Entry is a rate-limiting step for viral infection in a Drosophila melanogaster model of pathogenesis

DCV infection and entry in vitro

To establish the characteristics of DCV infection in D. melanogaster, we developed an assay to monitor infection in vitro. We generated infectious DCV (with a strain isolated originally in Charolles, France) by inoculating the D. melanogaster tissue culture cell line DL2 (ref. 11). DCV grew in the cells, as evidenced by cytopathic effect (CPE) and by the presence of infectious virions after biochemical purification from infected cells. CPE, in contrast to budding, is the mechanism of viral release used by many viruses, including mammalian picornaviruses¹². We monitored CPE by flow cytometry and distinguished dead cells from live cells by staining with propidium iodide. We noted increased cell death in DL2 cells by 4 d after infection (Fig. 1a). We used a modified CPE assay to determine the titer of the purified virus, and found that the concentration of the purified virus was 1 × 10⁹ to 1 × 10¹⁰ infectious units (i.u.)/µl (data not shown).

To monitor infection more directly, we generated polyclonal anti-bodies to DCV in chickens. Immunoblot analysis confirmed that the antibody recognized purified virions and cells infected with DCV but not uninfected controls (Fig. 1b). The sizes of the reactive bands were consistent with the molecular weights of the structural capsid proteins, between 28 and 38 kDa for DCV¹³.

Many viruses require transit through an endocytic compartment for productive infection, whereas others do not and directly fuse with the plasma membrane. To test whether DCV travels through a vesicular compartment for entry, we took advantage of the fact that endocytosis is blocked at low temperature. Thus, by infecting cells with DCV at 4 °C we allowed extracellular binding but not internalization by endocytic vesicles. Next, we removed free virions by washing the cells and returned them to 25 °C. We did not detect this plasma membrane–bound fraction of DCV (Fig. 2a, B). After 3 h, we noted virions inside a vesicular compartment (Fig. 2a, C). This would not be expected if the viruses fused directly with the plasma membrane. Furthermore, longer incubation allowed the virus to escape from this compartment and gain entry into the cytoplasm to produce new viral antigens (Fig. 2a, D). Thus, the formation of vesicles containing viruses preceded the production of new viral antigens.

To directly demonstrate that the cytoplasmic staining was due to the production of new viral antigens, we treated cells before infection with cycloheximide. This treatment blocked the late production of new viral antigens in the cytoplasm (Fig. 2a, D) but not the vesicular staining (Supplementary Fig. 1 online). These data indicate that DCV traffics through a vesicular compartment for entry in vitro.

To further demonstrate a dependence of DCV on endocytosis for viral entry, we assayed whether viral replication was sensitive to bafilomycin A1 or valinomycin. Bafilomycin A1, a vacuolar (v)-ATPase inhibitor, blocks acidification of vacuolar endosomal compartments used extensively in mammalian systems to demonstrate the dependence of viral entry on acidification¹⁴. Inhibition of v-ATPase also blocks the movement of vesicles from the early endosomal compartment to the endosomal carrier vesicles, an early step in vesicular trafficking¹⁵. Valinomycin is an ionophore that causes the collapse of the membrane potential of all membranes in the cell, including the endosomal vesicles¹⁶. This also inhibits the v-ATPase, among other targets. We confirmed that these drugs affected vesicular trafficking in D. melanogaster cells by monitoring the uptake of ‘LysoTracker’, a dye that is selectively retained in acidified compartments. In the presence of valinomycin or bafilomycin A1, the cells were unable to efficiently internalize the dye, thereby demonstrating a defect in endocytosis (Supplementary Fig. 2 online).

To test the effects of these small molecules on viral replication, we pretreated DL2 cells with bafilomycin A1 or valinomycin and then added DCV. We washed the cells after 3 h to remove the drugs and free virus, and assayed them 20 h later with an immunofluorescence assay for the production of viral antigens. Both bafilomycin A1 and valinomycin inhibited viral antigen production (Fig. 2b). We used cycloheximide as a positive control for these experiments because it blocks the synthesis of new viral proteins. Quantification of three independent experiments showed that bafilomycin A1 inhibited viral growth by 98.6% (P ≤ 0.001); valinomycin, by 100% (P ≤ 0.001); and cycloheximide, by 99.9% (P ≤ 0.001; Fig. 2c). These drugs had no effect on cellular viability in these conditions (data not shown). These results demonstrate a requirement for endocytosis in viral replication in vitro.

We next examined whether endocytosis is required to present the cellular receptor at the plasma membrane for viral binding and thus entry. We therefore tested whether the virus can enter cells in the presence of valinomycin, which blocks the movement of vesicles from the early endosomal compartment to the endosomal carrier vesicles. If the virus required endocytosis for entry, the virus would enter the cells and traffic to the early endosome but not gain entry to the cytoplasm to begin new replication. In contrast, if endocytosis were required to bring the receptor to the plasma membrane, no vesicular staining would be detected in these conditions. The virus was able to enter a vesicular compartment in the presence of valinomycin but did not express cytoplasmic antigen as in the untreated case, in which viral antigens are detected in the cytoplasm (Fig. 2a, G and H). This is consistent with a requirement for endocytosis in viral entry and not simply for the presentation of the cellular receptor to DCV.

Infection requires the endocytic machinery in vivo

To determine whether endocytosis is the mechanism for viral entry for DCV in vivo and, if so, to define essential components of this pathway, we microinjected the purified virus into wild-type flies. Infected flies succumbed to infection when injected with >100 i.u./µl of DCV (Fig. 3a and data not shown). Next, we tested whether flies that carry mutations in genes encoding various components involved in vesicular trafficking showed an altered susceptibility to DCV infection. We reasoned that flies deficient for components required for viral entry should be resistant to DCV infection, and thus we specifically tested mutants deficient in the clathrin-mediated endocytosis pathway (clathrin heavy chain, α-Adaptin, abnormal wing discs, synaptotagmin)¹⁷, vesicle coating (like AP180) and uncoating (hsc70–4 and cysteine string protein), vesicle fusion (syntaxin, synaptobrevin), the AP-3 adapter (orange, ruby, garnet), lysosomal transport (Vps homologs: deep orange and light), and late endosomal trafficking (hook)¹⁸. In addition, we also tested a class of eye color mutants (claret, lightoid and pink) on the assumption that they might encode additional trafficking proteins¹⁹.

To analyze the requirement of these genes in DCV infection of adult flies, we tested flies with homozygous viable mutant back-grounds when available or otherwise, heterozygous flies. Flies with mutations in genes encoding components of the clathrin-mediated endocytosis machinery were clearly protected from lethal doses of virus. Flies with mutations in α-Adaptin (α-Ada³/α-Ada⁴) were completely protected, whereas flies on backgrounds with other mutations such as clathrin heavy chain (chc⁴/+), abnormal wing discs (awd^Kpn/awd ^Kpn) and synaptotagmin (syt^T77/+) showed significant resistance to infection (Fig. 3a,b and data not shown; P < 0.05 in >75% independent tests). There was resistance in some cases in heterozygous mutation backgrounds, indicating that the genetic back-grounds in the stocks were not responsible for the resistance to DCV. Furthermore, the observation that flies with heterozygous α-Ada (α-Ada³/+) as well as awd (awd^Kpn/+) mutant backgrounds were more resistant to infection than were wild-type flies demonstrates the exquisite sensitivity of DCV to the levels of these genes. That loss of even one copy of these genes (heterozygous mutants) had a detectable effect on the resistance of the flies to the virus but not on adult viability demonstrates that viral entry is a rate-limiting step for infection in vivo, and that specific components of the endocytic machinery are required for viral pathogenesis in vivo.

None of the other mutations in trafficking genes that we tested (or, g, rb, dor, lt, hk, lap, hsc70–4, csp, syb, nsyb, syx, com, ltd, ca, pn or p) conferred DCV resistance (P > 0.05), indicating that they are dispensable for viral replication. However, because some genes were tested as heterozygous mutatations (lap, hsc70–4, csp, syb, nsyb and syx), we cannot exclude the possibility that a 50% reduction in the level of these genes is not sufficient to affect DCV susceptibility. In addition, the molecular nature of some of these alleles is not known, such that some mutations may be associated with residual gene activity.

To further demonstrate that these genes were required for viral entry, we tested for the production of actively replicating virus and viral antigens in vivo. First, we assayed infected flies for the production of viral antigens at various time points after infection by immunoblot analysis. Whereas viral proteins were detectable in wild-type flies infected with DCV by 48 h after infection, the α-Adaptin mutants had no detectable immunoreactive bands, even at 72 h after infection (Fig. 3c). The virus-mediated mortality correlates well with the detection of viral antigens. For example, flies that succumbed at a later time point (such as chc⁴/+ flies) produced detectable viral antigens later in the time course. We also found a decrease in viral antigen production in awd (awd^kpn/awd^kpn) and syt (syt^T77/+) mutants (data not shown). These data reinforce the idea that these endocytosis genes are required for viral entry and that decreasing the efficiency of the endocytic machinery can attenuate the infection.

Next, we examined the distribution of viral antigens in infected flies using immunofluorescence techniques. The tissue tropism of viruses affects the pathogenic consequences of infection. Immunostaining allowed us to assess whether the mutant proteins were affecting viral replication in all or a subset of susceptible tissues. Immunofluorescence analysis of wild-type flies showed that the antibody did not recognize uninfected tissues and that the virus infected a subset of tissues, indicating that not all cell types may be able to support viral replication²⁰. The restricted tissue expression may reflect the distribution of the receptor used by the virus for entry. The infected tissues included the fat body (Fig. 3d), the outer sheath of the egg chamber and sheaths surrounding individual ovarioles (Fig. 3d), visceral muscles along the midgut, and a subset of somatic muscles (data not shown). In contrast to the wide-spread infection of the fat body and egg chambers in infected wild-type flies, very few infected α-Ada³/α-Ada⁴ flies showed positive immunore-activity in the fat body and mutant egg chambers (Fig. 3d). Similarly, there was vastly reduced immunostaining in awd (awd^kpn/awd^Kpn), syt (syt^T77/+) and chc (chc⁴/+) mutant backgrounds (data not shown). In these mutants, all cell types normally infected were similarly protected from DCV replication. Moreover, the lack of staining in the mutant backgrounds demonstrates the exquisite sensitivity of this immunostaining assay for DCV infection. Altogether, the in vivo analysis of mutant flies demonstrated the general, essential and quantitative involvement of endocytosis in viral entry.

Autonomous requirement for endocytosis in vivo

To further characterize the requirements for endocytosis in viral entry, we tested whether there was an autonomous requirement for endocytosis in infection. For this we used a mosaic analysis in vivo by blocking endocytosis specifically in the fat body but not other tissues such as the egg chamber; both of these are normally infected by DCV. We took advantage of a dominant negative form of dynamin to specifically inactivate endocytosis in a tissue-specific way. Dynamin, encoded by the shibire gene, is the GTPase required for the fission of endocytic vesicles from the plasma membrane and thus should be required for endocytosis and entry in this system^21–23. We were unable to use temperature-sensitive shibire mutants because of the lethality associated with paralysis at the nonpermissive temperature. Instead we took advantage of the Gal4-UAS system, in which a transgene containing UAS transcription factor sites is expressed when the transcriptional activator Gal4 is expressed in a tissue-specific way. We generated mosaic flies that express a dominant negative form of dynamin (UAS-shi^K44A) only in the adult female fat body with a transgenic line that expresses Gal4 with a yolk promoter (yolk-Gal4 driver^24,25). Using this strategy we found that the infected female flies did not have detectable viral antigen production in their fat body (n = 15; Fig. 4), whereas the other tissues normally infected by DCV, such as the egg chamber, were positive by immunofluorescence (Fig. 4, data not shown). Thus, our results demonstrate that endocytosis is required autonomously for viral infection in vivo and that this pathway is amenable to inhibition.

Fly stocks.

All flies were maintained on standard medium at 24 °C. The following mutant strains were used in this study: clathrin heavy chain (chc⁴), α-Adaptin (α-Ada³, α-Ada⁴), nucleoside diphosphate kinase (awd^K–pn), shibire (UAS-shi.K44A), AP3-σ (orange: or¹), AP3-δ (garnet: g¹, g²), AP3-β (ruby: rb¹), Vps18p (deep orange: dor⁸, dor¹), Vps41p (light: lt¹), hook (hk¹), likeAP180 (lap¹), hsc70–4 (Hsc70–4⁰³⁵⁵⁰), cysteine string protein (Csp⁰³⁹⁸⁸), synaptobrevin (Syb^k07703), neuronal synaptobrevin (syb^I18), syntaxin (syx^L266), NEM-sensitive fusion protein (com^ST53, com¹⁹¹), synaptotagmin (syt^T77, syt^AD1), claret (ca¹), prune (pn¹), lightoid (ltd¹), pink (p^p, p^snb), Yolk protein1-Gal4 (yp1–64) and wild-type (WT) Oregon R (OreR). Alleles are described in the drosophila database (http://flybase.bio.indiana.edu/).

Infections.

Flies 2–4 d old were used for viral challenge. Glass needles ‘pulled’ as for microinjection into embryos were loaded with the inoculum (1 × 10⁵ i.u./µl and connected to an Eppendorf microinjection setup. We injected approximately 200 nl into the anterior ventral abdomen near the dorsal-ventral boundary. Flies were injected and were subsequently maintained at 24 °C. Groups of 25 or more flies were challenged, monitored for mortality and assigned scores daily. Statistical significance was determined with a log rank test. Only results of P < 0.05 in >75% of independent tests were considered significant in mutants that were affected.

CPE assays.

DL2 cells (a gift from P. Christian, CSIRO Entomology, Canberra, Australia) were maintained and propagated as described¹¹. Cells inoculated at a multiplicity of infection of one were assayed for cell death by flow cytometry (Beckton Dickenson) 4 d later by staining with propidium iodide (5 µg/ml). Cellular debris were electronically ‘gated out’ of the analysis. To determine the viral concentration, DL2 cells were plated at 30% confluence on tissue culture plastic and inoculated with serial dilutions of the virus. At 4 d after infection, the cells were stained for 10 min at room temperature with Hoescht 33342 (1:1,000 dilution; Molecular Probes) and Sytox green (1:5,000 dilution; Molecular Probes). The cells were washed twice with PBS and analyzed with an automated Nikon microscope and Metamorph analysis software to determine the minimum concentration of virus that resulted in CPE. This was defined as 1 i.u.

Immunoblot analysis.

For antibody production, virus purified as described above was injected into chickens (Covance). Antibody was purified from eggs following the manufacturer’s protocol (Promega). Adult flies or cells were lysed in radioimmunoprecipitation assay buffer supplemented with a protease inhibitor ‘cocktail’ (Boehringer). Samples were separated by 12% SDS-PAGE and blotted (Immobilon P). Membranes were incubated with antibody to DCV (anti-DCV; 1:50,000 dilution) followed by anti-chicken–horseradish peroxidase (1:200,000 dilution; Jackson Immunochemicals). Blots were visualized with ECL (Amersham).

Immunofluorescence assays.

Flies were dissected in 4% formaldehyde in PBS and fixed for 20 min. Flies were washed in PBS plus 0.1% Triton (PBST), blocked in 3% BSA in PBST and then incubated overnight at 4 °C in anti-DCV (1:5,000 dilution) in PBST. After the membrane was washed, anti-chicken–fluorescein isothiocyanate (1:400 dilution; Jackson Immunochemicals) 3% BSA in PBST was added and samples were incubated overnight at 4 °C. Samples were washed again and mounted in glycerol and Antifade (Molecular Probes). Samples were counterstained with Alexa Fluor 568–phalloidin (1:200 dilution; Molecular Probes) and analyzed with a Leica confocal microscope.

For endocytosis experiments, 3 × 10⁵ cells were plated and placed at 4 °C. Virions were added (1 × 10⁸) and samples were incubated for 3 h at 4 °C. Cells were washed and brought to 25 °C. Cells were washed in PBS and then fixed with 4% formaldehyde in PBS for 10 min at room temperature. Cells were washed twice with PBST, blocked with 3% BSA in PBST. Anti-DCV (1:5,000 dilution) was followed by anti-chicken–fluorescein isothiocyanate (1:200 dilution; Jackson Laboratories) and Hoescht 33342 (1:1000 dilution; Molecular Probes). Cells were analyzed with a Leica confocal microscope.

For drug treatments, 1 × 10⁴ cells in 100 µl were seeded into 96-well plates 24 h before addition of the drug. Drugs were added to the cells, and 15 min later 1 × 10⁵ virions were added. Then, 3 h later the wells were washed and fresh media was added. Plates were analyzed 20 h later as described above, except with an automated Nikon microscope and Metamorph analysis software. Bafilomycin A1 (0.02 µg/ml), valinomycin (0.01 mg/ml) and cycloheximide (0.01 mg/ml) were obtained from Sigma.