During infection of insect cells with the baculovirus Autographa californica nucleopolyhedrovirus (AcMNPV), budded virus (BV) attaches to susceptible insect cells via interactions of the major surface glycoprotein GP64 with host cellular receptors (25), following which the BV particle enters the cell via adsorptive endocytosis (72). GP64 facilitates fusion of the viral envelope with the endosomal membrane (7, 49, 50), releasing the nucleocapsid into the cytoplasm. Trafficking of nucleocapsids to the nucleus is thought to be mediated by the formation of actin cables (Factin), and major re-distribution of this cytoskeletal filament has been shown to occur during the early stages of infection (13, 14, 17, 34, 43, 44, 53, 70). Progeny nucleocapsids produced within the nucleus are trafficked to the plasma membrane via an unknown mechanism, to distinct areas of the membrane rich in GP64 (46, 72). Here, nucleocapsids bud through the plasma membrane, acquiring a host-derived envelope enriched in GP64 at one end (6, 72).
GP64 undergoes post-translational modifications including N-glycosylation (12, 28, 47, 72, 73) and palmitoylation (52, 75). Release of BV is concomitant with the localisation of GP64 in distinct loci at the plasma membrane (6) and deletion of gp64 results in a 98% reduction in BV production (46). Surface glycoproteins of other enveloped viruses such as HIV (29, 31, 38, 45), measles (39, 69) and influenza (36, 57, 66), are anchored within structured patches of the plasma membrane. These domains are enriched in cholesterol and sphingolipids and are commonly referred to as lipid rafts or, in the context of virus exit from cells, as budding domains (20, 59). Biochemical analysis of the proteins associated with lipid rafts show that post-translational modifications such as palmitoylation (15, 41, 51, 58) may be important for targeting to rafts and in some cases appears to be essential for raft localisation (5, 37, 54). Zhang et al. (75) has demonstrated palmitoylation of AcMNPV GP64, but showed that prevention of GP64 palmitoylation did not substantially affect synthesis or transport of GP64 to the plasma membrane.
During AcMNPV infection of Sf9 cells, we have shown GP64 co-localisation with areas of the infected cell plasma membrane enriched with the ganglioside GM1. Furthermore, we have demonstrated GP64 incorporation into detergent-resistant, buoyant membrane fractions derived from infected Sf9 cells at times up to 48 hpi. The demonstration that Factin also appeared to co-localise with GP64 at the plasma membrane led us to examine the role of F-actin during virus egress. A number of viruses are known to manipulate host F-actin during virus budding, including vaccinia virus (reviewed in 62) and HIV (22). Furthermore, the haemagglutinin (HA) surface glycoprotein of influenza relies on the host actin cytoskeleton to maintain correct organisation of lipid rafts for incorporation of HA into budding virus particles (61). Although the re-organisation and involvement of Factin during the early stages of AcMNPV infection of insect cells has been studied in detail (13, 17, 34, 43, 44, 53, 70), relatively little is known about the involvement, if any, of the actin cytoskeleton during virus egress and budding. We have demonstrated a reorganisation of F-actin to co-localise with GP64 at the plasma membrane during the late stages of AcMNPV infection. Depletion of F-actin using the drug Latrunculin B (LB) reduced BV production and prevented incorporation of GP64 within discrete domains of the plasma membrane. Examination of virus gene expression and the infectivity of occlusion bodies confirmed that the reduction of BV yield was not a consequence of reduced trafficking of nucleocapsids to the nucleus following infection nor of interference with the production of infectious nucleocapsids.
The glycoproteins of a number of enveloped viruses associate with lipid rafts during infection (29, 31, 36, 38, 45, 57, 66). These detergent-resistant membrane domains have been shown in a variety of insect cells (1, 25, 65, 74, 75). Flotation gradients were used to analyse lipid raft-associated proteins in AcMNPVinfected Sf9 cells. Western blot analysis confirmed the presence of caveolin-1, a protein known to localise within cholesterol-enriched raft domains in Sf9 cells (1), in both non-infected and AcMNPV-infected insect cells (data not shown). Flotation gradients of AcMNPVinfected insect cells harvested at 24 hpi (Fig. 1A) demonstrated localisation of caveolin-1 within fractions 1 and 5 of the flotation gradient, corresponding to the lipid raft fraction and soluble fraction respectively, a distribution typical of raft-associated proteins (9). Typically, proteins that become enriched within lipid rafts are detected in the upper-most fraction (fraction 1) of flotation gradients, due to the buoyant density of cholesterol-enriched rafts and their associated proteins. Conversely, membrane proteins not associated with raft domains are solubilised during detergent extraction and remain in the lower fractions. Analysis of the protein content of each fraction at 24 hpi (Fig. 1B) detected GP64 in all fractions of the gradient including fraction 1. By 48 hpi (Fig. 1C), the majority of GP64 was detected in fraction 1 and at 72 hpi (Fig. 1D) GP64 was only detected in fraction 1. The detection of GP64 in fraction 1 at all times post-infection suggests the incorporation of GP64 into detergent-resistant, cholesterol-enriched domains of the plasma membrane of AcMNPV-infected Sf9 cells. Detection of GP64 in other fractions of the flotation gradient at 24 and 48 hpi is likely to reflect the high levels of nonmembrane associated GP64 within AcMNPV-infected Sf9 cells at these times, as continuous synthesis and trafficking distributes high levels of this protein throughout the ER and Golgi during infection. At 72 hpi GP64 synthesis has all but ceased, and the remaining cellassociated GP64 has likely been transported to the plasma membrane. The presence of GP64 solely in fraction 1 of the flotation gradient at this time (Fig. 1D) suggests GP64 continues to be associated with lipid rafts.
Figure 1. Detection of AcMNPV GP64 in lipid raft-containing membrane fractions of Sf9 cells by flotation gradients. AcMNPV-infected Sf9 cells were harvested at 24 (A, B), 48 (C) or 72 (D) hpi and following lysis, cholesterol-rich membrane regions were separated by flotation analysis (fraction 1) from other cellular components (fractions 2-5). Fractions 1 – 5, obtained from flotation gradients for each cell sample, were analysed for protein content by Western blot detection with antibody specific for the lipid raft-associated protein caveolin-1 (A) or for AcMNPV GP64 (B to D). Protein molecular weight markers are indicated in kDa (L).
To confirm the localisation of GP64 within raft domains of AcMNPV-infected Sf9 cells, dual immunostaining and confocal microscopy were used to visualise the distribution of lipid rafts and GP64 throughout infection (Fig. 2). Cholera toxin subunit B (CTxB) binds to galactosyl moieties such as the ganglioside GM1, a well-characterised marker of raft domains (3, 33, 40), thereby indicating the distribution of raft domains across the membrane. Staining of non-infected Sf9 cells with Alexa Fluor 594-conjugated CTxB (Molecular Probes) revealed GM1-rich domains distributed across the plasma membrane in a random, punctate distribution (Fig. 2A), consistent with findings from mammalian cells (10). Dual immunostaining of AcMNPV-infected Sf9 cells at 24 hpi demonstrated both CTxB (raft domains) and GP64 distributed evenly across the membrane in both surfaceview (data not shown) and in cross-section (Fig. 2B). GP64 stained as a random, punctate distribution reminiscent of the CTxB staining pattern observed in uninfected cells. The CTxB staining in infected cells at 24 hpi appeared more intense than for uninfected cells and tended to obscure the punctate pattern, making true localisation difficult to discern. By 48 hpi (Fig. 2C), the levels of both CTxB and GP64 had fallen in infected cells, and both revealed a concentrated re-distribution at one edge of the cell where co-localisation of the two proteins was readily visible. CTxB was detectable in punctuate distribution at other parts of the plasma membrane but little GP64 was visible except at the region of co-localisation. By 72 hpi (Fig. 2D), staining for both proteins was weak, with the distribution of GM1-rich domains retaining a punctate distribution, and minimal GP64 associated with the cell periphery. These results concur with Fig. 1, suggesting the association of GP64 with cholesterol-enriched, detergent-resistant regions of the plasma membrane enriched in GM1.
Figure 2. Co-localisation of GP64 and GM1-rich domains in uninfected and AcMNPV-infected Sf9 cells by dual immunostaining and confocal microscopy. Uninfected (A) and AcMNPV-infected Sf9 cells at 24 (B), 48 (C) and 72 (D) hpi were labelled with cholera toxin subunit B Alexa Fluor 594 conjugate (CTxB; red; left panel), monoclonal anti-GP64 IgG and rabbit anti-IgG FITC conjugate (GP64; green, middle panel). Merging of the two images (merge, right panel) reveals co-localisation of anti-GP64 antibody and CTxB staining (yellow). (Scale bar = 10μm)
The actin cytoskeleton is known to interact directly with raft-associated proteins (11, 55, 61, 64) and thickens at the insect cell cortex during AcMNPVinfection. The distribution of F-actin and GP64 during late stages of AcMNPV infection were visualised with Alexa Fluor 594-conjugated phalloidin (Molecular Probes) and anti-GP64 antibody, respectively (Fig. 3). F-actin distribution in uninfected Sf9 cells (Fig. 3A) showed actin cables within microspikes projecting from the cell surface in an even distribution across the cell surface when viewed from a ventral plane (Fig. 3A). When viewed in a medial plane, uninfected cells displayed a faint ring of F-actin under the plasma membrane of variable intensity (Fig. 3B), with some degree of irregular cytoplasmic staining. No GP64 staining was visible in uninfected cells. At 24 hpi (Fig. 3C), Factin demonstrated a more universal distribution, with thickened staining under the cell cortex compared to uninfected samples (see Fig. 3B). GP64 at 24 hpi demonstrated a predominantly punctuate distribution across the plasma membrane, as in Figure 2B, and at several locations GP64 appeared to co-localise with F-actin under the lipid bilayer, although the thickened actin layer made co-localisation difficult to confirm. By 48 hpi (Fig. 3D) both F-actin and GP64 illustrated a polarisation toward one edge of the cell and GP64 appeared to completely co-localise with F-actin at this time. This phenomenon was also observed when staining for GP64 alone (data not shown). The results strongly reflect the data for lipid raft and GP64 redistribution at 48 hpi (Fig. 2C), indicating an association between GP64, actin and lipid rafts during infection, and supporting the observation that budding occurs from discrete areas of the plasma membrane enriched in GP64 (46, 72). The exact nature of the association between GP64 and F-actin is unknown but the colocalisation of GP64 and F-actin, and of GP64 and lipid rafts at the cell cortex suggest a possible role for both lipid rafts and F-actin during budded virus egress.
Figure 3. Co-localisation of F-actin and GP64 distribution in uninfected and AcMNPV-infected Sf9 cells by dual immunostaining and confocal microscopy. Uninfected Sf9 cells (A, B) and AcMNPV-infected cells at 24 (C), 48 (D) hpi were labelled with Alexa Fluor 594-conjugated phalloidin (Actin; red; left panel), monoclonal anti-GP64 IgG and rabbit anti-IgG FITC conjugate (GP64; green, middle panel). Merging of the two images (merge, right panel) reveals co-localisation of anti-GP64 antibody and phalloidin staining (yellow). Uninfected Sf9 cells are viewed in both the ventral (A) and medial (B) planes. Infected cells (C, D) are viewed medially. (Scale bar = 10μm)
Latrunculin B (LB) has been used to study the role of actin in the assembly, egress and release of a number of viruses (4, 16, 18, 23). LB rapidly and reversibly binds to G-actin monomer in cells in vitro (63), ultimately resulting in the net depolymerisation of the F-actin network within treated cells. Previous research into the role of F-actin in AcMNPV trafficking and morphogenesis has employed cytochalasin D (CD), an alternative actin-depolymerising drug. Treatment of Sf21 and Trichoplusi ni (T ni) 368 cells with CD does not result in the net depolymerisation of F-actin seen with Latrunculin A (LA) (30), but rather culminates in the formation of cytoplasmic aggregates (30, 67). To ensure complete depolymerisation of Factin within insect cells, non-infected Sf9 cells were treated with LB for 24 hours and stained with Alexa Fluor 594-conjugated phalloidin (Fig. 4A). Following LB treatment, minimal F-actin could be visualised within the cell cytoplasm or at the cell membrane compared to untreated cells (Fig. 4B), and no microspikes could be seen on the surface (data not shown). These results were comparable to insect cells treated with LA for 25 h (30) and illustrated the complete abrogation of cellular F-actin by LB in Sf9 cells.
Figure 4. Depletion of F-actin by Latrunculin B treatment of Sf9 cells. Uninfected Sf9 cells were treated for 24 h with 0.4μg/mL Latrunculin B (LB) (A) and stained with Alexa Fluor 594conjugated phalloidin (Molecular Probes, UK). Absence of staining, compared to untreated Sf9 cells (B), demonstrated LB treatment resulted in a total loss of F-actin from cells. (Scale bar = 10μm).
The effect of F-actin depolymerisation on GP64 distribution during infection is shown in Figure 5. Roncarati and Knebel-M rsdorf (53) demonstrated that F-actin cables, extending from the plasma membrane into the cytoplasm of insect cells as a consequence of baculovirus nucleocapsid entry, were no longer visible by 6 hpi suggesting that the trafficking of nucleocapsids to the nucleus had ceased by this time. We therefore predicted that the addition of LB at 7 hpi would be unlikely to adversely affect virus entry or trafficking of nucleocapsids to the nucleus. Sf9 cells were infected with AcMNPV and treated with LB from 7 hpi. Cells were fixed at 24 hpi and stained with anti-GP64 antibody and Alexa Fluor 594-conjugated phalloidin. Cells treated with LB from 7 hpi demonstrated an extremely low, randomlydistributed staining for F-actin (Fig. 5A). In comparison with untreated cells stained with Alexa Fluor 594 phalloidin (see Fig. 3), the low levels of fluorescence in these cells shows no filamentous network within the cytoplasm or under the cell cortex, indicating an almost complete absence of F-actin at 24 hpi. The fluorescence detected in Figure 5A may indicate some aggregation of actin within the cytoplasm, a phenomenon often described following the addition of the drug CD (67). GP64 appeared to accumulate within the cytoplasm of LB-treated cells (Fig. 5B), whereas in infected cells that had not been treated with LB (Fig. 2C, 3C) the distribution of GP64 was predominantly at the plasma membrane, as expected from Figure 3. Transmitted light microscopy confirmed the viability of the cell was maintained during drug treatment (Fig. 5C). The merged image indicated that co-localisation of F-actin and GP64 staining was absent (Fig. 5D). Treatment of Sf9 cells with LB disrupted the cellular F-actin network and resulted in a re-distribution of GP64, suggesting a role for actin in the targeting of GP64 to the plasma membrane.
Figure 5. Redistribution of GP64 in AcMNPV-infected Sf9 cells following F-actin depletion by treatment with Latrunculin B at 7 hpi. Sf9 cells were infected with AcMNPV at an MOI of 10 PFU/cell and treated with 0.4μg/mL Latrunculin B at 7 hpi. Cells at 24 hpi were fixed and stained with Alexa Fluor 594conjugated phalloidin (red; panel A), monoclonal anti-GP64 IgG and rabbit anti-IgG FITC conjugate (green; panel B). Panel C illustrates the same view using transmitted light. Panel D represents a merged image of panels A, B and C. (Scale bar = 10μm).
Little is known about the trafficking of nucleocapsids during baculovirus egress but a potential role is suggested for F-actin from our results above, which lead us to examine BV production in the presence (+) or absence (-) of LB. Sf9 cells were infected with AcMNPV and treated with LB at different times during the experiment. Yields of extracellular BV were determined by titration of culture media from infected cells at 24 hpi as previously described (32) (Fig. 6).
Figure 6. Budded virus titres at 24 hpi in the presence or absence of Latrunculin B. Sf9 cultures at 1.5×106 cells/mL were incubated with or without 0.4μg/mL LB for 24 h and cell densities adjusted to 2×106 cells/mL prior to infection with AcMNPV at MOI of 20 PFU/cell. Infected culture media were harvested at 24 hpi and BV titres determined by plaque assay. Control cultures remained free of drug during the whole experiment (LB--; A). Three drug treatment regimes were used. Cells were either (ⅰ) treated with LB only from 7 hpi (LB-+; B), (ⅱ) were treated with drug throughout the whole experiment (LB++; C) or (ⅲ) were treated with LB for 24 h prior to virus infection, but then incubated in medium free of drug from 1 hpi (LB+-; D). Titres are indicated in PFU/mL and are the mean of three replicates for each treatment.
Production of BV by infected cells that remained untreated throughout the experiment (LB--; Fig. 6A) reached an average titre of 1.18×107 PFU/mL at 24 hpi. Cells that had been treated with LB only from 7 hpi (LB-+; Fig. 6B) demonstrated significantly reduced BV yields, approximately 100-fold lower than untreated (LB--) control cells, with a mean titre of 1.2×105 PFU/ mL. In cells treated with LB continuously for the 24 h prior to infection and throughout the 24 h infection period, (LB++; Fig. 6C), BV production was also severely compromised, with a low mean titre of 5.7×104 PFU/mL, similar to that of cells treated with LB from 7 hpi (LB-+; Fig. 6B). Finally, cells that were treated with LB for 24 hours prior to infection, but were washed to remove the drug after incubation with virus and subsequently incubated free of LB for the remainder of the experiment (LB+-; Fig. 6D), produced a mean titre of 1.2×107 PFU/mL at 24 hpi, which was comparable to the untreated control samples LB--. ANOVA analysis of raw data showed a significant difference between all samples (Fig. 6 A to D) (P < 0.0001). T-test analysis for BV titres from LB-(untreated) and LB+(treated only prior to infection) samples demonstrated no significant difference between BV yields (Fig. 6A and 6D; P = 0.899), which remained high. Likewise, t-test analysis for BV titres from LB++ (treated both prior to and after infection) and LB-+ (treated only from 7 hpi) cells also demonstrated no significant difference between BV titres (Figure 6B and 6C; P = 0.213) although both gave severely reduced titres compared to the control. Treatment of cells with LB prior to infection followed by subsequent incubation in drug-free conditions (LB+-) demonstrated no reduction of BV egress, suggesting that the elimination of cellular F-actin prior to, and during the infection stage, was not detrimental to virus entry into these cells. Conversely, cells treated with LB throughout the course of the virus infection demonstrated drastic reductions in the titre of BV detected.
The inhibition of BV production by LB-induced Factin depletion occurred when the drug was present at all times during the experiment (LB++), but critically also occurred when the drug was only applied at several hours post-infection (LB-+; Fig. 6C). The results suggest that the absence of F-actin during late, but not early, stages of AcMNPV infection of Sf9 cells affected the production of BV at 24 hpi to the greatest degree. Reduced BV production was affected by the loss of F-actin from 7 hpi irrespective of prior drug treatment or otherwise and was associated most closely with the abrogation of F-actin formation by LB treatment of cells post infection (LB-+ and LB++; Fig. 6B and 6C), rather than with LB treatment prior to infection (LB+-; Fig. 6D). These data implied that reduced BV production at 24 hpi was a consequence of events following virus replication within the nucleus, rather than an inhibition of virus entry at the very stages times in infection. To attempt to verify this, virus gene expression was examined in these cells.
To determine if the reduction in BV production was a consequence of restricted trafficking of nucleocapsids to the nucleus following initial infection stages, reverse transcriptase PCR (RT-PCR) was employed to analyse transcription of selected viral genes. The genes identified for analysis were the immediate early 2 (ie2), DNA polymerase (dnapol) and viral capsid protein 39 (vp39) genes and transcription profiles of all cells treated with LB were compared to untreated LB-cells by agarose gel electrophoresis.
RT-PCR analysis of ie2 gene transcription for untreated cells (LB--; Fig. 7A) demonstrated transcription from 1 hpi to 24 hpi with an apparent peak of transcription at 4 hpi. All LB-treated samples demonstrated similar levels of ie2 transcription from 1 to 24 hpi compared to LB--, although transcription levels of ie2 appeared to be marginally greater at 4 hpi for LB+and marginally reduced at 1 hpi for LB++. Analysis of dnapol and vp39 for untreated cells demonstrated transcription from 4 hpi and 8 hpi, respectively (Fig. 7B and 7C) and revealed comparable timings and levels of transcription for these genes in all four samples. These results indicate that treatment with LB (and therefore depletion of F-actin) had no gross effects on selected virus gene transcription from 1 to 24 hpi. Cells treated either prior to infection (LB+-), after infection (LB-+), or throughout the experiment (LB++) demonstrated almost identical expression profiles for ie2, dnapol and vp39 genes compared to untreated control cells. These results confirmed that the absence of cytoplasmic and nuclear F-actin does not inhibit or restrict trafficking of nucleocapsids to the nucleus during initial infection processes, and that the reduction of BV yield associated with F-actin depletion was not a consequence of reduced, inhibited or delayed virus replication within the cell.
Figure 7. Selected virus gene expression in AcMNPV-infected cells in the presence or absence of Latrunculin B. Total RNA was extracted at various times post infection from Sf9 cells infected with AcMNPV under one of the three LB treatment regimes described (LB-+, LB++, LB+-), or from AcMNPV-infected Sf9 cells that had not been treated with LB (LB--). Reverse transcriptase was carried out on all RNA samples, and cDNA copies of viral transcripts were amplified by PCR using oligonucleotide primers designed to have specificity for the AcMNPV genes immediate early 2 (ie2, panel A), DNA polymerase (dnapol, panel B) and capsid (vp39, panel C). Controls include mock (uninfected Sf9 cells (M), virus DNA positive control (+), water negative control (-). Molecular weight marker is 100bp Ladder (L) from New England Biolabs.
Talhouk and Volkman (67) observed a delay in the synthesis of polyhedrin following addition of CD to AcMNPV-infected insect cells. To analyse the effects of LB on occlusion body (OB) production and infectivity, Sf9 cells were treated with LB as described above, and OBs harvested from cells at 48 hpi. The infectivity of OBs was determined using oral infection of third instar T. ni larvae. Statistical analyses of lethal dose (LD50 and LD95) and survival time revealed no significant differences between the infectivity of OBs produced in Sf9 cells treated with LB at any time prior to or during infection with AcMNPV, compared to untreated AcMNPV-infected cells (Table 1). No significant reduction of OB infectivity in vivo as determined by these parameters was found when Sf9 cells were treated with LB using any of the three regimes described (LB-+, LB++ or LB+-). Additionally, analysis of the DNA content of OBs produced by these cells revealed no detectable differences in the levels of viral DNA within the different OB sample types (data not shown). The absence of any detrimental effects of LB treatment on OB production and infectivity is strong evidence that the production and occlusion of infectious nucleocapsids is not affected in Sf9 cells by the absence of F-actin. These results support the conclusion that reduction of BV production following F-actin depletion was not due to the loss of virus genome replication or nucleocapsid assembly, indicating F-actin depletion did not affect virus replication.
Table 1. Infectivity of occlusion bodies produced in the presence or absence of Latrunculin B.