Nup358 and Transportin 1 Cooperate in Adenoviral Genome Import

To address whether Nup358 plays a specific role in adenoviral genome import, we first depleted U2OS cells using siRNAs against Nup358 or nonspecific siRNAs as a control. At 72 h after siRNA transfection, cells were infected with Ad5-wt-GFP (wild type-green fluorescent protein) vector particles (Ad5) and genomes and viral capsids were quantified at 20 min, 1 h, 2 h, and 4 h postinfection (hpi) (Fig. 1A). Note that during the time frame analyzed in this study (up to 4 hpi), no detectable GFP is expressed from the vectors used. Importantly, Nup358 depletion levels were analyzed at the individual cell level by immunofluorescence (IF) staining. Only cells with a ≥90% reduction of the Nup358 fluorescence signal were considered efficiently depleted (Fig. 1B). First, we analyzed Ad5 genome delivery through quantitative immunofluorescence analysis using specific antibodies against the adenoviral capsid and against protein VII as a marker for viral genomes (Fig. 1C). Incoming intact viral capsids not exposing the viral genome were negative for protein VII (Fig. 1C). Colocalization between Ad5 capsid and protein VII signals was exclusively observed at the nuclear periphery consistent with the disassembly of the viral capsid and exposure of the viral genome taking place at the NPC (Fig. 1C), as previously shown (25). Protein VII signals negative for Ad5 capsid colocalizing with DAPI (4′,6-diamidino-2-phenylindole) stain in z-projections were considered representative of released and imported genomes (Fig. 1C) (20, 24–26). We next determined the ratio of nuclear pVII dots and the total number of Ad5 capsid signals over time to estimate genome import efficiency in control cells and in Nup358-depleted cells (Fig. 1D). We observed a progressive increase in the levels of imported Ad5 genomes in control cells, which reached a plateau at 2 hpi. In Nup358-depleted cells, the number of imported genomes was reduced in a statistically significant manner compared to control cells at 2 hpi. This moderate reduction in import was also seen when shRNAs instead of siRNAs were used for the depletion of Nup358 (see below). At 4 hpi, similar levels of genome import were observed under both conditions. To corroborate our findings, we investigated the dependency of genome import rates on Nup358 levels in control cells in addition to depleted cells. We plotted the number of pVII dots after 2 hpi against the fluorescence intensities of Nup358 (Fig. 1E). Interestingly, control cells exhibited a range of endogenous Nup358 levels which strongly correlated with genome import efficiencies, a correlation that was also observed in Nup358-depleted cells (Fig. 1E). To exclude cell-type-specific effects, we next infected control and Nup358-depleted HeLa cells and analyzed nuclear import of Ad5 genomes (Fig. 1F). Similarly to our observations in U2OS cells, nuclear import of viral genomes was reduced at 2 hpi in cells that had been treated with siRNAs against Nup358 (Fig. 1F). Together, these results suggest that genome delivery is less efficient but not completely inhibited upon Nup358 depletion.

To confirm that depletion of Nup358 does affect genome delivery at the NPC and not upstream trafficking events, we monitored the exposure kinetic of internal capsid protein VI taking place during Ad5 endosomal escape (13). In both Nup358-depleted and control cells, levels of capsids positive for protein VI peaked about 30 min after virus addition (time point 0 min in Fig. 1A), followed by a decrease in the levels of protein VI-associated capsids over time (Fig. 2A). This reflects progressive endosomal escape of viral particles into the cytoplasm and shows that Ad5 entry steps prior to the arrival at the NPC are not affected by Nup358 depletion.

To discriminate between a role for Nup358 in capsid disassembly at the NPC and a role in genome import, we next investigated if Nup358 depletion would affect capsid disassembly kinetics. Capsid disassembly was calculated as the ratio of capsids exposing genomes (pVII-positive capsids) to total detectable capsids (pVII-positive and pVII-negative capsids, i.e., intact capsids). The ratio of capsids exposing genomes in control cells increased from 20 min to 1 hpi, coinciding with their arrival at the NPC, and remained at a constant level from 1 to 4 hpi (Fig. 2B). Strikingly, Ad5 capsids initially disassembled as efficiently in Nup358-depleted cells as in control cells and with similar kinetics (time points 20 min postinfection and 1 hpi). However, the ratio of capsids exposing genomes in Nup358-depleted cells was elevated at 2 hpi, while this ratio dropped back to the level seen with the control cells at 4 hpi. These results suggest that capsid disassembly was not affected by Nup358 depletion and that the differences observed at 2 hpi were due to the accumulation of nonimported genomes resulting from inefficient genome import at unaltered disassembly rates.

We next quantified the kinetics of genome import in the presence and absence of Nup358 on the same data set (Fig. 1A to E). We operationally defined genome import as the separation of the genome from the capsid. The value representing genome import was calculated as the ratio of capsid-associated genomes (pVII-positive capsids) to total detectable genomes (pVII-positive capsids plus free pVII) over time. We observed that at 20 min postinfection, most genomes were still associated with capsids in Nup358-depleted and control-depleted cells (Fig. 2C). The proportion of capsid-associated genomes decreased over time in the control cells, with most viral genomes having separated from Ad5 capsids as early as 2 hpi. In Nup358-depleted cells, larger amounts of genomes remained associated with Ad5 capsids at 1 and 2 hpi in a statistically significant manner. At 4 hpi, most genomes were no longer associated with capsids under both conditions, corroborating our previous findings (Fig. 1D).

Taken together, these results show that depletion of Nup358 delays adenoviral genome delivery by affecting specifically the import step, rather than arrival or disassembly of adenoviral capsids at the NPC.

To confirm a specific role for Nup358 in adenoviral genome import and to exclude off-target effects in the siRNA experiment, as well as to identify the region of Nup358 facilitating adenoviral genome import, we adapted a combined depletion/reconstitution approach previously used to reconstitute nuclear transport of proteins (27). Several siRNA-resistant Nup358 constructs, all associated with the N-terminal NPC-anchoring domain, were expressed in Nup358-depleted U2OS cells (Fig. 3A). To confirm efficient depletion of endogenous Nup358 and discriminate endogenous Nup358 from exogenous Nup358, we used antibodies against the C-terminal cyclophilin-like domain of Nup358, which was absent from the truncated fragments (Fig. 3A). All constructs included an N-terminal hemagglutinin (HA) epitope, allowing specific analysis of Nup358-depleted cells which were negative for the specific Nup358 antibody. For cells expressing full-length HA-Nup358, we analyzed only the cells in the vicinity of efficiently Nup358-depleted cells (Fig. 3B, third row). Following depletion-reconstitution, we quantified the import kinetics of adenoviral genomes upon infection. Transfected cells were infected with Ad5-wt-GFP and analyzed at 2 hpi, when the differences in the import efficiencies of the control and Nup358-depleted cells were most prominent (compare Fig. 1D). The absolute number of pVII dots in the nucleus (DAPI positive) of Nup358-depleted cells was significantly decreased compared to the number seen with the control-depleted cells, confirming that nuclear accumulation of viral genomes was less efficient in the absence of Nup358 (Fig. 3C). Strikingly, depleted cells expressing full-length reconstituted HA-Nup358 (amino acids 1 to 3223) accumulated Ad5 genomes as efficiently as nondepleted control cells. Statistically significant restoration of import was also achieved with shorter N-terminal Nup358 fragments (i.e., amino acids 1 to 2448, 1 to 2148, and 1 to 1810) (Fig. 3B and C). In contrast, fragments containing only the first 1,170 or 1,306 amino acids (aa) of Nup358 did not restore Ad5 genome import (Fig. 3B and C). These results show that Nup358 facilitates adenoviral genome import and identify the N-terminal half of Nup358 as a region that promotes adenoviral genome import. Importantly, the C-terminal part of Nup358, containing 3 of 4 Ran-binding domains (RanBDs), a region involved in SUMO binding, and the kinesin-1 binding domain, was not required to restore import.

It was previously reported that Nup358 affects capsid disassembly at the NPC through indirect binding to minor capsid protein IX via kinesin-1 to prime the genome for nuclear import (23). Our results obtained so far do not support such a model because the presence of the N-terminal part of Nup358 lacking the kinesin-1-binding domain is sufficient to facilitate genome import. To further address this issue, we compared results of analyses of genome import of viruses containing (Ad5-wt-GFP) or lacking (Ad5-ΔpIX-GFP) the minor capsid protein IX (Fig. 4). Ad5-ΔpIX-GFP-infected cells showed several cytoplasmic viral genomes separated from the capsid (i.e., pVII signals) as early as 20 min postinfection (Fig. 4A; bottom row) (28–30). This was not observed in Ad5-wt-GFP-infected cells (Fig. 4A; top row). From 1 hpi to 4 hpi, cytoplasmic genomes disappeared and nuclear genomes appeared progressively for Ad5-ΔpIX-GFP. For Ad5-wt-GFP, no cytoplasmic genomes were detected and nuclear genomes started to appear also from 1 hpi as shown before (Fig. 4B). Two explanations for the observed accumulation of nuclear genomes originating from pIX-lacking capsids are possible. First, the cytoplasmic genomes that we observed after 20 min of infection might have been imported into the nucleus. Second, these cytoplasmic genomes might have been degraded and the nuclear pVII signals might have been derived from a pool of intact viral particles. To distinguish between these possibilities, we infected cells with Ad5-ΔpIX-GFP in the presence of the CRM1 inhibitor LMB, which induces accumulation of capsids at the MTOC, preventing them from docking and disassembling at the NPC (31). In the presence of LMB, at 2 hpi Ad5-ΔpIX-GFP capsids were retained at the MTOC and no nuclear pVII was detected (Fig. 4C). Some cytoplasmic genomes were still observed under these conditions, suggesting that (i) they were not imported and (ii) they did not originate from capsid disassembly at the NPC because the LMB prevented NPC docking. Overexpression of the LMB-resistant CRM1 C528S mutant reversed the LMB effect and restored genome import, showing that genome delivery is strictly dependent on the presence of functional CRM1 as reported previously for Ad5-wt-GFP particles (Fig. 4C and D) (31). These data suggested that some Ad5-ΔpIX-GFP capsids disintegrate prematurely in the cytoplasm, exposing genomes which are rapidly degraded and that do not contribute to the pool of imported viral genomes. Instead, a subpopulation of intact Ad5-ΔpIX-GFP particles succeeds in reaching the nuclear periphery and depends on CRM1 to deliver the viral genome. These capsids are affected by CRM1 inhibition via LMB, suggesting that they use the same NPC-dependent disassembly mechanism as do wt capsids even though they lack the pIX capsid protein. Taking the results together, neither protein IX nor Nup358 plays a dominant role in adenoviral capsid disassembly at the NPC and genome import.

Our results showed that Nup358 depletion delayed but did not block adenoviral genome import. Nup358 could affect nuclear import at different levels. First, it could serve as a binding platform for import factors, facilitating the formation of karyopherin-containing transport complexes (6, 32). Hence, in the absence of Nup358, transport receptors may become rate limiting also for viral genome import. Second, Nup358 could interact directly with certain proteins, promoting their nuclear translocation (27). To discriminate between these possibilities, we set out to express HA-tagged importins in Nup358-depleted cells followed by infection with Ad5-wt-GFP. In this experiment, we used shRNA-mediated depletion of Nup358 using a validated protocol (20). Genome import at 2 hpi in shRNA Nup358-depleted cells was significantly reduced compared to control cell results (Fig. 5A), confirming our results described above using siRNAs for depletion (Fig. 1D). Next, we expressed HA-tagged transport factors such as transportin 1 (TRN1), importin β (Impβ), importin 9 (Imp9), and importin 7 (Imp7) in Nup358-depleted cells. Expression of the importins was examined by microscopy (Fig. 5B) and Western blotting (Fig. 5C). Strikingly, nuclear accumulation of Ad5 genomes at 2 hpi was restored to control levels in cells lacking Nup358 but expressing exogenous transportin 1 (Fig. 5D). To a lesser extent, exogenous importin 9 also promoted genome import in Nup358-depleted cells. Expression of exogenous importin β and importin 7, by contrast, hardly compensated for the lack of Nup358. Importantly, expression of exogenous transportin 1, importin β, or importin 9 in cells with endogenous Nup358 did not result in increased Ad5 genome import, suggesting that Nup358 regulates genome import through access to specific transport receptors (Fig. 5E).

Together, these results show that transport receptors are rate limiting for adenoviral genome import in the absence of Nup358. They suggest a role for Nup358 in maintaining a high local concentration of importins at the NPC to facilitate the formation of adenoviral-genome-associated import complexes, preferentially with transportin 1.

Our results described above showed that transportin 1 promotes adenoviral genome import. We next asked whether transportin 1 is the main import receptor for adenoviral genomes. To address this question, we used two different approaches: depletion of transportin 1 using specific siRNAs and pathway-specific inhibition using a competitive transport cargo. In the depletion approach, efficient knockdown of transportin 1 and/or Nup358 in U2OS cells was reached 72 h after siRNA transfection (Fig. 6A and B). Cells were then infected with Ad5-wt-GFP and analyzed 2 hpi. Only a slight reduction in Ad5 genome import was observed in transportin 1-depleted cells compared to control cells (Fig. 6C). In contrast, a significant decrease in genome import efficiency was clearly observed in both Nup358-depleted cells and cells lacking both Nup358 and transportin 1, compared with control cells (Fig. 6C).

Next, we transfected U2OS cells with a construct coding for a myc-tagged M9M peptide-encoding fusion protein, a high-affinity cargo that inhibits transportin 1-dependent import of other cargos (33). To control specific inhibition of the transportin 1 pathway, cells were transfected with a plasmid coding for GFP-M9, a reporter containing the NLS of hnRNP A1, which is specifically recognized by transportin 1. Alternatively, cells were transfected with a plasmid coding for GFP-cNLS as a reporter, which is imported by the importin α/β complex. All cells were cotransfected with the myc-tagged M9M inhibitor and subjected to IF analysis to assess GFP-tagged NLS constructs localization. Whereas expression of myc-M9M caused partial relocalization of GFP-M9 from the nucleus to the cytoplasm, GFP-cNLS was not affected, confirming the specific inhibition of transportin 1-dependent import (Fig. 6D). However, myc-M9M did not cause any reduction in Ad5 genome import in U2OS cells infected with Ad5 compared to mock-treated cells at 2 hpi (Fig. 6E and F). These results show that neither depletion nor functional inhibition of transportin 1 has a major impact on adenoviral genome import. One possible explanation is that in the presence of Nup358 other import receptors can compensate for the lack of transportin 1 to promote import of the adenoviral genome into the nucleus. Alternatively, the remaining pool of transportin 1 is sufficient to drive genome import.

Binding of transport receptors to FG repeats of nucleoporins is of low affinity and occurs transiently (34). In addition, RanGTP targets importin β to Nup358 via binding to the Ran-binding domains (RanBDs) (35, 36). To study the contribution of the different domains of Nup358 in the recruitment of transportin 1 (i.e., for adenoviral genome import), we first transfected cells with HA-tagged constructs coding for the following fragments of Nup358 lacking the NPC-anchoring region: Nup358(806–1306), comprising a coiled-coil region, a cluster of FG repeats and the first RanBD; Nup358(1350–2148), harboring several zinc fingers, another cluster of FG repeats, and the second RanBD; and Nup358(2307–3047), a C-terminal fragment, containing two FG-repeat clusters and RanBD-3 and RanBD-4 (Fig. 7A and B). The staining of endogenous transportin 1 was predominantly nuclear in control transfected cells (Fig. 7B). In contrast, in cells expressing HA-Nup358(806–1306), endogenous transportin 1 was partially redistributed to the cytoplasm and colocalized with the soluble nucleoporin fragment. In contrast, redistribution of transportin 1 was not observed in cells expressing Nup358(1350–2148) or Nup358(2307–3047). These results suggest that only the soluble fragment comprising the first patch of FG repeats and RanBD-1 recruits endogenous transportin 1. They are in agreement with our observation that this region can rescue Ad5 genome import when linked to the N-terminal NPC-anchoring fragment of Nup358. In contrast, overexpression of the soluble fragment Nup358(806–1306) did not affect Ad5 genome import (data not shown), which was expected, as it was previously shown to compete with import of neither cNLS-containing nor M9-containing reporter constructs (27). We then asked whether this recruitment was specific for transportin 1 or would also be observed for another prominent import receptor, importin β. We also tried to further narrow the search for identification of the region in the Nup358(806–1306) fragment driving the recruitment. Cells were transfected with HA-tagged fragments of Nup358 [Nup358(806–1306), Nup358(806–1170), Nup358(806–1133), and Nup358(806–1000)] and stained for endogenous transportin 1 or importin β (Fig. 7C). Localization of endogenous transportin 1 and importin β was largely nuclear in control transfected cells (Fig. 7C, top row). Expression of the largest fragment (aa 806–1306) caused a relocalization of both transportin 1 and importin β to cytoplasmic structures colocalizing with the HA-tagged fragment. The fragment lacking RanBD-1 (aa 806–1170) was still able to relocalize both endogenous importins. HA-Nup358(806–1133) caused only a partial redistribution of importin β and transportin 1. Finally, the fragment containing exclusively the coiled-coiled domain of Nup358(806–1000) itself strongly located in the nucleus but did not relocate any of the transport receptors tested. These results show that the N-terminal region (aa 806–1306) of Nup358 encoding a cluster of FG repeats and RanBD-1 is able to recruit the prominent importins importin β and transportin 1. This observation is in line with a role of the N terminus of Nup358 in adenoviral genome import by concentrating transport receptors at the NPC.

HEK-293 αvβ5 cells (human embryonic kidney 293 cells [ATCC CRL-1573], kindly provided by G. Nemerow, Scripps Research Institute, La Jolla, USA), HeLa cells (ATCC CCL-2, kindly provided by L. Gerace, Scripps Research Institute, La Jolla, USA), and U2OS cells (human bone osteosarcoma epithelial cells [ATCC HTB-96], kindly provided by M. Piechaczyk, IGMM, Montpellier, France) were maintained in a humidified incubator with 5% (vol/vol) CO₂ at 37°C, in Dulbecco’s modified Eagle medium (DMEM) GlutaMAX (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen), 100 U/ml of penicillin, and 100 μg/ml of streptomycin (Invitrogen). When indicated, cells were treated with 20 nM LMB (Sigma) for 30 min and then infected in the presence of 20 nM LMB.

U2OS cells were seeded to reach ∼60% to 80% confluence on the day of transfection. Cells were transfected using Lipofectamine 2000 and Opti-MEM (Invitrogen). To increase transfection efficiency of large constructs (e.g., constructs encoding long HA-tagged NPC-anchored fragments of Nup358), cells were treated with 5% dimethyl sulfoxide (DMSO) DMEM for 2 min and then gently washed with phosphate-buffered saline (PBS) prior to transfection.

Constructs coding for GFP₂-M9 core and GFP₂-cNLS (5), myc-MBP-M9M (27), HA-importin β and HA-transportin 1 (6), HA-importin 7 (37), HA-importin 9 (32), and HA-CRM1 C528S (45) and constructs coding for HA-tagged full-length Nup358 and its fragments (27) were described previously. A pcDNA3.1+ empty vector was purchased from Invitrogen and used as a DNA control.

U2OS were seeded in a 12-well plate (10⁵ cells/well). The next day, cells were transfected with 100 nM siRNA per condition using Lipofectamine RNAiMAX and Opti-MEM (Invitrogen) following the manufacturer’s protocol. In the case of Nup358 depletion/restitution experiments, a DNA-siRNA cotransfection of the Nup358 constructs with specific or nontargeting siRNAs was performed 48 h after the first siRNA depletion using the protocol for transient transfection. Seventy-two hours after the first siRNA depletion, cells were infected or directly fixed for further analysis. Validated siRNA oligonucleotides directed against the mRNA of Nup358 5′-CACAGACAAAGCCGUUGAA as described previously (27) or nonspecific 5′-AGGUAGUGUAAUCGCCUUG were obtained from Eurofins Genomics. siRNAs for the depletion of transportin were obtained from Santa Cruz (sc-35737) (32). Efficient depletion of both Nup358 and transportin 1 was monitored by immunofluorescence and reached a maximum 72 h after siRNA transfection. Gene silencing of Nup358 by the use of shRNA was done as previously described (20). Cells were transfected with pSUPER vector (Oligoengine) expressing shRNA directed against Nup358 (5′-GATCCCCCGAGGTCAATGGCAAACTATTCAAGAGATAGTTTGCCATTGACCTCGTTTTTA-3′) or with a peGFP-H1 vector containing two expression cassettes, one for enhanced GFP (eGFP) and one for nontargeting shRNA, as a control. An efficient knockdown of Nup358 was obtained 3 days after DNA transfection.

Amplification of replication-deficient E1-deleted GFP-expressing vectors based on genotypes HAdV-C5, Ad5-wt-GFP (46), and Ad5-ΔpIX-GFP (kindly provided by Christopher M. Wiethoff, Stritch School of Medicine, IL, USA) was done in HEK293-αvβ5 cells, and the resulting mixture was purified using double CsCl₂ banding (12, 46). Viruses were labeled by using an Alexa Fluor microscale labeling kit (Life Technologies) as detailed previously (47). Viral physical particle quantification was performed on the basis of the estimated copy numbers of viral genomes. Copy numbers were calculated according to the optical density at 260 (OD₂₆₀) method (OD₂₆₀ of 1 = 1.16 × 10¹² particles/ml) (48). The number of infectious particles was quantified by plaque assay. The viral titer of the stock sample was determined by taking the average number of GFP-fluorescent plaques for a dilution and the inverse of the total dilution factor.

Cells grown on 15-mm-diameter coverslips in a 12-well plate were seeded in order to have 1.5 × 10⁵ to 2 × 10⁵ cells the day of infection. For analysis of kinetic of adenoviral genome delivery, a time course infection was performed using a multiplicity of infection (MOI) of 30 and the Ad5-wt-GFP. Comparing different viruses (Ad5-wt-GFP versus Ad5-ΔpIX-GFP), a total of 5,000 pancreatic polypeptides (pp)/cell were used for infections. Time course infection was performed with 5,000 pp/cell for analysis of endosomal escape of Ad5 particles. In all cases, the number of internalized particles per cell at the time of analysis was in the range of 50 to 150. Viruses were diluted in prewarmed DMEM (100 μl/condition) (37°C) and added to cells for 30 min to allow Ad5 particle binding and internalization. After 30 min, the inoculum was removed; this was considered time point zero. Coverslips were incubated in 1 ml/well fresh prewarmed (37°C) DMEM and fixed at different time points using 4% paraformaldehyde (pH = 7.4). Cells were blocked/permeabilized using IF buffer (10% fetal calf serum [FCS]–0.5% saponin–PBS). Primary and secondary antibodies were diluted in IF buffer and applied to cells for 1 h in a humid chamber at 37°C. Cells were mounted using fluorescence mounting medium (Dako) containing 1 μg/ml DAPI (Sigma) to visualize the cellular DNA and were analyzed by microscopy.

The following antibodies used for IF in this study: serum of rabbit anti-HAd5-C5, dilution 1:1000 (kindly provided by R. Iggo, Institut Bergonie, Bordeaux, France); mouse anti-protein VI, dilution 1:400 (46); mouse anti-protein VII, dilution 1:100 (49); rat anti-protein VII, dilution 1:500 (50); mouse anti-human transportin 1, dilution 1:500 (BD Bioscience); rabbit anti-importin β, dilution 1:500; rabbit anti-human Nup358 (aa 3062 to 3223), goat anti-human Nup358 (aa 2553 to 2838) (8), both diluted 1:1,000; goat anti-human CRM1 (51), dilution 1:250; mouse anti-Myc tag, dilution 1:400 (Molecular Probes); rat anti-HA tag, dilution 1:500 (Roche); rabbit anti-RanBP1, dilution 1:250 (kindly provided by L. Gerace, Scripps Research Institute, La Jolla, USA). Secondary antibodies from donkey coupled with Alexa fluor were used at a dilution of 1:500 (Life Technologies). Antibodies used for Western blotting in this study were as follows: rabbit anti-chicken α-tubulin (Sigma), dilution 1:1,000; mouse anti-human transportin 1, dilution 1:1,000 (BD Bioscience); peroxidase-coupled secondary antibodies (Jackson Immuno Research) diluted 1:10,000.

Immunofluorescence preparations were imaged using confocal SP5 or SP8-CRYO microscopes (Leica) equipped with Leica software on the Bordeaux Imaging Center (BIC) platform. Confocal stacks (∼10 stacks) were taken every 0.3 μm with a pinhole setting of 1, using 3-fold oversampling for all channels. Images were acquired with settings of a resolution of 16 bits and a pixel size of 75 nm. Images were processed using Image J software (National Institutes of Health). Stacks from the channel of interest (e.g., fluorescence of Nup358) of confocal images were combined as Z-projections. Individual cells were selected from maximal-projection images by drawing a selection around the cell periphery. Individual cells were used for single signal quantifications. A threshold was applied to identify the area considered representative of positive signal in a given channel of interest (e.g., that corresponding to the fluorescence of DAPI to delineate the area of the nucleus and to the fluorescence of pVII to identify and quantify genomes within the DAPI area). Objects from binary images were quantified when their sizes exceeded a defined area of pixels (10 pixels for pVII and 5 pixels for Ad5 capsids). For quantification of colocalization between signals, superposition of two binary images from two channels of interest was used. Objects exceeding a defined area of pixels in common were quantified. The result of these quantifications gives the number of objects for each region of interest (for each cell). A semiautomated macro was developed to perform quantifications of all images. In the case of quantification of fluorescence intensity, a threshold was applied to highlight the area considered to represent positive signal for the control condition. The integrated density in the thresholded area for each one of the regions of interest (for each cell) was quantified. A macro was developed to automate quantifications of all images with settings resembling those for control cells.

Statistical analyses were performed using GraphPad Prism 6 software. Results are presented either in a scatterplot or in columns, as indicated in the figure legends. Bars indicate the standard deviations (SD). Horizontal lines or the lengths of the columns represent means. To compare the means of results from two populations, statistical analysis was done using unpaired two-tailed Student's t test. To compare the means of results from three or more sets of conditions, one-way analysis of variance (ANOVA) (one factor) or two-way ANOVA (two factors) was applied, as indicated in the figure legends. Multicomparison post hoc corrections were performed to figure out which groups in the sample differed. In the case of the one-way ANOVA post hoc tests, Tukey’s honestly significant difference test was used when the sample sizes of groups were equal; otherwise, Dunnett’s test was used. In the case of two-way ANOVA, Sidak’s correction was used for multicomparison tests (NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

Cells were harvested using 0.6 mM EDTA–PBS, pelleted by centrifugation, and resuspended in immunoprecipitation (IP) lysis buffer (Thermo Scientific) supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF) and cocktail protease inhibitors (cOmplete, mini, EDTA-free protease inhibitor cocktail; Roche). Lysates were clarified by centrifugation at 21,000 × g for 10 min at 4°C. Protein concentrations in the cell extracts were quantified by Bio-Rad protein assay (52). Clarified cell extracts were diluted in Laemmli 4× buffer (8% SDS, 40% glycerol, 0.01% bromophenol blue, 5% 2-mercaptoethanol, 200 mM Tris buffer, pH 6.8) and boiled at 95°C for 15 min. Samples were analyzed by SDS-PAGE followed by immunoblotting. Proteins were visualized by chemiluminescence using an ECL Femto kit (Millipore) and an ImageQuant LAS 4010 imaging system (GE Healthcare Life Sciences).