ntf-2 is an X-linked essential gene. Depending on the allele, animals die between the 2^nd larval instar and the pupal stage. Some alleles have an adult survival rate of 8–15% of expected (Table 1), and all survivors show a small eye phenotype, strongly reduced numbers of ommatidia 11 12 24 . The eye phenotype varies from 30% of normal size to a more severe phenotype displaying one or two small patches of 10–40 ommatidia (Fig. 1A).

The mutant eye-imaginal discs are smaller than wild-type and are often abnormally shaped (Fig. 1B). Overall, the structure of the mutant eye discs is perturbed and the organization of the actin cytoskeleton is strongly altered (compare Fig. 2A and 2B,C). Only few disorganized, irregularly spaced rabdomere-like structures are apparent in the posterior compartment of the eye disc (arrow in Fig. 2A–C).

We took advantage of the partial loss of function eye phenotype of ntf-2 alleles to identify genes functioning with ntf-2, and performed a dominant suppressor screen of the eye phenotype. Males from 2^nd and 3^rd chromosomal deficiency stocks (deficiency/balancer) uncovering 70% to 80% of the two autosomes, or about 60% of the Drosophila genome, were crossed with ntf-2^P7/FM7 females (Table 1 top). In the next generation the number of surviving ntf-2 males also carrying a deletion was counted and the survivors monitored for their eye phenotype. For our screen we set up 136 individual crosses, many of them repeatedly in order to obtain at least 150 adult progeny to screen for the eye phenotype. We only identified deletions and rearrangements in four regions of the second chromosome that showed suppression (Table 1). The suppression was confirmed using a second ntf-2 (P49) allele.

DNA rearrangements affecting regions 22A and 60B-D showed different results with the two ntf-2 alleles tested and were not pursued. Df(2l)cl-h2 (25D-F) appeared to rescue both viability and the eye phenotype, but the gene responsible for the suppression could not be identified. Df(2L)GpdhA (25D-26A) rescued the eye phenotype, but not viability. To identify the gene(s) responsible for the suppression of the eye phenotype we tested mutations in several genes that are uncovered by Df(2L)GpdhA and are available from the Drosophila stock center.

Mutants in one gene, chickadee (chic), encoding Drosophila Profilin 25 , uncovered by Df(2L)GpdhA, showed suppression of the ntf-2 eye phenotype. We tested several loss-of-function alleles of chic, including a complete lethal null allele (chic²²¹) and other partially viable alleles, that are either female, or male and female sterile. All chic alleles were crossed with at least 2 ntf-2 alleles, except chic²²¹ that was tested with 4 different ntf-2 alleles. The suppression of the eye phenotype was observed in all crosses and the majority of surviving trans-heterozygous males showed suppression of the ntf-2 eye phenotype, restoration of wild-type eyes (Fig. 1A). The percent of males with wild-type eyes varied in different allele combinations. Surprisingly, the eye phenotype was usually either small or wild-type and virtually no eyes of intermediate size were observed.

To investigate the cause underlying the suppression of the ntf-2 phenotype and possible function of Profilin in nuclear transport, we used a reporter gene approach. We assayed nuclear transport using UAS-NLS-NES reporter constructs C-terminally tagged with GFP in different mutant backgrounds. One construct contains a wild-type NLS and NES (UAS-NLS-NES-GFP), the other a wild-type NLS but a mutant NES that is not recognized by the nuclear export machinery (UAS-NLS-NES^P12-GFP; 16 26 ). Expression of the transgenes was driven by a heatshock-GAL4 driver, and the distribution of GFP was analyzed in salivary glands. As previously shown, the activity of the wild-type NES is stronger then that of the NLS 26 . Hence, in wild-type the NLS-NES-GFP is usually localized in the cytoplasm (Fig. 3B). In contrast, NLS-NES^P12-GFP has impaired nuclear export and strongly accumulates in nuclei (Fig. 3A). In homozygous chic⁰¹³²⁰ and the hetero-allelic combination chic²/chic²²¹, the distribution of the GFP reporter is altered. In contrast to the cytoplasmic distribution of NLS-NES-GFP in wild-type, in the chic mutant salivary glands the GFP reporter is found predominantly in the nucleus (Fig. 3D,F). The localization of NLS-NES^P12-GFP is similar in chic and wild-type (Fig. 3C,E), indicating that NLS-mediated import is not affected.

RanGAP functions in nuclear export of cargo and in Sd-RanGAP mutants the NLS-NES-GFP is found in the nucleus and NLS-NES^P12-GFP is distributed the same as in wild-type 16 19 . This failure of exporting NLS-NES-GFP in Sd-RanGAP mutants is reminiscent of what we observe in chic alleles (Fig. 3).

Given the similarity in nuclear export phenotypes in Sd and chic mutants, we tested if Sd would also suppress the eye phenotype of ntf-2 alleles. We crossed the Sd (Sd⁷², 27 ) chromosome with two ntf-2 alleles and found that the eye phenotype was suppressed in both of them. To confirm that the SD-RanGAP mutation, and not other genes on the Sd chromosome, is responsible for the suppression, we expressed a mutated Sd-RanGAP transgene (UAS-Sd-RanGAP12A-6; 16 ) driven by hsp70-GAL4 or arm-GAL4 in ntf-2^P7 and ntf-2^P49 males and observed similar levels of suppression as seen with Sd⁷² (Table 1).

The genetic interaction between Sd-RanGAP and ntf-2 is not altogether surprising because both RanGAP and NTF2 are known to function in the formation of the RanGTP-GDP gradient. To investigate if RanGAP is affected in ntf-2 mutants we studied the distribution of RanGAP in eye discs.

In wild-type cells Ran-Gap is present in low levels in the cytoplasm and forms a clearly visible punctuated circle around the nucleus (Fig. 2A,D). The punctuate pattern of RanGAP is due to its association with nuclear pores 17 18 . This distribution is different in ntf-2 discs. Patches of cells are observed in which RanGAP aggregates in small or large clumps near the nuclei (Fig 2E,F), but in other cells the distribution of the protein looks relatively normal. This observation suggests, that the clumping of RanGAP is an effect of the abnormal organization of the cells within the ntf-2 disc. The cells with clumped RanGAP are usually in close proximity to cells with high levels of F-actin.

To investigate a connection between Profilin, RanGAP, and actin, we next asked whether the function of Profilin or actin polymerization might have an effect on RanGAP localization. We generated clones in eye discs of null alleles of the two genes chic (chic²²¹) and, as a control, act up (acu^E636). Acu participates in actin de-polymerization, the opposite function of Profilin 28 .

In chic clones RanGAP protein is increased around the nuclear envelope and its distribution is uneven and patchy on the nuclear envelope surface (Fig. 4D arrows). In wild-type even, punctuated circles are observed (arrowheads). This abnormal distribution was found in 100% of examined clones (more then 50). In chic clones the level of F-actin was reduced as previously shown 28 . In the acu control clones high levels of F-actin are detected as expected (results not shown, 28 ), but the distribution of RanGAP is not significantly changed (Fig. 4E).

To test whether this patchy protein distribution of RanGAP on nuclear pores of chic²² cells is caused by problems in nuclear envelope assembly, we analyzed the distribution of Lamin and nuclear pore proteins (Nups) in chic²²¹ clones (Fig. 4A,B). The distribution of both Lamin and Nups is affected in about 30% of clones. This is likely due to the mislocalization of RanGAP. It has been shown previously that RanGTPase functions in nuclear pore and envelope formation 22 23 .

The staining experiments show higher levels of RanGAP around nuclei in chic eye disc clones. We investigated if this is due to overall higher levels of RanGAP in mutant cells. The chic alleles used in the clonal analysis are homozygous lethal therefore we prepared extracts from wild-type and mutant 1^st instar larvae. In western blots from extracts of chic²²¹ (lethal at first and early second larval instar) and chic⁰¹³²⁰ (viable and female sterile) larvae, the amount of RanGAP present in mutants is not dramatically changed compared to wild-type (results not shown). This may be because RanGAP and Profilin are maternally contributed and therefore at these early stages a difference in levels is not detected. We then dissected eye-antennal discs from normal larvae and larvae with chic clones (see experiments shown in Figure 4). The dissected tissues also contained some brain material because eye-antennal discs are next to the brain hemispheres and are difficult to separate. In two separate experiments we see an increase of 30–50% in the intensity of the RanGAP band in extracts from discs carrying chic²²¹ somatic clones compared to normal eye discs from chic²²¹/+ larvae. The intensity of the RanGAP bands were normalized to that of the control Bic-D band and equals 2.6 for discs with clones and 1.8 for wild-type discs (Fig. 5).

The w¹¹⁸ stock was used as wild-type stock (WT) in all experiments. All fly stocks were obtained from the Bloomington Stock Center 24 , except UAS-RanGAP12A-6, UAS-NLS-NES^P12-GFP and UAS-NLS-NES-GFP transgenic flies that were sent by Barry Ganetzky and Edwin Chan 16 31 .

The following ntf-2 stocks were used for suppression experiments: w ntf-2^P7/FM7i, P{w [+mC]=ActGFP}JMR3, y[1] f[1], w ntf-2^P49/FM7c, w ntf-2^G0086/FM7c, ntf-2^G0337/FM7c. All alleles carry different P-element insertions in the 5' UTR of the gene, 102–112 bases upstream from the start of transcription that cause reduced viability and the eye phenotype (see Table 1, 11 12 24 ).

To identify suppressors of ntf-2, ntf-2^P7/FM7 and ntf-2^P49/FM7 virgins were crossed with deficiency males. In the next generation the ntf-2 males not carrying a balancer chromosomes were analyzed. We tested 66 deficiencies on the second chromosome and about 70 deficiencies on the third chromosome, and about 25 existing mutations and P-element insertions mapping to the 22–30 cytological region of chromosome two. All were obtained from the Bloomington Stock center.

Alleles of chickadee, chic^K1332, chic²²¹, chic⁰¹³²⁰, chic², In(2R)SD72, In(2R)NS, Sd[72], GAL4 drivers and Balancer chromosomes were obtained from Bloomington Stock Center. UAS-RanGAP12A-6, UAS-NLS-NES^P12-GFP and UAS-NLS-NES-GFP transgenic flies were kindly sent to us by Barry Ganetzky and Edwin Chan 16 .

To identify mutant animals at larval stages we used the GFP marked balancers FM7i P{w [+mC]=ActGFP}JMR3 and CyO, P{w [+mW.hs]=Ubi-GFP.S65T}PAD1. Mutant 2^nd instar larvae were distinguished from their heterozygous siblings by the absence of green fluorescent protein. The expression of GFP-reporter genes were induced by 1 h incubation at 36°C. To diminish the possible effect of heating on nuclear trafficking we waited for 24 h before dissecting and fixing salivary glands. Localization of GFP was observed and fluorescence images obtained using a Zeizz Axioplan 2 (Zeiss) microscope and Image Pro Plus software.

FRT40 chic²²¹ and FRT40 acu^E636 chromosomes were obtained from Jessica Treisman. Clones in eye discs were generated as described 28 .

Rabbit polyclonal antibody against Drosophila RanGAP was provided by Sinthia Stabber and Barry Ganetzki 16 , mab414 recognizing nucleoporins with FXF repeats 32 33 was obtained from BAbCO (Richmond, CA), and anti-Lamin antibody was a gift from Paul Fischer, Stonybrook. Western analysis was performed as previously described 16 17 .

For antibody staining 3rd instar larvae were inverted in phosphate-buffered saline (PBS) and immediately fixed in 4% paraformaldehyde in PBS with 2% DMSO for 40 min and washed several times in PBT (PBS, 0.1% Triton X-100). Then tissues were blocked for 2 hours in PBS containing 1% bovine serum albumin (BSA) and 1% Triton X-100. Antibody incubations were done in PBT with 1% BSA overnight at 4°C. Anti-RanGAP rabbit serum was used at a 1:1000 or 1:800 dilution, and anti-RanGAP-1 monoclonal antibody was used at a 1:400 dilution, mab414 was used at a 1:300 dilution, anti-Lamin – 1:30. Secondary antibodies were used at a dilution of 1:500. Cy-3 conjugated anti-mouse and anti-rabbit IgG were purchased from Jackson Immuno Research Laboratories Inc. (West Grove, PA). F-actin was visualized by incubation with Alexa488 Phalloidin at 1:80 for 2 hours. DNA was stained using Hoechst 33258 (Molecular Probes, Eugene, OR). Samples were mounted in Vectashield (Vector Laboratories) and examined with a Leica DM IRBE (Leica) laser scanning confocal microscope. The images were analyzed with Leica Microsystems software and further processed using Adobe PhotoShop.