Endosomal processing limits gene transfer to polarized airway epithelia by adeno-associated virus (original) (raw)
Molecular analysis of rAAV genomes in polarized airway epithelia. Recent studies have revealed a lack of AAV-2 receptor (heparan sulfate proteoglycan) and coreceptors (FGFR-1 and αVβ5 integrin) at the apical surface of differentiated airway epithelia (6, 7, 22). However, differences in the binding of radioactive rAAV-2 virus at the apical and basolateral membranes were only four- to sevenfold (basolateral > apical). These differences in binding are insufficient to explain the 200-fold variance observed in the polarity of infection (basolateral >> apical) with rAAV-2 (6). These findings suggested that viral binding and/or uptake were not the sole limiting factors contributing to inefficient mucosal transduction in airway epithelia. We hypothesized that additional barriers, such as endosomal processing and/or the molecular conversion of viral genomes, might also contribute to the inefficient infection from the apical surface of airway epithelia. To this end, we evaluated the molecular state of rAAV DNA at 50 days after apical and basolateral infection of air-liquid interface–cultured human bronchial epithelia. At this time point, gene expression measured from an EGFP reporter was more than 200-fold higher in basolaterally infected cultures (data not shown) (6). Hirt DNA from the cultures was evaluated by Southern blot hybridization with [32P]-labeled EGFP probes. As shown in Figure 1, a significant amount of apically applied rAAV was able to infect airway cells. However, only single-stranded viral genomes (ssDNA) were detected at this time point (50 days). This result clearly suggests that rAAV can be endocytosed from the mucosal surface and that the endocytosed viral ssDNA was stably sequestered in some unknown subcellular compartment. In contrast, the majority of basolaterally applied rAAV was converted into double-stranded forms that migrated at 2.8 kb and greater than 12 kb in 1% nondenaturing agarose gels (Figure 1). Consistent with our previous reports (13, 23), subsequent restriction enzyme mapping with Hirt DNA Southern blots confirmed this 2.8-kb band to be a supercoiled, circular episomal molecule (data not shown). The identity of the greater than 12-kb band, which is significantly more intense after basolateral infection, is currently unknown but may represent episomal circular concatamers of the AAV genome. Taken together, our results suggest that inefficient molecular conversion of AAV viral DNA to circular genomes represents a significant obstacle for rAAV-mediated gene transfer from the apical surface of the airway. Furthermore, circularization, not linear replication through self-priming, appears to be the predominant pathway for rAAV gene conversion in polarized airway epithelia.
Southern blot analysis of AAV genomes in polarized airway epithelia. Differentiated human bronchial epithelial cultures were infected with AV.GFP3ori at an moi of 10,000 particles per cell from either the apical (lanes 1 and 3) or basolateral (lanes 2 and 4) surface. Fifty days after infection, Hirt DNA was harvested and electrophoresed in a 1% agarose gel. Each lane represents combined Hirt DNA from two transwell cultures, and the two gels are derived from two independent tissue samples. The Southern blot was hybridized to a [P32]-labeled EGFP probe and exposed to film for 48 hours. Molecular weight standards are marked to the left of the autoradiogram. The single-stranded and circular monomer forms of AAV viral DNA migrate at 1.6 kb and 2.8 kb, respectively. ssDNA, single-stranded viral DNA; CM, circular monomer; CC, circular concatamer.
Proteasome inhibitors dramatically enhance rAAV infection in polarized airway epithelia. Given that rAAV appears to remain latent within some cellular compartment(s) after apical infection in the airway, we sought to investigate whether agents that alter the molecular conversion of the viral genome might enhance rAAV transduction in airway epithelia. Several chemicals, including DNA-damaging agents (24), DNA synthesis and topoisomerase inhibitors (25), and cellular tyrosine kinases inhibitors (26, 27), have been shown to increase the AAV transduction efficiency in cultured primary fibroblasts and/or transformed cell lines. Application of camptothecin, etoposide, hydroxyurea, and genistein resulted a ten-to 60-fold enhancement in rAAV transduction from the basolateral surface (Figure 2). Interestingly, however, none of these reagents facilitated rAAV transduction from the apical surface (data not shown).
Effects of different chemical reagents on AAV transduction from the basolateral surface of polarized bronchial epithelia. Fully differentiated human bronchial airway epithelia were treated with the indicated chemicals at the time of AAV infection (moi = 10,000 particles per cell) from the basolateral surface, as outlined in Methods. Camp: camptothecin at 0.1 μM; Etop: etoposide at 10 μM; Aphi: aphidicolin at 5 μg/mL; HU: hydroxyurea at 40 mM; Geni: genistein at 50 μM; and LLnL at 40 μM. Because most of the tested chemicals (except for hydroxyurea) were dissolved in DMSO, a vehicle control of 1% (vol/vol) DMSO was also performed to exclude the possibility of a nonspecific effect. GFP expression was monitored at 48 hours after infection by counting the number of GFP-positive cells per ×10 field in the Millicell. The data represent the mean ± SEM of three independent infections for each tested chemical.
Because chemicals known to affect intranuclear events involved in rAAV transduction in other cell types (28) did not enhance rAAV apical infection in the airway, we sought instead to assess chemicals that might affect endocytic processing of internalized virions. Proteasome systems are known to modulate the intracellular processing of many foreign and endogenous molecules, including viruses such as HIV (29). Several specific, cell-permeable, peptide aldehyde inhibitors of proteasome pathways have recently been discovered (30, 31). These inhibitors bind to the active sites of proteolytic enzymes within the proteasome core and reversibly block their function (32). To test whether proteasomes represent an intracellular compartment that sequesters rAAV after infection, we applied the tripeptidyl aldehyde proteasome inhibitor (a cysteine protease inhibitor) LLnL to polarized cultures of human bronchial epithelial cells at the time of rAAV infection. To our surprise, a greater than 200-fold augmentation in transgene expression was obtained at 2 days after infection when 40 μM LLnL was applied to the serosal surface along with rAAV (Figure 2). A similar result was achieved when another ubiquitin-proteasome pathway inhibitor, Z-LLL (33), was tested (data not shown). However, as described later here, the most important finding was that these proteasome inhibitors also substantially increased rAAV transduction from the mucosal surface. When compared with other chemicals tested, proteasome inhibitors were found to be the most potent enhancers of rAAV transduction in airway epithelium.
Proteasome inhibitors augment rAAV transduction in airway epithelia in a polarized fashion. To test the effect of proteasome inhibitors on rAAV transduction from the apical membrane, we performed a side-by-side kinetic comparison of rAAV transduction from both mucosal and serosal surfaces of airway epithelia after treatment with LLnL. As shown in Figure 3a, coadministration of LLnL and rAAV to the mucosal surface resulted a sustained augmentation in AAV transduction, which peaked at 22 days after infection. In contrast to mucosal infection, rAAV infection from the serosal surface in the presence of LLnL resulted only in a transient peak in gene expression at 72 hours after infection, which returned to the levels equivalent to those of the untreated samples by 22 days (Figure 3b). These results suggested that the proteasome inhibitor LLnL produces different augmentation profiles when AAV virus is applied to either the apical or the basolateral membranes. To exclude potential effects caused by polarized uptake of LLnL by airway epithelia, we tested different combinations of rAAV and LLnL administration from both apical and basolateral surfaces. Similar augmentation patterns for AAV transduction were achieved, regardless of whether LLnL was applied to the same or opposite surface as rAAV during infections (data not shown).
Proteasome inhibitors differentially augment rAAV transduction from the apical or basolateral surfaces of airway epithelia. The efficiency and time course of rAAV transduction were evaluated in polarized airway epithelial cultures after infection with rAV.GFPori3 (moi = 10,000 particles per cell) in the presence or absence of 40 μM LLnL. Transgene expression was monitored by indirect fluorescence microscopy at the indicated time points by quantifying the mean number of GFP-positive cells per ×10 field (mean ± SEM of three independent samples for each time point). The effect of LLnL treatment was compared between matched sets of tissue samples at each time point after infection from the apical (a) or basolateral (b) surfaces. The photomicrographs on the right side of each panel illustrate representative ×20 fields for the 3- and 22-day time points after infection.
To expand these findings, we examined whether LLnL administration augmented rAAV transduction of particular airway cell types. We utilized a rAAV vector encoding the alkaline phosphatase gene (Alkphos) and evaluated transduced cell types by standard histochemical staining for Alkphos to address this question. In the absence of LLnL, rAAV preferentially transduced basal cells at 3 days after serosal application of virus (Figure 4, a and c). Consistent with previous findings utilizing AV.GFP3ori virus, coadministration of LLnL resulted a dramatic increase in AV.Alkphos transduction (Figure 4, b and d). Interestingly, ciliated cell transduction was most significantly increased by treatment with LLnL at the time of rAAV infection (Figure 4e). In contrast, basal cells were the least responsive to LLnL treatment. These findings indicated that the mechanisms of LLnL action may have some cell-specific components, which differ in polarized (i.e., ciliated) and nonpolarized (i.e., basal) cell types.
The proteasome inhibitor LLnL preferentially induces AAV transduction in ciliated cells. The cell types transduced by AAV were examined by alkaline phosphatase staining at 3 days after basolateral infection of polarized epithelial cultures with AV.Alkphos infection (moi = 10,000 particles per cell) in the presence or absence of 40 μM LLnL. En face photomicrographs in a (without LLnL) and b (with LLnL) depict enhanced rAAV transduction in the presence of LLnL. Paraffin sections (8 μm) were used to histologically quantify the types of cells transduced. Three classifications were used: ciliated cells (with alkphos expression localized to the apical membrane and cilia), basal cells residing in the lower half of the epithelium with no contact with the lumen, and nonciliated columnar cells. c (without LLnL) and d (with LLnL) show representative neutral red counterstained cross-sections for each condition. The numbers of alkphos-stained cells per 2,000 epithelial cells for the various cell types transduced by AAV are presented in e (mean ± SEM from three independent Millicell insert samples).
Optimization of LLnL enhanced rAAV transduction. With the aim of further improving the enhancement in rAAV transduction achieved in the presence of LLnL, we performed several detailed kinetic studies altering the timing and number of LLnL administrations after rAAV infection (Figure 5). Several important conclusions arose from these studies. First, after basolateral infection (Figure 5a), administration of LLnL once every 3 days increased the length of peak transgene expression, despite the fact that by the end of 30 days, levels were similar to those of cultures treated once at the time of infection. Second, continual administration of LLnL was toxic to cells and ablated transgene expression by 10 days. Third, reinfection of cultures with rAAV in the presence of LLnL at 7, 10, and 15 days resulted in a similar pattern of augmentation and, as expected, elevated the final level of transgene expression observed at 30 days (only data from the second infection at 15 days are shown). Most notably, however, all the cultures infected from the basolateral side in Figure 5a produced similar long-term transgene expression levels within two- or threefold of each other, regardless of whether LLnL was administered.
Optimization of LLnL-enhanced rAAV transduction in polarized bronchial epithelia. Differentiated airway epithelia were infected with AV.GFP3ori (10,000 particles per cell) from either the basolateral (a) or the apical (b) surface with the indicated treatments involving LLnL and/or EGTA. All infections were carried out for 24 hours, and GFP transgene expression was monitored by indirect fluorescent microscopy at the indicated times. Data represent the mean (± SEM; n = 6) for each experimental condition. Experiments were performed in triplicate on transwells derived from samples obtained from two different patients. The following conditions were evaluated for basolateral infection in a: single infection with AV.GFP3ori alone (solid line); single infection with AV.GFP3ori in the presence of 40 μM LLnL (filled circles); single infection with AV.GFP3ori in the presence of 40 μM LLnL followed by repeated 5-hour exposure to 40 μM LLnL in the basal compartment culture medium every third day thereafter (open circles); single infection with AV.GFP3ori in the presence of 40 μM LLnL followed by the continued exposure to 40 μM LLnL in the basal medium after rAAV was removed (filled triangles); and repeated infection with AV.GFP3ori on days 0 and 15 in the presence of 40 μM LLnL for 24 hours at the time of infection (open triangles). The following conditions were evaluated for apical infection in b: single infection with AV.GFP3ori alone (filled circles); single infection with AV.GFP3ori after pretreatment with 3 mM hypotonic EGTA before the viral infection (open circles); single infection with AV.GFP3ori in the presence of 40 μM LLnL (filled triangles); and single infection with AV.GFP3ori in the presence of 40 μM LLnL after pretreatment with 3 mM hypotonic EGTA before the viral infection (open triangles).
Despite the fact that LLnL administration at the time of the viral infection augmented rAAV transduction from both the apical and basolateral surfaces, the kinetics of this induction were significantly different. Enhancement after basolateral infection was transient, whereas enhancement after apical infection was long term (Figure 3). Furthermore, although induction with LLnL from the apical membrane was long-lasting, by 30 days the maximal level of transgene expression was only one eighth of that resulting from basal infection (Figure 3a). The application of hypotonic EGTA solution has been shown to increase AAV transduction from the apical surface by seven- to tenfold (6, 21). Therefore, we tested whether the combined administration of EGTA and LLnL could provide yet a further increase in rAAV transduction from the apical surface. Interestingly, treatment of airway cultures with EGTA before infection with rAAV in the presence of LLnL gave a transient peak in transduction within the first 3 days, and, a significantly increased (200-fold), prolonged level of transgene expression out to 30 days (Figure 5b). This prolonged level of transgene expression, while comparable to rAAV infection from the basal surface, was much above the level observed in apically infected epithelia treated with EGTA alone (Figure 5b). In summary, these results demonstrate that EGTA and LLnL have synergistic effects on rAAV transduction, allowing transduction from the apical surface at levels normally only seen after basolateral infection.
Viral binding and internalization are not affected by LLnL treatment. The action of LLnL has been typically attributed to its selective and reversible inhibition of the proteasome system. However, it was important to rule out any possible effect on viral binding and/or endocytosis. As has been found for type 1 herpes simplex virus (34), LLnL treatment had no significant effect on 4°C rAAV binding (Figure 6). Similarly, the uptake of [S35]-labeled rAAV for a 2-hour infection period at 37°C was not altered by LLnL treatment (Figure 6). Given these results, we conclude that LLnL acts at points distal to virus binding and entry. Interestingly, at 24 hours after infection, a very significant decrease in the amount of intracellular radioactivity was observed in epithelia treated with LLnL, regardless of which surface was infected (Figure 6). Given the concordant increase in transgene expression at this time point (Figure 6), we reasoned that LLnL may be accelerating processing and routing of the virus to the nucleus, where uncoating and clearance of [S35]-labeled capsid proteins occur. By this mechanism, S35 isotope would be diluted into the culture medium and could explain the decrease in cell-associated counts.
[![Binding and uptake of [S35]-labeled AV.GFP3ori in fully differentiated huma](https://dm5migu4zj3pb.cloudfront.net/manuscripts/8000/8317/small/JCI0008317.f6.gif)
](/articles/view/8317/figure/6)Figure 6
Binding and uptake of [S35]-labeled AV.GFP3ori in fully differentiated human bronchial epithelia. The ability of polarized bronchial epithelia to bind and internalize virus from the apical (a) and the basolateral (b) surfaces was quantified using [S35]-labeled rAAV. The binding assay was performed after incubation with virus at 4°C for 1 hour, followed by repeated washing in PBS. The combined bound and internalized virus was quantified after incubation with virus at 4°C for 1 hour, and subsequent incubation at 37°C for 2 hours or 24 hours. Nonspecific background binding of radiolabeled virus was determined in parallel studies on collagen-coated empty chambers not seeded with bronchial cells. Background counts (averaging 15.67 ± 5.17 cpm/well) were subtracted from experimental sample counts before analysis. Data in the left side of each panel are presented as the net cpm of bound/internalized virus (raw counts minus background counts of empty transwells). The results represent the mean (± SEM) of six independent transwells for each condition. Experiments were performed in triplicate from two independent tissue samples. The significance of the differences between each pair of samples (with or without LLnL) was evaluated using the Student’s t test, and P values are provided above the data for each condition. To correlate uptake of radioactive virus with the functional expression of the rAAV-encoded transgene, GFP expression from the same set of samples was quantified at 24 hours after infection by indirect fluorescent microscopy. The results (mean ± SEM; n = 6) are presented as a bar graph on the right side of each panel.
LLnL enhances endosomal processing and nuclear trafficking of rAAV. To test the hypothesis that LLnL increases trafficking of rAAV to the nucleus, we performed in situ localization of the [S35]-labeled rAAV particles after infection from the apical and basolateral surfaces in the presence and absence of LLnL. Because loss of intact radiolabeled capsid proteins occurred at 24 hours after infection, we chose a 2-hour time point for this analysis. Using photoemulsion overlay, we evaluated the subcellular distribution of [S35]-labeled rAAV particles by blinded morphometric analysis. As shown in Figure 7, the majority of viral particles localized to the cytoplasm in the absence of LLnL. This was the case regardless of whether infection was performed from the apical or basolateral surface. In contrast, LLnL treatment substantially changed the intracellular distribution of radiolabeled rAAV particles, resulting in a significant shift to nuclear associated grains. These results substantiated the findings from whole cell counts at 24 hours after infection, which suggested that LLnL increases viral uncoating and the subsequent loss of S35 isotope into the media.
[![In situ localization of rAAV in the polarized airway epithelia using [S35]-](https://dm5migu4zj3pb.cloudfront.net/manuscripts/8000/8317/small/JCI0008317.f7.gif)
](/articles/view/8317/figure/7)Figure 7
In situ localization of rAAV in the polarized airway epithelia using [S35]-labeled virus. Polarized human airway epithelia were infected with [S35]-labeled virus (moi = 50,000 particles per cell) from either the apical or basolateral side (± LLnL). At 2 hours after infection, Millicells were washed with media three times and fixed in 4% paraformaldehyde overnight before cryoprotection in sucrose and OCT embedding. Frozen sections (10 μm) were overlaid with photoemulsion and developed after 5 weeks of exposure. (a) The typical localization pattern after basolateral infection in the presence of LLnL. Arrows indicate nuclear-associated virus. (b) Enlargements of boxed regions in a. Blinded morphometric quantification was performed by counting the number of nuclear-associated and cytoplasmic radioactive grains of 10 random fields, as shown in a. A total of 60 cells was quantitated per field, to give a total of 600 cells per sample and 2,400 cells per condition. Results in c are the mean ± SEM of four independent samples for each condition.
LLnL augments rAAV transduction within a specific time frame after infection. Evidence thus far has suggested that LLnL may act to increase intracellular routing of rAAV to the nucleus. Additionally, LLnL action is independent of the epithelial surface to which it is administered (i.e., serosal application of LLnL augments mucosal infection and visa versa). This indicates that LLnL need not be endocytosed with AAV particles to enhance transduction. We therefore hypothesized that LLnL may act at a specific time after rAAV endocytosis but during endosomal processing. To provide functional support for this hypothesis, we performed a kinetic analysis of LLnL action at various times after infection from the basolateral surface. In these experiments, LLnL was added to the culture medium either at the time of AAV infection or at various time points after infection. Viral-mediated transgene expression was quantified at 24-hour intervals after infection. Augmentation was achieved regardless of whether LLnL was administrated at 0, 24, 48, or 72 hours after viral infection. However, addition of LLnL at 24 or 48 hours gave the strongest level of augmentation. The ability of LLnL to rescue AAV expression appeared to decline by 72 hours after infection (Figure 8) and was completely lost by 15 days after the initial AAV infection (data not shown). Taken together, we hypothesize that after rAAV enters the cell, it may be targeted to an intracellular compartment that is sensitive to proteasome inhibitor–facilitated liberation. In addition, the loss of an LLnL augmentation effect at 15 days after infection suggests that enhanced transcription, translation, and/or stability of the transgene products do not likely contribute to the mechanism responsible for our observation.
LLnL induces rAAV transduction within a specific time frame after infection. To study the kinetic effects of LLnL administration on AAV transduction, the proteasome inhibitor LLnL was added to the culture medium at different times after rAAV infection from the basolateral surface of airway epithelia. All cultures were infected with AV.GFP3ori (10,000 particles per cell) at 0 hours. LLnL (40 μM) was added at the time of infection (0 hours) or at 24-hour intervals after infection (24, 48, or 72 hours). The baseline of AAV transduction in the absence of LLnL treatment is also shown. GFP expression was quantified at 1, 2, 3, and 4 days after infection using indirect fluorescent microscopy and is presented as GFP-positive cells per ×10 field. The data in each panel represent the mean (± SEM) from three independent samples. The immediate increase in rAAV-mediated transgene expression after addition of LLnL is consistent with the hypothesis that LLnL acts by enhancing endosomal processing of viral particles from the cytoplasm to nucleus where it can be expressed. Furthermore, the addition of LLnL at 2–3 days after infection gave rise to higher levels of transgene gene expression than application of LLnL at the time of infection. These results support the notion that viral binding and internalization are not likely the steps enhanced by LLnL.
Combined treatment of LLnL and EGTA prevents degradation of internalized rAAV. To clarify further the molecular mechanism(s) responsible for augmentation of rAAV transduction by LLnL, we analyzed rAAV genomes in infected cells by Southern blotting Hirt DNA. Consistent with studies using [S35]-labeled virus (Figure 6), rAAV binding to either surface of polarized airway epithelia was not affected by LLnL treatment (Figure 9a, lanes 8 and 9 for apical infection, lanes 11 and 12 for basal infection). Southern blotting also demonstrated two- to sevenfold higher viral binding from the basal surface, which varied among different tissue samples (data not shown). The extent of virus internalization was compared after stripping surface-bound virus with trypsin. Confirming previous results (Figures 1, 6, and 7), a significant amount of rAAV was endocytosed from the apical surface during the infection period (Figure 9b, lanes 2, 3, 8, and 9), although viral uptake was more active from basolateral surface (Figure 9b, lanes 5, 6, 11, and 12). LLnL alone also did not substantially prevent enzymatic degradation of the internalized AAV viral DNA (Figure 9b; also see Figure 6), indicating that enhanced viral trafficking into the nucleus might be more important in the observed augmentation by LLnL. However, treatment with both hypotonic EGTA and LLnL substantially increased the amount of virus internalized from the apical surface (Figure 9b, compare lanes 2, 4 and lanes 8, 10; Figure 9c, compare lanes 1, 2, and 5, 6, and 10, 12). Given that hypotonic EGTA treatment alone only slightly increased persistence of AAV DNA (less than fourfold; Figure 9c, lane 11) or AAV-mediated gene expression (less than tenfold; Figure 5b) (6, 21) after apical infection, we reasoned that the predominant mechanism responsible for the combined effects of EGTA and LLnL might be due to reduced degradation of the internalized virus and an increased rate of endocytosis. These synergistic effects of EGTA and LLnL augment rAAV transduction from the apical membrane more than 200-fold. Additionally, the conversion of single-stranded viral genomes to linear replication or circular forms has been associated with enhanced AAV transduction by adenoviral early gene products or ultraviolet irradiation, respectively (12, 13, 23). As shown in Southern blots of Hirt DNA from cultures coinfected with Ad.dl802 and rAAV (Figure 9c, lane 9), LLnL-enhanced AAV transduction was clearly not mediated through the formation of linear replication intermediates (4.7-kb band) as seen in the presence of adenoviral E4orf6 protein produced by Ad.dl802 coinfection.
Examination of rAAV endocytosis by Southern blot analysis of viral DNA. Hirt DNA from AV.GFP3ori-infected or mock-infected (lanes 1 and 7 in both a and b) human bronchial epithelia were extracted for a direct examination of viral genomes by Southern blotting against a [P32]-labeled EGFP probe. (a) Viral binding studies in the presence and absence of LLnL with or without EGTA treatment before apical or basolateral infection for 1 hour at 4°C. Cell surface–bound virus was completely removed by trypsin (a, lanes 2 through 6). To determine the amount of the surface-bound rAAV, cells were infected with AV.GFP3ori for 1 hour at 4°C and were not treated with trypsin but rather washed before Hirt DNA extraction. a: Lanes 2 and 8: apical AAV infection; lanes 3 and 9: apical AAV infection in the presence of LLnL; lanes 4 and 10: cells were pretreated with hypotonic EGTA before apical infection in the presence of LLnL; lanes 5 and 11: basolateral infection; lanes 6 and 12: basolateral infection in the presence of LLnL. (b) Results of studies evaluating rAAV internalization from either the apical or the basolateral surface in the presence or absence of LLnL, and the internalization from the apical surface after combined treatment with hypotonic EGTA and LLnL. To detect the net amount of the internalized viral genome, all samples in b were treated with trypsin just before Hirt DNA was harvested. The extent of the internalized virus at 4 hours (b, lanes 2 through 6) and 24 hours (b, lanes 8 through 12) incubation at 37°C after infection is represented by the intensity of the 1.6-kb single-stranded viral genome band. b: Lane 2: apical AAV infection for 4 hours; lane 3: apical AAV infection in the presence of LLnL for 4 hours; lane 4: cells were pretreated with hypotonic EGTA before apical infection in the presence of LLnL for 4 hours; lane 5: basolateral infection for 4 hours; lane 6: basolateral infection in the presence of LLnL for 4 hours; lane 8: apical infection for 24 hours; lane 9: apical infection in the presence of LLnL for 24 hours; lane 10: cells were pretreated with hypotonic EGTA before apical infection in the presence of LLnL for 24 hours; lane 11: basolateral infection for 24 hours; lane 12: basolateral infection in the presence of LLnL for 24 hours. (c) Comparison of the effect of LLnL/EGTA on rAAV genomes at 2, 10, and 30 days after a 24-hour infection from the apical (lanes 1, 2, 5, 6, 10, 11, and 12) and basolateral (lanes 3, 4, 7, 8, 9, 13, and 14) membranes. Treatment conditions are noted above each lane; transwells were not treated with trypsin before harvesting Hirt DNA. An additional control included coinfection with Ad.dl802 (moi = 500 particles per cell) to demonstrate replication form monomers (lane 9, 4.7 kb). It should be noted that different exposure times were used for the three different panels in c (lanes 1–4, 3 hours; lanes 5–8, 15 hours; and lanes 10–14, 12 hours). Matched DNA samples from uninfected cultures did not demonstrate detectable hybridization (data not shown).
Ubiquitination of viral capsid proteins after rAAV infection in the airway alters the efficiency of transduction. Proteasome-dependent degradation of ubiquitinated molecules represents a major pathway for disposal of both endogenous and foreign proteins (35). Several distinct steps in the regulation of this pathway have been identified, including activation of ubiquitin by its activating enzyme (E1), transfer of the activated ubiquitin to the ubiquitin carrier protein (E2), and subsequent delivery of the activated ubiquitin to the protein substrate by ubiquitin ligase (E3). Ultimately, ubiquitinated proteins are degraded by the 26S proteasome through an ATP-dependent process. To test whether enhancement of rAAV transduction by proteasome inhibitor peptides involves liberation of ubiquitinated virus from an endosomal compartment, we first attempted to determine whether ubiquitination of viral capsid indeed occurs. To this end, we sought to examine the extent of ubiquitin side chains on AAV capsid proteins after infection and to determine whether treatment with proteasome inhibitors altered the extent of ubiquitination. AAV capsid proteins were immunoprecipitated using anti-VP 1, 2, and 3 antibody from rAAV-infected human polarized airway cells and human fibroblasts at 6 hours after viral infection. Subsequent Western analysis with antiubiquitin specific antibodies demonstrated a significant increase in the cellular level of ubiquitinated AAV capsid in fibroblasts after proteasome treatment (Figure 10c). Ubiquitination significantly increased the molecular weight of capsid proteins (63, 73, and 87 kDa) to 220–250 kDa and is consistent with the size change after ubiquitination for other molecules (36). Unfortunately, the limited amount of virus retrievable from air-liquid interface–cultured human airway cells precluded our ability to detect ubiquitinated capsid in this system (data not shown). Nonetheless, confluent primary fibroblasts also demonstrated augmentation (tenfold) of transgene expression after treatment with proteasome inhibitors (Figure 10, a and b). From these experiments, we hypothesize that proteasome inhibitors increase rAAV transduction by decreasing the targeting and/or degradation of ubiquitinated AAV in the proteasome. The net result of such proteasome inhibition would be expected to increase the abundance of ubiquitinated viral capsid and is consistent with our observations.
Modification of the viral ubiquitination state facilitates rAAV transduction. Similar to the polarized human airway cells, rAAV transduction in human primary fibroblasts was also augmented by tripeptide proteasome inhibitors. Human primary fibroblasts were infected with AV.GFP3ori at an moi of 1,000 DNA particles per cell. (a) GFP transgene expression in the absence (left photographs) and presence of 40 μM LLnL (right photographs) at 96 hours after infection. Similar effects were achieved with 4 μM Z-LLL (data not shown). Top and bottom panels represent bright-field and FITC-channel fluorescent photomicrographs, respectively. The mean (± SEM; n = 3) percentage of cells transduced with rAAV, as measured by FACS sorting of GFP expressing cells, is represented in the bar graph of (b). (c) The identification of the ubiquitinated AAV capsid proteins (marked by arrowhead) 6 hours after infection of primary fibroblasts. In this study, rAAV from infected cells was first immunoprecipitated with anti-VP1, 2, and 3 (AAV-2 capsid) mAb followed by Western blot detection of ubiquitin side chains using an antiubiquitin mAb. The two major background bands migrating at approximately 65 and 25 kDa represent heavy and light chain antibody subunits which cross-react with secondary antibodies. Additionally, the equal intensity of lower molecular weight cross-reactive bands (30–40 kDa) serve as internal controls for equal loading of protein. (d) Augmentation of rAAV transduction in polarized airway epithelia by inhibitors of ubiquitin E3 ligase. Epithelia were infected with AV.GFP3ori (10,000 particles per cell) from the basolateral surface after treatment with ubiquitin ligase inhibitor dipeptides (0.2 mM H-Leu-Ala-OH and 0.2 mM H-His-Ala-OH). Results demonstrate the mean (± SEM; n = 3) number of GFP-expressing cells per ×10 field at 1 and 15 days after infection.
Because a technical limitation in our polarized airway model prevented direct detection of ubiquitinated viral capsid, we sought to determine whether modulation of other steps in the ubiquitin-proteasome pathway could also increase rAAV transduction similarly to that seen with proteasome inhibitors LLnL and Z-LLL. Several dipeptides, such as H-Leu-Ala-OH and H-His-Ala-OH, are known to inhibit ubiquitin ligase E3 (37). Consistent with our hypothesis, application of these ubiquitin ligase inhibitors indeed enhanced rAAV transduction from the basolateral surface of human airway cells (Figure 10d). Taken together, data in both fibroblasts and polarized airway epithelia suggest that AAV capsid is ubiquitinated after endocytosis and that this process is a barrier to rAAV transduction. The most plausible mechanism responsible for the augmentation of rAAV transduction by tripeptide proteasome inhibitors involves the prevention of ubiquitinated virus degradation and/or targeting to the proteasome.
Long-term enhancement of rAAV transduction in lung by proteasome inhibitor in vivo. To evaluate the potential utility of proteasome inhibitors for in vivo gene therapy, we tested both the toxicity and efficacy of these compounds for in vivo rAAV-mediated gene transfer in mouse lung. To assess the toxicity of these proteasome inhibitors in mice, we administrated 10-, 100-, and 1,000-fold higher effective doses of LLnL or Z-LLL than used to induce gene transfer in polarized airway cells, using both intratracheal and systemic (intravenous) delivery. No toxicity was indicated by histological evaluation of the lung and liver (data not shown) or was evidenced by the death of animals. To investigate whether these proteasome inhibitors could improve rAAV transduction in vivo, we delivered AV.LacZ (5 × 1010 particles) either alone or in the presence of 400 μM Z-LLL by intranasal administration. Mouse lungs were harvested at 3, 10, and 150 days after infection to evaluate short- and long-term effects of the proteasome inhibitor on rAAV transduction. X-gal staining of the lung tissues at 3 and 10 days after infection demonstrated no detectable transgene expression in either proteasome inhibitor–treated or untreated groups (data not shown). In contrast, significant transduction was achieved at 150 days in Z-LLL–treated samples (Figure 11b). Targeted transgene expression was predominantly confined to the conducting airways, rather than in the parenchyma. Alveolar cells were rarely transduced. Although on average only about 5.88% of airway cells were transduced by AV.LacZ, and LacZ-positive cells were observed throughout the entire conducting airway, a characteristic transduction profile was evident. The transduction efficiency in larger bronchioles (> 350 μm) reached a mean of 10.36 ± 1.63% of the airway epithelium, whereas 1.37 ± 0.41% of airway cells in the smaller bronchioles (< 150 μm) expressed the β-galactosidase transgene (Figure 11c). The range of transgene expression in distal and proximal airways was 0–4% and 5–18%, respectively. This transduction profile demonstrating a higher and more consistent transduction in larger airways likely reflects a more uneven delivery of virus to regions of the lung encompassing the smaller airways. Examination of cryosections from lungs infected by AV.LacZ alone revealed only two LacZ-positive cells in a total of 315 airway sections (n = 3 animals).
Persistent induction of rAAV-mediated gene transfer in mouse conducting airways by proteasome inhibitors. Six-week-old BALB/c mice (n = 3 animals for each group) were infected with 5 × 1010 DNA particles of AV.LacZ in the absence (a) or presence (b) of 400 μM Z-LLL by nasal instillation. Representative examples of histochemical staining for LacZ expression in large bronchioles at 150 days after infection are shown in a and b. The right and left sides of each panel represent Nomarski and bright-field photomicrographs, respectively. The 100-μm scale bar applies to all photomicrographs. The mean (± SEM) percentage of LacZ-expressing epithelial cells at various levels of the airway was quantitated using the morphometric procedures outlined in the methods, and the analysis represents results from three independent animals for each group (c).
Proteasome inhibitors also facilitate rAAV-mediated gene transfer in liver, but not in muscle. It has been previously shown that AAV transduction in certain tissues such as muscle and retina is very efficient (16, 38), whereas other tissues including lung and liver are less favorably transduced (6, 39). Several factors could account for the variations in rAAV organ/tissue tropism. In addition to the abundance and distribution of the AAV receptor and/or coreceptors, differences in intracellular processing of the endocytosed virus might also affect AAV-mediated gene delivery. Results thus far have indicated that inefficient endosomal processing represents a major barrier for gene transfer in airway epithelia and that this limitation can be overcome by transient treatment with proteasome inhibitors. To extend further our findings in lung, we also examined the effect of the proteasome inhibitors on liver and muscle. As shown in Figure 12, rAAV alone resulted in only limited transduction in mouse liver, averaging less than 0.5% (0.49 ± 0.07; n = 3) of hepatocytes expressing the heat-resistant alkaline phosphatase transgene (Figure 12, a and g). In contrast, coadministration of proteasome inhibitors LLnL or Z-LLL with the virus significantly augmented gene transfer by 4 weeks after infection to an average of 5% hepatocytes (Figure 12, b, c, and g). In certain lobes of the liver, the level of AP expression with this vector reached 10% of hepatocytes in the presence of proteasome inhibitors. Despite findings that proteasome inhibitors augmented AAV transduction in both lung and liver, coadministration of proteasome inhibitors did not augment AAV-mediated gene transfer in either skeletal or cardiac muscle. Quantitative analysis of AAV transduction in mouse skeletal muscles (Figure 12, d–f and h) and rat cardiac muscles (data not shown) demonstrated that neither the kinetics nor the level of AAV transduction was affected. Taken together, these results suggest that ubiquitination and endosomal processing barriers to rAAV-2 transduction are likely to be organ specific and could explain the organ-specific tropisms observed with this virus.
Proteasome inhibitors augment rAAV transduction in liver, but not in muscle. The effect of proteasome inhibitors on rAAV-mediated gene transfer in liver was examined by portal vein administration of AV.Alkphos in the presence or absence of proteasome inhibitors. Representative photomicrographs of histochemical staining for AP in liver sections at 4 weeks after infection are shown in a (rAAV alone), b (rAAV+200 μM LLnL), and c (rAAV+ 40 μM Z-LLL). The scale bars in panels a, b, and c are 50 μm. rAAV-mediated gene transfer in mouse anterior tibialis muscle was evaluated at 1 week, 1 month, and 3 months after infection with AV.Alkphos. Representative photomicrographs of histochemical staining for AP in whole muscle cross-sections at 3 months after infection are shown in d (rAAV alone), e (rAAV+200 μM LLnL), and f (rAAV+ 40 μM Z-LLL). The scale bars in d, e, and f are 500 μm. (g) Mean (± SEM) percentage of AP-positive hepatocytes (n = 3 animals for each experimental condition, averaged for all six lobes in each animal). (h) Mean (± SEM) percentage of AP-positive myofibers at different times after infection (n = 4 muscle samples for each experimental condition).