Infectious Entry Pathway of Adeno-Associated Virus and Adeno-Associated Virus Vectors

Cell culture.

HeLa and 293 cells, which have been described elsewhere (12, 14), were obtained from the American Type Culture Collection, Manassas, Va. and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), penicillin (100 U/ml), and streptomycin (100 μg/ml) (GIBCO, Grand Island, N.Y.) at 37°C in a 5% CO₂ atmosphere. For microscopy, cells were seeded onto two-well chambered tissue culture treated glass slides (Falcon) at a density of 10⁴ cells per chamber or onto four-well chambered cover glasses at a density of 5 × 10³ cells per chamber and were used 24 to 48 h later. Chambered cover glasses were coated with poly-d-lysine (Sigma; 0.1 mg/ml; 1 h) prior to use. For confocal microscopy, HeLa S3 cells were grown in modified essential medium (S-MEM; GIBCO) containing 5% horse serum and 5% FBS.

HeLa cells expressing wild-type or the K44A mutant form of dynamin under tetracycline-inducible control (tTA-HeLa) were kind gifts from Sandra Schmid (Scripps Research Institute, La Jolla, Calif.) (9) and were initially maintained in DMEM supplemented with 10% heat-inactivated FBS, 400 μg of gentamicin/ml, 200 ng of puromycin/ml, and 1 μg of tetracycline/ml. For induction of dynamin overexpression, the cells were cultured in the absence of tetracycline for 2 days before being exposed to AAV.

Viruses and vector production and assay.

Adenovirus dl309 (19) has been described previously. Plasmid psub201, used to generate wild-type AAV, and plasmid pAB-11, used to generate AAVlacZ, have been described previously (13, 35). Plasmid pTRUF-5, kindly provided by Sergei Zolotukhin and Nicholas Muzyczka (University of Florida), was used to generate AAVEGFP. Both recombinant AAV vectors express transgenes under the control of the cytomegalovirus immediate-early promoter-enhancer. AAV vectors were produced in the absence of helper virus in 293 cells and purified either by successive bandings on CsCl gradients (23, 38) or by heparan sulfate (HS) affinity chromatography (8, 49). Wild-type AAV was prepared in adenovirus-infected cells by transfection of psub201 plasmid DNA as described previously (34). For preparation of fluorescently labeled AAV, the purity of virus was assessed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to ensure that adenovirus and cellular proteins had been completely removed. Preparations containing material other than the three AAV nonstructural proteins to the limit of detection either were further purified or were discarded because they proved unsatisfactory for fluorescence labeling. Purified virus particles were dialyzed into 10 mM Tris–150 mM NaCl (pH 7.8) containing 10% glycerol and frozen at −80°C until required for use. Particle numbers for wild-type AAV were determined by protein quantitation (7.5 μg of protein is equivalent to 10¹² particles). The titer of the recombinant virus, expressed as 293 cell transducing units, was determined by infecting 293 cells in the presence of adenovirus (dl309) and staining for β-galactosidase activity at 36 h (37). Radiolabeled AAV was prepared by adding [³H]methylthymidine (Amersham; 1 μCi/ml final concentration) to 293 cells 8 h after adenovirus (dl309) and wild-type AAV infection. The infected cells were harvested at 48 h postinfection, and the virus was purified as described above. The activity of the labeled virus was approximately 2 × 10⁻⁷ cpm per virus particle.

Fluorescent probes.

Purified virus was labeled with the carbocyanine dyes Cy2 and Cy3 (Amersham). Labeled AAV was prepared by adjusting AAV stocks to a concentration of 1 mg/ml (1.33 × 10¹⁴ particles/ml) in sodium carbonate-sodium bicarbonate buffer, pH 9.3. Buffer exchange was accomplished by rapid dialysis or by gel filtration on Sephadex G-50 (Pharmacia) spin columns. One milliliter of virus preparation was used to reconstitute the labeling reagent. After 30 min at room temperature, labeled virus was purified by extensive dialysis against 10 mM Tris (pH 7.8)–150 mM NaCl with 10% glycerol, by gel filtration on Sephadex G-50 using the same buffer, or by HS affinity chromatography. The purified virus was aliquoted into single-use vials and stored at −80°C prior to use. Labeled AAV was analyzed by SDS-PAGE to determine the specificity of the labeling reaction and the degree of protein cross-linking. Dye concentration was determined spectrophotometrically, using a molar extinction coefficient of 150,000 M⁻¹ cm⁻¹ at 552 nm for the Cy3 dye and an experimentally determined molar extinction coefficient at 280 nm for purified unlabeled AAV particles. Cy3-labeled adenovirus (Cy3Ad) was prepared as described by Leopold et al. (22).

Virus infections.

HeLa cells were infected with 1.25 × 10¹¹ particles of Cy3-conjugated AAV (Cy3AAV)/ml (approximately 10⁶ particles/cell) in binding buffer (DMEM containing 2 mM glucose, 10 mM HEPES [pH 7.3], and 1% bovine serum albumin) at 37°C unless otherwise noted. Cells were washed three times with binding buffer prior to infection. Lengths of viral exposure are noted but usually consisted of a short, 10-min pulse, after which the cells were washed twice with acid wash buffer [50 mM 2-(N-morpholino)ethanesulfonic acid (MES)–280 mM sucrose, pH 5] and three times with binding buffer. After being washed, the cells were either fixed immediately or maintained at 37°C during additional incubation periods. Prior to fixation, the binding buffer was removed and the cells were washed three times with ice-cold phosphate buffered saline (PBS; GIBCO, Grand Island, N.Y.). Cells were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature and washed three times with PBS. Where indicated, cells were either treated for 5 min at room temperature with 1-μg/ml DAPI (4′,6′-diamidino-2-phenylindole; Molecular Probes, Inc., Eugene, Oreg.) in PBS–0.1% Triton X-100 and washed three times with PBS or mounted in medium containing DAPI (VECTASHIELD; Vector Laboratories, Inc., Burlingame, Calif.). In some experiments, cells were treated with 25 nM ammonium chloride or bafilomycin A₁ (Sigma) as indicated.

Fluorescence microscopy.

Images were collected by the use of a Leica DMIRB microscope equipped with 40× NA 0.7 PlanApo DIC, 60× NA 1.4 PlanApo DIC, and 100× NA 1.4 PlanApo DIC objectives and a Hamamatsu intensified cooled charge-coupled device camera. The 16-bit images were digitally enhanced by subtraction of background and by gray-scale adjustment.

Characterization of fluorescently labeled virus.

Fluorescently labeled AAV was generated to study the viral entry pathway within the host cell. We characterized this reagent extensively in order to validate wild-type AAV entry steps. Virus labeled with cyanine-based reagents was readily visualized within HeLa cells after viral infection by standard fluorescence microscopy (Fig. 1). Since the fluorescent labeling of the virus involved the covalent modification of the capsid and exposure of virus to harsh conditions (pH 9.3), we were concerned that this process might alter the biological or physical properties of the viral particles. Although the attachment of fluorophore led to a small amount of protein cross-linking (less than 5%, as assessed by SDS-PAGE) (Fig. 2A), specificity for the viral receptor (Fig. 2B), virus attachment (Fig. 1 and 2C), and virus internalization (Fig. 1) were unchanged. Labeled virus particles that maintained normal physical and biological properties had dye-to-viral particle ratios in the range of 1.6 to 2.3 (n = 4). There were no detrimental effects on the physical properties of the viral particles (Fig. 2) or significant biological effects as evidenced by changes in viral titer (∼2 × 10¹¹ infectious units/ml both prior to and following fluorescence labeling). However, at dye-to-particle ratios above the specified range, e.g., >4 (n = 2), titers were severely affected and virus particles were prone to aggregation and precipitation. For this reason, all studies were carried out with AAV labeled at a dye-to-particle ratio of approximately 2. It should also be noted that by SDS-PAGE analysis it was shown that only viral capsid proteins were labeled with fluorophore. Fluorophore was not seen associated with cellular proteins or with any of the nonstructural viral proteins (Fig. 2a). However, it must be stressed that to achieve this level of specificity it was necessary to extensively purify AAV virions. Virus was purified either on three successive CsCl gradients or by a combination of CsCl gradient purification and HS affinity chromatography prior to labeling. Furthermore, all preparations were carefully monitored for the presence of degraded viral proteins or cellular proteins by SDS-PAGE to ensure specificity of the final labeled AAV preparation. The specificity and quality of the labeled virus are demonstrated by the ability to compete cell-associated Cy3AAV-2 fluorescence on HeLa cells with either excess unlabeled virus (Fig. 2C) or soluble HS (Fig. 2B). These results also demonstrate that the labeling reaction has not altered the ability of the virus to interact with its primary attachment receptor on the cell surface. Further evidence that fluorescent AAV has maintained its dependence on HSPG-mediated attachment was demonstrated by our ability to repurify labeled virus via HS affinity chromatography and by the cell binding profile of these reagents for HSPG-deficient cell lines (41). Validation of the labeled virions as described above provided a unique reagent to assay entry mechanisms of AAV infection.

AAV is rapidly internalized via clathrin-coated pits.

Virus internalization was monitored by using both fluorescently and radioactively labeled AAV-2. However, the use of ³H-labeled virus allowed easier quantification of AAV entry. HeLa cells, which express AAV attachment receptors (HSPG) at high levels, were incubated with [³H]AAV-2 at 4°C for 60 min, washed to eliminate unattached virus, and then incubated at 37°C for different lengths of time (0 to 90 min). Virus particles that had not been internalized were removed from the cell surface by washing with a mildly acidic buffer, and cell-associated radioactivity was determined by scintillation counting. More than 60% of the bound virus particles were taken up into the cells within the first 10 min of incubation at 37°C (Fig. 3). Thus, the internalization half-time for AAV is less than 10 min. Although this rapid entry of AAV into the host cells is indicative of receptor-mediated endocytosis, it was of interest to determine whether productive infection by AAV requires endocytosis from clathrin-coated pits. The recent identification of host cell proteins that regulate clathrin-mediated endocytosis provided an opportunity to more precisely define the mechanism of AAV entry.

Dynamin is a 100-kDa cytosolic GTPase that selectively regulates clathrin-mediated endocytosis. Dynamin associates with clathrin-coated membrane invaginations and has been proposed to mediate the constriction of coated pits and the budding of coated vesicles from the plasma membrane (18, 43). A dominant-negative mutant form of dynamin containing a point mutation in the GTP binding domain (Lys-44 to Ala-44 [K44A]) has been shown to block clathrin-mediated endocytosis (9). AAV infection of HeLa cells expressing the K44A mutant dynamin was measured by using the AAVlacZ vector. Uninduced tTA-HeLa cells or cells which had been induced by removal of tetracycline for 48 h were infected with AAVlacZ at a virus particle-to-cell ratio of 1,000:1. The number of cells expressing the lacZ reporter gene was then quantitated 48 h postinfection by 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) histochemistry analysis. A significant decrease in AAV-mediated lacZ transduction was observed in HeLa cells induced for K44A expression (minus tetracycline) compared to that evident in tetracycline-regulated HeLa cells (Fig. 4, left panel). The level of AAV-mediated gene delivery to HeLa cells in the presence of tetracycline was the same as that to HeLa cells overexpressing wild-type dynamin under the control of the tetracycline-regulated promoter (Fig. 4, left panel). Furthermore, induction of mutant dynamin 30 min following infection did not inhibit reporter gene expression (data not shown), supporting the involvement of this protein in AAV entry steps prior to gene expression. We also determined that expression of mutant K44A dynamin, or overexpression of wild-type dynamin, did not interfere with virus attachment by analyzing Cy3AAV-2 binding to tetracycline-induced and uninduced HeLa cell lines (Fig. 4, right panel). Therefore, inhibition of AAV-mediated gene delivery was due to decreased virus uptake and not to inhibition of virus attachment or vector gene expression. These findings demonstrate that the predominant route of AAV entry into HeLa cells is via receptor-mediated endocytosis.

AAV requires passage through an acidic compartment for productive infection.

Biochemical studies using the lysosomotropic drug ammonium chloride (28) or the proton pump inhibitor bafilomycin A₁ (6) showed that these drugs have a significant inhibitory effect on AAV infectivity and AAV-mediated gene expression (Fig. 5). These drugs essentially block infection of HeLa cells when present during the first 30 min after the onset of endocytosis. Ammonium chloride is known to raise the pH of intracellular organelles within 1 min following addition (28), making it possible to inhibit a low-pH-dependent endosomal escape mechanism at defined time points. The inhibitory effect of ammonium chloride on AAV-mediated gene transfer in HeLa cells is demonstrated in Fig. 5A. This drug did not significantly influence AAV-mediated gene transfer when present exclusively during the adsorption phase at 4°C, a temperature that blocks endocytosis, implying that the presence of ammonium chloride at this time point did not influence binding of AAV to the plasma membrane. We observed half-maximal AAV-mediated gene expression when the drug was added 30 min after the cells were warmed to 37°C, suggesting that endosomal escape had already begun by this time. This observation supports the occurrence of early endosome escape for AAV virions. Furthermore, AAV-mediated gene transfer was completely resistant to ammonium chloride by 90 min after the shift to 37°C, suggesting that penetration of the virus into the cytosol was complete by this time.

To independently confirm the occurrence of early endosome escape by AAV, we utilized a different drug, bafilomycin A₁, during AAV vector infection. Bafilomycin A₁ is a potent inhibitor of the vacuolar H⁺-ATPase responsible for acidification of endosomal vesicles. The influence of different bafilomycin A₁ concentrations on AAV infection of HeLa cells was determined. Cells were preincubated with 0, 20, or 200 nM bafilomycin A₁ and then infected with the AAVEGFP vector. Infection was monitored 24 h later by measuring green fluorescent protein fluorescence. Both 20 and 200 nM bafilomycin A₁ completely inhibited infection of HeLa cells by the AAV vector (Fig. 5B). These findings support the requirement of endosomal acidification and early endosomal escape for efficient AAV vector infection.

Entry of AAV and adenovirus via distinct endosomal pathways.

The previous data suggest that AAV is able to penetrate the endosomal membrane fairly quickly following internalization and that this process requires passage through an acidic compartment. Recently, we have shown that α_vβ₅ integrin is involved in the AAV entry process (40). It is interesting that adenovirus, the AAV helper virus necessary for a productive infection, also requires a slightly acidic pH and binding of the adenovirus penton base protein to α_vβ₅ integrin in order to efficiently escape from the endosome (7, 29, 47). Although there is substantial evidence demonstrating that AAV genomes can be expressed in host cells in the absence of adenovirus, implying that AAV alone is capable of penetration into the cytosol, it was of interest to determine synergism or competition of endosomal escape following adenovirus and AAV coinfection. Adenovirus has been shown to greatly enhance transduction of some cells by recombinant AAV vectors due to its effect on second-strand DNA synthesis (10, 11). By assessing the fate of fluorescently labeled adenovirus and AAV, we sought to determine whether these viruses ever colocalized within the host cell such that adenovirus might be in a position to physically assist AAV entry into the cytosol. HeLa cells were coinfected with Cy2AAV-3 and Cy3Ad, and the distribution of the viruses was monitored at 10, 20, and 40 min and at 1 h postinfection by fluorescence microscopy. At no time point was there significant colocalization of the AAV (green) and adenovirus (red) labels (Fig. 6A, top panel). In contrast, coinfection with Cy2AAV-3 and Cy3AAV-2 resulted in extensive colocalization of the red and green signals (Fig. 6A, lower panel). While these data do not provide direct evidence of the presence of multiple AAV particles within each endosome, they support it. These results further support the notion that endosomal escape of AAV is efficient and is unaffected by adenovirus. Interestingly, the very early endosomes all contain several AAV particles, as evidenced by label colocalization (yellow signal). This may be indicative of a mechanism of internalization requiring the grouping of multiple receptor-AAV complexes on the surface of the host cell prior to entry.

Translocation of AAV to the nucleus.

The efficiency of AAV infection relies to a large extent on the efficient targeting of the AAV genome to the host cell nucleus following infection. This property is reflected in the rapid translocation of virions to the nuclear envelope. Cy3AAV-2 was bound to HeLa cells for 2 h at 4°C. Cells were then washed to remove unbound virus and either fixed immediately or incubated for various periods of time at 37°C to examine redistribution of cell-associated virus (Fig. 6B). Prior to internalization, AAV virions were distributed evenly on the outside of the plasma membrane (Fig. 6B, 0 min). Following 30 min at 37°C, AAV exhibited a disperse, punctate distribution in the cell, likely reflecting virions in both endocytic compartments and free within the cytosol (Fig. 6B, 30 min). The distribution of staining had already started to shift toward the nucleus, with some virus beginning to accumulate at the nuclear envelope; however, the majority of the virions remained widely distributed in the cytoplasm. Later time points show a progressive perinuclear accumulation of AAV virions (Fig. 6B, 2 h and 4 h). By 2 h following internalization, nearly all of the virus particles had accumulated perinuclearly, and they remained at this location throughout the 4-h incubation period.

To more accurately access the intracellular distribution of AAV and to determine the potential for nuclear uptake of AAV virions, we used a laser scanning confocal microscope. In this manner, we were able to demonstrate fluorescent AAV particles within the nuclei of host cells within 2 h postinfection (Fig. 7). These observations support a rapid transport of capsid components to the nucleus with transgene expression detected within 3 to 4 h postinfection (data not shown).