Role of Integrin Αvß3 in the Production of Recombinant Adenoviruses In

Role of integrin ␣V␤3 in the production of recombinant adenoviruses in HEK-293 cells

WLW Ling1, RL Longley1, DL Brassard1, L Armstrong2,3 and EJ Schaefer1,4 1Department of Biotechnology Development, Schering-Plough Research Institute, Union, NJ, USA; and 2Department of Tumor Biology, Schering-Plough Research Institute, Kenilworth, NJ, USA

␣ ␤ Adenoviral infection is initiated by attachment of adenoviral 293 cells with integrin V 3 cDNAs increased rAd pro- fiber proteins to the CAR protein and subsequent internaliz- duction. Subjecting cells to a sudden reduction in serum ␣ ␣ ␤ ␣ ␤ ation aided by V -containing integrins, eg V 3 and V 5. (10% to 0.1% FCS) for 5 days, clone D cells maintained To further understand the process of infection and assembly 80% viability compared with 40% for HEK-293 cells. Further of recombinant adenoviral (rAd) vectors, we examined rAd indication of survival signaling involvement was provided by production in HEK-293 cells and one of its subclones, clone Western blot analysis demonstrating p38 and p44/42 D, isolated from the parental cells for high viral production. MAPKs were constitutively phosphorylated in HEK-293 cells. ␣ ␤ By flow cytometry, surface expression of integrin V 3 by However, for clone D cells, p38 MAPK was phosphorylated clone D cells was two-fold higher than by HEK-293 cells. only after rAd infection. The role of survival signaling ␣ ␤ However, clone D cells did not demonstrate greater trans- mediated by integrin V 3 in rAd production will be dis- lational efficiency or number of viral genome DNA copies cussed. shortly after rAd infection. Treating cells with inhibitors of Gene Therapy (2002) 9, 907–914. doi:10.1038/sj.gt.3301726 ␣ ␤ integrin V 3 reduced rAd production and transfecting HEK-

␣ ␤ Keywords: V 3; adenovirus; echistatin; HEK-293; integrin; RGD; survival

Introduction with a focus on the targeting of adenoviral vectors for gene therapy or protein expression. However, regulation The increasing use of recombinant adenoviral vectors for of post-internalization events such as adenoviral repli- gene therapy has focused more attention on ways to max- cation and assembly are poorly characterized, specifically imize volumetric production, as well as product quality. for HEK-293 cells. Complicating any studies such as these However, little is currently understood about the regu- is the effect of transgene expression on cellular metab- lation and control of adenoviral production in the most olism. In this work, we used a HEK-293 subclone, clone commonly used packaging cell line: human embryonic D, to investigate post-infection regulation of adenovirus kidney 293 (HEK-293) cells. They are the most readily production, and two recombinant adenoviruses, rAd- available cell line containing the complementary Ad5 E1 GFP and rAd-p21, to evaluate various aspects of protein region which is deleted in most replication-deficient expression. Clone D was selected originally for high virus adenoviral vectors. Recent work has identified the major production and was chosen in order to better elucidate cellular components involved in adenoviral infection and the mechanism which resulted in high virus yields. 1 entry. The primary receptor for the attachment of aden- Recombinant Ad-GFP is utilized for its ease in expression ovirus and subsequent infection has been identified as detection by FACS analysis, and the expression of 2 the coxsackie-adenovirus receptor protein (CAR). How- reporter gene GFP has been used routinely for monitor- ever, the CAR protein alone is not sufficient for efficient ing gene transfer and shows no interference with normal ␣ ␤ ␣ ␤ infection. The need for integrins V 3 or V 5 for effective cell function.5 Much of the work was done with rAd-p21 internalization of the attached virus has been well dem- which carries the p21WAF1/CIP1 transgene, a cell cycle 3 ␣ onstrated. Recently, the role of the V integrins in aden- regulator which represents a source of significant stress 4 oviral infection and gene delivery has been reviewed on the cell. Titers of recombinant virus expressing p21 are typically very low in the parental HEK-293 cell line. We had initially screened subclones of HEK-293 for Correspondence: WLW Ling, Department of Biotechnology Development, enhanced production of recombinant adenovirus when Schering-Plough Research Institute, 1011 Morris Avenue, Union, NJ, compared with the parental pool. One of the overproduc- 07083, USA ing subclones selected, clone D, was of interest because 3 Current addresses: Aventis Pharmaceuticals, Medical Affairs Oncology, these cells had an elevated expression of integrin ␣ ␤ PO Box 6977, 300 Somerset Corporate Blvd, Bridgewater, NJ, 08807, V 3 4 protein when analyzed by flow cytometry. Normally, USA; Bristol-Myers Squibb Company, Biopharmaceutical Science and ␣ ␤ ␣ ␤ HEK-293 express low levels of V 3 or V 5 integrins Technology, PO Box 5400, Princeton, NJ 08543, USA ␣ ␤ Received 9 August 2001; revised 20 February 2002; accepted while expressing integrin V 1 which can function as a 9 March 2002 vitronectin receptor6 and presumably also aids in the Role of intrgrin in rAd production WLW Ling et al 908 internalization of adenovirus. Thus, comparison of clone With the current understanding of adenoviral infection D cells, with their elevated expression of the co-receptor of cells,1 we initially hypothesized that the difference in ␣ ␤ for adenovirus entry – integrin V 3, with the parental rAd titer may be a result of better viral attachment pool allowed us to assess if increased adenoviral infection and/or entry into clone D cells. Therefore, we analyzed was occurring and what further role it had in increasing the endogenous expression levels of CAR and the key ␣ ␤ ␣ ␤ recombinant viral production. An alternative hypothesis integrins V 3 and V 5 by flow cytometry in the two cell was suggested recently by work demonstrating that the types (Figure 1b). We found that the levels of all three ␣ ␤ integrin V 3-mediated survival signaling pathway proteins are more highly expressed in clone D than in the (normally observed in endothelial cells) was active in HEK-293 cells, with expression of CAR in clone D at 1.4- ␣ ␤ ␣ ␤ ␣ ␤ integrin V 3-transfected HEK-293 cells (an epithelial cell fold, integrin V 3 at 2.2-fold and integrin V 5 at 1.1- 7 ␣ ␤ line). In this case, the proposed role of integrin V 3 fold that of HEK-293. It is noteworthy and consistent with would be to prolong adenoviral production via survival the results of Bodary and McLean,6 that cells from our ␣ ␤ signal transduction during infection. HEK-293 and clone D possess integrins V and 1 and the relative levels were comparable to each other (data ␣ not shown) suggesting that integrin V may be preferen- Results ␣ ␤ tially expressed in the complex of integrin V 3 than in ␣ ␤ Adenoviral productivity and surface expression of integrin V 5 in clone D cells. In addition, immunostain- ␣ ␤ integrins and CAR ing of HEK-293 and clone D showed surface integrin V 3 HEK-293 cells obtained from ATCC, the most common to be highly concentrated at focal contacts in clone D cell line used for producing recombinant adenoviral vec- cells, but diffused over the cell surface in HEK-293 cells tors, have no record of ever being subcloned, and thus, (data not shown). represent an uncloned pool of cells containing a hetero- To further assess whether it was the high expression ␣ ␤ geneous population with varying physiological charac- of integrin V 3 that was responsible for the increase in teristics. We have observed heterogeneity in cell shape rAd production, we transfected the parental HEK-293 ␣ ␤ and size during growth on tissue culture plasticware, cells with cDNAs carrying the genes for integrins V 3, ␣ ␤ which can be exacerbated by cultivation and passaging integrin V 5, or CAR. The rAd titers were determined conditions (data not shown). To identify a cell line with by HPLC (Figure 1c). When HEK-293 cells were trans- ␣ ␤ enhanced recombinant adenovirus (rAd) productivity, fected with equimolar amounts of V and 3 cDNAs, an clone D was subcloned from the parental HEK-293 cell increase of approximately 2.5-fold in rAd titer over the pool. Figure 1a shows the results comparing clone D with mock-transfected control was detected. However, trans- ␣ ␤ the parental HEK-293 control cells infected with rAd pos- fection with equimolar of V and 5 cDNAs or with CAR sessing identical backbones but expressing different cDNA resulted in a lower titer than the mock-transfected transgenes: green fluorescent protein (GFP) and p21. controls. Wickham et al3 have indicated that adenovirus Adenoviral titers from clone D were consistently greater attachment is primarily determined by levels of the CAR than those achieved by the parental cells independent of protein with integrins functioning primarily in post-CAR the transgene. However, the relative improvement in titer attachment. In this case, the HEK-293 parental cells trans- for clone D compared with the parental cell line tended fected with CAR cDNA may actually promote rAd entry to vary with the transgene being expressed. Notably, a to a greater degree than the mock-transfected control. modest increase (two- to three-fold) was seen with GFP However, high CAR expression may act a potent growth which generally has a minimal effect on cell physiology, inhibitor resulting in cell cycle arrest and/or apoptosis.8 whereas a greater impact was observed for p21 which The observed decrease of rAd titer in HEK-293 cells ␣ ␤ has a significant impact on cell cycle progression. In these transfected with integrin V 5 cDNAs relative to its ␣ ␤ cases, the actual rAd titers of the parental HEK-293 cells mock-transfected control suggests that integrin V 5 were significantly reduced while the rAd titer from clone needs to be expressed within a limited range for optimal D was unaffected over time (data not shown). rAd production. The results from these transient

␣ ␤ ␣ ␤ Figure 1 Recombinant adenoviral productivity and surface expression of integrins V 3, V 5 and CAR of parental HEK-293 and clone D cells. (a) rAd productivity of HEK-293 and clone D cells. Cells were grown in duplicate T25 flasks to near confluence, and incubated with adenovirus expressing indicated transgenes for 24 h. Spent media were replaced with fresh media on day 1 after infection. Cells were harvested on days 2 and 3 after infection, ␣ ␤ ␣ ␤ and adenoviral titer was quantified by HPLC. (b) Cell surface FACS analysis for integrins V 3, V 5 and CAR. Data presented as the ratio of median fluorescence for clone D cells compared with the median for HEK-293 cells. Each bar represents the average of at least two independent experiments. ␣ ␤ ␣ ␤ (c) Recombinant adenoviral-GFP production of HEK-293 transfected with cDNA for integrins V 3, V 5 or CAR. Each bar represents the average titer of two independent transfections and is normalized to mock-transfected HEK-293 controls.

Gene Therapy Role of intrgrin in rAd production WLW Ling et al 909 expression experiments indicated a significant impact on duction was negatively affected for both clone D cells and ␣ ␤ rAd production by high integrin V 3 expression. the parental HEK-293 cells (Figure 2a). The effect on total adenoviral titer was more dramatic for clone D cells since Inhibition of adenoviral productivity by interfering with it produced a three-fold higher level of adenovirus than integrin functions the parental cells in the absence of RGD peptide. Note The interaction between adenovirus and the integrin that the concentrations of RGD tested which were effec- receptors in viral internalization is mediated by the tive in reducing rAd production were in the range of amino acid sequence Arg-Gly-Asp (RGD) on the adenovi- approximately 30–60 ␮M. This concentration is lower ral penton base.3 Blocking this interaction using synthetic than concentrations previously reported to block aden- RGD peptides inhibits adenoviral internalization.3 To ovirus internalization.3 When rAd titers were normalized further confirm the involvement of integrins in rAd pro- to untreated controls, the degree of inhibition was similar duction, we examined the effect of these integrin antag- for both cell lines with clone D requiring higher concen- onists on rAd production (Figure 2). When a cyclic RGD trations of cRGD to obtain the same inhibitory effect on peptide similar to the ones used by Wickham et al3 to rAd titer as observed in HEK-293 cells (Figure 2b). The block adenovirus internalization was included during non-specific, inactive, scrambled RGE peptide did not our entire adenoviral production scheme, adenoviral pro- have an effect on rAd production in either cell type. While the use of cRGD does not correlate the specific ␣ ␤ involvement of integrin V 3, it nevertheless, indicates that surface expression of integrin(s) may have a role in regulating rAd production. ␣ ␤ To further examine the role of integrin V 3, we treated the two cell types with the natural disintegrin, echistatin. Natural disintegrins have been reported to bind with high specificity to integrins via an RGD-motif.9 The effects most frequently reported include inhibition of cell adhesion,10 platelet aggregation,11 sperm-oolemmal adhesion,12 osteoclast function13 and FAK and paxillin phosphorylation.14 When echistatin was incubated with HEK-293 and clone D cells during rAd-p21 infection, rAd production was inhibited (Figure 2c). The results for each cell line were normalized to the maximum titer achieved in the absence of inhibitor for that experiment. Similar to the results observed with RGD, recombinant adenovirus production in HEK-293 cells was inhibited at a slightly lower concentration of echistatin compared with clone D cells. However, in HEK-293 cells the inhibitory effect of echistatin on rAd production was observed until it hit a plateau at approximately 50% of maximum production. While in clone D cells, inhibition of rAd production con- tinued with increasing concentrations of echistatin. These ␣ ␤ results demonstrate that antagonists of integrin V 3 function effectively interfered with adenoviral productivity. In fact, consistent with their increased speci- ␣ ␤ ficity for V 3 integrin, echistatin inhibited recombinant adenoviral production at concentrations about 1000-fold lower than that which we observed for RGD peptide: 10– 50 nM compared with 10–50 ␮M.

Effects of varying rAd contact time To understand how integrins may affect rAd productivity in the HEK-293 cells, we speculated that integrins may function during one or more intracellular events. These could include any combination of the fol- lowing mechanisms: (1) integrins facilitate adenoviral entry which leads to better infectivity and productivity; Figure 2 rAd-p21 production in the presence of integrin antagonists. Par- (2) integrins mediate signals to improve protein trans- ental HEK-293 and clone D cells were seeded in triplicates and grown on six-well plates, infected with rAd-p21 in the presence of varying concen- lational efficiency which may enhance rAd productivity; trations of the integrin inhibitor, cyclic RGD (cRGD) or non-specific (3) integrins promote adenoviral assembly; (4) integrins scrambled RGE (sRGE) peptides. Adenoviral titers quantified by HPLC prolong cell survival through various signaling pathways were normalized to those from either the untreated HEK-293 controls (a) to allow an increased number of assembled adenoviruses. or its own untreated controls (b). (c) Inhibition of rAd-p21 production by When viral genome copies were measured by real-time echistatin. Parental HEK-293 and clone D cells were grown in duplicate quantitative PCR after 1.5 and 3 h of rAd-p21 incubation, T25 flasks, and infected with rAd-p21 in the presence of varying concentrations of the disintegrin, echistatin. Cells were harvested on day 2 post both parental HEK-293 and clone D cells possessed com- infection and adenoviral titer was determined by HPLC. Each point rep- parable viral copy number per cell when the standard resents the average of either the triplicate wells or duplicate flasks. deviations were taken into account: 4.6 ± 2.1 for HEK-293

Gene Therapy Role of intrgrin in rAd production WLW Ling et al 910 Table 1 Determination of viral genome copies as a function of varying viral contact timea

rAd contact time Viral genome copies per cell

1.5 h 3 h

HEK-293 4.6 ± 2.1 9.9 ± 4.8 Clone D 6 ± 2.1 3.8 ± 1.4

a HEK-293 and clone D cells were seeded into triplicate wells of six-well plates, infected with rAd-p21, and harvested at 1.5 h and 3 h. DNA was extracted from the cell pellets and copy numbers of the adenoviral genomic DNA were measured by quantitative real- time PCR speciﬁc for the viral p21 transgene. Data are expressed as the average viral copy per cells ± s.d.

and 6 ± 2.1 for clone D at 1.5 h, and 9.9 ± 4.8 for HEK- 293 and 3.8 ± 1.4 for clone D at 3 h (Table 1). This result suggests that the enhancement in rAd production observed for clone D is not due to higher viral particles present in the cell during the initial infection state. To investigate the possibility that high surface ␣ ␤ expression of integrin V 3 may stimulate transgene expression resulting in increased rAd production, we determined GFP expression by flow cytometry when HEK-293 and clone D cells were incubated with a saturating concentration of rAd-GFP for 0.5 to 24 h (Figure 3). At the indicated times, cells were washed to remove loosely associated adenoviruses before replacing with fresh medium, and all samples of cells were harvested 24 h after viral exposure and quantified for GFP expression by FACS analysis. The fraction of cells expressing GFP reached a maximum after approximately 3 h of adenovi- Figure 3 Effects of varying contact time during adenoviral infection. Par- ral exposure (Figure 3a). We observed little difference ental HEK-293 and clone D cells were grown in five replicates of T25 between the cell lines in both the percent GFP-expressing flasks to confluency and exposed to a saturating concentration (MOI of cells (Figure 3a) and the mean fluorescence measured 80) of rAd-GFP for 0.5 to 24 h. At the indicated times, cells were washed to remove cell surface associated adenovirus, and medium was replaced. (Figure 3b) after 24 h. Assuming equal efficiency of tran- (a) On day 1 after infection, one T25 flask was harvested and the % of scription of the heterologous GFP transgene and GFP cells expressing GFP was quantified by FACS analysis. (b) The mean stability between the two cell types, these data indicated fluorescence intensity of GFP expression was determined for the same that GFP expression over time was also equivalent. samples analyzed in (a). (c) rAd-GFP viral titers of samples from HEK- Samples of cells infected for these times were harvested 293 and clone D cells harvested on days 2 and 3 after infection were at both 2 and 3 days after infection and adenoviral titers determined by HPLC. Each point represents the average of two T25 flasks. were determined by HPLC. Consistent with the GFP expression data at 24 h obtained by FACS analysis, rAd mediated inhibition of cell cycle progression. Thus, it is titers did not significantly improve with increased viral possible that in addition to assisting in adenoviral entry, ␣ ␤ exposure time after 3 h. Also, the relative titers obtained integrin V 3 functions in clone D through these sig- for the HEK-293 parental cells and clone D cells were naling pathways to reduce the stress and cytotoxic effects consistent with those presented in Figure 1a with the rAd of infection imposed on the cell. To examine this hypoth- titer produced by clone D cells being approximately esis, we subjected HEK-293 and clone D cells to serum three-fold higher than that produced by the parental withdrawal and monitored cell death (Figure 4a) in an HEK-293 cells (Figure 3c). These data taken together indi- experiment similar to the study demonstrated by Brass- cated that while clone D cells gave higher rAd titers than ard et al7 in a stably transfected HEK-293 cell line over- ␣ ␤ the parental HEK-293 cells, the enhanced productivity expressing integrin V 3. Both clone D cells and the par- observed for clone D did not appear to be due to a more ental HEK-293 cells were seeded on to vitronectin-coated efficient protein translation of the transgene in these cells. plates. For one set of plates the serum content of the These data, however, do not rule out the presence of a medium was reduced from 10% to 0.1%. Dead cells, more efficient viral assembly process in clone D cells. defined by their permeability to propidium iodide, were measured after 3 and 5 days of growth. The results after ␣ ␤ Survival signaling mediated by integrin V 3 3 days suggested a trend that became more prominent by Aside from its role as a co-receptor for adenovirus intern- 5 days: when subjected to the stress of serum withdrawal, ␣ ␤ alization, integrin V 3 activation has been shown to pro- clone D cells had a distinct advantage in survival over mote cell survival15 and protect cells from apoptosis.16 the HEK-293 parental cells. These results were confirmed One mechanism that has been proposed by Stromblad et by an alternative assessment of apoptotic cells defined by al17 for this survival signaling is to suppress p53 and p21- measuring annexin V staining (data not shown).

Gene Therapy Role of intrgrin in rAd production WLW Ling et al 911

Figure 4 Involvement of survival signaling in rAd production. (a) Survival in low serum. Parental HEK-293 and clone D cells were seeded on to vitronectin-coated plates. For one set of plates, the serum content of the medium was reduced from 10% to 0.1%. Dead cells, deﬁned by their permeability to propidium iodide, were measured after 3 and 5 days of growth by FACS analysis. Each bar represents the average of at least two independent samples. (b–f) Activation of p44/42 and p38 MAPK, and expression of p21. HEK-293 and clone D were plated on to 60-mm dishes and were either non-infected (lanes 1 and 4) or incubated for indicated times with rAd-p21 (lanes 2, 3, 5 and 6). Equal amounts of lysates were prepared for SDS-PAGE and Western blotting with antibodies speciﬁc for phospho-p44/42 MAPK (b), p44/42 MAPK (c), phospho-p38 MAPK (d), p38 MAPK (e) and p21 (f).

In addition, uninfected, and 0.5 h and 24 h post- the two cell types. These results taken together indicate infected lysates from HEK-293 and clone D cells were that the increase in rAd titer observed in clone D cells ␣ ␤ prepared to identify expression differences in proteins, involves an integrin V 3 mediated signaling pathway such as p44/42 and p38 MAPKs, which are involved in that permits sustained cell survival during viral infection cell survival and integrin-mediated signaling (Figure 4b– and allows higher viral productivity. f). The p44/42 MAPK, which mediates integrin signaling and regulates cell growth and differentiation, is generally Discussion considered to be a pro-survival mediator.18,19 In contrast, p38 MAPK, activated by stress and cytokines, is a pro- In the production of adenoviral vectors for gene therapy, apoptotic molecule.20,21 In our experiments, we found the consistently high specific productivity is important for phosphorylation state of p44/42 MAPK at steady state ensuring product quality. Yet the production of recombi- for the two cell types were vastly different with the par- nant adenoviruses in HEK-293 cells is a complex process ental HEK-293 cell being constitutively active regardless that is poorly understood in terms of its regulation and of viral infection (Figure 4b). Since HEK-293 cells are control. In this study, we provided an initial characteriz- more susceptible to serum deprivation than clone D cells, ation of a subclone of HEK-293, clone D, which demon- it may be possible that high endogenous phospho-p44/42 strated high adenoviral productivity. We found that MAPK signaling needs to be sustained in order to main- enhanced viral titer in clone D did not necessarily corre- tain normal cell function. However, the absence of the spond to an increased number of viral particles internal- phospho-p44/42 MAPK in clone D lysate in the times ized by the cells during the initial infection. Viral pro- examined may suggest cell survival in these cells does duction by HEK-293 cells was reduced by incubation not depend on p44/42 MAPK signaling. Alternatively, with integrin antagonists, but elevated by transfecting ␣ ␤ phosphorylation of p44/42 MAPK of clone D may occur with integrin V 3 cDNAs. In particular, we demon- after 24 h post-infection. The phosphorylated form of p38 strated that for this subclone, increased adenovirus pro- ␣ ␤ MAPK was detected at 24 h post-infection in clone D duction seemed to be linked to integrin V 3-mediated cells, but was not present at the non-infected or 0.5 h survival signaling events. post-infected states (Figure 4d). However, p38 MAPK In addition to its physical roles in cell adhesion and ␣ ␤ was constitutively active in HEK-293 cells even under adenoviral internalization, integrin V 3 has been linked non-infected condition further suggests that without suf- with a variety of signal transduction pathways,22 includ- ficient integrin signaling, HEK-293 may need the continu- ing PI3K,23 ERK/MAPK,24 CAMP,25 and NF-␬B.26 How- ␣ ␤ ous activation of p44/42 and p38 MAPK to sustain cell ever, the relative importance of the role of integrin V 3 survival. in triggering these signaling pathways compared with its While activity differences were detected for p44/42 role in adenoviral internalization needs to be explored and p38 MAPKs between HEK-293 and clone D cells, further with respect to adenovirus production. The rela- both cell types expressed the transgene, p21, equally at tive importance of the signaling pathways may be differ- 24 h after infection suggesting that protein synthesis was ent for the sequential stages of adenovirus attachment, not affected (Figure 4f). This is consistent with the data internalization, replication and assembly. For example, Li obtained by FACS analysis of cells infected with rAd-GFP et al23 suggest that integrin signaling through the PI3K (Figure 3a, b) where the number of GFP-positive cells and pathway is important for the step of adenovirus endo- the mean GFP fluorescence level are comparable between cytosis, but not signaling through the MAPK pathway.

Gene Therapy Role of intrgrin in rAd production WLW Ling et al 912 Moreover, Bruder et al27 have shown that adenovirus recombinant adenoviruses, such as rAd-p21 (SCH infection will stimulate Raf/MAPK signaling. Overall, 412499) and rAd-GFP, were generated as per the method ␣ ␤ 30 although integrin V 3 acts as one of the receptors for described previously for rAd-p53 (SCH 58500). adenoviral entry and internalization, our data suggest that it may also act via one or more signal transduction Cell culture and subcloning of HEK-293 cells pathways important to the regulation of cellular homeo- HEK-293 cells were routinely cultured at 37°C and 5% stasis and thus, impact adenoviral production. Our data CO2 in DMEM medium containing 10% calf serum and suggest that it may be worthwhile to explore this 4 mM glutamine. Subclones were isolated from HEK-293 relationship further. by limiting dilution and selected for further characteriz- Infection by adenovirus represents a source of stress ation based on results from plaque-forming assays and for the infected cells. For example, Brand et al28 demon- adenoviral titer measurements based on HPLC quantifi- strated the induction of apoptosis was caused by infec- cation of total viral particles.31 tion of various cell lines with a replication deficient adenovirus. In that work, the authors also showed that UV- Adenoviral infection and cell harvest inactivation of the adenovirus did not significantly Cells were grown to 85–95% confluence, infected with reduce post-infection apoptosis. It is very plausible that fresh medium containing adenovirus (at MOI of 100 and the production of recombinant adenoviruses, along with at approximately 1 × 108 adenoviral particles/ml) for the any concomitant transgene expression, may represent an indicated times. Medium was replaced with fresh overly stressful burden for the HEK-293 cells. Con- medium 24 h after initial adenoviral exposure. Cell lys- versely, infection by recombinant adenoviruses has also ates were harvested on days 2 and 3 after infection. For been shown to promote the survival of endothelial cells.29 experiments with integrin antagonists, inhibitors were ␣ ␤ In our system, the high integrin V 3 expressing clone included in medium containing the adenovirus, as well D cells maintained cell viability under low serum con- as during the medium exchange at 24 h after the adenovi- ditions, while HEK-293 cells showed a decrease in cell ral exposure. Cell lysates were harvested on days 2 and viability (Figure 4a). To investigate the potential presence 3 as indicated. of an intrinsic survival capacity in clone D cells, we For adenoviral titer determination, infected cell lysates explored the phosphorylation status of the p44/42 and were generated by three cycles of freeze/thaw, clarified p38 MAPKs. In fact, the phosphorylation status of p44/42 by centrifugation, and analyzed by HPLC30 (Amersham and p38 MAPKs were vastly different between the two Pharmacia, Uppsala, Sweden). The adenoviral titer for cell types at both non-infected and infected states. Unlike each lysate was compared with the elution profile and the clone D cells, the constitutively phosphorylated concentration of adenovirus standard. p44/42 MAPK observed in HEK-293 cells indicates the For %GFP-positive cells and mean GFP fluorescence, phospho-p44/42 MAPK state needs to be maintained for adenoviral infected cells were washed twice with PBS, cell survival. The exact time course of p44/42 MAPK acti- and scraped into cold PBS. Cell pellets were obtained by vation in clone D cells upon viral exposure needs to be centrifugation and washed once more with PBS. To deter- further explored. Similar to p44/42 MAPK, p38 MAPK mine %GFP-positive cells and mean fluorescence, is also constitutively phosphorylated in HEK-293 cells at washed cells were resuspended in PBS and fluorescence steady state, but not phosphorylated in clone D until after was measured by flow cytometry (FACS Calibur, Becton rAd exposure. The presence of phospho-p38 MAPK in Dickinson, Bedford, MA, USA). the non-infected state indicates that the HEK-293 cells are For rAd-p21 genomic DNA copy number determi- under stress during normal growth. However, the acti- nations, cells were seeded in triplicate wells of six-well vation of this stress signal by clone D cells was not plates with 1.5 ml of medium. Once cell confluence was detected until viral infection. Data presented here dem- reached, cells were kept on ice and infected with rAd at onstrate that: (1) an increase in the basal rAd productivity MOI of 100. Cells were warmed at 37°C for indicated of HEK-293 cells is associated with increased integrin times before harvest. Adenoviral infected cells were ␣ ␤ V 3 expression, achieved either by its presence in a washed twice with ice-cold PBS, and scraped into cold stable subcloned cell line or by transient transfection; and PBS. Cell pellets were obtained by centrifugation at 4°C ␣ ␤ (2) integrin V 3-mediated survival signaling represents and washed once more with ice-cold PBS. To quantify a way of prolonging cell homeostasis or delaying stress copy number per cell, DNA was extracted from the pel- signals, and thus, allowing more adenovirus production lets and analyzed by real-time quantitative PCR (7700 by the infected cell. Sequence Detection System; Applied Biosystems, Foster City, CA, USA) using a primer pair specific for the virally encoded p21 transgene. Materials and methods Transfections of HEK293 Reagents HEK-293 cells were seeded into T-25 culture flasks and Synthetic cyclic RGD (G(Pen)GRGDSPCA peptide, MW grown to 80% confluence over a 3-day period. The cells = 948) and scrambled RGE (GRGESP peptide, MW = 602) were then transfected using 18 ␮l of Mirus TransIT-293 peptides were purchased from Life Tech (Grand Island, reagent (PanVera Corporation, Madison, WI, USA) and 2 NY, USA), integrin antibodies from Chemicon (Temecula, ␮g total DNA, according to the manufacturer’s instruc- CA, USA), echistatin from Bachem Bioscience (King of tions. Transfected constructs included pcDNA3.1 contain- ␣ Prussia, PA, USA), antibodies specific for phospho- ing full length human CAR cDNA, human v integrin ␤ ␤ ␣ ␤ p44/p42 MAPK, p44/42 MAPK, phospho-p38 MAPK cDNA, human 3 cDNA or human 5 cDNA; v and 3 ␣ ␤ and p38 MAPK from Cell Signaling (Beverly, MA, USA), integrin constructs, and v and 5 integrin constructs anti-p21 from Pharmingen (San Diego, CA, USA). All were transfected in 1:1 ratio. Empty pcDNA3.1 vector

Gene Therapy Role of intrgrin in rAd production WLW Ling et al 913 was transfected as a control. Cells were incubated in the Acknowledgements transfection medium for 24 h, after which the medium was removed and the cells allowed to recover for 8 h in We would like to thank Dr Mary Ann Fritz for the HPLC normal growth medium. The cells were then infected at analysis, Dr Xiao-Yan Cai for the quantitative PCR analy- an MOI of 80 with rAd-GFP virus for 24 h. The virus was sis. Integrin cDNAs were generous gift of Dr Chandra removed and HPLC samples taken 3 days after infection. Kumar and CAR cDNA was a generous gift of Dr Monica Zepeda and Dr William Demers. Anti-CAR antibody was a generous gift of Dr Barry Sugarman. Analysis of integrin and CAR expression by flow cytometry Cells (HEK-293 or clone D), grown in T-flasks to 85–95% References × confluency, were detached with 0.5 mM EDTA in 1 PBS, 1 Bergelson JM. Receptors mediating adenovirus attachment and washed twice with cold PBS and pelleted by centrifug- internalization. Biochem Pharmacol 1999; 57: 975–979. ation. Cells were resuspended at 1 × 106 cells/ml in block- 2 Bergelson JM et al. Isolation of a common receptor for coxsackie ing buffer (0.5% BSA, 1% calf serum, 1× PBS) with anti- B viruses and adenoviruses 2 and 5. Science 1997; 275: 1320–1323. integrin (Pharmingen) or anti-CAR antibodies and incu- 3 Wickham TJ, Mathias P, Cheresh DA, Nemerow GR. Integrins ␣ ␤ ␣ ␤ bated for 2 h at 4°C to minimize internalization. At the V 3 and V 5 promote adenovirus internalization but not virus attachment. Cell 1993; 73: 309–319. end of the incubation, cells were centrifuged and washed ␣ with blocking buffer twice. Cells were resuspended again 4 Nemerow GR, Stewart P. Role of V integrins in adenovirus cell entry and gene delivery. Microbiol Mol Biol Rev 1999; 63: 725–734. at 1 × 106 cells/ml in blocking buffer containing second- 5 Welch S, Kay SA. Reporter gene expression for monitoring gene ary antibodies (PE-conjugated anti-mouse antibody; Mol- transfer. Curr Opin Biotechnol 1997; 8: 617–622. ecular Probes, Eugene, OR, USA) and incubated for ␤ 6 Bodary SC, McLean JW. The integrin 1 subunit associates with ° ␣ another 1 h at 4 C. Cells were then washed twice with the vitronectin receptor V subunit to form a novel vitronectin PBS and centrifuged. The cell pellet was resuspended in receptor in a human embryonic kidney cell line. J Biol Chem 3.5% p-formaldehyde in PBS for 20 min, followed by two 1990; 265: 5938–5941. ␣ ␤ washes with PBS, and resuspended in PBS. The level of 7 Brassard DL et al. Integrin V 3-mediated activation of surface integrin expression was analyzed by flow cyto- apoptosis. Exp Cell Res 1999; 251:33–45. metry (FACS-Calibur, Becton Dickinson). 8 Okegawa T et al. The mechanism of the growth-inhibitory effect of coxsackie and adenovirus receptor (CAR) on human bladder cancer: a functional analysis of CAR protein structure. Cancer Western analysis Res 2001; 61: 6592–6600. Non-infected and infected cells were washed twice and 9 Kumar CC et al. Biochemical characterization of the binding of ␣ ␤ scraped with ice-cold PBS containing 1 mM Na3VO4, and echistatin to integrin V 3 receptor. J Pharmacol Exp Ther 1997; cell number of each sample was determined. Cells were 283: 843–853. pelleted at 1000 g for 5 min at 4°C, lyzed on ice with lysis 10 Spalletti-Cernia D et al. Echistatin inhibits Lewis lung carcinoma buffer (1% Na deoxycholate, 1% NP-40, 2 mM PMSF, 1 cell-matrix adhesion in vitro and experimental metastasis in vivo. Int J Oncol 1997; 11: 757–763. mM Na3VO4), and lysates were by centrifuged at 14 000 syk FAK ° 11 Staiano N et al. Echistatin inhibits pp72 and pp125 phos- g for 10 min at 4 C. SDS-sample buffer was added to each phorylation in fibrinogen-adherent platelets. Biochimie 1997; 79: of the clarified lysates. Samples were analyzed by SDS- 769–773. PAGE, transferred on to PVDF membranes and probed 12 Bronson RA, Galit J, Bronson S, Oula L. Echistatin, a disintegrin, with indicated antibodies followed by chemilumi- inhibits sperm-oolemmal adhesion but not oocyte penetration. nescence detection with ECL-Plus (Amersham Ferti Steril 1995; 64: 414–420. Pharmacia). 13 Vanderpluijm G et al. Integrins and osteoclastic resorption in 3 bone organ cultures – differential sensitivity to synthetic arg- gly-asp peptides during osteoclast formation. J Bone Min Res Determination of apoptosis and cell death 1994; 9: 1021–1028. To examine differences in susceptibility to apoptosis, 14 Della Morte R et al. Echistatin inhibits pp125FAK autophos- HEK-293 and clone D were serum-starved using a pro- phorylation, paxillin phosphorylation, and pp125FAK-paxillin cedure similar to one previously described.7 Briefly, 35 interaction in fibronectin-adherent melanoma cells. Eur J mm culture dishes (Becton Dickinson) were coated with Biochem 2000; 267: 5047–5054. 40 ␮g of human vitronectin (Life Tech) in serum-free 15 Howe A, Aplin AE, Alahari SK, Juliano RL. Integrin signaling DMEM for 2 h at room temperature and then washed and cell growth control. Curr Opin Cell Biol 1998; 10: 220–231. 16 Malik RK. Regulation of apoptosis by integrin receptors. J Ped three times with serum-free medium. Cells were tryp- Hematol/Oncol 1997; 19: 541–545. × 4 sinized, seeded into dishes at a density of 3 10 17 Stromblad S et al. Suppression of p53 activity and p21WAF1/CIP1 cells/cm2 in normal growth medium, and allowed to ␣ ␤ expression by vascular cell integrin v 3 during angiogenesis. J attach and spread for 4 h at 37°C. The cells were then Clin Invest 1996; 98: 426–433. washed twice with serum-free medium and DMEM 18 Dent P et al. The roles of signaling by the p42/44 mitogenic- medium supplemented with 0.1% FCS added. Cultures activated protein (MAP) kinase pathway; a potential route to were allowed to grow for either 3 or 5 days. The samples radio- and chemo-sensitization of tumor cells resulting in the were then trypsinized and stained either for Annexin V induction of apoptosis and loss of clonogenicity. Leukemia 1998; or with propidium iodide using the Vybrant Apoptosis 12: 1843–1850. Assay Kit 2 (Molecular Probes, Eugene, OR, USA) accord- 19 Bonni A et al. Cell survival promoted by the Ras-MAPK signaling pathway by transcription-dependent and -independent ing to the manufacturer’s instructions. The total apoptotic mechanisms. Science 1999; 286: 1358–1362. (Annexin V-positive) and dead (propidium iodide- 20 Berra E, Diaz-meco MT, Moscat J. The activation of p38 and positive) cells were estimated by FACS measurements. apoptosis by the inhibition of Erk is antagonized by the phos- Cells grown on vitronectin-coated dishes in normal phoinositide 3-kinase/Akt pathway. J Biol Chem 1998; 273: growth medium were used as controls. 10792–10797.

Gene Therapy Role of intrgrin in rAd production WLW Ling et al 914 21 Birkenkamp KU et al. A dual function of p38 MAP kinase in 27 Bruder JT, Kovesdi I. Adenovirus infection stimulates hemopoietic cells: involvement in apoptosis and cell activation. Raf/MAPK signaling pathway and induces interleukin-8 Leukemia 1999; 13: 1038–1045. expression. J Virol 1997; 71: 398–404. 22 Coppolino MG, Dedhar S. Bi-directional signal transduction by 28 Brand K et al. Induction of apoptosis and G2/M arrest by infec- integrin receptors. Int J Biochem Cell Biol 2000; 32: 171–188. tion with replication-deﬁcient adenovirus at high multiplicity of 23 Li E et al. Adenovirus endocytosis via ␣V integrins requires pho- infection. Gene Therapy 1999; 6: 1054–1063. sphoinositide-3-OH kinase. J Virol 1998; 72: 2055–2061. 29 Ramalingam R et al. E1-E4+ adenoviral gene transfer vectors 24 Klemke RL et al. Regulation of cell motility by mitogen-activated function as a ‘pro-life’ signal to promote survival of primary protein kinase. J Cell Biol 1997; 137: 481–492. human endothelial cells. Blood 1999; 93: 2936–2944. 25 Faucheux N, Haye B, Nagel MD. Activation of the cyclic AMP 30 Wills KN et al. Development and characterization of recombi- pathway in cells adhering to biomaterials: regulation by nant adenoviruses encoding human p53 for gene therapy of can- vitronectin and ﬁbronectin-integrin binding. Biomaterials 2000; cer. Hum Gene Ther 1994; 5: 1079–1088. 21: 1031–1038. 31 Shabram PW. Analytical anion-exchange HPLC of recombinant 26 Scatena M et al. NF-kappaB mediates alphaV beta3 integrin- type-5 adenoviral particles. Hum Gene Ther 1997; 8: 453–465. induced endothelial cell survival. J Cell Biol 1998; 141: 1083–1093.

Gene Therapy