Microtubule-Mediated and Microtubule-Independent Transport of Adenovirus Type 5 in HEK293 Cells

ABSTRACT Adenovirus serotypes 2 and 5 are taken into cells by receptor-mediated endocytosis, and following release from endosomes, destabilized virions travel along microtubules to accumulate around the nucleus. The entry process culminates in delivery of the viral genome through nuclear pores. This model is based on studies with conventional cell lines, such as HeLa and HEp-2, but in HEK293 cells, which are routinely used in this laboratory because they are permissive for replication of multiple adenovirus serotypes, a different trafficking pattern has been observed. Nuclei of 293 cells have an irregular shape, with an indented region, and virions directly labeled with carboxyfluorescein accumulate in a cluster within that indented region. The clusters, which form in close proximity to the microtubule organizing center (MTOC) and to the Golgi apparatus, are remarkably stable; a fluorescent signal can be seen in the MTOC region up to 16 h postinfection. Furthermore, if cells are infected and then undergo mitosis after the cluster is formed, the signal is found at each spindle pole. Despite the sequestration of virions near the MTOC, 293 cells are no less sensitive than other cells to productive infection with adenovirus. Even though cluster formation depends on intact microtubules, infectivity is not compromised by disruption of microtubules with either nocodazole or colchicine, as determined by expression of an enhanced green fluorescent protein reporter gene inserted in the viral genome. These results indicate that virion clusters do not represent the infectious pathway and suggest an alternative route to the nucleus that does not depend on nocodazole-sensitive microtubules.

Fifty-one serotypes of human adenoviruses have been identified to date and classified into six species (A to F) based on the degree of DNA relatedness (42). In the natural host, adenoviruses infect predominantly epithelial cells of the respiratory tract (species B, C, and E), urinary tract (species B), small intestine (species A and F) and the conjunctiva (species B, D, and E) (20,43). Species C serotypes (adenovirus serotype 2 [Ad2] and Ad5) can be propagated readily in conventional cell lines (HeLa, HEp-2, and A549) and have been well characterized in many respects, including virus entry. According to the current model (29), the process of adenovirus entry begins with attachment of the fiber knob (on the virion) to the host cell receptor. Interaction of the viral penton base with integrins on the cell surface then triggers endocytosis, and following release from endosomes, destabilized virions travel along microtubules to accumulate around the nucleus. The entry process culminates in delivery of the viral genome through nuclear pores. However, not all adenoviruses utilize the same pathway. Most species B serotypes recognize CD46 (14,34,39) rather than the Coxsackie and adenovirus receptor (CAR). Two species B serotypes (Ad3 and Ad7) bind to other receptors (33), identified as CD80 and CD86, in the case of Ad3 (38). Unlike Ad2 and Ad5, which rapidly escape endosomes soon after uptake, virions of species B serotypes are transported within vesicles towards the nucleus, being released from late endosomes or lysosomes (32,36). Species D serotypes also appear to use a receptor other than CAR. In different studies, Ad37 (species D) has been shown to utilize CD46 (45), like species B serotypes, and sialic acid (1,2,3,9).
Direct labeling of virions with fluorescent dyes, like carboxyfluorescein, has made it possible to follow virions by microscopy during transport to the host cell nucleus. In experiments initiated to study the transport of different adenovirus serotypes, we observed an atypical trafficking pattern in 293 cells, even with Ad5. Further investigation focused on Ad5 and addressed microtubule-mediated transport in relation to the pattern observed. The juxtanuclear virion clusters in 293 cells revealed a pathway which is microtubule dependent but is not required for successful genome delivery. virus 40 poly(A) signal) inserted in reverse orientation, in place of the E1 region. Ad5.CMV-LacZ was purchased from Quantum Biotechnologies, Inc., Montreal, Canada (now Qbiogene, Inc., Carlsbad, CA). Ad5EGFP was generated in this laboratory by homologous recombination in 293 cells. All virions were propagated in 293 cells at an adsorbed multiplicity of infection (MOI) of about 0.2 to 2 and were used in experiments at passage level 3. Mature virions ( ϭ 1.34 g/cm 3 ) were purified by two cycles of cesium chloride density gradient centrifugation. The virus band from a step gradient of CsCl (1.2 g/ml in 50 mM Tris-HCl, pH 8.1, underlaid with 1.4 g/ml in 50 mM Tris-HCl, pH 8.1), spun at 120,000 ϫ g for 1 h, was recovered and further purified on a self-forming gradient of CsCl (initial density, 1.4 g/ml) spun at 120,000 ϫ g for 20 h. Purified virus was desalted by dialysis against three changes of buffer containing 50 mM Tris-HCl, pH 7.8, 150 mM NaCl, 10 mM MgCl 2 , and 10% glycerol. The virion concentration was determined by disrupting the virions in a 50-l sample (30) and then measuring the absorbance of the suspension at 260 nm, using the following equation: 1 A 260 ϭ 1.1 ϫ 10 12 particles/ml (27). Infectivity was determined by endpoint dilution in 60-well Terasaki plates (Nunc) as described previously (7). Intracellular trafficking of labeled virions. Virions were labeled with 5-(and 6-) carboxyfluorescein, succinimidyl ester (Molecular Probes) as described by Miyazawa et al. (31). Infectivity was not reduced by the labeling process (results not shown). For the intracellular trafficking experiments, cells were seeded in LabTek chamber slides (Nunc) at densities of 4 ϫ 10 4 to 8 ϫ 10 4 cells per chamber or on 12-mm coverslips in a 24-well plate at a density of 10 5 cells per well. Cells were infected 2 days after seeding using labeled virus (ϳ10 11 particles/ml.; ϳ10 5 particles/cell) in an adsorption volume of 0.15 ml per chamber of the chamber slide or 0.25 ml per well of a 24-well plate. Under these conditions, typically less than 10% of the inoculum is adsorbed (data not shown). Prior to infection, cells were rinsed with cold binding buffer (minimal essential medium with 10% FCS, an additional 1% bovine serum albumin [BSA], and 10 mM HEPES, pH 7.3), and then virus, diluted in cold binding buffer, was applied. The cells were held at 4°C for 60 min to allow virus adsorption but not entry and thereby synchronize the infection. Unadsorbed virus was removed by washing the cells three times with cold binding buffer, and slides were fixed immediately with 4% paraformaldehyde or incubated with warm binding buffer at 37°C to allow the entry process to continue. Slides were usually fixed 60 min after warming, but incubation times as short as 20 min and as long as 16 h were used in some experiments. Slides were mounted with 50% glycerol in phosphate-buffered saline (PBS) containing 1,4-diazabicyclo[2,2,2]-octane (20 mg/ml) and 4,6-diamidino-2-phenylindole (DAPI) (0.25 g/ml) and examined with a Leica DMA fluorescence microscope, using the 63ϫ oil objective and the 1.0ϫ, 1.25ϫ, or 1.6ϫ magnifier. Images were captured with a Hamamatsu charge-coupled-device camera (C4742-95) and processed using Openlab imaging software, version 2.0.7 (Improvision, Inc.) and Adobe Photoshop, version 6.0.
Immunofluorescence microscopy. Primary monoclonal antibody against the Golgi protein with a molecular weight of 58,000 (58K) and secondary tetramethyl rhodamine isothiocyanate-conjugated antimurine antibody, as well as Cy3-conjugated monoclonal antibody against ␤-tubulin, were purchased from Sigma-Aldrich Canada, Ltd. Polyclonal antibody against ␥-tubulin and Texas Redconjugated antirabbit antibody were gifts from Andrew Wilde, University of Toronto. Monoclonal antibody (clone FK2) against multiubiquitin chains (13) and polyclonal antibody against the 20s proteasomal subunit were gifts from John Brumell, The Hospital for Sick Children.
Cells fixed with 4% paraformaldehyde were permeabilized with PBT (0.1% Triton X in PBS) for 15 min at room temperature and then blocked with 5% BSA-PBT at room temperature for 45 min before incubation with primary antibody (diluted in 5% BSA-PBT) for 45 to 60 min at 37°C. After three 5-min washes with PBS at room temperature, cells were incubated with secondary antibody (also diluted in 5% BSA-PBT) for 45 min at 37°C. Cells were then washed with PBT (two 5-min washes), followed by PBS (two 5-min washes). Coverslips were mounted in PBS containing DAPI (0.25 g/ml) with or without 1,4-diazabicyclo[2,2,2]-octane.
Electron microscopy. 293 cells were seeded onto 18-mm coverslips in six-well tissue culture dishes (3 ϫ 10 5 cells/well). The next day, cells were rinsed three times with ice-cold binding buffer and infected with 0.6 ml per well of the same labeled virus suspension used in the tracking experiments (ϳ10 11 particles/ml; ϳ10 5 particles/cell) at 4°C for 60 min to synchronize infection. Unbound viruses were then removed by rinsing the cells with ice-cold binding buffer. Subsequently, cells were incubated in warm binding buffer at 37°C for 90 min. After infection, cells were rinsed twice with PBS, fixed with 2% glutaraldehyde for 30 min at room temperature, washed with PBS and then 0.1 M phosphate buffer, pH 7.4, and stored in the same buffer until further processed.
To continue processing, cells were fixed with 1% osmium tetroxide in 0. Microtubule disruption experiments. Nocodazole (Sigma-Aldrich) was prepared as a 10 mM stock solution in dimethyl sulfoxide, stored in aliquots at Ϫ20°C, and diluted to a concentration of 30 M in culture medium or in binding buffer immediately prior to use. Colchicine (Sigma-Aldrich) was prepared as a 20 mM stock solution in distilled deionized water, stored in aliquots at Ϫ20°C, and diluted to a concentration of 100 M in culture medium immediately prior to use. 293 and A549 cells were seeded on 12-mm coverslips in 24-well plates (1.5 ϫ 10 5 cells/well) for microscopy or in 6-well plates (6 ϫ 10 5 cells/well) for analysis by flow cytometry and used the following day.
For analysis of virion trafficking by microscopy, cells were treated for 60 min at 37°C with 30 M nocodazole in binding buffer prior to infection with carboxyfluorescein-labeled Ad5 diluted in binding buffer containing 30 M nocodazole. Following adsorption at 4°C for 60 min, the inoculum was removed, cells were washed with binding buffer containing 30 M nocodazole, and fresh binding buffer containing 30 M nocodazole was added. Cells were incubated at 37°C for 90 min and then fixed with 4% paraformaldehyde and stained for ␤-tubulin to confirm the absence of microtubules.
For experiments to monitor the effect of microtubule disruption on genome delivery, nocodazole pretreatment was extended to 2 h to minimize the number of surviving microtubules, and adsorption took place at 37°C to avoid any alterations to the cells induced by exposure to low temperature. The input MOI was based on results of preliminary dose-response experiments and was chosen to give a detectable number of EGFP-positive cells, but less than 50% of the population, at 12 h postinfection (p.i.). Accordingly, cells were pretreated with 30 M nocodazole, diluted in complete medium containing 10% FCS, for 2 h at 37°C prior to infection with 0.6 ml virus inoculum (ϳ2 ϫ 10 7 particles/ml of Ad5EGFP diluted in binding buffer with 30 M nocodazole; ϳ20 particles/cell). Following adsorption at 37°C for 60 min, the inoculum was removed and cells were rinsed twice with complete medium containing 30 M nocodazole. Fresh medium containing 30 M nocodazole was added, and cells were returned to 37°C. Mock-treated cells, infected but not exposed to nocodazole, served as controls. At 12 h p.i., Ad5EGFP-infected cells on coverslips were fixed with 4% paraformaldehyde and stained for ␤-tubulin to confirm the absence of microtubules. For flow cytometry, cells were trypsinized at 12 h p.i., collected by centrifugation at 300 ϫ g for 5 min, resuspended in 4% paraformaldehyde, pH 7.2, held for 15 min at room temperature, and collected by centrifugation at 300 ϫ g for 5 min. Cells were then resuspended in 500 l PBS and passed through a 70-m nylon cell strainer (BD Falcon) to remove clumps. Samples were analyzed on an EPICS Elite cell sorter (Beckman-Coulter).
In flow cytometry experiments using colchicine to disrupt microtubules, 293 cells were pretreated with 100 M colchicine for 2 h on ice and then for 2 h at 37°C prior to infection. The colchicine concentration was maintained at 100 M during the virus adsorption period at 4°C and the subsequent incubation period at 37°C. Conditions for infection and subsequent processing were the same as those described for the flow cytometry experiments using nocodazole. In the colchicine experiments, cells on coverslips were treated with Triton X (0.1% in PEM buffer [100 mM piperazine-N,NЈ-bis(2-ethanesulfonic acid), 1 mM EGTA, 1 mM MgCl 2, pH 6.9]) for 5 min and then rinsed three times with PEM buffer to extract soluble tubulin prior to staining residual microtubules with anti-␤-tubulin (28).
Ad5.CMV-LacZ was also used to monitor genome delivery in the presence or absence of nocodazole by microscopic examination of infected 293 cell cultures stained with 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-Gal). The pretreatment and infection conditions were the same as those used with Ad5EGFP. At 12 h p.i., cells were fixed with 2% formaldehyde for 5 min at room temperature and then washed with PBS and incubated with X-Gal (diluted from a stock solution of 20 mg/ml in dimethylformamide to 1 mg/ml in PBS containing 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 2 mM magnesium chloride) for 1 h at 37°C. The X-Gal solution was replaced with PBS, and cultures were examined, using a Leica DMIL microscope, for blue cells, indicative of ␤-galactosidase activity and successful genome delivery.

RESULTS
Trafficking patterns. The trafficking pattern of labeled Ad5 virions in HeLa, HEp-2, and A549 cells was consistent with published images of HeLa cells (19) and A549 cells (23,31), showing virus bound to the cell surface after adsorption at 4°C (Fig. 1b to d) and accumulation of virions around the nucleus after warming to 37°C for 60 to 90 min ( Fig. 1f to h). In the case of 293 cells, virions were bound to the cell surface at 4°C (Fig. 1a), as seen with the other cell lines, but the trafficking pattern of labeled Ad5 virions at 37°C was strikingly different. The nuclei of 293 cells were found to have an irregular shape, most with an "indented" region ( Fig. 1a and e and 2A to C), and 60 min after warming the cells to 37°C, the majority of virions had accumulated in a cluster within the "indented" region ( Fig. 1e). A minor population of virions could be seen in a perinuclear distribution around the nucleus in some cells following enhancement of the signal ( Fig. 2A).
Immunospecific staining of 293 cells, following internalization of labeled virus, showed that the clusters are found near the Golgi apparatus (Fig. 2B) and the microtubule organizing center (MTOC) (Fig. 2C). In cells which were infected and then underwent mitosis, the signal remained with the centrioles (Fig. 2D). Time course experiments showed that clusters were beginning to form by 20 min after warming to 37°C, and by 40 min, they were comparable in appearance to clusters seen after 60 min (not shown). By 6 h after warming, clusters were still apparent, and signal could be detected up to 16 h (not shown).
To determine whether the juxtanuclear clusters represented intact virions rather than degradation products, thin sections of infected cells were examined by electron microscopy. Virions which appeared to be intact were readily found in the "indented region" of the nucleus. Figure 3 (upper and lower panels) shows two representative fields, each at low, medium, and high magnification. In one field, multiple virions are con-tained within a single vesicle (Fig. 3a to c), whereas in the other field, virions appear to be in the cytoplasm in close proximity to small empty vesicles (Fig. 3d to f) whose identity is unknown but which might represent Golgi vesicles sectioned on a tangent.
Effect of microtubule disruption on virion trafficking. When cells were treated with 30 M nocodazole (to disrupt microtubules) prior to infection and were maintained in the presence of 30 M nocodazole during virus adsorption and incubation, virions did not accumulate in juxtanuclear clusters but were predominantly scattered throughout the cytoplasm (Fig. 4). Homogeneous staining for ␤-tubulin confirmed that microtubules were not intact. Under these conditions, a perinuclear distribution of virions around the nuclei of 293 cells was more evident than in cultures whose microtubules were intact (Fig.  4A). In A549 cells, often used for trafficking studies because they are flat and provide high-quality images in fluorescence microscopy, perinuclear accumulation of labeled virions was particularly striking in the absence of microtubules (Fig. 4B).
Although formation of the juxtanuclear clusters was dependent on intact microtubules, it appeared that a population of virions was transported to the perinuclear region even in the absence of detectable microtubules. Ad5EGFP was then used to determine whether infectivity was compromised under these conditions. Both 293 and A549 cells were treated with 30 M nocodazole prior to infection with Ad5EGFP at a MOI sufficient to infect less than 50% of the cells. The nocodazole concentration was maintained at 30 M during virus adsorption and after the cells were warmed to 37°C. It was reasoned that if microtubules were important for transport of virions along the infectious pathway to the nucleus, then the number of cells expressing EGFP should be lower in cultures treated with nocodazole. When infected cultures were examined microscopically 12 h p.i., the numbers of EGFP-expressing cells were comparable in treated and nontreated cultures (Fig. 5), FIG. 1. Labeled virions before and after internalization. Cells were incubated with labeled Ad5 (ϳ10 11 particles/ml; ϳ10 5 particles/cell) for 60 min at 4°C. Unadsorbed virus was removed, and cells were fixed immediately (a to d) or fresh binding buffer was added and cells were transferred to 37°C for 60 min before fixation (e to h). Each field was photographed with a filter for fluorescein (green signal) and a filter for DAPI (blue signal), and the images were overlaid using Openlab imaging software, version 2.0.7.

VOL. 81, 2007
Ad5 ENTRY IN HEK293 CELLS 6901 even though microtubules were disrupted by nocodazole treatment. Moreover, a brighter signal was observed in the treated cultures (i.e., an absence of microtubules) by microscopy and was confirmed by flow cytometry (Fig. 5). Nocodazole treatment had no effect on the number of fluorescent cells, but the curves representing both 293 and A549 cells treated with nocodazole were shifted to the right, indicating an increase in fluorescence intensity. The mean fluorescence intensities of nocodazole-treated and untreated cells infected with Ad5EGFP are compared in Table 1. In both 293 and A549 cells, the mean fluorescence intensity of infected, treated cells was about double that of infected, untreated cells.
When experiments were repeated with 293 cells, using colchicine to disrupt microtubules, the proportions of EGFPpositive cells were again comparable in treated and nontreated cultures (Table 1). However, the increased intensity of fluorescence seen with nocodazole treatment was not seen with colchicine treatment ( Table 1). Extraction of soluble tubulin from colchicine-treated cells prior to staining of the cells with anti-␤-tubulin showed that there were still some surviving microtubules immediately prior to infection, but most of these appeared as disconnected fragments (Fig. 6).
Genome delivery in 293 cells was also monitored microscopically after X-Gal staining of nocodazole-treated and untreated cultures infected with Ad5.CMV-LacZ. As seen with Ad5EGFP, using the same pretreatment conditions and the same low MOI, the numbers of cells expressing signal were comparable in the presence or absence of nocodazole ( Table 2), indicating that genome delivery was not compromised by disruption of microtubules.

DISCUSSION
This study draws attention to the irregular shape of 293 cell nuclei and to an intracellular trafficking pattern for Ad5 which is strikingly different from the perinuclear accumulation characteristic of HeLa, HEp-2, and A549 cells. Accumulation of virions in clusters near the MTOC proved to be dependent on microtubule-mediated transport, but microtubules were not essential for successful genome delivery to the nucleus.
Localization of virions near the MTOC, within the "indented" region formed by the nucleus, in 293 cells is consistent with the minus end-directed transport of incoming virions along microtubules, as shown by Suomalainen et al. (41) and Leopold et al. (24). However, it is not clear why the majority of incoming virions remain in stable clusters near the MTOC in 293 cells (Fig. 1e) and some A549 cells (14; data not shown) but not other cells. Initially, it was thought that clustering of the virions might reflect a cytoskeletal organization related to the irregularly shaped nuclei or to the neuronal origin of 293 cells (35). However, Shayakhmetov et al. (37) showed juxtanuclear clusters of chimeric Ad5 (with Ad35 fiber knobs) in hematopoietic cells with regular oval-shaped nuclei. This observation is meaningful in that the fiber knobs from Ad35 (species B) confer a binding specificity for CD46, rather than CAR, as the primary receptor. Clustering at the MTOC therefore is not unique to cells with irregularly shaped nuclei or to infections mediated by virion binding to CAR.
It is not known whether the fluorescent clusters represent virions contained within vesicles or free in the cytoplasm, since both types were seen by electron microscopy (Fig. 3). Given the current model for adenovirus entry (29), the Ad5 clusters likely represent virions released from vesicles and then carried along microtubules to the MTOC. No colocalization was evident by immunostaining with antibody against EEA-1, transferrin receptor, or LAMP-1 as markers for early endosomes, recycling endosomes, and late endosomes/lysosomes, respectively (data not shown). Formation of virion clusters near the MTOC may reflect the activation of signal transduction pathways by high levels of input virus binding to the surfaces of 293 cells. Increasing dilution of the labeled Ad5 inoculum gave a fluorescent signal that was less intense and still localized to the "nuclear indentation" of 293 cells, but even at the limit of detection, the adsorbed MOI was several hundred particles per cell (not shown). The virion clusters are similar in appearance to aggresomes, which represent accumulations of protein near the MTOC, often in association with components of the protein degradation pathway (15,21). Aggresome formation is microtubule dependent and typically is accompanied by a major reorganization of vimentin filaments to form a cage around the protein inclusion (15,21,22). Although aggresome formation is enhanced in the presence of proteasome inhibitors or when certain proteins are overexpressed (44), the adenovirus E1B 55K protein, along with p53, has been reported to form aggresomes in uninfected 293 cells at normal expression levels (25). In the current study, formation of the virion clusters, like that of aggresomes, was microtubule dependent (Fig. 4), but vimentin remained as extended filaments throughout the cells and did not form a cage around the virion cluster (not shown). Moreover, the virion clusters did not recruit the 20s proteasome subunit, nor were they ubiquitinated, as determined by immunofluorescent staining with anti-20s and anti-FK2 antibodies, respectively (not shown). Brown et al. (6) reported that the p53/E1B 55K inclusions, which form at the MTOC, dissociate from the spindle poles in mitotic 293 cells. In contrast, virions that have already become associated with the MTOC remain attached at both spindle poles during mitosis (Fig. 2). Taken together, the properties of the virion clusters suggest that they likely do not represent classical aggresomes.
Stable association of Ad5 virions with spindle poles of dividing 293 cells (Fig. 2D) is consistent with strong association of virions with the MTOC. Other studies have reported virion clusters at the spindle poles of mitotic A549 cells (17,24) and at the MTOC of enucleated A549 cells (4). Virions were more likely to accumulate at the nuclei of A549 cells rather than the MTOC when the nuclear membrane was intact (4), but even so, virion clusters were noted at the MTOC in some cells (17). Strunze et al. (40) have linked virion accumulation at the MTOC to the absence of functional CRM1, a nuclear export factor proposed to facilitate detachment of virions from microtubules at the nuclear periphery and thus allow attachment to the nuclear pores. In the case of 293 cells, most of the virions clustered at the MTOC even in the presence of intact nuclei and with no inhibitors (Fig. 1 and 2). The stable association of virions with the MTOC might be expected to compromise infectivity, making 293 cells less sensitive to infection, but in fact, 293 cells are no less sensitive than other cells to productive infection (8). This observation, coupled with the ability to detect clusters up to 16 h p.i., a time when progeny virus is already being produced (8), suggested that the clusters did not with DAPI signal (blue) in cells treated with nocodazole (ϩ Noc). represent virions en route to successful genome delivery. The use of nocodazole and colchicine showed that intact microtubules are required for cluster formation but not for successful genome delivery, thereby confirming that the clusters are not an essential part of the infectious pathway in 293 cells and showing that the pathway leading to genome delivery could be uncoupled from bulk virion transport along microtubules. These results support the notion put forward by Glotzer et al. (17), based on studies with A549 and HeLa cells, that virions can be transported to the nucleus along a microtubuleindependent pathway. As noted by many authors, microtubule disruption resulted in dispersion of the majority of virions throughout the cell (11,17,24,26) (Fig. 4). Even so, a detectable proportion of the virion population could still be found at the nuclear periphery, as determined by tracking fluorescent virions (17) (Fig. 4 of current study) and by electron microscopy (11). Indeed, nuclear targeting of the bulk virion population appeared to improve in the absence of microtubules following nocodazole treatment (Fig. 4). Importantly, viral gene expression was not compromised when microtubules were disrupted (17), as reported for 293 and A549 cells in the current study ( Fig. 5; Tables 1 and 2). The increased fluores-cence intensity from EGFP expression in repeated experiments with nocodazole-treated cells is unexplained; a similar increase was not seen when microtubules were disrupted with colchicine. Giannakakou et al. (16) reported enhanced nucleus-directed transport of labeled Ad, as well as enhanced nuclear delivery of p53, in A549 and HeLa cells when microtubule dynamics were suppressed by low doses of nocodazole (1 to 3 nM). In that case, however, the low dose of nocodazole (1 to 3 nM) was not sufficient to depolymerize microtubules, whereas in the current study, a higher concentration of nocodazole (30 M) clearly disrupted the microtubules.
A high MOI, typically used for trafficking experiments, facilitates tracking of the bulk virion population but can obscure the pathway taken by the minor virion population en route to successful genome delivery. For this reason, experiments to determine the effect of microtubule disruption on infectivity were done with a low MOI, sufficient to infect only a small proportion of the cells, and cells were scored for genome delivery before progeny virus could spread to infect neighboring cells. The rationale was that if intact microtubules are required for virion transport, depolymerization of microtubules should result in a lower proportion of productively in-  Table 1). The apparent discrepancy in sensitivity between 293 and A549 cells likely reflects the fact that endogenous E1 expression by 293 cells facilitates replication of the Ad5EGFP genome (whose E1 region is deleted), thus increasing the copy number of the reporter gene. In A549 cells, the Ad5EGFP genome does not replicate, and thus, the number of EGFP gene copies is the number of genomes successfully delivered to the nucleus. Given that less than 10% of the inoculum virus is adsorbed under the infection conditions used, an input MOI of ϳ20 to 40 particles per cell should result in an adsorbed MOI of ϳ2 to 4 particles per cell. The Poisson distribution predicts that at an adsorbed MOI of three particles per cell, 38% of the cells are likely to be infected with four or more particles while 4% of the cells are likely to be infected with seven or more particles (12). Accordingly, the proportion of EGFP-positive cells suggests that a detectable EGFP signal at 12 h p.i. depends on ϳ4 particles for 293 cells and ϳ7 particles for A549 cells. These  a Values for mean fluorescence and percent fluorescent cells represent the mean Ϯ standard deviation for triplicate samples. Uninfected, untreated cells were used to determine the cutoff between background fluorescence and fluorescence due to EGFP. The value for mean fluorescence is based on the population detected above the cutoff. The total number of cells counted per sample was ϳ90,000 for A549 cells and ranged from ϳ30,000 to 70,000 for 293 cells. The values for nocodazole are taken from the flow cytometry data shown graphically in Fig. 5.
b MF T/U, Mean fluorescence of treated cells/mean fluorescence of untreated cells. a Cells were infected with Ad5.CMV-LacZ, as described in Materials and Methods, and then fixed at 12 h p.i. and stained with X-Gal. Cells were examined microscopically, and the number of blue cells per field was counted with the 10ϫ objective. Data are shown as mean no. of counts per field Ϯ standard deviation. Noc, treated with nocodazole. b n, no. of random fields per well.
few virions are not likely to encounter the few surviving microtubules in treated cells, and thus, genome delivery in most of the infected cells in treated cultures is not likely explained by surviving microtubules. The data shown in Fig. 5 and in Tables 1 and 2 do not support an essential role for microtubules in successful genome delivery in either 293 or A549 cells. In contrast, Mabit et al. (26) concluded that microtubules play a key role in virion transport along the infectious pathway, based on their findings that reporter gene expression was compromised when microtubules were disrupted. The same authors did note the existence of an alternative, less-efficient pathway and suggested actin or short nocodazole-resistant microtubules but favored diffusion of virions over short distances (26). While diffusion might explain successful infection of some cells in the population, it seems unlikely to compensate for microtubules in all of the infected cells treated with nocodazole or colchicine. Furthermore, diffusion seems unlikely to account for the nuclear targeting of so many labeled virions in treated A549 cells and 293 cells (Fig. 4). The apparent discrepancy is not explained by the use of different reporter genes. Since Mabit et al. (26) used Ad5LacZ and assessed infectivity by expression of ␤-galactosidase in a batch assay, the nocodazole experiments described here were repeated with 293 cells using Ad5.CMV-LacZ in place of Ad5EGFP. Disruption of microtubules with nocodazole did not compromise genome delivery by Ad5.CMV-LacZ as determined microscopically (Table 2) or in batch assays (not shown) using X-Gal or o-nitrophenyl-␤-D-galactopyranoside, respectively, as the substrate.
The mechanism for microtubule-independent transport is not known. "Cross talk" between microtubules and other cytoskeletal components (5, 10) might signal transport along a different pathway upon disruption of microtubules. In any case, it is difficult to draw conclusions about the normal infectious pathway on the basis of chemical disruption experiments. The experiments described here, along with those reported by Glotzer et al. (17), indicate that microtubules are not essential for nuclear targeting of virions, but they do not rule out the possibility that microtubules normally play a role in virion transport to the nucleus in untreated cells.
It is clear from multiple reports in the literature that adenoviruses are transported along microtubules towards the center of the cell (11,24,41). In the current study, the striking formation of juxtanuclear virion clusters in 293 cells highlights a seemingly nonproductive minus-end directed transport of virions along microtubules to accumulate near the MTOC and raises the question as to whether much of the documented transport along microtubules might also be nonproductive. It is interesting to speculate whether this pathway might represent a host defense mechanism by which virions are diverted from the infectious pathway. Segregation of virion accumulation at the MTOC from successful genome delivery, using nocodazole to disrupt microtubules, emphasizes that microtubule-mediated trafficking of the bulk virion population to the MTOC should not necessarily be interpreted as the route taken by virions en route to genome delivery. The current study indicates that successful genome delivery does not require intact microtubules for virion transport, at least in 293 and A549 cells. These findings reflect the complex nature of the adeno-virus entry pathway and draw attention to the need for further investigation.