Transient Blocking of Both B7.1 (CD80) and B7.2 (CD86) in Addition

Gene Therapy (2002) 9, 537–546  2002 Nature Publishing Group All rights reserved 0969-7128/02 $25.00 www.nature.com/gt RESEARCH ARTICLE Transient blocking of both B7.1 (CD80) and B7.2 (CD86) in addition to CD40–CD40L interaction fully abrogates the immune response following systemic injection of adenovirus vector

C Ziller1, F Stoeckel1, L Boon2 and H Haegel-Kronenberger1 1TRANSGENE, Strasbourg, France; and 2Tanox Pharma BV, Amsterdam, The Netherlands

Blockade of the CD40–CD40L and CD80/CD86–CD28 of the immune response against Ad. Using either of these costimulatory pathways represents a strategy to inhibit the mAb pairs, a second vector could be administered 1 month immune response against Ad vectors designed for gene after the first injection but with lower efficiency than in naive therapy applications. Since most previous studies have used animals. Thus, CD86 stands as the pivotal B7 molecule a CTLA4-Ig fusion molecule binding to both CD80 and involved in the development of the immune response against CD86, the respective roles of these B7 molecules remained Ad. However, only the blockade of both CD80 and CD86 undefined. We have studied the effect of blocking mono- in addition to CD40L fully inhibited the humoral and cellular clonal Abs (mAbs) directed against the costimulatory mol- responses against the Ad vector, such that readministration ecules CD40L, CD80 and CD86, alone or in different combi- after 1 month was as efficient as in naive animals. At the nations, on the humoral and cellular immune responses time of readministration, treated animals had regained their against Ad. Groups of mice were transiently treated with ability to mount a normal immune response to the second each combination of blocking mAbs upon systemic injection Ad vector, showing that tolerance was not induced. of a first Ad vector. Combinations of anti-CD80 + anti-CD86 Gene Therapy (2002) 9, 537–546. DOI: 10.1038/sj/gt/3301684 or anti-CD40L + anti-CD86 mAbs resulted in strong inhibition

Keywords: costimulatory molecules; gene therapy; adenovirus vector

Introduction repeated systemic administrations do not achieve efficient re-transduction of the target tissues due to the The immune response directed against Ad vectors rep- production of circulating neutralizing antibodies resents a major limitation for their use in gene therapy (NAbs).1,6–9 In mouse models, transient depletion or applications which require repeated administrations. blocking of the CD4+ T cell population decreases the pro- Transduction of target tissues such as the liver or the duction of anti-Ad antibodies (Abs) and permits vector lung by non-replicative Ad vectors is inflammatory and readministration.10,11 However, these strategies induce induces humoral and cellular responses involving CD4+ + profound immunosuppression, potentially detrimental to and CD8 T cells which can be directed against adenovi- patients with genetic diseases. Furthermore, it may not 1–4 ral proteins and the transgene product. Cytotoxic T apply to primates as suggested by a recent study show- lymphocytes (CTLs) directed against immunogenic trans- ing that blocking of CD4+ cell responses did not decrease genes have been shown to limit the persistence of trans- the humoral response following vector instillation in the 3,4 duced cells. CTLs directed against adenoviral proteins lung of Rhesus monkeys.12 Ideally, a non-depleting treat- have also been suspected to affect the long-term persist- ment inducing only transient suppression of the immune ence of the viral genome by destroying transduced cells, response against the virus should be achieved. Such 5 due to leaky expression of late viral genes. However, approach should not lead to immunological tolerance to expression of a non-immunogenic transgene can be rela- Ad proteins, since patients must remain capable of tively stable and long-lasting, suggesting that anti-Ad responding against infection by wild-type Ad. CTLs only play a minor role in the elimination of Antigen presentation leads to productive lymphocyte 2–4 transduced cells. activation when co-stimulatory signals are delivered by The humoral response to Ad proteins constitutes the the binding of surface molecules such as CD40, main obstacle to vector readministration. Particularly, CD80/B7.1 and CD86/B7.2, present on antigen- presenting cells (APCs) and up-regulated in the context of inflammation, with their ligands on T cells.13–15 In the Correspondence: H Haegel-Kronenberger, Experimental and Clinical Immunology, TRANSGENE, 11 rue de Molsheim, 67082 Strasbourg absence of co-stimulation, antigen recognition by the TCR Cedex, France has been shown to induce T cell anergy or unresponsive- Received 3 August 2001; accepted 28 January 2002 ness.15–19 CD28 and CTLA4 present on the surface of T Costimulation blockade and Ad vector readministration C Ziller et al 538 cells can bind indifferently CD80 and CD86 molecules on ated 4 weeks later, following administration of an Ad APCs, but deliver opposite signals: the binding of CD28 vector encoding a second reporter transgene (Ad-h␥IFN) contributes to T cell activation while CTLA4, which is in the absence of a costimulation-blocking treatment. The induced later during activation, down-regulates the T cell immune response was further analyzed after readminis- response and participates in the induction of peripheral tration to investigate the effect of the costimulation-block- tolerance.20–25 CD40L/CD154 is the ligand for CD40 and ing treatment on the capacity to elicit a posteriori an is transiently induced on T cells upon antigen recog- immune response against Ad. nition. Stimulation of CD40 molecules, present on the surface of B cells, by CD40L plays a crucial role in the Results development of the humoral response, including Ig isotype switching and regulation of B cell proliferation. In Abrogation of the humoral response against Ad vector addition, activation of APCs through CD40 results in by blocking CD40L, CD80 and CD86 increased expression of CD80 and CD86 co-stimulatory We first tested the effect of all combinations of blocking molecules.13 Importantly, CD40 signalling in dendritic mAbs directed against CD40L, B7.1/CD80 and cells induces their functional maturation by up-reg- B7.2/CD86 on the anti-Ad immune response when ulating the expression of MHC and co-stimulatory mol- injected at the time of first vector administration. Figure ecules on their surface and the secretion of cytokines, 1 describes the experimental scheme comprising 10 such as IL-12.26 groups of 10 C57Bl/6 mice each. At day 0, an Ad vector In mouse models, a large body of evidence has sug- encoding human factor IX (Ad-hFIX, 109 IU/mouse) was gested that the blockade of either or both of the CD40– injected into the tail vein of animals treated with the CD40L and B7–CD28 co-stimulatory pathways can blocking mAbs that were injected intraperitoneally. We inhibit the development of an immune response against chose to use the C57Bl/6 mouse strain because it was Ad vectors.27–30 Cellular and humoral reponses against shown to be tolerant to the hFIX transgene product after Ad were partially inhibited by treatment with anti- i.v. injection of Ad-hFIX, resulting in sustained hFIX CD40L mAb, which allowed vector readministration to expression in the serum.4,37 Levels of this transgene pro- the lungs in mice and also in primates.29–31 Most encour- duct in the circulation therefore reflect the number of aging results have been obtained in the study of Kay et transduced cells in the targeted tissues. A second vector, al,28 where simultaneous blocking of CD40 and of B7 mol- encoding human ␥IFN (Ad-h␥IFN, 109 IU/mouse), was ecules using an anti-CD40L mAb and a CTLA4-Ig fusion administered intravenously 4 weeks later (day 28). Dos- molecule resulted in successful secondary Ad-mediated age of circulating h␥IFN was used as read-out to assess gene transfer to the mouse liver. These results showed the efficiency of readministration. that blockade of both co-stimulatory pathways is neces- The humoral response against Ad was measured fol- sary to efficiently inhibit the anti-Ad immune response lowing the first administration of Ad-hFIX and the following systemic administration. efficiency of immunosuppression was evaluated by com- Because the CTLA4-Ig molecule binds both CD80 and paring mouse groups that received the mAbs to the CD86, studies using this molecule do not allow the untreated group (Figure 1b, group 8, first vector injected respective contribution of B7 molecules to the immune response against Ad to be deciphered. The requirement of blocking either or both of CD80 and CD86 costimulatory molecules in addition to CD40–CD40L interaction to achieve optimal suppression of the immune response remains to be resolved. Distinct and seemingly contradic- tory roles have been reported for CD80 and CD86 molecules notably in the context of autoimmune diseases and tolerance induction.14–17,25,32–35 Transgenic mice over- expressing CD80 on mature B cells make reduced humoral responses to T-dependent antigens.36 Allowing CTLA4 engagement by CD80 has been shown to inhibit T cell activation in the absence of CD28.24 Since CD80 is up-regulated later than CD86 on the surface of activated APCs,15,17 it could be hypothesized that leaving CD80 molecules available could favor their interaction with CTLA4 on T cells and thereby contribute to down-regulation of T cell responses. CTLA4 shows around 10-fold greater affinity for B7 molecules than CD28 does, which is reflected by the capacity of the CTLA4-Ig fusion pro- Figure 1 Experimental design. (a) C57Bl/6 mice were treated or not with blocking mAbs against CD40L, CD80 and/or CD86 and injected i.v. with tein to inhibit CD28-mediated T cell responses. 109 IU of Ad-hFIX. The immune response against the Ad vector was To address these points and attempt to down-regulate evaluated and detection of the transgene product was followed in the the immune response against Ad with maximal efficacy, serum. Four weeks later, mice were re-injected i.v. with 109 IU of Ad- we investigated the effect of transiently blocking CD40L, h␥IFN. Levels of h␥IFN in the serum was used as a read-out to determine CD80 and/or CD86 at the time of first vector adminis- the efficiency of Ad vector readministration. (b) MAbs against CD40L, tration. The effect of all combinations of blocking mAbs CD80 and/or CD86 were used alone or in different combinations to treat the indicated groups of C57Bl/6 mice. Control groups included untreated on humoral and cell-mediated immune responses to an mice or animals that received all three blocking mAbs injected with the Ad vector (Ad-hFIX) injected systemically were exam- second Ad vector alone and animals injected with both vectors without ined. The efficiency of secondary gene transfer was evalu- costimulation-blocking treatment.

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 539 effect (group 1), since when this mAb was given only at days –2 and 0, no significant reduction in anti-Ad NAb titers was observed (data not shown). Therefore, CD40– CD40L interactions must be blocked for a few days after administration of the Ad vector in order to decrease immune stimulation. While blocking of CD86 alone did not show any inhibitory effect (compare groups 2 and 8, Figures 2 and 3), the combination of anti-CD86 mAb with either anti-CD40L or anti-CD80 mAbs resulted in a strong reduction in anti- Ad Abs (groups 4 and 5, Figures 2 and 3). These two combinations appeared equally potent in dampening the humoral response, since only low to intermediate NAb titers were observed at day 27 (groups 4 and 5, Figure 3). However, as shown in Figure 3 (group 7), the most efficient immunosuppression was achieved using the combination of all three blocking mAbs anti-CD40L + + Figure 2 Inhibition of the humoral response against Ad vector by treat- -CD80 -CD86, resulting in a complete lack of anti-Ad ment with mAb anti-CD80 + -CD86, anti-CD40L + -CD86 or all three NAbs 4 weeks after injection. In another experiment, blocking mAbs. Sera were taken at the indicated time-points following mice exposed to the same regimen were found not to injection of Ad-hFIX. Serial dilutions were tested by ELISA for the pres- develop an anti-Ad humoral response for at least 3 ence of total Abs directed against the Ad vector. Revelation was performed months (result not shown). using an anti-mouse IgG+M Ab. Results obtained from each mouse group with sera diluted 1/400 are shown. Efficiency of Ad vector readministration As shown in Figure 4, administration of the second vector without immunosuppressive treatment) and to naive ani- (Ad-h␥IFN) 28 days after the first injection and costimu- mals (group 10, injected with PBS only). Figure 2 shows lation blockade resulted in levels of h␥IFN in the serum the kinetics of appearance of total anti-Ad Abs, as that were inversely correlated with pre-existing anti-Ad determined by Elisa. Figure 3 shows the titration of anti- Ab titers. Like preimmunized animals (group 8, Figure Ad NAbs at day 27 after injection, 1 day before adminis- 4), mouse groups treated with anti-CD80 or anti-CD86 tration of the second vector. Anti-CD80 or anti-CD86 mAb alone (groups 2 and 3, Figure 4) did not have detect- mAbs used alone did not affect the humoral response able levels of the transgene product in their serum. (groups 2 and 3, Figures 2 and 3). In contrast, treatment Groups 1 (treated with anti-CD40L only) and 6 (treated with anti-CD40L partially reduced the production of anti- with anti-CD40L + -CD80), which showed a slight Ad Abs (group 1, Figures 2 and 3), as reported in former reduction in anti-Ad NAb titers before readministration, studies.29,30 It should be noted that the anti-CD40L mAb had only low levels of circulating h␥IFN. Groups treated had to be injected at days Ϫ2, 0, +3 and +6 to show an with the anti-CD86 mAb in combination with either anti-

Figure 3 The neutralizing Ab response against Ad vector is totally abrogated at day 27 in mice treated with anti-CD40L + -CD80 + -CD86 upon first Ad vector administration, and strongly inhibited with anti-CD40L + -CD86 or anti-CD80 + -CD86. Ad-specific NAb were titrated 27 days after administration of the first vector Ad-hFIX (1 day before injection of the second Ad vector, Ad-h␥IFN, at day 28). Each graph includes a positive control (pooled serum from immune mice, bold curves with black squares) and a negative control (pooled serum from naive mice, bold curves with black circles). Control groups show the titration curves obtained from animals injected with Ad-hFIX in the absence of costimulation-blocking mAbs (immunized, upper right graph), or from untreated mice (naive, lower right graph).

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 540

Figure 4 Fully efficient readministration of Ad-h␥IFN in mice pre-treated with anti-CD40L + -CD80 + -CD86. Following injection of the second vector (Ad-h␥IFN at day 28), serum levels of transgene hgIFN were measured in 10 mice/group. Each graph represents the mean values of serum h␥IFN (mean ± s.d.) at day 3’,7’,14’,21’ and 42’ post-readministration, in the groups previously treated or not with blocking mAbs against CD40L, CD80 and/or CD86 upon first vector administration (Ad-hFIX at day 0). Without costimulation blockade, readministration is totally inefficient (group 8) while mice treated with anti-CD40L + -CD80 + -CD86 show the same kinetics of transgene expression as naive mice injected for the first time with Ad-h␥IFN (group 10).

CD40L or anti-CD80 mAb (groups 4 and 5, Figure 4) dis- viral antigens is intact 1 month after the mAb treatments, played levels of circulating h␥IFN approximately one that are therefore non-tolerogenic in these conditions. half log lower than naive animals (group 10, Figure 4). Mice that had been treated with all three blocking mAbs at the time of first vector injection exhibited serum levels Inhibition of Ig class switching of h␥IFN as high as those observed in animals that We next sought to analyze the IgM to IgG isotype switch received only the second vector (compare groups 7 and in the different mouse groups. In all groups characterized 10, Figure 4). These results show that blocking all three by efficient immunosuppression and readministration, CD40L, CD80 and CD86 costimulatory molecules abro- levels of serum anti-Ad IgG were low or undetectable 2 gates the humoral response against Ad and allows vector weeks after the first administration (Figure 6, IgG at day readministration with the highest efficiency. 13 in groups 1, 4, 5, 6 and 7). In contrast, IgM (predominating in a primary response) were detected in Blocking of the immune response is transient all groups, even in those that were efficiently immuno- In all mouse groups, total anti-Ad Abs as measured by suppressed (Figure 6, IgM at day 13). This shows that our Elisa increased following injection of the second vector Elisa test for the detection of total anti-Ad Abs (Figure 2) in the absence of costimulation blockade (data not did not recognize IgM, and that groups 4, 5 and 7 actually shown). As shown in Figure 5, 15 days after the second produced low levels of Abs of the IgM isotype before injection (day 15’), NAb titers in groups 4, 5 and 7 readministration. Two weeks after the second injection (treated with anti-CD80 + -CD86, anti-CD40L + -CD86 or in the absence of costimulation blockade, IgG/IgM ratios the three blocking mAbs, respectively) were similar to were increased in all mouse groups (Figure 6, day 14’). those of group 10 that received only the second vector, In groups 4, 5 and 7, IgG levels at day 14’ after readminis- reflecting a primary humoral response. Groups 1 and 6, tration were similar to those observed in naive animals in which the primary immune response had been decreased by treatment with anti-CD40L in the presence 2 weeks after the first administration (group 8 at day 13 or absence of anti-CD80 mAb, exhibited intermediate or group 10 at day 14’). NAb titers (Figure 5). Groups 2 and 3, in which treatment Blocking CD40L, CD80 and CD86 upon administration with anti-CD80 or anti-CD86 mAbs alone did not inhibit of the first vector blocks the switch in Ig isotype. IgM do the humoral response against the first vector, had high not seem to influence the efficiency of Ad vector read- levels of anti-Ad NAbs at day 15’ similarly to the pre- ministration, since mouse group 7 (treated with the three immunized group, suggesting a secondary humoral blocking mAbs, but showing some anti-Ad IgM, Figure response upon readministration (not shown). 6) showed similar expression and kinetics of the second These results show that inhibiting the humoral transgene as compared with naive animals (group 10, response with blocking mAbs against CD40L, CD80 and Figure 6, absence of IgM). Moreover, we found that these CD86 did not lead to the deletion or inactivation of Ad- IgM were unable to neutralize adenoviral transduction of reactive lymphocytes. The capacity to respond to adeno- 293 cells in vitro (Figure 3).

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 541

Figure 5 Blocking costimulatory molecules upon first Ad vector administration does not lead to immunological tolerance. Fifteen days after injection of the second vector (Ad-h␥IFN), neutralizing anti-Ad Abs were titrated in groups that showed partial or total inhibition of anti-Ad humoral response after the first vector administration. Each graph includes a positive control (pooled serum from immune mice, bold curves with black squares) and a negative control (pooled serum from naive mice, bold curves with black circles). Control graphs show titration curves obtained from groups injected with Ad-hFIX in the absence of costimulation-blocking mAbs and readministered with Ad-h␥IFN (group 8), or from naive mice injected for the first time with Ad-h␥IFN (group 10).

Effect of treatment with anti-CD40L, -CD80 and -CD86 bility that a cellular response had actually been initiated mAbs on the cellular immune response against Ad before readministration, but remained undetected in the In a second experiment, we evaluated the effect of the groups treated with anti-CD40L + anti-CD86. In contrast, blocking mAbs anti-CD40L, -CD80 and -CD86 on the animals treated either with anti-CD80 + -CD86 or with cellular arm of the immune response. To this end, groups anti-CD40L + -CD80 + -CD86 (groups 4 and 7, Figure 7) of 20 mice were treated with the mAb combinations that mounted an anti-Ad cellular response which was compa- were the most potent in repressing the humoral response rable to that observed in the naive mouse group, at day against Ad. The effect of these blocking mAb combi- 3’ and at day 14’. Therefore, for the cell-mediated as well nations on the repression of the humoral response were as for the humoral response, these groups behaved as if similar in this experiment to the one described above they had not been primed upon administration of the first (result not shown). Spleens were taken from five mice Ad vector. per group at days 14 and 27 after administration and the ␥ The presence of immune cell infiltrates in the liver, the number of m IFN-producing splenocytes was determ- main target organ following systemic injection of Ad vec- ined in Elispot assays that used Ad as antigen. The tors, was found to correlate with the cellular response remaining animals were used to measure the anti-Ad measured by ELISPOT. In particular, perivascular infil- cellular response at days 3’ and 14’ following admini- trates of CD4+ lymphocytes were rarely detected 2 weeks stration of the second vector. after injection of the first Ad vector in the mouse groups The cell-mediated immune response was inhibited fol- treated with anti-CD80 + -CD86 or with the three block- lowing Ad vector administration in the presence of the ing mAbs, while they were frequent in the liver of blocking mAb combinations (Figure 7). Before readmini- untreated animals or of mice treated with anti-CD40L stration, at day 27, groups treated with either anti-CD80 + CD86, anti-CD40L + CD86 or anti-CD40L + CD80 + only (results not shown). CD86 had the same numbers of Ad-specific, m␥IFN-pro- Altogether these data show that transient blocking of ducing splenocytes as found in the naive controls (Figure CD40L, CD80 and CD86 inhibits the development of a 7). In the group treated with anti-CD40L alone, the num- cellular immune response against the Ad vector. This ber of Ad-reactive splenocytes was only partially reduced treatment does not seem to induce the depletion or inacti- when compared with the control group (no immuno- vation (anergy) of Ad-reactive lymphocytes, since the suppressive treatment). cellular response observed following readministration in After administration of the second vector, mice treated the absence of costimulation blockade is comparable in ␥ with anti-CD40L +/Ϫ anti-CD86 developed comparable pre-treated and in naive mice, in terms of m IFN-produc- numbers of m␥IFN-producing splenocytes as in the pre- ing, Ad-specific splenocytes. Rather, it seems to prevent immune group (group 8, Figure 7), higher than in the the initiation of the immune response against Ad, which naive group (group 10, Figure 7). This raises the possi- could be considered as temporary ‘ignorance’.

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 542

Figure 7 Reversible inhibition of the cellular immune response against Ad vector in mice treated with blocking mAbs to CD40L, CD80 and/or CD86. Splenocytes from five mice per group were analyzed in ELISPOT assays for the detection of Ad-reactive, m␥IFN-producing cells at days 14 and 27 following the first administration of Ad vector with or without anti-CD40L, -CD80 and/or -CD86 mAbs, and at days 3’ and 14’ following second vector injection in the absence of costimulation-blocking treatment (days 31 and 42 after the first administration).

Figure 6 Reversible inhibition of IgM to IgG isotype switch by treatment with costimulation-blocking mAbs. IgM and IgG directed against the Ad vector were measured by ELISA at day 13 following ﬁrst administration of Ad-hFIX in the mouse groups treated or not with blocking mAbs against CD40L, CD80 and/or CD86 (upper panel) and at day 14’ following administration of the second vector Ad-h␥IFN (lower panel).

Expression of non-immunogenic transgene hFIX is not affected by the development of a specific cellular immune response against Ad after first vector administration In both experiments performed, the kinetics of first transgene hFIX expression in the serum was comparable in all mouse groups, whether treated or not with the costimulation-blocking mAbs (Figure 8). After peaking at day 15, hFIX expression stabilized at around 10 ␮g/ml for at least 28 days after readministration (not shown). Mice of the C57Bl/6 strain are known to be tolerant to the hFIX transgene product after i.v. injection of Ad-hFIX.4,37 As shown above, the cell-mediated immune response against the Ad vector is drastically if not totally abrogated in mice treated with blocking mAbs anti-CD40L, -CD80 and - CD86 (Figure 7). Thus, we have observed similar persistence of transgene expression in animals characterized by the development or the absence of Ad-specificm␥IFN- Figure 8 Expression of non-immunogenic transgene hFIX is not affected secreting lymphocytes. This suggests that at least in by the development of a specific cellular immune response against Ad after C57Bl/6 mice injected with an Ad vector, the persistence first vector administration. Levels of hFIX transgene product in the serum were measured following injection of first vector Ad-hFIX in the different of expression of a non-immunogenic transgene is not mouse groups, characterized by the presence or absence of a cellular dependent on the development of a specific cellular immune response against Ad. Data shown are from one of two experi- immune response against Ad. ments giving similar results.

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 543 Discussion immune response to the Ad vector so as to allow readministration, even partially, after 8 weeks. In our study, The present work shows that transient blocking of the the combination of anti-CD80 + -CD86 efficiently down- interactions mediated by CD80, CD86 and CD40L with regulated the anti-Ad immune response and readminis- their ligands is necessary to fully abrogate humoral and tration at day 28 was efficient, though not totally so. cellular immune responses otherwise induced by sys- While this discrepancy may also reflect a strain-specific temic injection of an Ad vector. While injection of a difference, it suggests that anti-CD80 + anti-CD86 mAb second Ad vector was totally ineffective in untreated, may be more effective than CTLA4Ig for blocking the pre-immunized mice, treatment with the three blocking immune response to Ad. mAbs at the time of first vector injection allowed read- Blocking the interactions of both CD80 and CD86 with ministration with the same efficiency as in naive mice, as their ligands produces a strong inhibitory effect on the determined by the levels of the second transgene secreted anti-Ad immune response. Thus, CD40–CD40L interac- in the serum. When used individually, none of the mAbs tion cannot drive efficient humoral or cellular responses anti-CD40L, -CD80 or -CD86 could affect the immune against the Ad vector in the absence of B7 molecules. response to Ad, despite a weak effect observed with anti- Complementary functions have been described for the B7 CD40L. Blocking of CD86 along with any of the two other molecules and the CD40–CD40L pathway in the costimu- costimulatory molecules resulted in strong, but incom- lation of T cell responses, CD40L-mediated interactions plete, inhibition of the immune response. This indicates playing a role in the maintenance of T cell responses, that CD86/B7.2 plays a central role in the immune while CD28/B7 interactions are required at their response against the Ad vector. When anti-CD86 was initiation.13,14,40,41 This correlates with the observation that used in combination with anti-CD40L or anti-CD80 at the treatment with anti-CD40L alone results in a very limited first administration, injection of a second Ad vector decrease of the immune response. The immune response resulted in levels of transgene expression that were about following systemic injection of the Ad vector was signifi- half an order of magnitude lower than those found in cantly inhibited only when anti-CD40L was combined naive mice. One month after treatment, animals were with the anti-CD86 mAb. In the presence of anti-CD40L able to develop a normal immune response to an Ad vec- + -CD80 + -CD86, both humoral and cellular responses tor. This suggests that repeated administrations of Ad- against the first Ad vector were fully abrogated. The only derived vectors can be envisaged with this strategy, at detectable response was the secretion of IgM, which is least for gene therapy applications in which short-term known to occur independently of CD40–CD40L interac- immunosuppression does not compromize the patient’s tion.13 IgGs were totally absent, as a consequence of the status. essential roles of CD40/CD40L and of B7/CD28 on Ig We chose the C57Bl/6 mouse strain in our model, isotype switching.13,42 because it has been shown to be tolerant to hFIX encoded While CD80 has been described as having an activating by Ad vectors injected systemically.4,37 We found no dif- effect on T cell responses, studies reporting the pheno- ference in the kinetics of hFIX expression between the type of CD80-transgenic mice, which show markedly different groups, whether or not they were treated with impaired immune response,36 and the exacerbation of blocking mAbs and were shown to develop Ad-reactive, disease caused by the anti-CD80 blocking mAb in the m␥IFN-producing lymphocytes. This indicates that in NOD mouse model34 suggested that this B7 molecule C57Bl/6 mice injected for the first time with an Ad vec- may also have a down-regulatory function.14,15 In our tor, the development of a specific cellular immune study, leaving CD80 molecules free in order to favor their response against Ad does not modulate the kinetics of interaction with CTLA4 receptors on T cells did not result expression of a non-immunogenic transgene. The cellular in stronger inhibition of the immune response. Since immune response includes CD8+ cells, which have been CTLA4 is normally induced on the surface of activated previously implicated in the rapid decline in transgene T cells following binding of CD28,23 it is possible that lack expression.1–3,27–30 In mouse models in which the trans- of CD28 stimulation by CD86 may result in lack of gene product is immunogenic, expression was found to CTLA4 expression (which would then not be available persist over a longer time period when vectors were for interaction with CD80). Alternatively, early CD80 administered in the context of costimulation block- expression on APC (or T cell) surfaces may be sufficient ade.27–29,38,39 Our results suggest that in these studies, the to stimulate CD28 and up-regulate T cell responses. prolonged transgene expression may be due to inhibition Blocking of either CD80 or CD86 is ineffective in of the immune response against the transgene product decreasing the immune response triggered by Ad vector rather than against the Ad vector. administration. However, CD86 stands as the pivotal B7 The use of a blocking CTLA4-Ig fusion molecule molecule for the development of the anti-Ad immune (binding to both B7 molecules) along with an anti-CD40L response. Blocking of CD80 along with the CD40–CD40L mAb was shown to be a powerful means to inhibit the pathway has little inhibitory effect on the anti-Ad specific immune response triggered after intravenous immune response, showing that CD86 may compensate injection of an Ad vector.28 Blocking only the almost completely for the absence of CD40L- and CD80- CD40L/CD40 pathway is sufficient to prevent the devel- mediated interactions. In contrast, CD80 alone is a very opment of a humoral response following Ad vector weak co-stimulator of the immune response to Ad vector, administration in the lung, but insufficient when the vec- as shown in mice treated with mAbs blocking CD86 and tor is injected systemically.27–31 In the latter situation and CD40-CD40L interactions: the latter mAb combination as reported here, blocking both CD40/CD40L and B7- dramatically inhibits both humoral and cellular mediated interactions is necessary to inhibit the immune responses, resulting in fairly efficient readministration of response. However, in the study of Kay et al,28 it was Ad vector. However, the role of CD80 as a costimulator found that treatment with CTLA4Ig could not inhibit the is reflected by the fact that while the anti-CD86 mAb

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 544 alone has absolutely no effect on the immune response, of DMEM without FCS was added in the nutrient com- additional blocking of CD80 dramatically down-regulates partment. Cells were diluted and the medium renewed this response. Overlapping as well as complementary every 3 days. MAbs were purified from the culture super- functions have indeed been reported for B7 molecules in natants using AFFI-T Thiophilic Agarose (KEM-EN-TEC, T cell activation.14,15,17,32,42–45 Copenhagen, Denmark) following the procedure T cell stimulation in the absence of a second signal pro- described by the manufacturer. The eluted fractions were vided by B7/CD28 or CD40/CD40L interactions has been dialyzed overnight against PBS (Sigma) in a sterile dialy- shown to induce a state of anergy.16–19 Allograft rejection sis cassette (Slide-A-Lyser 10, Pierce, Rockford, ME, can be efficiently inhibited by blocking B7 molecules and USA). MAb concentrations were determined by measur- CD40L for a limited time period following grafting, sug- ing the absorbance at 280 nm. The quality of the products gesting that tolerance can be achieved using this strat- was assessed by controlling the absence of endotoxin egy.40,46,47 In gene therapy applications, it would be prob- contamination and performing SDS-Page gel (mini-gel lematic to induce long-term tolerance to an Ad-derived NOVEX, San Diego, CA, USA). vector, since this would render patients unable to mount a response against infection by wild-type Ads. The fate of Recombinant Ad vectors Ad-reactive lymphocytes in the context of costimulation All viral genomes derived from Ad serotype 5 were first blockade is therefore an important question. When constructed as infectious plasmids by homologous treated mice were readministered a second Ad vector, recombination in Escherichia coli as described.49 They they developed humoral and cellular responses that were were deleted of viral E1 and E3 regions. Expression cas- of similar amplitude as animals injected for the first time settes were inserted in place of the E1 region, in sense with the same vector. This suggested that the Ad-reactive orientation. AdTG9397 (Ad-hFIX) contained the cDNA of lymphocyte population was neither depleted nor inacti- human factor IX (hFIX) under the control of the human vated following antigen encounter in the absence of CMV promoter and followed by the SV40 polyadenyl- costimulation. It appears unlikely that anergy is induced ation signal. Vector AdTG13378 (Ad-h␥IFN) comprises in our system, even though an anergic state might be the cDNA of human ␥IFN preceded by the CMV pro- reversed upon readministration of Ad vector in the pres- moter and a ␤globin/IgG chimeric intron and followed ence of costimulatory molecules and of inflammatory by a bovine growth hormone polyadenylation signal. For cytokines. Most importantly, in our study, the obser- the generation of viruses, the viral genomes were vation that 1 month after treatment, animals showed no released from the respective recombinant plasmids by alterations in the immune response to Ad suggests that PacI digestion and transfected into 293 cells.49 Virus costimulation blockade in our system does not result in propagation, purification and titration of the viral par- tolerance to Ad, but rather in ignorance. ticles were carried out as described previously.50,51 For Transient blockade of B7/CD28 and CD40/CD40L each virus preparation, the concentration of infectious using mAbs fully inhibits the development of immunity units (IU) was measured by indirect immunofluorescence to an Ad vector injected systemically, resulting in a lack of the viral DNA-binding protein.51 Purified virus was of both anti-Ad NAb formation and cell-mediated stored in viral storage buffer containing 1 M sucrose, 10 responses. Consequently, an Ad vector can be readminis- mM Tris-HCl (pH 8.5), 1 mM MgCl2, 150 mM NaCl, tered with the same efficacy as in naive mice. Our study 0.005% Tween 80. has further dissected the respective roles of B7 molecules in the development of the immune response to the Ad Animal studies vector, showing that both CD80 and CD86 exert a parti- C57Bl/6 female mice were purchased from Iffa Credo cular stimulatory function on humoral and cellular (l’Arbresle, France). Mice were maintained in our spe- responses. Although we found that their roles were cific-pathogen-free animal facility and used at 7 weeks of complementary, CD86/B7.2 stands as the main player in age. MAbs (100 ␮g anti-CD80, 250 ␮g anti-CD86 and 100 the development of the immune response against current ␮g anti-CD40L at days –2 and +3; 100 ␮g anti-CD80, 250 Ad vectors designed for gene therapy applications. ␮g anti-CD86 and 200 ␮g anti-CD40L at day 0; and 100 ␮g anti-CD40L at day +6) were mixed and diluted in PBS (Sigma), such that a volume of 100 ␮l was injected intrap- Materials and methods eritoneally. Ad vectors were diluted in the viral storage buffer described above. 109 IU in a volume of 100 ␮l were Hybridoma culture and production of mAbs injected into the tail vein of mice. Ad-hFIX was adminis- GL1 (anti-CD86), 16.10A1 (anti-CD80) and MR1 (anti- tered at day 0 of the experiment and readministration of CD40L) hybridomas were kindly provided by Dr Peter the second vector (Ad-h(IFN) was performed at day 28 Van Kooten (Utrecht, Faculty of Veterinary Medicine). (see Figure 1). Ocular punctions were done every week These mAbs have been widely used in costimulation- for the analysis of anti-Ad Abs and transgene expression blockade experiments.24,25,28–30,34,38,40,45,48 and GL1 and in the serum. At the indicated time-points (Figure 7), 1610A1 have been shown to block equally well CD86 and mice were killed and their spleens were used in CD80, respectively.48 Hybridomas were cultured in Inte- ELISPOT assays. gra CL350 flasks (Integra Biosciences, Cergy le Haut, France). Viable cells (8 × 106) were inoculated in the culti- Dosage of transgene products in the serum vation chamber containing 5 ml of DMEM supplemented The concentration of hFIX protein in mouse serum with 4 mM glutamine, 4.5 g/l glucose, non-essential ami- samples was determined using a commercial enzyme- noacids (Life Technologies, Paisley, UK), sodium pyru- linked immunosorbent assay (ELISA) (Asserachrom, vate (Sigma, St Louis, MO, USA), 20% FCS (Sigma) for Diagnostica Stago, Asnie`res, France). The concentration MR1 and 16.10A1, 15% FCS for GL1. A volume of 125 ml of hIFN␥ was measured using an ELISA assay (Duoset

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 545 hIFN␥, R&D Systems, Minneapolis, MN, USA) following taining 0.05% Tween 20 (PBS-Tween). Wells were incu- the manufacturer’s instructions. bated with 0.3 ␮g per 100 ␮l of a biotin-conjugated rat anti-m␥IFN mAb (Pharmingen, San Diego, CA, USA) for Measurement of total anti-Ad Abs 2 h at room temperature. Plates were then washed five Levels of total anti-Ad Abs were measured by ELISA. times with PBS-Tween and alkaline-phosphatase-conju- Ninety-six-well plates were coated overnight at 4°Cin gated Extravidin (BioRad Laboratories, Hercules, CA, ␮ 100 l of buffer (200 mM NaHCO3,80mMNa2CO3,pH USA) diluted 1/5000 in PBS-Tween-1% FCS was added. 9.5) containing UV-inactivated Ad-LacZ (108 viral After 45-min incubation, plates were washed five times particles). Wells were washed five times with a buffer with PBS-Tween and the presence of mIFN␥-secreting composed of PBS, 0.1% Tween 20 and 10 mM EDTA, cells was revealed by adding 100 ␮l per well of the sub- saturated with 150 ␮l of 3% BSA in PBS, rinsed once and strate solution (AP Conjugate Substrate Kit, BioRad incubated with serial dilutions of serum (1/200 to 1/25 Laboratories). After 30 min, the substrate solution was 600) for 2 h at room temperature. Wells were washed four discarded and the plates were washed with water and times. 100 ␮l per well of a peroxidase-conjugated goat air-dried. Blue spots were counted using a stereomicro- anti-mouse IgG+M(H+L), an anti-mouse IgG or an anti- scope. mouse IgM (Jackson Immunoresearch Laboratories, West Grove, PA, USA) (1/5000 dilution) were added and incubated for 1 h at room temperature. After washing, 100 ␮l Acknowledgements of substrate solution (0.05 M citric acid, 0.05 M sodium We gratefully acknowledge Peter van Kooten (Faculty of acetate, 1% tetramethylbenzidine (TMB, Sigma), 0.015% Veterinary Medicine, Utrecht, NL) for the gift of H2O2) was added and the reaction was stopped by ␮ hybridoma cell lines and for his help in producing and adding 100 l per well of 0.8 M H2SO4. Absorbance was purifying the mAbs. We thank Majid Mehtali, Ronald measured at 450 nm. Results shown represent OD values Rooke for critical reading of the manuscript, Bruce Acres of sera diluted at 1/400 and normalized according the and Monika Lusky for helpful discussions. absorbance obtained using standard immune sera. Titration of anti-Ad NAbs References NAbs were titrated by measuring the ability of serum dilutions to inhibit infection of 293 cells by Ad-LacZ. 293 1 Yang Y, Lin Q, Ertl HCJ, Wilson JM. Cellular and humoral cells (104 per well) in 96-well plates were cultured in immune responses to viral antigens create barriers to lung- DMEM medium supplemented with 10% FCS, for 24 h directed gene therapy with recombinant adenoviruses. J Virol ° 1995; 69: 2004–2015. at 37 C, 5% CO2. Serial dilutions (1/10 to 1/20 480) of ° 2 Yang Y et al. Immune responses to viral antigens versus trans- decomplemented mouse sera (15 min at 56 C) were pre- gene product in the elimination of recombinant adenovirus- incubated for 1 h with Ad-LacZ in PBS containing 0.1% infected hepatocytes in vivo. Gene Therapy 1996; 3: 137–144. BSA and added to 60–70% confluent cells. Control sera 3 Tripathy SK, Black HB, Goldwasser E, Leiden JM. Immune from naive or immune mice were tested on each plate. responses to transgene-encoded proteins limit the stability of ° ␤ After incubation for 16 h at 37 C, 5% CO2, -gal activity gene expression after injection of replication-defective adeno- was determined by washing the wells with PBS, lysing virus vectors. Nat Med 1996; 2: 545–550. 4 Michou A-I et al. Adenovirus-mediated gene transfer: influence the cells in 0.5 % Triton X-100, 6 mM Na2HPO4,4mM ␤ of transgene, mouse strain and type of immune response on per- NaH2PO4, 10 mM KCl, 0.1 mM MgSO4,50mM -mer- capto-ethanol, 170 ␮g/ml 4-methyl-umbelliferyl ␤-D gal- sistence of transgene expression. Gene Therapy 1997; 4: 473–482. 5 Yang Y, Ertl HCJ, Wilson JM. MHC class I-restricted cytotoxic actoside (Sigma) and measuring fluorescence at 340/460 T lymphocytes to viral antigens destroy hepatocytes in mice nm (Victor, Wallac, Turku, Finland). The neutralizing Ab infected with E1-deleted recombinant adenoviruses. Immunity titer is determined as the highest serum dilution giving 1994; 1: 433–442. Ն50% inhibition of infection, relative to infection in the 6 Juillard V et al. Long-term humoral and cellular immunity absence of serum. induced by a single immunization with replication-defective adenovirus recombinant vector. Eur J Virol 1995; 25: 3467–3473. ELISPOT assay for detection of Ad-specific, mouse 7 Van Ginkel FW et al. Intratracheal gene delivery with adenoviral ␥IFN (m␥IFN)-secreting splenocytes vector induces elevated systemic IgG and mucosal IgA anti- Ninety-six-well nitrocellulose plates (Millipore, Molsh- bodies to adenovirus and beta-galactosidase. Hum Gene Ther 1995; 6: 895–903. eim, France) were coated by adding 100 ␮l per well of a 8 Christ M et al. Gene therapy with recombinant adenovirus vec- rat anti-m␥IFN mAb (Pharmingen, San Diego, CA, USA) ␮ tors: evaluation of the host immune response. Immunol Lett 1997; diluted at 3 g/ml in carbonate buffer (15.7 mM 57:19–25. NaHCO3, 15.7 mM Na2CO3) and incubated overnight at 9 Gahe´ry-Se´gard H et al. Humoral immune response to the capsid 4°C. Plates were washed five times with DMEM 10% FCS components of recombinant adenoviruses: routes of immuni- and satured 2 h at 37°C with 100 ␮l/well of DMEM 10% zation modulate virus-induced Ig subclass shifts. Eur J Immunol FCS. Splenocytes from C57Bl/6 mice were plated at a 1997; 27: 653–659. concentration of 106 cells/100 ␮l/well. A non-replicative 10 Kolls JK et al. Use of transient CD4 lymphocyte depletion to Ad vector devoid of a transgene expression cassette was prolong transgene expression of E1-deleted adenoviral vectors. added at a multiplicity of infection of 100. In each well, Hum Gene Ther 1996; 7: 489–497. 11 Chirmule N et al. Repeated administration of adenoviral vectors medium was supplemented with 6 units IL-2 (R&D ␮ in lungs of human CD4 transgenic mice treated with a non- Systems). Concanavalin A (5 g/ml)-stimulated spleno- depleting CD4 antibody. J Immunol 1999; 163: 448–455. cytes were used as positive controls (performed for each 12 Shean MK et al. Immunomodulation and adenoviral gene trans- ° spleen). After 48 h incubation at 37 C, 5% CO2, plates fer to the lungs of nonhuman primates. Hum Gene Ther 2000; 11: were washed once with PBS and five times with PBS con- 1047–1055.

Gene Therapy Costimulation blockade and Ad vector readministration C Ziller et al 546 13 Grewal IS, Flavell RA. CD40 and CD154 in cell-mediated 33 Kuchroo VK et al. B7-1 and B7-2 costimulatory molecules acti- immunity. Annu Rev Immunol 1998; 16: 111–135. vate differentially the Th1/Th2 development pathways : appli- 14 Lenschow DJ, Walunas TL, Bluestone JA. CD28/B7 system of T cation to autoimmune disease therapy. Cell 1995; 80: 707–718. cell costimulation. Annu Rev Immunol 1996; 14: 233–258. 34 Lenschow DJ et al. Differential effects of anti-B7-1 and anti-B7- 15 Coyle AJ, Gutierrez-Ramos J-C. The expanding B7 superfamily: 2 monoclonal antibody treatment on the development of dia- increasing complexity in costimulatory signals regulating T cell betes in the nonobese diabetic mouse. J Exp Med 1995; 181: function. Nature Immunol 2001; 2: 203–209. 1145–1155. 16 Powell JD, Ragheb JA, Kitagawa-Sakakida S, Schwartz R. Mol- 35 King C et al. Interleukin-4 acts at the locus of the antigen- ecular regulation of interleukin-2 expression by CD28 co-stimu- presenting dendritic cell to counter-regulate cytotoxic CD8+ T- lation and anergy. Immunol Rev 1998; 165: 287–300. cell responses. Nat Med 2001; 7: 206–214. 17 Lu P, Wang YL, Linsley PS. Regulation of self-tolerance by 36 Sethna MP et al. A negative regulatory function of B7 revealed CD80/CD86 interactions. Curr Opin Immunol 1997; 9: 858–862. in B7-1 transgenic mice. Immunity 1994; 1: 415–421. 18 Van Parijs L et al. Role of interleukin 12 and costimulators in T 37 Fields PA et al. Intravenous administration of an E1/E3-deleted cell anergy in vivo. J Exp Med 1997; 186: 1119–1128. adenoviral vector induces tolerance to factor IX in C57Bl/6 mice. 19 Van Gool SW. Blocking CD40–CD154 and CD80/CD86–CD28 Gene Therapy 2001; 8: 354–361. interactions during primary allogeneic stimulation results in T 38 Stein CS, Martins I, Davidson BL. Long-term reversal of hyper- cell anergy and high IL-10 production. Eur J Immunol 1999; 29: cholesterolemia in low density lipoprotein receptor (LDLR)- 2367–2375. deficient mice by adenovirus-mediated LDLR gene transfer 20 Kearney ER et al. Antigen-dependent clonal expansion of a trace combined with CD154 blockade. J. Gene Med 2000; 2:41–51. population of antigen-specific CD4+ T cells in vivo is dependent 39 Kosuga M et al. Adenovirus-mediated gene therapy for mucopo- on CD28 costimulation and inhibited by CTLA4. J Immunol 1995; lysaccharidosis VII: involvement of cross-correction in wide- 155: 1032–1036. spread distribution of the gene products and long-term effects 21 Walunas TL, Bakker CY, Bluestone JA. CTLA4 ligation blocks of CTLA4-Ig coexpression. Mol Ther 2000; 1: 406–413. CD28-dependent T cell activation. J Exp Med 1996; 184: 2541– 40 Larsen CP et al. Long-term acceptance of skin and cardiac allo- 2550. grafts after blocking CD40 and CD28 pathways. Nature 1996; 22 Perez VL et al. Induction of peripheral tolerance in vivo requires 381: 434–438. CTLA4 engagement. Immunity 1997; 6: 411–417. 41 Howland KC, Ausubel LJ, London CA, Abbas AK. The roles of 23 Thompson CB, Allison JP. The emerging role of CTLA4 as an CD28 and CD40 ligand in T cell activation and tolerance. J immune attenuator. Immunity 1997; 7: 445–450. Immunol 2000; 164: 4465–4470. et al 24 Fallarino F, Fields PE, Gajewski TF. B7-1 engagement of cyto- 42 Borriello F . B7-1 and B7-2 have overlapping, critical roles in immunoglobulin class switching and germinal center formation. toxic T lymphocyte antigen 4 inhibits T cell activation in the Immunity 1997; 6: 303–313. absence of CD28. J Exp Med 1998; 188: 205–210. 43 Blazar BR et al. Infusion of anti-B7.1 (CD80) and anti-B7.2 (CD86) 25 Judge TA. The role of CD80, CD86 and CTLA4 in alloimmune monoclonal antibodies inhibits murine graft-versus-host disease responses and the induction of long-term allograft survival. J lethality in part via direct effects on CD4+ and CD8+ T cells. J Immunol 1999; 162: 1947–1951. Immunol 1996; 157: 3250–3259. 26 Cella M et al. Ligation of CD40 on dendritic cells triggers pro- 44 Schweitzer AN et al. Role of costimulators in T cell differen- duction of high levels of IL-12 and enhances T cell stimulatory tiation: studies using antigen-presenting cells lacking expression J Exp Med 1996 capacity: T-T help via APC activation. 1996; : 184. of CD80 or CD86. J Immunol 1997; 158: 2713–2722. 27 Kay MA et al. Long-term hepatic adenovirus-mediated gene 45 Sigal LJ, Reiser H, Rock KL. The role of B7-1 and B7-2 costimu- expression in mice following CTLA4Ig administration. Nat Genet lation for the generation of CTL responses in vivo. J Immunol 1995; 11: 191–197. 1998; 161: 2740–2745. 28 Kay MA et al. Transient immunomodulation with anti-CD40 46 Kirk AD et al. CTLA4-Ig and anti-CD40 ligand prevent allograft ligand antibody and CTLA4Ig enhances persistence and second- rejection in primates. Proc Natl Acad Sci USA 1997; 94: 8789–8794. ary adenovirus-mediated gene transfer into mouse liver. Proc 47 Li Y et al. Blocking both signal 1 and signal 2 of T-cell activation Natl Acad Sci USA 1997; 94: 4686–4691. prevents apoptosis of alloreactive T cells and induction of per- 29 Yang Y et al. Transient subversion of CD40 ligand function ipheral allograft tolerance. Nat Med 1999; 5: 1298–1302. diminishes immune response to adenovirus vectors in mouse 48 Reiser H, Schneeberger EE. Expression and function of B7-1 and liver and lung tissues. J Virol 1996; 70: 6370–6377. B7-2 in hapten-induced contact sensitivity. Eur J Immunol 1996; 30 Scaria A et al. Antibody to CD40L inhibits both humoral and 26: 880–885. cellular immune responses to adenoviral vectors and facilitates 49 Chartier C et al.Efficient generation of recombinant adenovirus repeated administration to mouse airways. Gene Therapy 1997; vectors by homologous recombination in Escherichia coli. J Virol 4: 611–617. 1996; 70: 4805–4810. 31 Chirmule N et al. Readministration of adenovirus vector in non- 50 Lusky M et al. In vitro and in vivo biology of recombinant aden- human primate lungs by blockade of CD40–CD40 ligand inter- ovirus vectors with E1, E1/E2A, or E1/E4 deleted. J Virol 1998; actions. J Virol 2000; 74: 3345–3352. 72: 2022–2032. 32 Freeman GJ et al. B7.1 and B7-2 do not deliver identical costimu- 51 Mittereder N, March KL, Trapnell BC. Evaluation of the concen- latory signals, since B7-2 but not B7-1 preferentially costimulates tration and bioactivity of adenovectors for gene therapy. J Virol the initial production of IL-4. Immunity 1995; 2: 523–532. 1996; 70: 7498–7509.

Gene Therapy