswild-type females. Although the DIBB eggs exhibited nor-mal size, only 14% of them displayed a normal morphology(Fig. 2), whereas 86% showed abnormally fused dorsalappendages. In a few cases the eggs were shorter than nor-mal. The DIM� transgene restored normal oogenesis inD14 flies but the eggs of normal size were unable to sustainnormal development. From our analysis we concluded thatmutations modifying putative phosphorylated residuesexerted no visible effect on oogenesis but were critical forembryogenesis. Similarly the DIM� and DIBB mutationsallowed oogenesis to proceed but radically blockedembryogenesis.

Examination of imp-a2S56A embryos by DNA anda-tubulin staining revealed that meiosis normally tookplace in the majority of the eggs with formation of polarbodies (Figs. 3A and D). However, by comparison towild-type, mitosis was considerably delayed and asynchro-nous (Figs. 3B and E). Usually we found that more than�35% of the 90–165 min-old imp-a2S56A embryos were atthe second or third nuclear divisions, whereas correspond-ing wild-type embryos have reached divisions 8–13. How-ever, �20% of imp-a2S56A embryos were able to attain amore advanced stage with nuclei migrating under the eggcortex. At this location the nuclei displayed an irregular

Fig. 2. imp-a2 egg shell morphology. Eggs laid by D14 females containing theUAS::DIBB/nos::Gal4, (D) UAS::NLSB�/nos::Gal4, and (E) UAS::CASB�/n3xSA, DIM�, and SNLSB� transgenic D14 females displayed normal morpho

pattern of distribution with empty regions and regions inwhich their distribution was apparently denser than normal(Fig. 3E). With a few exceptions, the development of theseembryos was later blocked since we found that only 3.4%of the laid eggs were able to give rise to larvae. Examina-tion of the other mutants affecting only embryogenesisshowed a similar pattern of early arrest of nuclear divisionsor abnormal pattern of nuclear divisions.

The second group of imp-a2 mutants including NLSB�

and CASB� is characterized by their inability to undertakenormal oogenesis. NLSB� and CASB� females laid eggs insimilar amount (Fig. 1A) and shape, as in the case of D14

flies (Figs. 2B, D, and E). To determine more preciselywhich part of the NLS-binding domain is criticallyinvolved in oogenesis, we used the LNLSB� and SNLSB�

transgenes in which conserved W and N residues weresubstituted by A. We found that substitutions in theLNLSB domain impaired oogenesis, whereas mutationsin the SNLSB domain exerted no effect on oogenesis butcompletely blocked early embryonic development. Thesedata showed that we could define two categories of Imp-a2 mutations; one category of mutations affects oogenesis,whereas the other type results in developmental arrest dur-ing early embryogenesis.

3.3. F-actin binding to wild-type and mutant Imp-a2

To determine whether the mutations exert no majorconformation change in the modified Imp-a2 proteins, wedetermined their ability to bind to F-actin in an NLS-de-pendent interaction, as previously shown in the case ofwild-type Imp-a2 (Gorjanacz et al., 2002). For this purposewe prepared soluble nuclear fractions of wild-type andmutant CASB�, NLSB�, and DIBB proteins, and assayedtheir binding to in vitro polymerized actin filaments inpresence of an SV40-NLS peptide. As shown in Fig. 4, this

following transgenes: (A) UAS::imp-a2+/nos::Gal4, (B) UAS::imp-a2+, (C)os::Gal4. (F) Egg laid by a kelchDE1 female. Eggs laid by S37A, S56A, S98A,logy (data not shown).

Fig. 3. Delayed and abnormal pattern of mitosis in imp-a2S56A embryos. (A–C) Wild-type embryos collected for 20 min and aged for an additional periodof (A) 20 min, or (B and C) 40 min. (D–F) imp-a2S56A embryos collected for 75 min and aged for 90 min. (C and F) Enlargements of a portion of theembryos shown in (B and E). Arrows in (A and D) indicate polar body position and arrow heads in (F) point out anaphases whereas the other mitoticfigures were in metaphases. Embryos were stained with DAPI for DNA and with monoclonal antibody for a-tubulin.


analysis revealed that the CASB�, and DIBB proteins,albeit not the NLSB� protein, were able to bind polymer-ized actin in the presence of an NLS-peptide. These resultsshow that the amino acid substitutions in the CAS bindingdomain or the deletion of the IBB domain exerted no majoralteration impairing the NLS-binding function, whereassubstitutions in the NLS-binding domain affected the abil-ity to bind polymerized actin.

3.4. Ring canal assembly depends on kelch association withring canals

The fact that �10% of the eggs laid by DIBB femaleswere moderately shorter than normal suggests that themutation affects ring canal assembly. To test this hypothe-sis, we stained ovaries of DIBB flies with antibodiesrecognizing specific ring canal components, such as Kelchand F-actin (Fig. 5), as well as Filamin and Hts-RC (datanot shown). Examination of DIBB ring canals showed that73% of the ring canals displayed a nearly normal morphol-ogy with actin forming a sharp ring and Kelch depositedon the internal rim (Fig. 5D), albeit in a lesser amount thanin wild-type. The other 27% were fully occluded (Fig. 5E)with a typical kelch- (Xue and Cooley, 1993) or D14-like(Fig. 5B) morphology (Gorjanacz et al., 2002). Interesting-ly both types of structures could be observed in the same

egg chamber. Nevertheless the opening of a sufficient num-ber of DIBB ring canals appeared to ensure a nearly fulltransport of nurse cell components into the oocyte, produc-ing only a weak dumpless phenotype (Fig. 2C). In all cases,the pattern of Filamin and Hts-RC distribution was similarto that of F-actin (data not shown).

We performed similar analyses on the other transgeniclines and found that the CASB� (Fig. 5F), NLSB�

(Fig. 5G), and LNSLB� mutations (Fig. 5H) led to actin-obliterated ring canals with no Kelch deposition. In con-trast, the SNLSB� and DIM� mutations (Figs. 5I and J),as well as the mutations substituting putative phosphoryla-tion residues (data not shown), produced normal ringcanals. These data showed that the occlusion of the ringcanals is closely related to the CASB and LNLSB domainsof Imp-a2. When these two domains are mutated, no depo-sition of Kelch occurred on the inner rim of the ring canals.

3.5. Anisotropy of actin filaments in imp-a2D14 and kelchDE1

ring canals

Conventional laser scanning microscopy gave us valu-able information on the overall ring canal structure inwild-type and mutant imp-a2 egg chambers. However, noinformation could be obtained by this technique on theorganization of micro-filaments in the ring canals or on

Fig. 4. NLS-dependent binding of Imp-a2 mutant proteins with micro-filaments. Protein extracts from ovaries expressing either the (A) CASB�,(B) DIBB, or (C) NLSB� transgenes were centrifuged at high speed(200000g) and aliquots (lane 1) were incubated without added supplement(lane 2), with F-actin alone (lane 3), with F-actin and SV40-NLS peptide(lane 4), or with SV40-NLS peptide alone (lane 5), and pelleted through a10% glycerol cushion by ultra-centrifugation at 150000g. The CASB� andDIBB Imp-a2 proteins can be recovered with F-actin in the presence of theSV40-NLS peptide. Immunoblots for Imp-a2 (upper panels) and actin(lower panels).

Fig. 5. Domains of Imp-a2 required for Kelch deposition onto the ringcanals. Ring canals were double-stained for actin and Kelch. Actinfilaments co-localize with Kelch in (A) wild-type ring canals, and ringcanals of D14 ovaries expressing (C) an imp-a2+, (I) SNLSB�, or (J)DIM� transgene. The Kelch protein fails to associate with (B) D14 ringcanals and ring canals of D14 ovaries expressing the (F) CASB�, (G)NLSB�, or (H) LNLSB� transgenes. (D and E) In the case of D14 ovariesexpressing a DIBB transgene (D) 73% of the ring canals display a normalaperture with a detectable, albeit weak, Kelch staining, whereas (E) 27% ofthe ring canals are occluded with F-actin and devoid of Kelch. Scale bar is10 lm.


their arrangement in the lumen of the mutant ring canals.To obtain more information on the molecular organizationof actin filaments we performed fluorescence anisotropyimaging on Alexa Phalloidin-stained samples. The specificbinding of Phalloidin to F-actin (Wulf et al., 1979) allowsthe visualization of actin filament structures and birefrin-gence imaging (Katoh et al., 1999; Oldenbourg et al.,2000) demonstrates that actin is assembled in anisotropi-cally organized macro-structures. Since the position andorientation of Phalloidin with respect to F-actin is welldefined (Oda et al., 2005), it is possible to employ laserscanning confocal fluorescence anisotropy imaging tounravel the molecular architecture of actin macro-structures.

Anisotropy displayed by ring canals using DP-LSMrevealed that the precession of Alexa Fluor Phalloidinwas small. Consequently, this dye was used to map anisot-ropy values of F-actin in the ring canals. We also verifiedthat the anisotropy of the fluorescence was independentfrom the intensity of the emission, which showed onlyminor variations along the ring.


Anisotropy distribution in the ring canals displayed acharacteristic pattern, which is dependent on the positionof the orthogonal plane defined by the polarization stateanalyzer. Our analysis showed that the orientation of theemission dipoles varied periodically and followed the ringcircle. Fig. 6A2 displays, in false colors, the variation ofdipole orientation with respect to the laboratory-fixed co-ordinate system for a wild-type ring canal on a circularcross section (face-aligned position). This image wasobtained from two independent fluorescence measurementsand showed that the dipoles were tangentially organizedaround the ring circle. The uncertainty of the measureddirection was smaller than 3� and its variation remainedsmall, indicating homogeneity in the sample.

In contrast, no preferential orientation of the dipolescould be obtained for ‘edge-aligned’ positions in wild-typeand mutant rings (Fig. 6B2, data not shown). In this case,the magnitude of the anisotropy was very small, lower than0.05, compared to the typical 0.35–0.4 values for the face-aligned position. These data indicate that one or morecomponents randomize the orientation angles with respectto the axis perpendicular to the ring plane. The structuralbasis of a lack of sizeable anisotropy in this plane remainsunknown, but the occurrence of simple twisted/coiledstructures may readily explain the randomization of thedipole orientation.

As shown in Figs. 6C2 and D2, the D14 and kelch

mutant ring canals also retained the characteristic patternof anisotropy, and a preferential tangent orientation ofthe emission dipoles. Although the orientation pattern ofthe dipoles was somewhat weaker than in wild-type ringcanals, as revealed by reading the values along the circle(Table 1), our data showed that D14 and kelch mutationsexerted no major effect on the preferential orientation ofthe emission dipoles with respect to the tangentialdirection.

We also determined the magnitude of anisotropy inthe face-aligned plane and found that this value essen-tially displayed a uniform distribution (data not shown),also suggesting homogeneity of the samples. However, acomparison of mutant and wild-type data revealed smallbut significant differences in the value of the magnitude.In particular we noticed that this value was decreased inmutant rings (Table 1). The magnitude measured foreach pixel is indicative of the local order of the emis-sion dipoles with respect to the preferential orientation

Table 1Anisotropy of actin filaments and angle fluctuation of their orientation inwild-type and mutant ring canals

Sample Number of datapoints (sample)

Anisotropymean value ±SE

Fluctuationangle (�)

Wild-type 1531 (3) 0.42 ± 0.006 17.4imp-a2D14 2701 (6) 0.31 ± 0.003 23.6kelchDE1 3308 (6) 0.31 ± 0.003 22.9

These values were measured on face-aligned ring canals.

angle, i.e., with respect to the tangent direction. Inother terms, variations in the magnitude could be attrib-uted to fluctuations of the orientation angle around thetangency. Such variations might arise from a release ofbound Alexa Phalloidin or a disorder at the level of theactin-based macro-structures, i.e., in the bundles of actinfilaments. Using a simple model calculation, similar tothat used earlier for thylakoid membranes (Szito et al.,1984) we determined the angular interval of the fluctua-tion of the orientation angle with respect to the tangen-tial direction. In wild-type ring canals, the angularinterval was 17.4�, whereas in the mutants this valuewas increased by about 5� (Table 1). Although the ori-gin and nature of this fluctuation remains unknown, wecan conclude that the D14 and kelch mutations result ina small but significant loss in the anisotropic organiza-tion of the ring canals. This disorganization may resultfrom an excess of F-actin present in the mutant ringcanals.

3.6. Deletion of the IBB domain changes the intracellular

localization of Imp-a2 without affecting its function in ringcanal assembly

Since Imp-a can shuttle from cytoplasm to nucleus,and reciprocally, it is possible to envisage that mutationscausing ring canal occlusion may modify more the intra-cellular localization and dynamics of the mutant proteinsrather than their binding to Kelch or a factor regulatingKelch. In particular, we found that treatment of wild-type ovaries with Cytochalasin D, which disrupts actinfilaments, leads to a nuclear accumulation of Imp-a2(Gorjanacz et al., 2002). Therefore, we investigated theintracellular distribution of the various mutant Imp-a2proteins. As shown in Fig. 7, Imp-a2 was predominantlydetected in the cortical periphery of oocytes and nursecells in wild-type (Fig. 7A) and imp-a2+D14 egg cham-bers (Fig. 7C), whereas no Imp-a2 could be detected intheir nuclei.

Analysis of the pattern of Imp-a2 distribution in themutant lines revealed that DIBB (Fig. 7D), CASB�

(Fig. 7E), NLSB� (Fig. 7F), and LNLSB� proteins(Fig. 7G) predominantly accumulated in nuclei of bothnurse cells and oocytes. In contrast, the SNLSB�

(Fig. 7H) and DIM� Imp-a2 proteins (Fig. 7I) werelocalized in the cytoplasm of these cells with a preferentiallocalization at the periphery similar to that in wild-type.Since both proteins exerted no deleterious effect on ringcanal assembly, their occurrence in the cytoplasmwas expected. Similarly, the nuclear localization of theCASB�, NLSB�, and LNLSB� proteins would explaintheir inability to contribute to ring canal formation. Incontrast, the nuclear localization of DIBB, which supportsthe assembly of normal or nearly normal ring canals andthe deposition of Kelch on these structures, contradictedthis assumption and showed that Imp-a2 function isindependent from its intracellular localization.



4. Discussion

Previous investigations reveal that imp-a2 function iscritical for female and male gametogenesis (Giarreet al., 2002; Gorjanacz et al., 2002; Mason et al.,2002). In females, imp-a2 inactivation leads to occlusionof the ring canals (Gorjanacz et al., 2002), which linkthe nurse cells to the oocyte (Cooley, 1998; Robinsonand Cooley, 1996). This occlusion hinders the transferof the nurse cell cytoplasm into the oocyte. This defectcan be traced to the inability of the Kelch protein, anobligate constituent of the ring canals (Hudson andCooley, 2002), to be deposited on their inner rim (Gor-janacz et al., 2002). Although D14 male sterility couldbe rescued by expression of either imp-a1+ or imp-a3+

transgenes, the aperture of the ring canal lumen duringoogenesis is strictly dependent on imp-a2 function, asshown by complementation studies with every Drosoph-

ila imp-a gene (Mason et al., 2002). Moreover, in con-trast to Imp-a1 (data not shown) and Imp-a3 (Matheet al., 2000), Imp-a2 is essentially a cytoplasmic protein,which remains undetectable in nurse cell nuclei, and isnever found in association with ring canals. These find-ings suggest that Imp-a2 should interact in the cyto-plasm with specific factor(s) acting in ring canalformation. Furthermore, no defect in nuclear–cytoplas-mic transport could be attributed to imp-a2 since (1)NLS-fused reporter molecules (Gorjanacz et al., 2002)or endogenous nuclear proteins, including Pavarottiand Prod, can normally accumulate in D14 nurse cellnuclei (data not shown), and (2) D14 egg chambersexhibit a normal development until dumping. Further-more, our previous observation that Drosophila eggscontain large maternal stockpiles of Imp-a2, whose rateof decay follows the cumulative number of mitosis dur-ing the first 13 nuclear divisions (Torok et al., 1995),raises the possibility that Imp-a2 plays a critical roleduring early Drosophila development. Therefore, we con-ducted a mutational analysis to determine the Imp-a2domains involved in ring canal formation, and examinedwhether Imp-a2 may exert other critical functions dur-ing embryogenesis.

Fig. 6. Anisotropy of actin filaments in wild-type, imp-a2D14 and kelchDE1 ringfemales were stained with Alexa 488 Fluor Phalloidin. (A1–D1) The intensityconfocal microscope, and (A2–D2) the anisotropy of the polarized fluorescenceD2 indicate the preferential orientation of the emission dipoles with respect tomade of double-headed arrows. The white circle with an arrowhead illustratesalong any arbitrarily chosen set of pixels in the sample. In the face-aligned plaperiodical distribution of the emission dipoles indicating that actin filaments ano preferential orientation of actin filaments could be detected.

Fig. 7. Distribution of wild-type and mutant Imp-a2 proteins in Drosophila egD14 ovaries expressing the (C) imp-a2+, (D) DIBB, (E) CASB�, (F) NLSB�, (GImp-a2 and DNA. Imp-a2 displays a sub-cortical distribution in wild-type otransgenes. In D14 ovaries expressing the (D) DIBB, (E) CASB�, (F) NLSB�,nurse cell and oocyte nuclei. Scale bar is 50 lm.

b

4.1. imp-a2 function is required during oogenesis and early

embryogenesis

In this report, we distinguish imp-a2 mutations thatdirectly affect ring canal formation, and thus oogenesis,from those producing normal egg chamber developmentbut causing embryonic arrest. Our results indicate thatmutations affecting embryogenesis perturb mitoses witheither arrest in the first nuclear divisions, or mitotic asyn-chrony in escaping embryos. Recently, a role for Imp-ain regulating spindle formation has been described (Grusset al., 2001; Nachury et al., 2001; Wiese et al., 2001) andinvolves the occurrence of a high concentration of RanG-TP near the mitotic chromosomes to release spindle-pro-moting factors from Imp-a (Kalab et al., 1999),indicating that Imp-a plays a critical role during mitosis.Thus, our data confirm that Imp-a2 exerts a critical roleduring nuclear divisions, which may be particularly sensi-tive to any loss of factors involved in mitotic progression.In subsequent somatic cell divisions, which are insensitiveto the inactivation of imp-a2, the duration of mitosesmay be extended without affecting the development ofhomozygous D14 animals. Current investigations of muta-tions causing arrest during early embryogenesis will pro-vide further insights into the role of imp-a2 duringmitosis. Interestingly, imp-a2 inactivation causes no alter-ation in female meiosis, as indicated by the occurrence ofpolar bodies in the majority of eggs laid by mutant femalesin which ring canal assembly occurs normally. These resultsare in contrast to those of RNA interference studies con-ducted in Caenorhabditis elegans showing that a decreasein ima-3 gene activity results in the failure of germ cellsto progress through meiotic prophase I during oocytedevelopment (Geles and Adam, 2001), but in general theimp-a genes of C. elegans are evolutionarily distinct fromany known imp-a and may have thus acquired new func-tions unique to nematode.

4.2. Imp-a2 domains regulating ring canal morphogenesis

Our major concern was to identify Imp-a2 domainsinvolved in the process of ring canal assembly. We found

canals. Ovaries from wild-type (A and B), imp-a2D14 (C), and kelchDE1 (D)of fluorescence staining in ring canal was collected with a laser scanningemission was determined as described in Section 2. The false colors in A2–the laboratory-fixed co-ordinate system, as indicated in the star-diagram

that the orientation of the emission dipoles and its magnitude can be readne both wild-type and mutant imp-a2D14 or kelchDE1 ring canals display are tangentially organized around the ring canal. In the edge-aligned plane,

g chambers. Ovaries from (A) wild-type, (B) mutant D14 flies, and mutant) LNLSB�, (H) SNLSB�or (I) DIM� transgenes, were double-stained for

r D14 nurse cells expressing the (C) imp-a2+, (H) SNLSB�, or (I) DIM�

or (G) LNLSB� transgenes, Imp-a2 proteins are predominantly found in


that essentially two domains of Imp-a2, namely the LNLS-and CAS-binding domains, were required for normal ringcanal formation. As in D14, the ring canals of LNLSB�

and CASB� were nearly fully occluded by an F-actin‘‘plug’’, blocking the transfer of nurse cell cytoplasm atthe time of dumping but exerting no effect on the transportof factors involved in the establishment of egg axes, as indi-cated by the formation of dorsal appendages. In DIBB thestructure of the ring canals was also affected, albeit to alesser degree than in D14. During early vitellogenesis, wefound that the DIBB ring canals were heterogeneous in sizewith the smallest canals fully obstructed by actin filamentsand devoid of Kelch, whereas in pre-dumping egg cham-bers we observed a majority of ring canals displaying nor-mal or nearly normal morphology and size. MoreoverKelch, albeit in lower amount than in wild type, was pres-ent on these canals indicating that the DIBB mutation mayessentially delay Kelch deposition. This hindrance is how-ever insufficient to prevent Kelch to fulfil its function,resulting in the transfer of nurse cell components into theoocyte at the time of dumping.

The finding that mutations in the LNLS- and CAS-bind-ing domains of Imp-a2 affect ring canal assembly suggeststhat Imp-a2 binds through these domains to one or morefactors regulating Kelch function. However, the accumula-tion of the mutant proteins in nurse cell nuclei constitutes aremarkable feature that can be interpreted in two ways.This result can be interpreted if we envisage that theLNLSB� and CASB� mutations prevent the interactionof the mutant proteins with specific cytoplasmic compo-nents, through which Imp-a2 is normally maintained inthe cytoplasm. Second we can also conceive that thesemutations block the mutant proteins in the nuclear com-partment, provided that during interphase the nuclearexport of Imp-a2 is much faster than its import, since weare unable to detect Imp-a2 in the nucleus. In view of themolecular interactions occurring between Imp-a2 withitself and with CAS (Herold et al., 1998; Matsuura andStewart, 2004) this hypothesis is conceivable. However,analysis of Imp-a2 distribution in dcasts1 and dcasts28

mutant ovaries (Tekotte et al., 2002) revealed a normalcytoplasmic Imp-a2 localization (data not shown). Wewere unable to detect any nuclear Imp-a2 accumulationindicating the absence of Imp-a2 shuttling betweencytoplasm and nucleus during oogenesis.

In addition, the occurrence of large amounts of DIBB innurse cell nuclei refutes the idea of a rapid nucleo-cytoplas-mic shuttling of Imp-a2, since DIBB retains its ring canalfunction. Thus, a nuclear sequestration is not sufficient toprevent DIBB to function in the cytoplasm.

The finding that the DIBB protein accumulates in thenucleus constitutes also an intriguing observation becausethe deletion of the Imp-b binding domain should preventthe DIBB binding with Imp-b and thus block its transferinto the nucleus. However, in Drosophila a directinteraction between Imp-a2 and Imp-b remains to bedemonstrated. Furthermore, the findings that the DIBB

and CASB� proteins retain the ability to bind polymerizedactin indicate that the modifications introduced in Imp-a2exert no major structural change but affect specific domainsof Imp-a2.

Similar to the findings that heat-shock and stress condi-tions induce nuclear retention and inhibition of Importin-arecycling in human cells (Furuta et al., 2004; Miyamotoet al., 2004), it is possible to envisage that mutations abol-ishing interaction with specific cytoplasmic partners resultin a nuclear translocation of the mutant proteins. Our pre-vious finding that treatment with Cytochalasin D, an agentpreventing actin polymerization (Lin et al., 1980), leads toImp-a2 nuclear accumulation (Gorjanacz et al., 2002) indi-cates that nuclear sequestration may result from the desta-bilization of protein complexes involving F-actin.However, our knowledge on the interaction between F-ac-tin and Imp-a2 remains conflicting. On the one hand, wecan show an NLS-dependent physical interaction betweenF-actin and Imp-a2 and, accordingly, find that mutationsin the NLS-binding domain of Imp-a2 block this interac-tion. On the other hand, we detect a cytosolic distributionof Imp-a2 in nurse cells and never observe Imp-a2 associ-ated with either the ring canals or the actin bundles main-taining the nucleus in the center of nurse cells, all structurescontaining high concentrations of F-actin. The lack ofImp-a2 association with F-actin macro-structures doesnot however preclude that Imp-a2 may preferentially inter-act with F-actin under the form of filaments dispersed innurse cell cytoplasm and thus barely visible under Phalloi-din staining. The nuclear accumulation of CASB� andDIBB, which are able to bind F-actin in vitro, may indicatethat these proteins are unable to interact with proteinsanchored to actin filaments. Taking into consideration thatthe association between Imp-a2 and F-actin reflects a bonafide interaction, we propose that Imp-a2 associates withNLS-proteins anchored to micro-filaments. This anchoragemay be necessary for targeting specific proteins to the cyto-plasm or for storing nuclear proteins that will be laterrequired during early embryonic development. Identifica-tion of factors that may bind to the NLSB and CASBdomains would thus provide valuable insights into themechanism of Imp-a2 function in nurse cells.

4.3. Regulation of kelch function by imp-a2

Although Kelch can interact with Imp-a1 in yeast cells(Giot et al., 2003), we were unable to find any interactionbetween Kelch and Imp-a2 in these cells. In addition wewere equally unable to detect Kelch in Imp-a2 immuno-complexes, or reciprocally Imp-a2 in Kelch complexes(data not shown). Thus, our data preclude a direct interac-tion between Imp-a2 and Kelch. On this basis, we concludethat Imp-a2 should indirectly regulate Kelch through oneor more distinct proteins.

Identification of Imp-a2 partners was achieved by usingthe yeast two-hybrid system or by physically isolatingImp-a2 protein complexes and determining the molecular


nature of their components. By using both procedures wecharacterized numerous proteins able to interact withImp-a2 in yeast cells or present in ovarian Imp-a2 com-plexes (data not shown). Among these proteins we foundthat Grip71 could also interact with Kelch in yeast cells,confirming the independent finding that Grip71 is a highconfidence partner of Kelch (Giot et al., 2003; Gunawar-dane et al., 2003). However, analyses of ovarian proteinextracts revealed no direct in vivo interaction betweenGrip71 and Imp-a2 or Kelch. In addition, over-expressionof a grip71+ transgene carried no effect on ring canal for-mation (data not shown), indicating that Grip71 is unlikelyto function in ring canal assembly. Other proteins knownto interact with Imp-a2 or regulate its function, such asCAS and Ran-GTP, can be also excluded since they pre-dominantly accumulate in nurse cell nuclei during lateegg chamber development (Tekotte et al., 2002; Triesel-mann and Wilde, 2002). Further works will be requiredto determine the factor(s) mediating Imp-a2 function inKelch deposition onto the ring canals.

Multiple factors are involved in the assembly of ringcanals that is initiated at the position of arrested cleavagefurrows formed during the mitotic divisions of germ linecells. In Drosophila, once egg chambers are formed, ringcanal growth can be divided in two phases. First the actinrim increases to a thickness of �0.3 lm and the diameterof the ring grows to 2 lm. Then, the thickness of theactin rim and the density of actin filaments remain con-stant as the ring canals expand to a diameter of �10–12 lm (Robinson and Cooley, 1996). The net increaseof actin within the ring canals reaches 134-fold (Tilneyet al., 1996). During the rapid phase of ring canal growth,the actin filaments are polymerized on the plasma mem-brane side and disassembled on the inner part of therim to widen the lumen. Furthermore, actin filamentstransform from a single continuous bundle into severalinterwoven cables. The accumulation of actin during theinitial phase of ring canal assembly is independent fromKelch. Two genes, including cheerio and hts, regulatethe initial stage of ring canal assembly and encode theactin filament-cross-linking protein ABP280/filamin (Liet al., 1999; Sokol and Cooley, 1999) and an adducing-like protein (Yue and Spradling, 1992), respectively.Then, phosphotyrosine-containing proteins appear onring canals and persist throughout egg chamber develop-ment. Two tyrosine kinases, Src64 and Tec29, are opera-tive in ring canal expansion. Both Scr64 and Tec29

mutants share similar phenotypes displaying small ringcanals lacking the characteristic phosphotyrosine content(Dodson et al., 1998; Roulier et al., 1998). Src64 activityis necessary for ring canal localization and full activationof Tec29 (Guarnieri et al., 1998; Roulier et al., 1998),although Tec29 by itself localizes to ring canals (Luet al., 2004). One of the major targets of Src64 is Kelchwhose function is required for the completion of ringcanal assembly. Src64 can specifically phosphorylate Kel-ch at tyrosine residue 627 and this phosphorylation

appears to regulate actin filament organization (Kelsoet al., 2002). In particular the expression of a kelchY627A

transgene producing a nonphosphorylatable Kelch pro-tein mimics src64 mutations and gives rise to small ringcanals characterized by two parallel rims of F-actin andno F-actin in the lumen (Kelso et al., 2002). The pheno-type of kelchY627A ring canals characterized by a smallerdiameter, a reduced thickness, and the withdrawal ofactin from their lumen (Kelso et al., 2002) suggest thatKelchY627A can still de-polymerize actin filaments butremains unable to promote actin polymerization. Sincetyrosine phosphorylation takes normally place on D14

ring canals it is very unlikely that Imp-a2 regulates theactivity of Src64 and Tec29, and through them the depo-sition of Kelch on ring canals.

Laser scanning confocal fluorescence anisotropy imag-ing shows that the organization and architecture of actinfilaments in the ring canals as well as the orientation ofthe emission dipoles along the ring canals are essentiallysimilar in both wild-type and mutant D14 or kelch ringcanals of similar size. The mutant ring canals exhibit how-ever a lesser degree of order that may explain their slight-ly greater thickness. In general, the mutant ring canalsessentially differed from wild-type by an increased accu-mulation of actin filaments in their lumen, indicating thatKelch might be predominantly required for depolymeriza-tion of F-actin and apparently unnecessary for ring canalgrowth.

Similar to the role of Importin-a in the regulation ofspindle formation, we envisage that Imp-a2 releases Kel-ch from one or more factors present in the cytoplasm.These factors, yet to be identified, may prevent Kelchto function during the early stages of egg chamberdevelopment, and should be inactivated, or sequestratedby Imp-a2, to allow Kelch deposition on the ring canalsand completion of their assembly. We would predictthat in the absence of Imp-a2 these factors are associat-ed with Kelch and prevent Kelch to access the ringcanals. Analysis of the components of Kelch-complexesin D14 egg chambers may provide insights into themolecular nature of the factor(s) sequestering Kelch inthe cytoplasm.

In conclusion our analysis of the imp-a2 mutant pheno-types demonstrate that Imp-a2 is essential for gametogen-esis (Gorjanacz et al., 2002; Giarre et al., 2002) and earlyembryonic mitoses (this work). In contrast, the functionof imp-a3 is required for the development of larval andadult tissues (Mason et al., 2003), although the precise nat-ure of the cellular and molecular defects remains yet to bedetermined. Moreover no mutation is yet available for imp-a1 to undertake in vivo genetic studies. Our studies demon-strate that the house-keeping function of imp-a2 in nucleartransport is dispensable, albeit not its functions in theassembly of ring canals and mitosis, and show that, as anadaptor protein, Imp-a2 has acquired specific functionsin the assembly of macro-molecular structures during theevolution of metazoan animals.


Acknowledgments

The authors thank D. Albrecht, G. Robinson, R. Sch-mitt, O. Zsıros, and I. Velkeyne Krausz for technical assis-tance and K. Helm for secretarial help. This work wassupported by grants to B.M.M. from the Deutsche Fors-chungsgemeinschaft (436UNG113/81/26–5), fellowshipsto M.G., and I.K., from the Deutsches Krebsforschungs-zentrum, to I.K. from the Hungarian Fund for Basic Re-search (OTKA T034393), to G.G. for the design andconstruction of DP-LSM from the Ministry of Education(OMALK-00219; co-financed by Carl Zeiss Jena GmbH)and to I.P. from the Ministry for Economy (TST).

References

Andrade, M.A., Petosa, C., O’Donoghue, S.I., Muller, C.W., Bork, P.,2001. Comparison of ARM and HEAT protein repeats. J. Mol. Biol.309, 1–18.

Chook, Y.M., Blobel, G., 2001. Karyopherins and nuclear import. Curr.Opin. Struct. Biol. 11, 703–715.

Coates, J.C., 2003. Armadillo repeat proteins: beyond the animalkingdom. Trends Cell Biol. 13, 463–471.

Conti, E., Izaurralde, E., 2001. Nucleocytoplasmic transport enters theatomic age. Curr. Opin. Cell Biol. 13, 310–319.

Conti, E., Kuriyan, J., 2000. Crystallographic analysis of the specific yetversatile recognition of distinct nuclear localization signals by kary-opherin alpha. Struct. Fold. Des. 8, 329–338.

Conti, E., Uy, M., Leighton, L., Blobel, G., Kuriyan, J., 1998. Crystal-lographic analysis of the recognition of a nuclear localization signal bythe nuclear import factor karyopherin alpha. Cell 94, 193–204.

Cooley, L., 1998. Drosophila ring canal growth requires Src and Teckinases. Cell 93, 913–915.

Dodson, G.S., Guarnieri, D.J., Simon, M.A., 1998. Src64 is required forovarian ring canal morphogenesis during Drosophila oogenesis.Development 125, 2883–2892.

Fontes, M.R., Teh, T., Kobe, B., 2000. Structural basis of recognition ofmonopartite and bipartite nuclear localization sequences by mamma-lian importin-alpha. J. Mol. Biol. 297, 1183–1194.

Furuta, M., Kose, S., Koike, M., Shimi, T., Hiraoka, Y., Yoneda, Y.,Haraguchi, T., Imamoto, N., 2004. Heat-shock induced nuclear reten-tion and recycling inhibition of importin alpha. Genes Cells 9, 429–441.

Gama-Carvalho, M., Carmo-Fonseca, M., 2001. The rules and roles ofnucleocytoplasmic shuttling proteins. FEBS Lett. 498, 157–163.

Geles, K.G., Adam, S.A., 2001. Germline and developmental roles of thenuclear transport factor importin alpha3 in C. elegans. Development128, 1817–1830.

Giarre, M., Torok, I., Schmitt, R., Gorjanacz, M., Kiss, I., Mechler, B.M.,2002. Patterns of importin-alpha expression during Drosophila sper-matogenesis. J. Struct. Biol. 140, 279–290.

Giot, L., Bader, J.S., Brouwer, C., Chaudhuri, A., Kuang, B., Li, Y., Hao,Y.L., Ooi, C.E., Godwin, B., Vitols, E., et al., 2003. A proteininteraction map of Drosophila melanogaster. Science 302, 1727–1736.

Goldfarb, D.S., Corbett, A.H., Mason, D.A., Harreman, M.T., Adam,S.A., 2004. Importin alpha: a multipurpose nuclear-transport receptor.Trends Cell Biol. 14, 505–514.

Gorjanacz, M., Adam, G., Torok, I., Mechler, B.M., Szlanka, T., Kiss, I.,2002. Importin-alpha 2 is critically required for the assembly of ringcanals during Drosophila oogenesis. Dev. Biol. 251, 271–282.

Gorlich, D., Kutay, U., 1999. Transport between the cell nucleus and thecytoplasm. Annu. Rev. Cell Dev. Biol. 15, 607–660.

Gruss, O.J., Carazo-Salas, R.E., Schatz, C.A., Guarguaglini, G., Kast, J.,Wilm, M., Le Bot, N., Vernos, I., Karsenti, E., Mattaj, I.W., 2001.Ran induces spindle assembly by reversing the inhibitory effect ofimportin alpha on TPX2 activity. Cell 104, 83–93.

Guarnieri, D.J., Dodson, G.S., Simon, M.A., 1998. SRC64 regulates thelocalization of a Tec-family kinase required for Drosophila ring canalgrowth. Mol. Cell 1, 831–840.

Gunawardane, R.N., Martin, O.C., Zheng, Y., 2003. Characterization ofa new gammaTuRC subunit with WD repeats. Mol. Biol. Cell 14,1017–1026.

Gupta, V.P., Kornfield, J.A., 1994. Polarization modulation laser-scan-ning microscopy: a powerful tool to image molecular orientation andorder. Rev. Sci. Instrum. 65, 2823–2828.

Gupta, V.P., Kornfield, J.A., Ferencz, A., Wegner, G., 1994. Controllingmolecular order in ‘‘Hairy-rod’’ Langmuir Blodgett films: polarizationmodulation microscopy study. Science 265, 940–942.

Hachet, V., Kocher, T., Wilm, M., Mattaj, I.W., 2004. Importin alphaassociates with membranes and participates in nuclear envelopeassembly in vitro. EMBO J. 23, 1526–1535.

Herold, A., Truant, R., Wiegand, H., Cullen, B.R., 1998. Determinationof the functional domain organization of the importin alpha nuclearimport factor. J. Cell Biol. 143, 309–318.

Hood, J.K., Silver, P.A., 1998. Cse1p is required for export of Srp1p/importin-alpha from the nucleus in Saccharomyces cerevisiae. J. Biol.Chem. 273, 35142–35146.

Hudson, A.M., Cooley, L., 2002. Understanding the function of actin-binding proteins through genetic analysis of Drosophila oogenesis.Annu. Rev. Genet. 36, 455–488.

Jans, D.A., Xiao, C.Y., Lam, M.H., 2000. Nuclear targeting signalrecognition: a key control point in nuclear transport? Bioessays 22,532–544.

Kalab, P., Pu, R.T., Dasso, M., 1999. The Ran GTPase regulates mitoticspindle assembly. Curr. Biol. 9, 481–484.

Kamakaka, R.T., Tyree, C.M., Kadonaga, J.T., 1991. Accurate and efficientRNA polymerase II transcription with a soluble nuclear fraction derivedfrom Drosophila embryos. Proc. Natl. Acad. Sci. USA 88, 1024–1028.

Katoh, K., Hammar, K., Smith, P.J., Oldenbourg, R., 1999. Birefringenceimaging directly reveals architectural dynamics of filamentous actin inliving growth cones. Mol. Biol. Cell 10, 197–210.

Kelso, R.J., Hudson, A.M., Cooley, L., 2002. Drosophila Kelch regulatesactin organization via Src64-dependent tyrosine phosphorylation. J.Cell Biol. 156, 703–713.

Kobe, B., 1999. Autoinhibition by an internal nuclear localization signalrevealed by the crystal structure of mammalian importin alpha. Nat.Struct. Biol. 6, 388–397.

Kohler, M., Ansieau, S., Prehn, S., Leutz, A., Haller, H., Hartmann, E.,1997. Cloning of two novel human importin-alpha subunits andanalysis of the expression pattern of the importin-alpha protein family.FEBS Lett. 417, 104–108.

Kohler, M., Speck, C., Christiansen, M., Bischoff, F.R., Prehn, S., Haller,H., Gorlich, D., Hartmann, E., 1999. Evidence for distinct substratespecificities of importin alpha family members in nuclear proteinimport. Mol. Cell. Biol. 19, 7782–7791.

Kuersten, S., Ohno, M., Mattaj, I.W., 2001. Nucleocytoplasmic transport:Ran, beta and beyond. Trends Cell Biol. 11, 497–503.

Kutay, U., Bischoff, F.R., Kostka, S., Kraft, R., Gorlich, D., 1997. Exportof importin alpha from the nucleus is mediated by a specific nucleartransport factor. Cell 90, 1061–1071.

Leung, S.W., Harreman, M.T., Hodel, M.R., Hodel, A.E., Corbett, A.H.,2003. Dissection of the karyopherin alpha nuclear localization signal(NLS)-binding groove: functional requirements for NLS binding. J.Biol. Chem. 278, 41947–41953.

Li, M.G., Serr, M., Edwards, K., Ludmann, S., Yamamoto, D., Tilney,L.G., Field, C.M., Hays, T.S., 1999. Filamin is required for ring canalassembly and actin organization during Drosophila oogenesis. J. CellBiol. 146, 1061–1074.

Lin, D.C., Tobin, K.D., Grumet, M., Lin, S., 1980. Cytochalasins inhibitnuclei-induced actin polymerization by blocking filament elongation. J.Cell Biol. 84, 455–460.

Lu, N., Guarnieri, D.J., Simon, M.A., 2004. Localization of Tec29 to ringcanals is mediated by Src64 and PtdIns(3,4,5)P(3)-dependent mecha-nisms. EMBO J. 23, 1089–1100.


Mahajan-Miklos, S., Cooley, L., 1994. Intercellular cytoplasm transportduring Drosophila oogenesis. Dev. Biol. 165, 336–351.

Malik, H.S., Eickbush, T.H., Goldfarb, D.S., 1997. Evolutionary special-ization of the nuclear targeting apparatus. Proc. Natl. Acad. Sci. USA94, 13738–13742.

Mason, D.A., Fleming, R.J., Goldfarb, D.S., 2002. Drosophila

melanogaster importin alpha1 and alpha3 can replace importinalpha2 during spermatogenesis but not oogenesis. Genetics 161,157–170.

Mason, D.A., Endre, M., Fleming, R.J., Goldfarb, D.S., 2003. TheDrosophila melanogaster importin a3 locus encodes an essential generequired for the development of both larval and adult tissues. Genetics165, 1943–1958.

Mathe, E., Bates, H., Huikeshoven, H., Deak, P., Glover, D.M., Cotterill,S., 2000. Importin-alpha3 is required at multiple stages of Drosophila

development and has a role in the completion of oogenesis. Dev. Biol.223, 307–322.

Matsuura, Y., Stewart, M., 2004. Structural basis for the assembly of anuclear export complex. Nature 432, 872–877.

Mattaj, I.W., Englmeier, L., 1998. Nucleocytoplasmic transport: thesoluble phase. Annu. Rev. Biochem. 67, 265–306.

Miyamoto, Y., Saiwaki, T., Yamashita, J., Yasuda, Y., Kotera, I.,Shibata, S., Shigeta, M., Hiraoka, Y., Haraguchi, T., Yoneda, Y.,2004. Cellular stresses induce the nuclear accumulation of importinalpha and cause a conventional nuclear import block. J. Cell Biol. 165,617–623.

Nachury, M.V., Maresca, T.J., Salmon, W.C., Waterman-Storer, C.M.,Heald, R., Weis, K., 2001. Importin beta is a mitotic target of the smallGTPase Ran in spindle assembly. Cell 104, 95–106.

Oda, T., Namba, K., Maeda, Y., 2005. Position and orientation ofphalloidin in F-actin determined by X-ray fiber diffraction analysis.Biophys. J. 88, 2727–2736.

Oldenbourg, R., Katoh, K., Danuser, G., 2000. Mechanism of lateralmovement of filopodia and radial actin bundles across neuronalgrowth cones. Biophys. J. 78, 1176–1182.

Robinson, D.N., Cant, K., Cooley, L., 1994. Morphogenesis of Drosophila

ovarian ring canals. Development 120, 2015–2025.Robinson, D.N., Cooley, L., 1996. Stable intercellular bridges in

development: the cytoskeleton lining the tunnel. Trends Cell Biol. 6,474–479.

Robinson, D.N., Cooley, L., 1997a. Examination of the function of twoKelch proteins generated by stop codon suppression. Development124, 1405–1417.

Robinson, D.N., Cooley, L., 1997b. Genetic analysis of the actincytoskeleton in the Drosophila ovary. Annu. Rev. Cell Dev. Biol. 13,147–170.

Rorth, P., 1998. Gal4 in the Drosophila female germline. Mech. Dev. 78,113–118.

Roulier, E.M., Panzer, S., Beckendorf, S.K., 1998. The Tec29 tyrosinekinase is required during Drosophila embryogenesis and interacts withSrc64 in ring canal development. Mol. Cell 1, 819–829.

Sokol, N.S., Cooley, L., 1999. Drosophila filamin encoded by the cheeriolocus is a component of ovarian ring canals. Curr. Biol. 9, 1221–1230.

Stewart, M., Baker, R.P., Bayliss, R., Clayton, L., Grant, R.P.,Littlewood, T., Matsuura, Y., 2001. Molecular mechanism of trans-location through nuclear pore complexes during nuclear proteinimport. FEBS Lett. 498, 145–149.

Szito, T., Garab, G.I., Mustardy, L.A., Kiss, J.G., Faludi-Daniel, A.,1984. Increasing fluctuation in orientation of pigment protein complexwithin photosynthetic membranes treated with linolenic acid. Photo-biochem. Photobiophys. 8, 239–249.

Tekotte, H., Berdnik, D., Torok, T., Buszczak, M., Jones, L.M., Cooley,L., Knoblich, J.A., Davis, I., 2002. Dcas is required for importin-alpha3 nuclear export and mechano-sensory organ cell fate specifica-tion in Drosophila. Dev. Biol. 244, 396–406.

Tilney, L.G., Tilney, M.S., Guild, G.M., 1996. Formation of actinfilament bundles in the ring canals of developing Drosophila follicles. J.Cell Biol. 133, 61–74.

Torok, I., Herrmann-Horle, D., Kiss, I., Tick, G., Speer, G., Schmitt, R.,Mechler, B.M., 1999. Down-regulation of RpS21, a putative transla-tion initiation factor interacting with P40, produces viable minuteimagos and larval lethality with overgrown hematopoietic organs andimaginal discs. Mol. Cell. Biol. 19, 2308–2321.

Torok, I., Strand, D., Schmitt, R., Tick, G., Torok, T., Kiss, I., Mechler,B.M., 1995. The overgrown hematopoietic organs-31 tumor suppres-sor gene of Drosophila encodes an Importin-like protein accumulatingin the nucleus at the onset of mitosis. J. Cell Biol. 129, 1473–1489.

Trieselmann, N., Wilde, A., 2002. Ran localizes around the microtubulespindle in vivo during mitosis in Drosophila embryos. Curr. Biol. 12,1124–1129.

Van Doren, M., Williamson, A.L., Lehmann, R., 1998. Regulation ofzygotic gene expression in Drosophila primordial germ cells. Curr. Biol.8, 243–246.

Weis, K., 2003. Regulating access to the genome: nucleocytoplasmictransport throughout the cell cycle. Cell 112, 441–451.

Wiese, C., Wilde, A., Moore, M.S., Adam, S.A., Merdes, A., Zheng, Y.,2001. Role of importin-beta in coupling Ran to downstream targets inmicrotubule assembly. Science 291, 653–656.

Wulf, E., Deboben, A., Bautz, F.A., Faulstich, H., Wieland, T., 1979.Fluorescent phallotoxin, a tool for the visualization of cellular actin.Proc. Natl. Acad. Sci. USA 76, 4498–4502.

Xue, F., Cooley, L., 1993. kelch encodes a component of intercellularbridges in Drosophila egg chambers. Cell 72, 681–693.

Yue, L., Spradling, A.C., 1992. hu-li tai shao, a gene required for ringcanal formation during Drosophila oogenesis, encodes a homolog ofadducin. Genes Dev. 6, 2443–2454.

Domains of Importin-[alpha] 2 required for ring canal assembly during Drosophila oogenesis

Documents