Based Vaccination and Nonviral Cytokine Gene Transfer

Home , Immunogen, T helper cell

Gene Therapy (2000) 7, 481–492  2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00 www.nature.com/gt NONVIRAL TRANSFER TECHNOLOGY RESEARCH ARTICLE Regulation of T-helper-1 versus T-helper-2 activity and enhancement of tumor immunity by combined DNA- based vaccination and nonviral cytokine gene transfer

K Song, Y Chang, and GJ Prud’homme Department of Pathology, McGill University, 3775 University Street, Room 13, Montreal, Quebec, Canada, H3A 2B4

Intramuscular (i.m.) injections of a plasmid encoding human type 2 (Th2) response. Antitumor immunity was enhanced carcinoembryonic antigen (CEA) elicited both humoral and when the CEA and IL-12 plasmids were coinjected at the cellular immune responses in mice, but only partial inhibition same muscle site, but not at separate sites despite of the growth of transplanted syngeneic CEA-positive P815 increased serum IL-12 levels. Though the tumor cells tumor cells (CEA/P815). Coinjection of the CEA vector with a expressed neomycin phosphotransferase, mice immunized vector encoding either interferon-␥ (IFN␥) or IL-12 promoted with vectors encoding that protein (without CEA) were not IgG2a isotype anti-CEA antibody production, anti-CEA/P815 protected against tumor growth, and produced no CTLs CTL activity and greater resistance to CEA/P815 tumor chal- except for low levels when coinjected with an IL-12 vector. lenge. As well, CEA/P815-stimulated IFN␥ secretion in vitro Thus, we show that immunity elicited by DNA vaccination was increased, but IL-4 diminished, consistent with a T- against CEA can be biased to a protective type (high Th1 helper type 1 (Th1) response. In contrast, coinjection of the and CTL activity) or nonprotective type (high Th2 and low CEA vector with an IL-4 vector increased IgG1 production, CTL activity) by i.m. coinjection of cytokine-expressing plas- but reduced CTL activity and resistance to tumor challenge. mids. IL-12 appears to act locally, but not systemically, The latter treatment inhibited CEA/P815-dependent IFN␥ through an adjuvant effect. Gene Therapy (2000) 7, 481– production but enhanced IL-4 secretion, consistent with a Th 492.

Keywords: DNA vaccination; carcinoembryonic antigen; cytokine plasmids; T-helper cells; cytotoxic T lymphocytes; cancer immunotherapy

Introduction Immunization usually requires use of adjuvants;15 however, many adjuvants cause undesirable side-effects, It has been shown that DNA vaccination can elicit both such as severe inflammation, which limits or precludes antigen-specific humoral and cell-mediated immune their use in humans.16 We investigated the adjuvant effect responses, including T-helper (Th) cell and cytotoxic T- of plasmids encoding a single cytokine on the immune lymphocyte (CTL) responses. The immune responses responses elicited by a plasmid encoding human CEA. generated by this technique in many experimental animal Thus, we coinjected the CEA plasmid with plasmids models can provide protection against infectious agents encoding mouse IFN␥, IL-12, or IL-4. Though cytokine or malignant tumor cells (reviewed in Ref. 1). A major expression plasmids have been found to enhance or mod- limitation of immunization with conventional vaccines is ify immune responses, most of this work has been perfor- the lack of CTL stimulation. The conventional vaccines med in infectious disease models,17–20 and relatively little stimulate mostly antibody and to varying degrees, Th cell in tumor models. In our study, a transplantable mouse 2–4 responses. The latter responses frequently fail to elim- mastocytoma cell line P815 was stably transfected to inate tumors or virally infected cells, while CTLs are express human CEA (CEA/P815). After five weekly plas- 5–8 more effective. On the other hand, DNA vaccines excel mid i.m. injections, anti-CEA humoral and anti- 9–11 at stimulating CTLs. CEA/P815 cellular immune responses were detected. We used human carcinoembryonic antigen (CEA) as an This immunity, however, only partially inhibited the antigen in mice, where it is not an endogenous protein, growth of CEA/P815 cells. Anti-CEA IgG isotypes, as with the purpose of studying strategies of DNA vacci- well as cytokine secretion by spleen cells were analyzed nation. In humans, CEA is one of the best characterized to reveal the dominant type of the immune response (Th1 tumor-associated antigens in terms of its tissue distri- versus Th2). By this method, we could readily bias the 12,13 bution and biochemistry. It is considered a potential anti-CEA/P815 T cell response to a more protective type 14 target antigen for cancer immunotherapy. (high Th1 and CTL activity) with either IFN␥ or IL-12 vectors, or to a nonprotective type (high Th2 and low CTL activity) with an IL-4 vector. Though circulating lev- Correspondence: GJ Prud’homme els of IL-12 were detectable after IL-12-vector treatment, Received 6 May 1999; accepted 5 November 1999 coinjection of that vector at the same site as the CEA vec- Regulation of anti-tumor response by cytokine vectors K Song et al 482 tor was required for optimal immunization. The use of preliminary experiments, we found that two DNA injec- cytokine vectors to regulate immunity is discussed. tions were not sufficient to generate CTL responses, while anti-CEA antibodies were detectable but could be further Results enhanced by additional injections. Four or five DNA injections were required consistently to generate CTLs and optimal antibody titers. For this reason, all experi- Construction of expression vectors ments in this study were performed with five DNA injec- Human CEA cDNA (full length) was inserted into the tions at weekly intervals. One week after the last injec- vector pCI-neo, and the resulting plasmid was desig- tion, serum samples were obtained for quantitation of nated pCI-CEA (Figure 1). This vector has the neo gene, IgG anti-CEA antibody. It has been shown that the Th1 encoding neomycin phosphotransferase (NPT). In some + and Th2 subsets of CD4 T cells promote different pro- experiments we also used pCMV and pCMV-CEA, that files of IgG isotype production. Strong IgG1 production do not encode NPT, but are otherwise identical to pCI- is an important indication of a Th2-dominant immune neo and pCI-CEA, respectively. Cell surface expression of response, whereas strong IgG2a production reflects a CEA directed by pCI-CEA or pCMV-CEA was confirmed Th1-dominant immune response.22 initially by transient transfection of COS-7 cells with As shown in Figure 2a, production of IgG anti-CEA these vectors, and analyzed by flow cytometry with anti- antibodies was elicited by i.m. injection of the CEA vector CEA mAb (data not shown). Mouse IFN␥ cDNA and into mice. Coinjection of pCI-CEA and VR-IL12 increased mouse IL-4 cDNA were also cloned into the pCI-neo vec- the production of IgG anti-CEA antibody compared with tor, and the resulting vectors were designated pCI-IFN␥ the group injected with pCI-CEA and control vector. and pCI-IL4, respectively. The plasmid encoding mouse Injection of a mixture of pCI-CEA and pCI-IL4, but not IL-12, designated VR-IL12, was constructed by cloning a pCI-IFN␥, also induced an increase in the total IgG anti- segment consisting of p35 cDNA, a synthetic intron (IVS), CEA antibody production. an internal ribosome entry site (IRES) sequence and p40 In all cases, the profiles of IgG1 and IgG2a isotypes in cDNA into the VR1255 plasmid (Figure 1). the serum of mice were influenced by injection of a mix- The in vitro production and secretion of IFN␥, IL-4, and ture of pCI-CEA and a cytokine vector as compared with IL-12 were confirmed by transient transfection of COS-7 a control vector (Figure 2b). A mixture of pCI-CEA with cells with the respective vectors. In all cases the respect- either pCI-IFN␥ or VR-IL12 induced higher IgG2a levels ive cytokines, as detected by ELISA, were secreted at lev- and lower IgG1 levels, while coinjection of pCI-CEA and els ranging from 100 to 200 ng/ml. The biological activity pCI-IL4 had the opposite effect (higher IgG1, lower of these cytokines was confirmed with appropriate bio- IgG2a), and these changes were statistically significant assays (data not shown). (Figure 2b). These results are consistent with Th1 (IgG2a) and Th2 (IgG1) responses, respectively. Detection of anti-CEA IgG antibody and isotyping To study the effects of the plasmid pCI-CEA and the cytokine-encoding vectors on production of anti-CEA Detection of CTL response against CEA-expressing antibodies, groups of DBA/2 mice were injected i.m. with tumor cells in vitro a mixture of pCI-CEA and either a control vector or a CTLs play a critical role in cell-mediated immune cytokine-encoding vector. To induce an optimal immune responses against virally infected cells and tumor response to CEA, we adapted a protocol previously dem- cells.7,8,11 The generation of CTLs is highly influenced by onstrated to be effective by Conry and colleagues.21 In the cytokines produced during an immune response.

Figure 1 Schematic diagram of the expression vectors. CEA was cloned into pCI-Neo or pCMV (lacking the neo gene), to produce pCI-CEA and pCMV-CEA (not shown), respectively. The major transcriptional control elements are shown: in pCI-CEA, cytomegalovirus (CMV) immediate–early enhancer/promoter (CMV-I-E/E-P); a chimeric intron; the human CEA cDNA sequence; the SV40 late polyadenylation and transcriptional termination sequence (SV40 poly A). Mouse IFN␥ and IL4 cDNAs were cloned into pCI-neo vector respectively to generate cytokine-expressing vectors pCI-IFN␥, and pCI-IL4. Mouse IL-12 p35 cDNA, a synthetic intron (IVS), an internal ribosome entry site sequence (IRES) and p40 cDNA were cloned into the VR1255 vector, which also has the CMV-I-E/E-P, CMV intron A, and the minimal rabbit beta globin terminator sequence (RBG poly A), to generate a IL-12-expressing vector, VR-IL12.

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 483

Figure 3 CTL activity of spleen cells of mice injected with DNA vectors. Figure 2 IgG anti-CEA serum levels following DNA vaccination. DBA/2 (a and b) DBA/2 mice were immunized with DNA vectors using the same mice were injected i.m. in each leg (rectus femoris) with a mixture of protocol as described in the legend of Figure 2 (five mice per group). One 50 ␮g of pCI-CEA vector and 100 ␮g of either blank vector VR (pCI- + ␮ week after the last DNA injection, spleen cells were isolated and restimu- CEA VR) or 100 g of a cytokine encoding vector, five times at weekly lated in vitro with irradiated CEA/P815 cells for 5 days in IL-2-sup- intervals. Serum samples were collected 7 days after the last DNA injec- plemented medium. Responder cells were recovered and tested for their tion and analyzed for CEA-specific antibody response. Each value rep- ability to kill CEA/P815 cells. Percentage of specific lysis was measured resents the mean ± s.d. for eight mice, from one representative experiment. 51 after a 6-h incubation of Cr-Na2CrO4-labeled CEA/P815 cells with the Three experiments yielded similar results. (a) Serum levels of IgG anti- effector cells. Results are reported at an effector to target ratio = 100:1. CEA antibody. Values that are significantly different are indicated as fol- Each value represents the mean ± s.d. of quadruplicate cultures from one a Ͻ + + lows: P 0.001 versus control vectors pCI-neo VR (pCI VR); representative experiment. In these assays, no specific lysis was detected bP Ͻ 0.001 versus all the other groups; cP Ͻ 0.001 versus pCI-CEA + VR, + ␥ with parental P815 (CEA-negative) target cells (data not shown). Five pCI-CEA pCI-IFN (b) Serum levels of IgG1 (hatched bars) and IgG2a experiments yielded similar results. Values that are statistically significant anti-CEA antibodies (black bars). Values that are significantly different in Figure 3a are as follows: aP Ͻ 0.001 versus either pCI-neo + VR are indicated as follows: in comparison to IgG1 levels among all groups, (pCI + VR), or saline (no DNA); bP Ͻ 0.001 versus either pCI- a Ͻ b Ͻ c Ͻ d Ͻ P 0.001; P 0.001; P 0.01; P 0.05; in comparison to IgG2a CEA + VR, or pCI-CEA + pCI-IL4; cP Ͻ 0.01 versus pCI-CEA + VR. e Ͻ levels among all groups, P 0.001; in comparison of IgG1 versus IgG2a Values that are statistically significant in Figure 3b are as follows. f Ͻ g Ͻ levels within each group, P 0.05; P 0.001. Immunization did not aP Ͻ 0.001 versus pCMV + VR, or pCMV + VR-IL12; bP Ͻ 0.001 versus induce IgG antibodies reactive to P815 (CEA-negative) cells (not shown). pCMV-CEA + VR, or pCI-neo + VR-IL12 (pCI + VR-IL12). cP Ͻ 0.01 versus pCMV-CEA + VR. Values for VR-IL12 injected alone (not shown) were not statistically different from the groups receiving blank vectors or + + Thus, production of IFN␥ by Th1 CD4 cells, and IL-12 pCMV VR-IL12. by other cells such as macrophages, has been shown to promote CTL activity.23–25 DBA/2 mice were injected i.m. with plasmids in various combinations as shown in (NPT), having been selected for G418 resistance, and that Figure 3. Seven days after the last of five DNA injections, protein is known to be antigenic in mice. pCI-CEA, pCI- spleen cells were recovered and incubated with IFN␥ and pCI-IL4 encode NPT, while pCMV, pCMV- CEA/P815 cells for 5 days in medium supplemented CEA and vectors derived from VR1255 (ie blank VR and with IL-2. The cells were then tested for CTL activity. As VR-IL12) do not. pCMV-CEA and pCI-CEA induced shown in Figure 3a, treatment with a mixture of pCI-CEA comparable levels of CTL activity against CEA/P815 and a blank vector induced a CTL response against (Figure 3a, b). Injection of pCI-neo (with blank VR) did CEA/P815. Coinjection of pCI-CEA with either pCI-IFN␥ not induce CTLs reactive to CEA/P815 (Figure 3a). How- or VR-IL12 increased anti-CEA/P815 CTL activity con- ever, when pCI-neo was coinjected with VR-IL12, a low siderably, while coinjection of pCI-CEA with pCI-IL4 level of CTL activity was noted (Figure 3b), though much decreased the CTL activity. In all groups, no CTL activity less than with CEA-encoding vectors following similar was seen against parental P815 target cells (not shown). immunization. Thus, it appears that CEA is the main CEA/P815 cells express neomycin phosphotransferase target antigen for CTLs in the anti-tumor response.

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 484 Secretion of IL-4 and IFN␥ by spleen cells in vitro later time. On day 28, the average volume of the tumors To determine whether codelivery of cytokine-expressing was smaller than that observed in the groups injected vectors and CEA vector could influence differentiation of with either no DNA, null vectors alone, or IL-12-encoding Th cells, and thus influence production of cytokines, we vectors alone (no CEA) (Figure 5a, c). compared the amount of IL-4 and IFN␥ released by P815 cells are highly tumorigenic. We also observed spleen cells stimulated with CEA-expressing syngeneic this feature with the CEA/P815 cell line. The high tumo- tumor cells. DBA/2 mice were injected i.m. with a mix- rigenic potential may explain why immunization with ture of plasmids, ie pCI-CEA and other vectors as shown the pCI-CEA vector alone could not completely inhibit in Figure 4. Spleen cells from the immunized mice were tumor growth. recovered and stimulated in culture with CEA/P815 Mice that received a CEA-encoding vector plus either cells. Analysis of supernatants by ELISA revealed that pCI-IFN␥ or VR-IL12 showed much enhanced protection spleen cells from mice coinjected with pCI-CEA and con- against the tumor challenge. Many of these mice (63% trol vector produced substantially more IFN␥ (27-fold and 75%, respectively) were completely tumor free over increase), and slightly more IL-4, when compared with 90 days after tumor challenge (Figure 6a). As expected, control mice (pCI-neo + VR). Spleen cells from mice coin- CEA-immunized mice were not protected from native jected with pCI-CEA and either pCI-IFN␥ or VR-IL12 had P815 cells (Figure 5b). a further increase in the levels of IFN␥ and produced As demonstrated by average tumor volume very low levels of IL-4. Spleen cells from mice coinjected (Figure 5a), mice that received pCI-CEA plus pCI-IL4 with pCI-CEA and pCI-IL4 secreted high levels of IL-4, showed reduced tumor immunity compared with those but reduced levels of IFN␥. In all groups, parental P815 that received pCI-CEA plus blank vector, and all died by cells did not stimulate cytokine production above spleen day 45 (Figure 6a). cells grown in medium alone (not shown). In these experiments, the possibility that NPT contrib- utes to tumor rejection must be considered. However, Protective tumor immunity induced by DNA vaccination immunization with pCI-neo which encodes NPT but not We immunized DBA/2 mice with the pCI-CEA vector, CEA (delivered with VR which encodes neither NPT nor with or without a cytokine-encoding vector, by coinjec- CEA and is added as a control for the amount of bacterial tion of vectors in both the right and the left rectus femoris DNA injected) did not protect from tumor growth muscles. Mice were challenged with CEA/P815, the syn- (Figure 5a) or improve survival (Figure 6a) more than geneic tumor cells expressing CEA (Figures 5a, b, c and injection of saline (no DNA). Moreover, when pCI-neo 6a, b). Mice injected with saline (no DNA), or injected was combined with VR-IL12 there was only a small with null vectors, had tumor growth (Figure 5a) and decline in tumor growth (Figure 5b), which was not stat- started to die around 30 days after tumor transplantation istically significant versus either blank pCMV + VR treat- (Figure 6a). Mice that received either pCI-CEA or pCMV- ment (Figure 5c) or CEA/IL-12-immunized mice injected CEA and blank vector showed partial inhibition of tumor with native P815 cells (expressing neither CEA nor NPT) growth (Figure 5a, c), with the tumors appearing at a (Figure 5b). Thus, we found no evidence that immunity directed to NPT was protective.

Circulating IL-12 does not enhance immunity to CEA/P815 Since it has been reported that IL-12 has strong anti- tumor effects,26 we also tested a different manner of delivery of the DNA vaccine and the VR-IL12 vector. In previous experiments, both vectors were injected at the same muscle site; however, mice that received pCI-CEA injected into muscle of one leg and VR-IL12 into muscle of the other leg did not show enhancement of tumor rejection over mice not receiving VR-IL12 (Figures 5a, b, and 6a, b). Delivery of IL-12-expressing plasmids alone, without pCI-CEA immunization, could not inhibit tumor growth either (Figures 5a and 6b). Serum samples from mice injected with VR-IL12, with or without other vectors, Figure 4 In vitro cytokine production by CEA/P815-stimulated spleen were tested for the presence of serum IL-12 by ELISA. In cells. Spleen cells from groups of mice treated as described in the legend all cases, VR-IL12 treatment significantly increased both to Figure 3 were cultured with irradiated CEA/P815 cells for 3 days. Cell- total IL-12 and IL-12p70 serum levels. Before DNA injec- free supernatants were then harvested and assayed for the presence of = ± ␥ ± = tions mice had levels of total IL-12 224.0 9.3 pg/ml IFN and IL-4 by ELISA. Each value represents the mean s.e.m. (n 5 = ± = per group). Black bars, IFN␥; open bars, IL-4. Values that are statistically and IL-12p70 2.4 0.1 pg/ml (n 25). One week after a different are as follows: in comparison to IFN␥ levels among groups, series of five weekly vector injections the following aP Ͻ 0.001 versus blank vectors pCI-neo + VR (pCI + VR), pCI- serum levels were recorded, expressed as total IL-12 and CEA + pCI-IL4; bP Ͻ 0.001 versus pCI-CEA + VR; cP Ͻ 0.001 versus IL-12p70 (the latter in parentheses) in pg/ml ± s.e.m. + + ␥ pCI-CEA VR, pCI-CEA pCI-IFN ; in comparison to IL-4 levels (n = 5 per group): pCI-neo + VR treatment = 235.4 ± 6.5 d Ͻ among groups, P 0.001 versus all the other groups. In these assays, (2.3 ± 0.2); pCI-CEA + VR-IL12 (same muscle site) cytokine secretion of spleen cells cultured with parental P815 cells (CEA- = ± ± + negative) was not significantly different from spleen cell grown in medium treatment 298.2 4.7 (3.7 1.6); pCI-CEA VR-IL12 alone (Ͻ10 pg/ml of IL-4 and Ͻ15 pg/ml of IFN␥ in these cultures) (data (separate muscle sites) treatment = 389.4 ± 2.2 (5.7 ± 0.3); not shown). VR-IL12 alone treatment = 398.4 ± 16.2 (5.5 ± 0.2). Only

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 485

Figure 5 Effect of DNA vaccination on subcutaneous growth of CEA/P815 tumor cells in DBA/2 mice. (a) DBA/2 mice were immunized as described in the legend to Figure 2 and challenged by s.c. injection of CEA/P815 cells. ᭺, no DNA; ᭹, control vectors pCI-neo + blank VR (pCI + VR); ̆, pCI- CEA + pCI-IL4; ᭿, pCI-CEA + VR; ᮀ, pCI-CEA + pCI-IFN␥; ̄, pCI-CEA + same site VR-IL12; ᭛, injection of 100 ␮g of VR-IL12 in each leg (VR- IL12). (b) DBA/2 mice were challenged with either CEA/P815 or native P815 cells. They were injected with pCI-neo + VR-IL-12 (ٌ, pCI + same site VR-IL-12; challenged with CEA/P815), or 100 ␮g of pCI-CEA in the left leg and 200 ␮g of VR-IL12 in the right leg (᭜, pCI-CEA + separate site VR- IL12, challenged with CEA/P815). To facilitate comparison, the group injected with pCI-CEA + same site VR-IL12 (Figure 5a) is also shown in Figure 5b (̄). Another group (̅) was immunized with pCI-CEA + same site VR-IL12 and challenged with native P815 cells. *, P Ͻ 0.01 versus groups without an asterisk. (c) DBA/2 mice were immunized as in panel a, but with vectors that do not encode NPT, and challenged with CEA/P815. ᭹, pCMV + blank VR; +, pCMV + same site VR-IL-12; ᭿, pCMV-CEA + blank VR; ̄, pCMV-CEA + same site VR-IL-12. *, P Ͻ 0.01 versus groups without an asterisk. In panels a, b and c, mice received 5 weekly DNA injections. Seven days after the last DNA injection, 106 tumor cells were injected s.c. into the left flank of each mouse. Size of the tumors was determined every 2 days by caliper measurements. Each value represents the mean ± s.e.m. (n = 8 per group). The injection of pCI-CEA alone gave results similar to injection pCI-CEA + VR (not shown). Similar results were obtained in one repeat experiment. Statistical analysis of values on day 28 after tumor transplantation was performed. Significant values in panel a are as follows: aP Ͻ 0.001 versus groups injected with no DNA, with control vector pCI + VR, and with VR-IL12 alone; bP Ͻ 0.01 versus groups injected with either no DNA, pCI- CEA + pCI-IFN␥, or pCI-CEA + VR-IL12; cP Ͻ 0.05 versus groups injected with either pCI-CEA + VR, pCI-CEA + separate site VR-IL12, and with VR-IL12 injected alone. groups of mice receiving VR-IL12 had significantly elev- As reported by others,27 total IL-12 levels (includes ated values of IL-12 (both total and p70 in all cases) versus heterodimeric p70, as well as monomeric and dimeric preimmunization levels in the same groups (P Ͻ 0.001). p40) were always much higher than the active IL-12p70 It is unclear why coinjection of pCI-CEA with VR-IL12 form. resulted in lower IL-12 levels than separate site injection. These results indicate that the IL-12 vector must be

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 486

Figure 6 Effect of DNA vaccination on survival of DBA/2 mice following CEA/P815 tumor challenge. Figure 6a shows the survival rate of mice described in Figure 5a. ᭺, no DNA; ᭹, control vectors pCI-neo + VR (pCI + VR); ̆, pCI-CEA + pCI-IL4; ᭿, pCI-CEA + blank vector VR (pCI-CEA+VR); ᮀ, pCI-CEA + pCI-IFN␥; ̄, pCI-CEA + same site VR-IL12. Figure 6b shows the survival rate in groups of mice described in Figure 5a, b. ᭛, VR-IL12; ᭜, pCI-CEA + separate site VR-IL12. To facilitate comparison, the pCI-CEA + same site VR-IL12 group is shown in both panels (̄). In both panels a and b, the percent survival of mice at various time-points is plotted for each group (n = 8 per group). The injection of pCI-CEA alone gave results similar to injection pCI-CEA + VR (not shown). Similar results were obtained in one repeat experiment. Values that are statistically different are indicated as follows: aP Ͻ 0.01 versus either no DNA injection, pCI + VR, pCI-CEA + pCI-IL4, or VR-IL12; bP Ͻ 0.05 versus pCI-CEA + VR, pCI- CEA + separate VR-IL12; cP Ͻ 0.05 versus all groups except pCI-CEA + same site VR-IL12.

injected at the same muscle site as the CEA vector for improved the immune responses.28 In mice, Irvine et al29 optimal immunization. Moreover, in the absence of CEA reported that they could not detect cytotoxic T cell immunization, circulating IL-12 was insufficient to inhibit responses to CEA-transduced tumor cells following tumor growth in this model. immunization with purified native CEA emulsified in complete Freund’s adjuvant. Bei et al30 also reported that Discussion three injections of bV-CEA emulsified in Detox adjuvant did not induce the rejection of a transduced CEA-positive CEA seems to be a weak immunogen in humans and in cell line. In several murine models, multiple immuni- mice.14,28–33 Low levels of CEA-specific antibody and T zations were required to achieve immunity against syn- cell responses have been found in some of the patients geneic tumor cells expressing CEA.21,31 Thus, CEA rep- immunized with recombinant baculovirus CEA (bV- resents an example of a weak immunogen for which CEA) in alum, while combination with granulocyte– effective immunization strategies need to be explored. macrophage colony-stimulating factor (GM-CSF) Consequently, efforts have been made to improve anti-

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 487 CEA immune responses. For example, McLaughlin et al32 observations). Furthermore, cytokine secretion by spleen reported that codelivery of interleukin-2 (IL-2) protein cells in vitro required stimulation with CEA/P815 cells with a recombinant CEA vaccinia vaccine enhanced anti- as a source of antigen. Cytokines were only present at tumor immunity. Bei et al33 showed that use of a cationic background levels in the supernatants of spleen cells liposome formulation (DOTAP) as an adjuvant increased grown without CEA/P815, which argues against vector antitumor effect elicited by the bV-CEA vaccine. How- driven production independent of T cell stimulation. ever, DNA vaccines provide a potent alternative to these To test the anti-tumor efficacy of the vaccination, CEA- methods, as shown by Conry and colleagues.21 expressing P815 mastocytoma cells of DBA/2 origin In this respect, codelivery of cytokine genes may rep- (CEA/P815) were transplanted into DBA/2 mice. The resent an effective approach to enhance or regulate mice injected with saline (no DNA) or with control vec- immune responses. Notably, It has been shown that plas- tors had tumor growth, which could be observed as early mids encoding a single cytokine such as IL-12, IL-2, or as 10 days. None of these mice survived, and evidently GM-CSF were effective at augmenting antigen-specific transplantation of CEA-expressing P815 tumor cells into immune responses elicited by DNA vaccines.34–36 the unimmunized syngeneic mice could not elicit an In this study, we addressed the following issues: (1) effective protective immune response. The mice immu- can the immune responses elicited by a DNA vaccine for nized with the CEA-encoding DNA vaccine, without a a weak immunogen, such as CEA, be enhanced by i.m. cytokine-expressing plasmid, had only partial inhibition coinjection of plasmids encoding various cytokines; (2) of tumor growth and most died by 60 days. Mice injected can the type of T cell response (eg Th1, Th2, or CTLs) with pCI-CEA plus pCI-IL4 had reduced tumor immu- and protective tumor immunity elicited by the DNA vac- nity, compared with pCI-CEA treatment alone. More- cine be modulated by this method; and (3) can circulating over, weak tumor immunity in IL-4-treated mice was IL-12 alone enhance antitumor immunity in this model? accompanied by lowered anti-CEA/P815 Th1 and CTL We constructed a plasmid encoding human CEA for use responses. Clearly, IL-4 inhibited the T cell-mediated as a DNA vaccine (pCI-CEA), and plasmids expressing responses thought to be protective in this model. the murine cytokines IFN␥, IL-12, and IL-4 to be used as In contrast, mice immunized with pCI-CEA plus either genetic adjuvants in the context of pCI-CEA DNA- IFN␥- or IL-12-expressing plasmids showed significantly based vaccination. enhanced inhibition of tumor growth. Many of these mice Our choice of cytokines was influenced by the fact that (62.5% and 75%, respectively) were completely tumor IL-12 and IFN␥ are important enhancers of Th1-type free over 90 days after tumor challenge. These results par- responses.37 Moreover, IL-12 inhibits angiogenesis which alleled the increased Th1 and CTL activity induced by is required for tumor growth,38 enhances macrophage the cytokine-encoding vectors. In our experiments, IFN␥ activity via increased production of IFN␥,39 and stimu- and IL-12 vectors were almost equally effective at lates CTL and NK cell activity.40 Conversely, IL-4 stimu- inducing Th1 responses and CTL activity, or inhibiting lates the differentiation of Th2 cells and inhibits the tumor growth. Although we did not address this ques- development of Th1 cells.41 It also inhibits activation of tion, we speculate that IL-12 acts at least partially by macrophages by IFN␥.42,43 IL-4 promotes B cell growth enhancing IFN␥ secretion. and induces isotype shift in B cells (IgG1 and IgE The plasmids pCI-CEA, pCI-IL4 and pCI-IFN␥ contain production).44 the NPT gene (neomycin resistance or neo), and NPT is We found that coinjection of the CEA DNA vaccine an immunogenic protein in mice. The CEA-positive P815 with a plasmid encoding IFN␥ or IL-12 promoted CEA- stable transfectant cell line (CEA/P815) used in our specific IgG2a production, and enhanced anti-CEA/P815 experiments produces this molecule, since it was selected CTL activity. In vitro, spleen cells of treated mice showed for G418 resistance. Therefore, the question arises as to increased CEA/P815-stimulated IFN␥ production. On the whether NPT is a rejection antigen in our experiments. other hand, coinjection of the DNA vaccine with a plas- Vaccination with either pCMV-CEA (NPT-negative) or mid encoding IL-4 resulted in an increase in anti-CEA pCI-CEA was equally effective at inhibiting tumor IgG1 production, but reduced IgG2a, and reduced anti- growth and inducing anti-CEA/P815 CTLs. Immuni- CEA/P815 CTL responses. There was increased IL-4 zation of mice with the pCI-neo vector did not induce secretion by CEA/P815-stimulated spleen cells. Thus, CTL activity against the CEA/P815 cell line, except for codelivery of the DNA vaccine with a vector encoding low levels when it was coinjected with VR-IL12. This CTL IFN␥ or IL-12 promoted the development of a Th1-domi- activity was much lower than observed with CEA-enco- nant response, while codelivery of the DNA vaccine with ding vectors following similar immunization. Spleen cells a vector encoding IL-4 induced a Th2-dominant response. of pCI-neo-immunized mice only produced low inter- In the CTL assay, killing is unlikely to be mediated by feron-gamma levels in vitro following stimulation with NK cells for two reasons. First, P815 cells are NK resist- these cells, unlike the cells of mice immunized with pCI- ant, though they can be killed by lymphokine-activated CEA. Moreover, pCI-neo, even when coinjected with VR- killer (LAK) cells. Second, cell lysis was only observed IL12, did not significantly enhance resistance to a tumor with CEA-transfected P815 (antigen-positive) target cells. challenge. In these experiments, VR is a non-CEA/non- Since NK and LAK cells are not antigen specific, they NPT containing vector which was added as a control for should kill both native P815 and CEA/P815 cells. In the amount of bacterial DNA injected. interpreting our cytokine secretion results we must con- Thus, it appears that CEA is the main target antigen of sider the possibility that plasmids could have gained the anti-tumor response. The reason for the lower access to the spleen and influenced in vitro results. How- immunogenicity of NPT in our experiments is unclear, ever, this seems unlikely, since by PCR analysis we find though the fact that it is strictly a cytoplasmic antigen that only small amounts of plasmid can be detected out- (unlike CEA that can also be expressed on the membrane) side muscle following i.m. injection (our unpublished may be a factor. A recent study10 revealed that following

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 488 i.m. injection of ovalbumin (OVA) expression plasmids, phenomena. By immunohistochemical analysis, we find secreted or membrane-bound forms of OVA generated that CEA is produced by muscle cells for 2–4 weeks fol- stronger CTL responses than a cytoplasmic form. It is lowing a single DNA injection of pCI-CEA (unpublished postulated that these differences result from altered anti- observation). Usually, at the site of injection, only 3–5% gen uptake or processing by APCs. of muscle cells produce the antigen at a detectable level. Since P815 is itself an immunogenic tumor, part of the Following i.m. injection of antigen-encoding plasmids, antitumor response could be directed against native P815 we find that the decline in protein expression correlates tumor antigens. Such responses may have occurred but with accumulation of a lymphocytic infiltrate, can be did not protect the mice from tumor progression and delayed by an anti-inflammatory cytokine, and is much death. Recently, we have transplanted a different tumor slower in immunodeficient SCID mice (C Piccirillo and cell line, ie CEA-transfected Lewis lung carcinoma of the G Prud’homme, manuscript submitted for publication). mouse. Though this tumor cell line is aggressive and not Several reports have demonstrated the important role notably immunogenic on its own, pCI-CEA immuni- of bone marrow-derived cells in antigen presentation and zation of syngeneic C57BL/6 mice (with or without cyto- CTL induction following DNA vaccination.50–52 It seems kine vectors) yielded protective immunity very similar to likely that the adjuvant effect of the plasmids requires that reported here against CEA/P815 (K Song and G secretion of these cytokines in close proximity to the Prud’homme, manuscript in preparation). Thus, P815 APC, or interacting T cell/APC aggregates, or perhaps tumor antigens are not required for tumor rejection in directly by the same APC that is presenting CEA to T our model. A limitation of our study is that CEA is not cells. This is akin to the findings of Conry et al,53 who an endogenous antigen in mice. In future studies, it will reported that the costimulatory molecule B7–1 had to be be interesting to determine if immune responses can be expressed by the same vector as CEA, and presumably generated against murine tumor-associated antigens by the same APCs, for enhancement of immunity. However, this DNA vaccination method. in that case B7–1 is a membrane-bound molecule. Inter- IL-12 is known to be a potent anti-cancer agent. The estingly, Behrens et al54 found that after stimulation with antitumor efficacy of IL-12 has been demonstrated in a IFN␥, human muscle cells express a functional costimu- variety of tumor models by all of the following: systemic latory molecule BB-1 that plays a role in antigen presen- administration of recombinant IL-12 protein;45 local deliv- tation. This molecule is similar but not identical to B7–1 ery of IL-12 at the tumor site using genetically modified and B7–2. However, the significance of this is unclear, fibroblasts46 or tumor cells;47 recombinant viral vector since professional APCs (macrophages, dendritic cells, B delivery;48 or gene gun-delivered IL-12-encoding plasmid cells) are thought to be essential for DNA vaccination. In DNA.49 To determine whether i.m. injection of the IL-12- any case, it is well established that cytokine stimulation encoding vector could act solely through elevated circul- of APCs influences their costimulatory activity.55 Further- ating IL-12 levels, we injected the IL-12-encoding vector, more, locally secreted cytokines may act directly on CD4+ with or without pCI-CEA, in various ways (eg DNA vac- T cells to induce differentiation into Th1 or Th2 cells. cine in the right leg, IL-12 vector in the left leg). We found Similarly, locally produced cytokines could induce differ- that when the CEA-expressing and IL-12-expressing plas- entiation of CTL precursors into mature CTLs. mids were simultaneously delivered at different muscle It has been found that unmethylated CpG dinucleotide sites (opposite legs), there was no enhancement of tumor motifs present in plasmid can function as an adjuvant.56 immunity. Similarly, injection of the IL-12 vector alone However, in our study blank vectors do not mimic the (no pCI-CEA) was ineffective, even though these mice immunostimulatory effects of cytokine-encoding vectors. had higher circulating IL-12 than the mice coinjected with Thus, though bacterial DNA is known to have an adju- both pCI-CEA and IL-12 vector at the same site. IFN␥ vant effect, this cannot replace or match the strong adju- was not detected in the serum of any of these mice. How- vant effects we are reporting with cytokine vectors. ever, though serum total IL-12 levels were increased by To our knowledge, this is the first report showing that 70–170 pg/ml, the levels of the active heterodimeric form the i.m. coinjection of a CEA-expressing plasmid with of IL-12, IL-12p70, were only increased by a few pg/ml, cytokine-expressing plasmids biased the type of CEA- which is perhaps too little to inhibit tumor growth. specific immune response (Th1 versus Th2). The CEA Recently, Schults J et al27 have reported that i.m. injec- cDNA and cytokine cDNA were encoded by separate tion of an IL-12-encoding plasmid different from ours expression plasmids. Moreover, enhancement of CEA- generates strong antimetastatic activity to the malignant specific CTL activity was detected in our model, but not melanoma cell line B16-F10. I.m. injection with 100 ␮gof in others.53 Our ability to generate CTLs may be related their IL-12 expression plasmid produced approximately to many factors including the type of vectors used for 35 pg/ml of serum IL-12p70. These levels are much immunization, method of in vitro restimulation of T cells, higher than achieved in our study. However, as in our and the type of target tumor cells. Interestingly, IFN␥ and study, total IL-12 levels (IL-12p70, plus monomeric and IL-12 were almost equally effective at inducing these dimeric p40) were much higher than the active IL-12p70 cells, while IL-4 was inhibitory. form. The significance of a high total IL-12/IL-12p70 ratio Our results showing negative effects of IL-4 on tumor is unclear, though p40 dimer has been shown to have immunity contrast with some reports of protective inhibitory effects. At any rate, it is clear that in our model effects.57,58 However, in most of these experiments IL-4 is the IL-12-induced enhancement of immunity cannot be produced locally by the tumor cells. It is unclear how IL- attributed to circulating IL-12. Local effects, presumably 4 protect tumors in these cases, but this is a situation either in the muscle or draining lymph nodes, appear to clearly different from DNA-based immunization in mus- be important. cle. For example, local production of IL-4 by tumor cells, The local effects of cytokine-expressing plasmids are even in nude mice, results in eosinophil accumulation not well understood, but could be related to a number of and eosinophil-mediated cytotoxicity.57 For unclear

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 489 reasons, IL-4 can display a direct antitumor effect as segment, consisting of p35 cDNA, the synthetic intron shown in human colonic, renal, gastric, and breast (IVS), the IRES sequence, and p40 cDNA, was then sub- tumors, as well as leukemia.58 This cytokine can also cloned into the plasmid VR125560 (kindly provided by stimulate B cells to produce antibodies which may be rel- VICAL Inc, San Diego, CA, USA), after removal of the evant in some tumor models. In our case, however, pro- luciferase gene, to produce a vector designated VR-IL12. tective immunity probably depends on effector T cells The VR1255 vector also contains the human CMV induced by vaccination either in the muscle or (more immediate–early enhancer/promoter, a CMV intron A, likely) draining lymph nodes. If Th1 cells and CTLs are and the minimal rabbit beta globin poly (A) signal the main mediators of tumor rejection it is not surprising sequence. An empty VR1255-derived vector, designated that IL-4 has detrimental effects. VR, was used as a control in several experiments. Cytokines such as GM-CSF, IFN␥, IL-4 and others can skew immune responses by acting on the APC. This may Transfection of COS-7 cells and detection of CEA and alter antigen presentation, cytokine/chemokine secretion, cytokines and either stimulate or inhibit macrophage effector func- To confirm production of the CEA, IFN␥, IL-4, and IL-12 tions. Some cytokines (particularly IL-12 and IFN␥) have directed by the respective plasmids, COS-7 cells (ATCC pleiotropic proinflammatory effects that can contribute to CRL 1651) were transiently transfected with each of the tumor cell killing. Thus, in situ cytokine production by vectors. The presence of IFN␥, IL-4, or IL-12 p70, in tumor cells may have several effects not duplicated by supernatant was confirmed and quantitated by ELISA DNA vaccination. assays (see next section). On the other hand, it is clear that inflammatory cyto- Cell surface expression of CEA by pCI-CEA-trans- kines such as IFN␥ are not always beneficial in DNA vac- fected COS-7 cells was analyzed by immunofluorescence cination studies. Notably, in an i.m. DNA vaccination staining and flow cytometry analysis using an anti-CEA study Xiang and Ertl59 found that while GM-CSF mAb (from hybridoma clone B18, kindly provided by Dr enhanced B cell and T cell responses to a rabies virus Abraham Fuks, McGill University, Montreal, Canada) antigen, IFN␥ had the opposite effect. It is difficult to and a goat anti-mouse IgG-FITC conjugate (Sigma compare this situation with a tumor model, but it has to Chemical, St Louis, MO, USA). be considered that the positive effects of DNA vaccination in various models may be mediated primarily by ELISA assays either antibodies, Th1 cells, Th2 cells, CTLs, or macro- Cytokine levels in culture supernatants or mouse serum phages. Thus, IFN␥ may inhibit a protective Th2- samples were determined by ELISA as follows: concen- dependent response. trations of IFN-gg and IL-4 were measured by ELISA kits Administration of cytokine-encoding plasmids in com- (R&D Systems, Minneapolis, MN, USA; detection limit bination with DNA vaccines may serve as a simple and Ͻ2 pg/ml, both assays); according to the manufacturer’s effective way of generating an adjuvant effect. This may instructions. IL-12 levels were measured either with the provide a sustained but relatively low level of cytokine mouse total IL-12 ELISA kit (Genzyme, Cambridge, MA, secretion, without the undesirable side-effects frequently USA; detection limit = 10 pg/ml), which detects all three caused by the systemic administration of boluses of cyto- forms of IL-12 (p70 heterodimer, p40 homodimer, and kine protein. In our experiments circulating levels of p40 monomer), or with the mouse IL-12 p70 ELISA kit active IL-12 were very low. This is relevant, since this (R&D Systems; detection limit Ͻ2.5 pg/ml), which cytokine can have severe toxic effects. Furthermore, cyto- detects the p70 heterodimer only. kine therapy allows regulation of immunity to either pre- dominant Th1 or Th2 types. This approach to vaccination Plasmid DNA preparation has the potential for applications in many infectious, Plasmid DNA was prepared by the alkaline lysis method neoplastic or autoimmune diseases. using the Qiagen kits (Qiagen, Santa Clarina, CA, USA), according to the manufacturer’s instructions. DNA prepared for i.m. injections was dissolved in sterile 0.85% Materials and methods NaCl and stored at −80°C. Spectrophotometric analysis revealed 260/280 nm ratios у1.80. Purity of DNA Plasmid construction preparations and conformation was confirmed by cDNA containing the complete coding sequence for CEA electrophoresis with 1% agarose gel. was inserted into pCI-neo (Promega, Madison, WI, USA), which contains the human cytomegalovirus (CMV) Establishment of the CEA transfected mouse tumor immediate–early enhancer/promoter, a chimeric intron, cells and the SV40 poly (A) signal sequence. The resulting pCI A mouse mastocytoma cell line, P815 (ATCC TIB64), of vector encoding CEA was then designated pCI-CEA DBA/2 origin was used for establishment of a CEA- (Figure 1). pCMV and pCMV-CEA lack the neo expressing stable transfectant cell line. To facilitate expression cassette but are otherwise identical to pCI-neo efficient production of stable transfectant lines, CEA and pCI-CEA, respectively. Complete cDNA sequences cDNA from pCI-CEA vector was transferred into the bici- of mouse IFN␥ and IL-4 were inserted into pCI-neo to stronic expression vector pIRES1neo. Stable transfectant produce vectors designated pCI-IFN␥ and pCI-IL4, clones were generated from CEA-positive cultures, and respectively (Figure 1). the P815 cells expressing CEA were designated The p35 and p40 cDNA sequences of murine IL-12 CEA/P815. These cells were used in cytotoxicity assays were cloned into the bicistronic expression vector pIRES- and in in vivo tumor challenge experiments. Expression 1neo (Clontech Bio/Can Scientific, Mississauga, Ontario, of CEA on at least two-thirds of cells was reconfirmed Canada) to produce a vector designated pIRES-IL-12. The before use in each experiment.

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 490 Immunization of mice using the following equation: percent specific lysis = Six-week old female DBA/2 mice (H-2d) were purchased [(experimental release − spontaneous release)/(maximum from the Jackson Laboratory (Bar Harbor, ME, USA). release − spontaneous release)] × 100. Spontaneous Unless indicated otherwise, 6- to 8-week old mice were release was determined by incubation of target cells in anesthetized and injected i.m. in the rectus femoris the absence of effector cells. Maximum release was muscles, with one injection in each leg consisting of 50 ␮g determined by incubation of target cells in 1% Triton X- CEA-plasmid (100 ␮g of DNA per mouse), mixed with 100. P815 cells are not lysed by NK cells. either 100 ␮g of a blank VR (200 ␮g of DNA per mouse) or 100 ␮g of a cytokine-encoding vector (200 ␮g of DNA Cytokine release assay per mouse), in saline, injected at a volume of 75 ␮l per For measuring cytokine release by lymphoid cells of muscle site using a disposable sterile plastic insulin mice, spleen cells from the experimental or control ani- syringe and 29G1/2 needle (VWR Canlab, Mississauga, mals were cultured with irradiated CEA/P815 or par- Ontario, Canada). ental P815 cells in 24-well plates, in the same conditions described above for the cytotoxicity assay, except that the Quantitation of CEA antibody culture was maintained for 3–4 days. Then cell-free Serum samples were assayed for the presence of anti- supernatants were harvested and stored at −80°C. The CEA antibodies by a cellular enzyme-linked immuno- supernatant samples were assayed for the presence of absorbent assay (CELISA). The CEA-producing human IFN␥ and IL-4 by ELISA. adenocarcinoma cell line LS180 (kindly provided by Dr Abraham Fuks, McGill University) was used as a CEA Tumor cell challenge antigen source in the assay. Briefly, 96-well plates were One week after the last of five immunizing DNA injec- coated with 2 × 105 of LS180 cells in 50 ␮l of HBSS con- tions, DBA/2 strain mice were transplanted with 1 × 106 taining 0.5% formalin (Fisher Scientific, Los Angeles, CA, CEA/P815 cells by subcutaneous injection into the left USA) and incubated at 37°C overnight. The rest of the flank. Tumors were measured by caliper in two dimen- steps were performed at room temperature. After wash- sions and the volumes were calculated using the formula ings with washing buffer (PBS supplemented with 0.5% (width2 × length)/2. In addition, survival of mice was Tween 20), wells were blocked with 3% BSA in PBS con- determined. Moribund mice with large tumors were taining 0.5% Tween 20 for 1 h. As a second layer, diluted killed. serum samples, or as a standard dilutions of an anti-CEA mAb, were applied to the plate for 1 h. After additional Statistical analysis washings, a third layer consisting of either goat-anti The statistical significance of differences among various mouse IgG (Cedarlane Laboratories, Hornby, Ontario, groups of mice for all data, except survival, was deter- Canada), goat-anti mouse IgG1 (Cedarlane Laboratories) mined by analysis of variance. Survival of mice injected or goat-anti mouse IgG2a (Cedarlane Laboratories) was with tumor cells was plotted using the Kaplan–Meier applied to the plate. The plate was washed again, and method (nonparametric cumulated survival plot), and the fourth layer consisting of rabbit anti-goat IgG conju- statistical comparison between the curves obtained using gated with alkaline phosphatase (BioRad Laboratories, the log rank test. All analyses were done with the SYS- Hercules, CA, USA) was added to the plate. The reaction TAT statistical software (version 8.0, SPSS Inc, Chicago, was developed and read according to the manufacturer’s IL, USA). instructions. Concentrations of total IgG anti-CEA Ab and the isotypes IgG1 and IgG2a Abs were calculated Acknowledgements from the standard curve generated by appropriate anti- CEA mAbs and results are reported as units (U), where We thank Dr Abraham Fuks (McGill University, Mon- 1 U of anti-CEA antibody = 1 ␮g of anti-CEA mAb. treal, Canada) and Dr Serge Jothy (Sunnybrook Hospital, Toronto, Ontario, Canada) for kindly providing reagents Cytotoxicity assay and cell lines, Dr Fawaz Halwani (McGill University) for Spleen cells (4 × 106 per well) were first cultured with assistance with FACS analyses, and Dr Fu Hu (Montreal irradiated (12 Gy) CEA/P815 cells (4 × 106 per well) in General Hospital, Montreal, Canada) for assistance with medium consisting of RPMI 1640 (Gibco BRL), containing statistical analyses. This study was funded by the Fraser 10% FBS, and supplemented with 20 U/ml human rIL-2 Fund (Royal Victoria Hospital, Montreal, Canada). Keli (MEDICORP, Montreal, Quebec, Canada) in 24-well Song was supported in part by Royal Victoria Hospital plates at 37°C for 5 days, for in vitro stimulation of the Research Institute Fellowship. responder spleen cells. To perform the chromium-release assay, target CEA/P815 cells and parental P815 tumor × 6 ␮ 51 References cells (2 10 cells) were labeled with 200 Ci of Na2 CrO4 (Amersham, Oakville, Ontario, Canada) for 90 min at 1 Cohen AD, Boyer JD, Paul DB. Modulating the immune 37°C, followed by three washes. The target cells and the response to genetic immunization. FASEB J 1998; 12: 1611–1626. in vitro stimulated responder cells, both resuspended in 2 Rock KL. A new foreign policy: MHC class I molecules monitor medium, were then combined at various effector-to target the outside world. Immunol Today 1996; 17: 131–138. ratios in 96-well, U-bottom plates (Fisher Scientific, 3 Jondal M, Schirmbeck R, Reimann J. MHC class I-restricted CTL responses to exogenous antigens. Immunity 1996; 5: 295–302. Whitby, Ontario, Canada) in a volume of 200 ␮l per well. + ° ␮ 4 Raychaudhuri S, Morrow WJ. Can soluble antigen induce CD8 After a 6-h incubation at 37 C, 100 l of supernatant was cytotoxic T cell responses? A paradox revisited. Immunol Today harvested from each well and radioactivity was quantit- 1993; 14: 344–348. ated in a gamma counter (Beckman Instrument, Fuller- 5 Engers HD et al. Functional activity in vivo of effector T cell ton, CA, USA). The percent specific lysis was calculated populations. J Immunol 1984; 133: 1664–1670.

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 491 6 Yee C, Riddell SR, Greenberg PD. Prospects for adoptive T cell 28 Fagerberg J et al. Recombinant carcinoembryonic antigen and therapy. Curr Opin Immunol 1997; 9: 705–708. granulocyte–macrophage colony-stimulating factor for active 7 Carmichael A, Jin X, Sissons P, Borysiewicz L. Quantitative immunization of colorectal carcinoma patients. Proceedings of analysis of human immunodeficiency virus type 1 (HIV-1)-spe- the 11th Annual Meeting of the Society for Biological Therapy, cific cytotoxic T lymphocyte (CTL) response at different stages Madison, WI: Society for Biological Therapy, 1996. of HIV-1 infection: differential CTL responses to HIV-1 and 29 Irvine K, Kantor J, Schlom J. Comparison of a CEA-recombinant Epstein–Barr virus in late disease. J Exp Med 1993; 177: 249–256. vaccinia virus, purified CEA and an anti-idiotype bearing the 8Bo¨hm W et al. T cell-mediated, IFN␥-facilitated rejection of image of a CEA epitope in the treatment and prevention of murine B16 melanomas. J Immunol 1998; 161: 897–908. CEA-expressing tumors. Vaccine Res 1993; 2: 79–94. 9 Xu D, Liew FY. Protection against leishmaniasis by injection of 30 Bei R et al. Enhanced immune responses and anti-tumor activity DNA encoding a major surface glycoprotein, gp63, of 1-major. by baculovirus recombinant carcinoembryonic antigen (CEA) in Immunol 1995; 84: 173–176. mice primed with the recombinant vaccinia CEA. J Immunother 10 Boyle JS, Koniaras C, Lew AM. Influence of cellular location and 1994; 16: 275–282. expressed antigen on the efficacy of DNA vaccination: cytotoxic 31 Kantor J et al. Antitumor activity and immune responses T lymphocyte and antibody responses are suboptimal when induced by a recombinant carcinoembryonic antigen-vaccinia antigen is cytoplasmic after intramuscular DNA injection. Int virus vaccine. J Natl Can Inst 1992; 84 : 1084–1091. Immunol 1997; 9: 1897–1906. 32 McLaughlin JP, Schlom J, Kontor JA, Greiner JW. Improved 11 Schirmbeck R, Bo¨hm W, Reimann J. DNA vaccination primes immunotherapy of a recombinant carcinoembryonic antigen MHC class I-restricted, Simian virus 40 large tumor antigen-spe- vaccinia vaccine when given in combination with interleukin-2. cific CTL in H-2d mice that reject syngeneic tumors. J Immunol Cancer Res 1996; 56: 2361–2367. 1996; 157: 3550–3558. 33 Bei R et al. The use of a cationic liposome formulation (DOTAP) 12 Thompson J, Zimmermann W. The carcinoembryonic antigen mixed with a recombinant tumor-associated antigen to induce gene family: structure, expression and evolution. Tumor Biol immune responses and protective immunity in mice. J Immu- 1988; 9: 63–83. nother 1998; 21: 159–169. 13 Thompson JA, Grunert F, Zimmermann W. Carcinoembryonic 34 Tsuji T et al. Enhancement of cell-mediated immunity against antigen gene family: molecular biology and clinical perspec- HIV-1 induced by coinoculation of plasmid-encoded HIV-1 anti- tives. J Clin Lab Anal 1991; 5: 344–366. gen with plasmid expressing IL-12. J Immunol 1997; 158: 4008– 14 Hodge JW. Carcinoembryonic antigen as a target for cancer vac- 4013. cines. Cancer Immunol Immunother 1996; 43: 127–134. 35 Kim JJ et al. In vivo engineering of a cellar immune response by 15 Hunter RL, Olsen MR, Bennett B. Copolymer adjuvants and tit- coadministration of IL-12 expression vector with a DNA immu- ermax. In: Stewart-Tull DES (ed). The Theory and Practical Appli- nogen. J Immunol 1997; 158: 816–826. cation of Adjuvants. John Wiley: Chichester, 1995, pp 51–94. 36 Michael G, Gesien A, Tokushige K, Wands JR. Enhancement of 16 Allsopp CE et al. Comparison of numerous delivery systems for cellular and humoral immune responses to hepatitis C virus the induction of cytotoxic T lymphocytes by immunization. Eur core protein using DNA-based vaccines augmented with cyto- J Immunol 1996; 25: 1137–1140. kine-expressing plasmids. J Immunol 1997; 158: 1231–1237. 17 Ishii KJ et al. Activity and safety of DNA plasmids encoding IL- 37 OGarra A. Cytokines induce the development of functionally 4 and IFNgamma. Gene Therapy 1999; 6: 237–244. heterogeneous T helper cell subsets. Immunity 1998; 8: 275–283. 18 Sin JI et al. In vivo modulation of vaccine-induced immune 38 Sgadari C, Angiolillo AL, Tosato G. Inhibition of angiogenesis responses toward a Th1 phenotype increases potency and vac- by interleukin-12 is mediated by the interferon-inducible protein cine effectiveness in a herpes simplex virus type 2 mouse model. 10. Blood 1996; 87–9: 3877–3882. J Virol 1999; 73: 501–509. 39 Macatonia SE, Hsieh C, Murphy KM, O’Garra A. Dendritic cells 19 Kim JJ et al. Modulation of amplitude and direction of in vivo and macrophages are required for Th1development of CD4+ T immune responses by co-administration of cytokine gene cells from ␣␤-TCR transgenic mice: IL-12 substitution for macro- expression cassettes with DNA immunogens. Euro J Immunol phages to stimulate IFN␥ production is IFN␥-dependent. Int 1998; 28: 1089–1103. Immunol 1993; 5: 1119–1128. 20 Chow YH et al. Development of Th1 and Th2 populations and 40 Kobayashi M et al. Identification and purification of natural the nature of immune responses to hepatitis B virus DNA vac- killer cell stimulatory factor (NKSF), a cytokine with multiple cines can be modulated by codelivery of various cytokine genes. biologic effects on human lymphocytes. J Exp Med 1989; 170: J Immunol 1998; 160: 1320–1329. 827–845. 21 Conry RM et al. A carcinoembryonic antigen polynucleotide 41 Chatelain R, Varkila K, Coffman RL. IL-4 induces a Th2 vaccine has in vivo antitumor activity. Gene Therapy 1995; 2: response in Leishmania major-infected mice. J Immunol 1992; 59–65. 148: 1182–1187. 22 Coffman RL, Lebman DA, Rothman P. Mechanism and regu- 42 Seder RA et al. The presence of interleukin 4 during in vitro lation of immunoglobulin isotype switching. Adv Immunol 1993; priming determines the lymphokine-producing potential of 54: 229–270. CD4+ T cells from T cell receptor transgenic mice. J Exp Med 23 Bloom ET, Horvath JA. Cellular and molecular mechanisms of 1992; 176: 1091–1098. the IL-12-induced increase in allospecific murine cytolytic T cell 43 Liew FY et al. Macrophage activation by interferon-gamma from activity: implications for the age-related decline in CTL. J Immu- host-protective T cells is inhibited by interleukin (IL)3 and IL4 nol 1994; 152: 4242–4254. produced by disease-promoting T cells in leishmaniasis. Eur J 24 Seder RA, Gassinelli R, Sher A, Paul WE. IL-12 acts directly on Immunol 1989; 19: 1227–1232. CD4+ T cells to enhance priming for IFN␥ production and 44 Paul WE, Ohara J. B-cell stimulatory factor-1/interleukin-4. diminishes IL-4 inhibition in such priming. Proc Natl Acad Sci Annu Rev Immunol 1987; 5: 429–459. USA 1993; 90: 10188–10192. 45 Nastala CL et al. Recombinant IL-12 administration induces 25 Abbas AK, Murphy KM, Sher A. Functional diversity of helper tumor regression in association with IFN␥ production. J Immu- T lymphocytes. Nature 1996; 383: 787–793. nol 1994; 153: 1697–1706. 26 Tahara H, Lotze MT. Antitumor effects of interleukin-12 (IL-12): 46 Zitvogel L et al. Cancer immunotherapy of established tumors applications for the immunotherapy and gene therapy of cancer. with IL-12. Effective delivery by genetically engineered fibro- Gene Therapy 1995; 2: 96–106. blasts. J Immunol 1995; 155: 1393–1403. 27 Schultz J et al. Long-lasting anti-metastatic efficiency of interleu- 47 Martinotti A et al. CD4 T cells inhibit in vivo the CD8-mediated kin 12-encoding plasmid DNA. Hum Gene Ther 1999; 10: 407– immune response against murine colon carcinoma cells transduced 417. with interleukin-12 genes. Eur J Immunol 1995; 25: 137–146.

Gene Therapy Regulation of anti-tumor response by cytokine vectors K Song et al 492 48 Bramson JL et al. Direct intratumoral injection of an adenovirus (CD86) in vitro and in inﬂammatory lesions. J Immunol 1998; 161: expressing interleukin-12 induces regression and long-lasting 5943–5951. immunity that is associated with highly localized expression of 55 Erwing LP, Kluth DC, Walsh GM, Ress AJ. Initial cytokine interleukin-12. Hum Gene Ther 1996; 7: 1995–2002. exposure determines function of macrophages and renders them 49 Rakhmilevich AL et al. Cytokine gene therapy of cancer using corresponsive to other cytokines. J Immunol 1998; 161: 1983– gene gun technology: superior antitumor activity of interleukin- 1988. 12. Hum Gene Ther 1997; 8: 1303–1311. 56 Sato Y et al. Immune-stimulatory DNA sequences necessary for 50 Coor M, Lee DJ, Carson DA, Tighe H. Gene vaccination with effective intradermal gene immunization. Science 1996; 273: naked plasmid DNA: mechanism of CTL priming. J Exp Med 352–354. 1996; 184: 1555–1560. 57 Tepper RI, Coffman RL, Leder P. An eosinophil-dependent 51 Ulmer JB et al. Generation of MHC class I-restricted cytotoxic T mechanism for the anti-tumor effect of interleukin-4. Science lymphocytes by expression of a viral protein in muscle cells: 1992; 257: 548–551. antigen presentation by non-muscle cells. Immunol 1996; 89: 58 Chomarat P, Rybak ME, Banchereau J. Interkeukin-4. In: Thom- 59–67. son A (ed). The Cytokine Handbook. Academic Press: San Diego, 52 Iwasaki A et al. The dominant role of bone marrow-derived cells 1998, pp 133–174. in CTL induction following plasmid DNA immunization at dif- 59 I. Xiang Z, Ertl HC. Manipulation of the immune response to a ferent sites. J Immunol 1997; 159: 11–14. plasmid-encoded viral antigen by coinoculation with plasmids 53 Conry RM et al. Selected strategies to augment polynucleotide expressing cytokines. Immunity. 1995; 2: 129–135. immunization. Gene Therapy 1996; 3: 67–74. 60 Hartikka J et al. An improved plasmid DNA expression vector 54 Behrens L et al. Human muscle cells express a functional for direct injection into skeletal muscle. Hum Gene Ther 1996; 7: costimulatory molecule distinct from B7.1 (CD80) and B7.2 1205–1217.

Gene Therapy