Gene Therapy (1999) 6, 865–872  1999 Stockton Press All rights reserved 0969-7128/99 $12.00 http://www.stockton-press.co.uk/gt Cytotoxic T lymphocyte response against non- immunoselected tumor antigens predicts the outcome of gene therapy with IL-12-transduced tumor cell

M Rodolfo, C Zilocchi, B Cappetti, G Parmiani, C Melani and MP Colombo Experimental Oncology D, Istituto Nazionale per lo Studio e la Cura dei Tumori, Milan, Italy

The colon adenocarcinoma C26, carrying two endogenous related tumor antigens, whereas nonresponders had CTL tumor-associated antigens (TAA) recognized by CTL, has that recognized preferentially the FR␣ antigen. CD8 from been transduced with the gene coding for the human folate responder mice were characterized to release high levels receptor ␣ (FR␣) as an additional antigen in order to study of granulocyte–macrophage (GM)-CSF upon antigen the efficacy of vaccination against a tumor expressing mul- stimulation. Tumors obtained from mice that died despite tiple antigens. A dicistronic vector was used to transduce vaccination lost expression of the FR␣ transgene but main- the IL-12 genes to create C26/IL-12/FR␣ that has been tained expression of endogenous C26 antigens. Immuno- used as a cellular vaccine to treat mice bearing lung met- selection against FR␣ antigen was not observed in tumors astases of C26/FR␣. After vaccination mice were partially from non-vaccinated controls and from CD8-depleted vac- splenectomized and splenic lymphocytes frozen and used cinated mice. Down-regulation of FR␣ antigen expression retrospectively to study in vitro CD8 T cell response related was due, at least in part, to methylation of retroviral vector to the treatment outcome. Vaccination cured 50% of mice long terminal repeat promoter since FR␣ expression was and the effect was CD8 T cell dependent. Mice either cured partially restored, ex vivo, by treatment with 5-aza-2Ј- (responders) or not cured (nonresponders) by vaccination deoxy-cytidine (aza). These results indicate that CD8 T developed tumor-specific CTL. However, analysis of CTL cell-mediated immunoselection and production of GM-CSF specificity and pCTL frequencies revealed that responders are determining factors for the efficacy of tumor . had a predominant CTL activity against endogenous C26-

Keywords: gene transfer; tumor immunity; CTL

Introduction against immunodominant antigens, such as melanoma- associated MART-1/MelanA or Pmel17/gp100, can Active tumor by vaccination with mediate in vivo destruction of antigen-positive tumor irradiated neoplastic cells is expected to activate an cells but also select for antigen loss variants. It has been against the entire repertoire of tumor- reported that during tumor progression, antigenic vari- associated antigens (TAA), a few of which have been ants can appear which may loose not only the HLA class 1,2 molecularly defined. Data have emerged that encour- I molecules, but also T cell epitopes.7–9 age the use of cellular-based vaccines for melanoma In this study we have analyzed the CD8 T cell response 3,4 patients. Active immunotherapy can be further associated with the therapeutic activity following IL-12- improved by genetic modification of tumor cells through transduced tumor cell vaccination. Considering that 5 insertion and expression of cytokine genes. In fact, pre- immunization may direct the immune response against clinical studies have shown that the injection of such an antigen that is therapeutically irrelevant or lost engineered tumor cells induces a local inflammatory because of immune selection, we have set up a model reaction leading to the activation of a systemic antitumor system in which vaccination therapy is directed to a memory response and, in some cases, to the cure of estab- tumor carrying more than one known antigen. This 6 lished tumors. enables testing which antigen is the relevant target of A major issue in the use of cell-based vaccines is the host-induced immune response but also the relationship role of the different antigens or subsets of antigens between immunodominance and immunoselection. We expressed by tumor cells in eliciting a T cell response used the C26 colon carcinoma which expresses as immu- determining tumor regression. In fact, CTL responses nodominant TAA the Ld restricted peptide gp70 (423– 431), called AH1, encoded by env gene of the endogenous ecotropic murine leukemia virus (MuLV) emv-1.10 In Correspondence: M Rodolfo, Experimental Oncology D, Istituto Nazionale addition, C26 express at least one additional, but as yet Tumori, via Venezian, 1-20133 Milan, Italy molecularly undefined, subdominant tumor rejection Received 12 August 1998; accepted 25 November 1998 antigen recognized by a CTL clone,11 TA9A (tumor anti- CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 866 gen 9A). The antigenic repertoire of C26 cells was further increased by transducing the human gene coding for the folate receptor ␣ (FR␣), a gp38 binding protein which is overexpressed on human carcinomas especially of the ovary,12,13 which encodes for at least two different pep- tides recognized by CTL in the context of either Kd or DdLd MHC class I of BALB/c mice (our unpublished data). IL-12 was chosen because of its potent antitumor effects.14 We have previously reported that vaccination with irradiated C26 cells engineered to produce IL-12 were more active than C26 cells transduced with IL-2 in curing mice bearing C51, another colon carcinoma shar- ing the emv-1-derived TAA.15 We have now examined whether the antigen specificity and cytokine profile of CTL generated by IL-12 vaccine can be associated with tumor eradication or with progression of antigen loss variants.

Results

Active immunotherapy with C26/IL-12/FR␣ tumor cell is dependent on CD8+ T lymphocytes C26 cells engineered to produce IL-12 and to express the human FR␣ (C26/IL-12/FR␣) were used to treat mice bearing lung metastases of C26/FR␣. Cell vaccination, given s.c. biweekly for 2 weeks, cured 13 of 26 mice while all 29 untreated counterparts died of lung metastases within 50 days (Figure 1a, P Ͻ 0.0001 by log-rank test). IL-12 production determined the efficacy of the vaccine, since treatment with C26/FR␣ cells (not transduced with IL-12 genes) did not change mice survival significantly (Figure 1b). The therapeutic effect was dependent on host CD8 T cells. In fact, no cures or increase of survival time were observed in CD8-depleted, vaccinated mice (Figure 1c). In contrast, the effectiveness of C26/IL-12/FR␣ cell vaccination was not changed by depletion of CD4 lym- phocytes: in fact five of 13 vaccinated mice were cured while 13 of 13 controls died of metastatic disease (Figure 1d, P Ͻ 0.001). We then tested whether partial splenectomy performed at the end of vaccination therapy affected the treatment outcome. Figure 1e shows survival curves of controls and vaccinated mice that underwent hemisplenectomy 3 days after the last vaccine injection (day 16). Partial spleen Figure 1 Survival of mice bearing C26/FR␣ lung metastases after removal by surgery did not change the rate of survivors vaccination with C26/IL-12/FR␣ (a), or with C26/FR␣ cells (b), after CD8 (five of 10 cured mice, P Ͻ 0.01) or the survival time from (c) or CD4 (d) depletion, and after partial splenectomy (e). Mice injected that of non-splectomized mice reported in Figure 1a. This i.v. with 5 × 103 C26/FR␣ cells (day 0) were vaccinated with 3 × 106 150 surgical procedure enabled us to collect individual lym- Gy irradiated tumor cells injected s.c. on days 3, 6, 9 and 13. Treatment phocyte samples after vaccination and to analyze the with anti-CD4 and anti-CD8 mAbs were initiated on day 1 and repeated antitumor response in mice identified as responders or every 2 weeks. Partial splenectomy was performed on day 16. nonresponders on the basis of the therapeutic outcome. directed against FR␣-transduced BALB/c, HTG and 5R CTL from cured mice (responders) display a preferential fibroblast lines (indicating that FR␣ codes for antigenic recognition of C26-related TAA/s on C26/FR␣ cells than peptides associated with either H-2Dd or Kd, CTL from nonresponder mice respectively), while nontransduced HTG and 5R were not Lymphocytes from mice vaccinated with C26/IL-12/FR␣ lysed (Figure 2, upper panel). CTL activity was MHC were tested for CTL activity against a panel of target cells restricted as inhibited by mAbs to either CD3 or CD8 expressing different TAAs of the C26/FR␣ tumor. (data not shown). Untreated tumor-bearing controls did Cytolysis was directed against C26 and syngeneic F1- not show significant CTL activity. transformed fibroblasts pulsed with the AH1 peptide, the Splenocytes from mice cured (responders) or not cured immunodominant CTL antigen eluted from the C26 vari- (nonresponders) by C26/IL-12/FR␣ cell vaccination were ant CT26,10 but not against F1 cells not pulsed or pulsed then separately tested for CTL activity against the same with an irrelevant peptide (not shown). CTL were also panel of target cells. Although lymphocytes from CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 867

Figure 2 In vitro lytic activity of lymphocytes from mice treated with C26/IL-12/FR␣ vaccines and from untreated controls (upper panel), and from mice cured (responders) or not cured (nonresponders) (lower panel) against a panel of target cells expressing the different tumor antigens. Splenocytes obtained from four to six mice in each group were pooled and tested after 6 days MTLC with irradiated C26/FR␣ cells. responders and nonresponders lysed C26/FR␣ tumor anti-C26/FR␣ CTLp frequencies were measured (data not cells to the same extent, responders’ lymphocytes dis- shown). On the contrary the frequencies of CTLp played a stronger response to C26-associated antigens detected after restimulation with the AH1 peptide were (Figure 2, lower panel), a result confirmed by comparing shown to be similar between the two groups (166 CTL activity of individual mice (not shown). CTLp/107 responder lymphocytes versus 142 CTLp/107 Comparison of anti-C26 CTLp frequencies between nonresponder cells, Figure 3b), indicating that CTL reco- responders and nonresponders by LDA further con- gnizing C26 TAA/s other than the immunodominant firmed these data (Figure 3a). Frequencies of CTLp AH1 were activated in responder but not in non- directed to C26 evaluated simultaneously in three differ- responder mice. ent experiments ranged between 87 and 141 CTLp/107 Altogether, these data indicate that following vacci- lymphocytes for nonresponders and between 266 and 419 nation with irradiated C26/IL-12/FR␣ tumor cells both CTLp/107 lymphocytes for responder mice, while similar responders and nonresponders developed a cytolytic T

Figure 3 Frequencies of anti-C26 (a) and anti-AH1 (b) CTLp in lymphocytes from responder and nonresponder mice, and frequencies of CTLp producing GM-CSF obtained in the same LDA (c). The assays were carried out in parallel with the same pool of lymphocytes obtained from responder and nonresponder mice. For determination of anti-C26 CTLp frequency, lymphocytes were cultured with C26 cells, while for anti-AH1 CTLp lymphocytes were stimulated with 1 ␮g/ml AH1 peptide. On day 10 LDA plates were split in two replicate plates and 51Cr-labeled or unlabeled C26 cells were added; supernatants were recovered after 6 h for counting 51Cr release or after 24 h for GM-CSF quantification in ELISA, respectively. Results expressed as CTLp/107 lymphocytes; numbers in parentheses represent the upper and lower 95% confidence limits. A representative experiment is shown. CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 868 cell response against the tumor, but such response was further confirmed by LDA. Although the frequencies of either directed to different TAAs or to a different TAA anti-AH1 CTLp between responders and nonresponders array. was similar (Figure 3b), the frequency of T cell precursors producing GM-CSF measured within the same assay was Tumor-specific CD8 lymphocytes produce more GM- higher in responders than in nonresponders (Figure 3c). CSF than lymphocytes from nonresponder mice To compare the pattern of produced by CD8 Tumor cells isolated from pulmonary metastases of lymphocytes from responder and nonresponder mice, nonresponder mice lack FR␣ but retain C26 TAA MLTC-activated lymphocytes were depleted of CD4+ expression cells, cocultured with tumor cells and the cytokines Since a strong antitumor CTL response was shown in released in the supernatant evaluated by specific ELISA. nonresponder mice which did not benefit from vacci- Of the tested cytokines (IL-2, IL-4, IL-10, TNF␣, IFN␥ and nation therapy, we asked whether antigen loss tumor GM-CSF), only IFN␥ and GM-CSF were reproducibly variants outgrew and killed these mice. Tumors obtained detected. The levels of IFN␥ released by CD8 T lympho- at death from vaccinated mice showed a complete lack cytes after stimulation with C26/FR␣ or with F1/FR␣ (seven of nine tumors) or reduction below 50% positivity cells or with AH1 peptide were similar in responder and (two of nine tumors) of FR␣ expression as detected by nonresponder mice, while stimulation with C26 cells flow cytometry with FR␣-specific mAb. The same analy- induced a higher IFN␥ release by responder lymphocytes sis carried out on tumors obtained at death from non- (Figure 4, upper panel), a result consistent with the differ- vaccinated controls, showed that only two of nine tumors ent CTL response to C26-related TAA described above. were negative for FR␣ expression (Figure 5a, P Ͻ 0.0001 On the contrary, the levels of GM-CSF produced by CD8 by Student’s t test). lymphocytes from responder mice were significantly The same type of analysis was performed at the end higher than those of lymphocytes from nonresponder of vaccination in mice depleted of CD8 or CD4 cells. Mice mice regardless of the type of stimulating cells (Figure 4, were killed and lung tumors were tested for FR␣ lower panel). In addition, these data indicate that C26- expression. In CD8-depleted mice tumor progression related TAA other than AH1 were recognized by CD8 occurred despite vaccination (see Figure 1c) and FR␣ on cells from responders. these tumors remained highly expressed (Figure 5b), sug- The difference in GM-CSF release by antigen-specific gesting that in the absence of CD8 immunoselection did CD8 cells between responders and nonresponders was not take place. On the contrary, CD4 depletion, which did not reduce the efficacy of vaccination (see Figure 1d) allowed selection of FR␣-negative variants; in fact FR␣ was detected in less than 50% of tumor cells (Figure 5c). Since tumors from cured mice cannot be examined, the data suggest that immunoselection of FR␣ is likely to occur in all vaccinated mice. To test whether the ‘endogenous’ set of TAA present on C26 underwent immunoselection, their expression was tested using the CTL clone TIL/26 recognizing AH1 peptide and the CTL clone 9A/89 specific for the as yet undefined, subdominant C26 antigen TAA9A. Six tumors removed at death from vaccinated mice, four showing loss and two showing reduction of FR␣ expression in the experiment reported in Figure 5a, were tested as stimul- ating cells for induction of IFN␥ release by anti-C26 and anti-FR␣ CTL clones. All tumors were recognized by CTL

Figure 4 IFN␥ and GM-CSF release by CD8 lymphocytes from responder Figure 5 FR␣ antigen expression by tumor cells obtained from pulmonary and nonresponder mice upon stimulation with C26/FR␣, C26 or F1/FR␣ metastases of control and vaccinated mice as assessed by indirect cells, or with AH1 peptide pulsed F1 cells. Lymphocytes from 7 day MLTC immunofluorescence and FACS analysis. Tumors obtained from mice at were depleted of CD4+ cells and tested for cytokine production by 24 h death (a), and at the end of vaccinations in mice depleted of CD8 (b) or coculture with the indicated stimulating cells. Supernatants were of CD4 (c) lymphocytes are shown. Each dot represent a sample obtained recovered and cytokine release quantified by specific ELISA. Results as from a single mouse. For tumors shown in (a), survival times of vaccinated mean ± standard deviation of three independent experiments are shown. and control mice were within the same range. CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 869 immunologic parameters before and after treatment; however, which of such parameters correlates with clini- cal response remains undetermined.16 We and others have shown that the induction of antitumor CTL does not correlate with tumor regression in both humans17–19 and mice.15,20 The present study corroborates this con- clusion and shows that activation of CD8 cells is neces- sary but not sufficient to induce the eradication of lung metastases. In this tumor model antitumor CD8 T lym- phocytes displayed a different pattern of reactivity between mice that were cured or not cured by vacci- nation, and efficiently caused tumor immunoselection. Successful therapy was clearly associated with activation ␥ of CD8 T cells endowed with GM-CSF production and Figure 6 IFN production by CTL clones upon stimulation with tumors with specificity for antigens that are not lost during obtained at death from vaccinated nonresponder mice. Six tumors analyzed for FR␣ antigen expression in the experiment reported in Figure tumor progression. 5a were used. IFN␥ release upon stimulation with C26/FR␣ cells kept in Although this model is artificial because of transgene culture was included as positive control. expression under an heterogous promoter, our finding may offer a possible interpretation of events occurring following vaccination with cytokine-transduced tumor clones TIL/26 and 9A/89, indicating that C26 TAAs were cells. It is likely that the outgrowth of antigen loss tumor expressed, whereas only two of six tumors were recog- variants is causally associated with the therapeutic effi- nized by anti-FR␣ CTL clone IF1 (Figure 6), thus con- cacy of the treatment. If the target antigen of the domi- firming immunostaining data. All CTL clones were nant response is immunoselected, the weaker response to shown to recognize C26/FR␣ but not F1 cells maintained the remaining antigen(s) may not be sufficient to cause in culture (not shown). These data indicate that FR␣ but tumor eradication. In this view, tumor escapes if CTL are not C26-related TAAs were immunoselected in vivo primed against antigens that can be lost. When the after vaccination. response directed against the C26-related TAAs was We tested the possibility that antigen loss could also strong enough all mice were cured whereas a dominant be due to in vivo outgrowth of FR␣ nontransduced cells. response to the FR␣ antigen resulted in the outgrowth of Four tumors lacking FR␣ expression were found to be antigen loss tumors. In fact, tumors from mice which G418 resistant when recultured in vitro, suggesting that died despite vaccination were negative for FR␣ but posi- the transduced genes were not lost but rather down-regu- tive for C26-TAAs. The different specificity of the CTL lated. In fact, treatment of tumor cells with the demethyl- response can be due to the early expansion of the few ating agent aza restored FR␣ expression, albeit in a lim- CTLp that first encounter a certain tumor antigen. In the ited fraction of cells, in all tested tumors as detected by P815 tumor model CTL response has been shown to be FACS analysis (Table 1). directed against different TAAs when individual responses were considered.21 Of course, the expression of Discussion FR␣ cannot be examined in tumors from cured mice, thus antigen loss cannot be causally associated with the failure Several clinical phase I studies have been initiated to test of immunotherapeutic treatment. Reduction of FR␣ posi- the effects of vaccinations with gene-modified tumor cells tivity below 50% was observed when tumors were exam- or peptides.2,5 The efficacy of vaccination on the gener- ined after vaccination, while FR␣ expression remained ation of immune response is evaluated by comparing Ͼ50% in CD8-depleted mice. Since it cannot be excluded that both rejected and non-rejected tumors had lost expression of FR␣, tumor eradication may ultimately Table 1 Induction of FR␣ expression in FR␣-negative tumors depend on the occurrence of ‘epitope spreading’ includ- after aza treatment ing other TAA following the response against FR␣.22 In this view CD8-released GM-CSF may also be instrumen- Tumors Aza treatment % Cell staining with anti-FR␣ tal for the initiation of a response against subdominant (mean) epitopes, through APC recruitment, activation and processing of other antigens. The higher frequency of T 1 − 1 (6) cells producing GM-CSF may reflect the stronger CTL + 48 (47) activation occurring in responders. 2 − 1 (15) + 26 (36) Heterogeneity of tumors may represent a substantial 3 − 1 (10) problem for immunotherapy. Different mechanisms of + 28 (47) tumor escape from immune attack are likely to operate 4 − 3 (18) in vivo and tumor immunoselection has been documented + 35 (58) in patients after treatment and in preclinical stud- ␣ − C26/FR 100 (170) ies also.7–9,23–25 Our data indicate that CD8 response can be efficient in selecting antigen loss tumor variants. The Cell staining with mAb MOV19 and FITC anti-mouse Ig was use of an artificial system hampered the difficulties of determined by FACS analysis. Aza was added to cultured ␮ ␣ investigating the mechanism of immunoselection. How- tumors at the dose of 1 m for 72 h. C26/FR cells maintained ␣ in culture were tested as control. ever, transducing human FR gene into C26 allowed the increase of the antigenic repertoire of the tumor such to CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 870 include a strong, known antigen which may undergo Materials and methods immunoselection. Other tumor models which display more than two endogenous known or detectable antigens FR␣ vector construction and retroviral transduction ␣ are not available. Furthermore, by placing the FR gene A 910-bp EcoRI fragment of human gp38 FR␣ was cloned under the LTR promoter we were able to document that into the retroviral vector LXSN, that was oriented by immunoselection was associated with promoter down- restriction analysis and transfected into the amphotropic regulation since treatment with the demethylating agent packaging cell line gp+AM12. Transiently produced virus ␣ aza partially restored FR TAA expression by immuno- particles were used to infect the ecotropic packaging selected tumors. This finding may be relevant since aza gp+E86 as described.15 C26 tumor cells, B10.HTG, treatment has been shown to induce the expression of B10.A(5R) (thereafter called HTG and 5R, respectively) MAGE-1 in melanoma and melanocytes26,27 indicating and F1 BALB/c fibroblast cell lines were infected with that demethylation can affect the expression of TAA the viral supernatant in the presence of polybrene and which are silent in normal tissues. We have also observed transduced cells were isolated by G418 selection. FR␣ that aza treatment induced the expression of AH1 TAA expression was assayed by flow cytometry using the in F1 transformed fibroblasts (M Rodolfo, unpublished). MOV19 mAb38 and was shown to be stable over several Further experiments are needed to see whether CD8 months of continuous in vitro culture. immunoselection may occur through methylation of TAA promoters other than retroviral LTR. Transcriptional Cell lines inactivation of proviral LTR by methylation has been C26/IL-12/FR␣ cells were obtained by infecting IL-12- described in hematopoietic stem cells, in fibroblasts and producing C26 cells15 by cocultivation with irradiated in embryonic stem cells.28 E86/FR␣ producer cells in the presence of polybrene. IL-12 production by cell vaccine was shown to be Transduced cells were separated by two rounds of mag- instrumental in obtaining a curative response in different netic cell sorting with paramagnetic beads coupled with tumor models; furthermore, its presence ensures that a anti-mouse IgG (MiniMacs; Miltenyi Biotec, Bergisch Th1 type response is induced independently by the nat- Gladbach, Germany) after incubation with mAb MOV19. ure of the stimulating antigens.14 In fact, lymphocytes C26/IL-12/FR␣ cells were shown to produce 1–2 ng/ml from both responder and nonresponder mice were cyto- of IL-12 (106 cells/ml/48h) by specific ELISA and to toxic and able to produce IFN␥. Although local pro- express high levels of FR␣. The CTL line IF1 specific for duction of IFN␥ has been associated with a favorable out- FR␣ was obtained from spleen lymphocytes of mice ␣ come of therapy in different models,29,30 in the present immunized with F1/FR cells. The method of obtaining, study production of GM-CSF was the hallmark of lym- maintenance and reactivity of anti-C26 CTL clones 11 phocytes from responder mice. The ability to produce TIL/26 and 9A/89 has been previously reported. CTL d GM-CSF has been associated with the therapeutic activity TIL/26 was shown to recognize the L -restricted peptide 10 31 AH1, while CTL 9A/89 recognizes C26-specific TAA9A of CD8 lymphocytes in other preclinical studies, and d GM-CSF production by infused T cells appeared to be an in association with H-2 K . important predictor of tumor response in melanoma and renal cancer patients.32,33 GM-CSF released by antigen- Immunotherapy experiments specific T cells which reach the tumor site may either Two-month-old female BALB/cnAnCr mice obtained potentiate the effect of concomitant produced IFN␥ and from Charles River (Calco, Italy) and maintained accord- other secondary cytokines, and recruit and activate ing to institutional and EC guidelines were used through- out the study. Immunotherapy experiments were carried macrophages and granulocytes that are associated with 15 tumor killing through nonspecific bystander effects.6,34,35 out as described. Lung metastases were induced by the × 3 ␣ In addition, we showed that GM-CSF can substitute for i.v. injection of 5 10 C26/FR cells and mice were vac- × 6 IFN␥ in inducing rejection of C26/IL-12 injected into cinated on days 3, 6, 9 and 13, by injecting s.c. 3 10 150 IFN␥ KO mice.36 Gy irradiated tumor cells. CD4 and CD8 cell depletion were carried out by treating mice with mAbs GK1.5 or IL-12 has also been reported to modulate immunodom- 2.43 (0.3 mg i.p.) on day 1 after i.v. tumor injection and inance of cytotoxic T lymphocytes epitopes coinjected in every 2 weeks thereafter. For survival follow-up mice vivo.37 The detection of CTL specific for C26-related TAA were killed when showing dyspnea. Vaccinated mice other than the immunodominant AH1 in lymphocytes were tagged and individual lymphocyte samples were from responder mice suggests that IL-12 produced by obtained on day 16 by hemisplenectomy under tumor vaccines may modulate TAA immunodominance. anesthesia. AH1 peptide has been eluted from CT26 and described as immunodominant for CTL response but insufficient to Cell-mediated cytotoxicity assay and precursor (p) CTL stimulate antitumor response, probably because of the determination in limiting dilution analysis (LDA) assay lack of a helper signal.10 We may indirectly confirm its CTL response and specificity were assessed in standard immunodominance, since 19 of 20 CTL clones obtained 4h 51Cr-release assays after 6 days mixed lymphocyte from C26 tumor-immune or tumor-bearing mice after tumor cell cultures (MLTC) as previously described.15 repeated in vitro stimulations with irradiated C26 cells Pools of lymphocytes (from responder and non- recognized the AH1 peptide. The remaining clone responders) were obtained with the same numbers of (9A/89), however, detects a different, as yet undefined cells from each individual sample (10–30 × 106 cells). The TAA thus indicating the existence of additional antigens synthetic peptide AH1 (SPSYVYHQF), gp70 env 423– on C26 cells. Studies aimed at the identification of this 431,10 was used at 1 ␮g/ml. For LDA lymphocytes were TAA are in progress. serially 1:2 diluted (from 8 × 104/w to 500/w) and cul- CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 871 tured with irradiated tumor cells (5 × 103/w), 20 Gy transduction in the immunotherapy of cancer. Adv Pharmacol irradiated syngeneic splenocytes as feeder cells 1997; 40: 259–308. (5 × 105/w) and IL-2 (50 U/ml), in 32 identical replicates, 6 Musiani P et al. Cytokines, tumor-cell death and immuno- in 96 U-well plates for 10 days. 51Cr-release assays were genicity: a question of choice. Immunol Today 1997; 18: 32–36. performed with 103 target cells/w. Values of lysis 7 Jager E et al. Inverse relationship of melanocyte differentiation antigen expression in melanoma tissues and CD8+ cytotoxic-T exceeding three times the standard deviation of the mean cell responses: evidence for immunoselection of antigen-loss spontaneous release were used as a threshold for scoring variants in vivo. Int J Cancer 1996; 66: 470–476. positive cytolysis. Calculation of the frequencies from 8 Lehmann M et al. Differences in the antigens recognized by cyto- limiting dilution data was based on the single-hit Poisson lytic T cells on two successive metastases of a melanoma patient 39 model. Frequencies given by the number of responder are consistent with immune selection. Eur J Immunol 1995; 25: cells corresponding to 37% negative cultures, were then 343–347. calculated from the slope of the regression line. The same 9 Cormier JN et al. Heterogeneous expression of melanoma- method was used for calculation of frequencies of GM- associated antigens and HLA-A2 in metastatic melanoma in CSF producing CTLp. Fig P software (Biosoft, Cam- vivo. Int J Cancer 1998; 75: 517–524. bridge, UK) was used to calculate the confidence limits 10 Huang AS et al. The immunodominant major histocompatibility of the regression line. complex class I-restricted antigen of a murine colon tumor derives from an endogenous retroviral gene product. Proc Natl Acad Sci USA 1996; 93: 9730–9735. Cytokine production assay 11 Rodolfo M et al. Cytotoxic T lymphocytes recognize tumor anti- CD8 cells were enriched to 75–85% from MLTC lympho- gens of a murine colonic carcinoma by using different T cell + cytes by complement lysis of CD4 cells after incubation receptors. Int J Cancer 1994; 57: 440–447. with anti-CD4 mAb (GK1.5). In vitro cytokine production 12 Coney LR et al. Cloning of a tumor-associated antigen: MOv18 by CD8 cells from MLTC and by CTL clones was induced and MOv19 recognize a folate-binding protein. by incubating lymphocytes (1–2 × 105/w) with stimulat- Cancer Res 1991; 51: 6125–6132. ing tumor cells (105/w) in 96-well plates for 24 h. Super- 13 Mantovani LT et al. Folate binding protein distribution in nor- natants were assayed for cytokine content by specific mal tissues and biological fluids from ovarian carcinoma ELISA (PharMingen, San Diego, CA, USA). patients as detected by the monoclonal antibodies MOv18 and MOv19. Eur J Cancer 1994; 30: 363–369. 14 Trinchieri G. -12: a proinflammatory cytokine with Ex vivo tumor cell analysis immunoregulatory functions that bridge innate resistance and Lungs were removed from mice found dead during the antigen-specific adaptative immunity. Annu Rev Immunol 1995; survival follow-up or those killed after the course of vac- 13: 251–276. cinations, and tumor cells were obtained by enzymatic 15 Rodolfo M et al. Immunotherapy of experimental metastases by digestion. Cells were kept in culture for at least three vaccination with interleukin gene-transduced adenocarcinoma passages in order to eliminate contaminating normal cells sharing tumor-associated antigens. Comparison between cells. Treatment with 5-aza-2Ј-deoxycytidine (Sigma- IL-12 and IL-2 gene-transduced tumor cell vaccines. J Immunol Aldrich, Milan, Italy) was set on semiconfluent tumor 1996; 157: 5536–5542. cells (105/ml) at the dose of 1 ␮m for 72 h. 16 Salgaller ML. Monitoring of cancer patients undergoing active or passive immunotherapy. J Immunother 1997; 20: 1–14. 17 Arienti F et al. Limited antitumor T cell response in melanoma Acknowledgements patients vaccinated with IL-2 gene-transduced allogeneic mela- noma cells. Hum Gene Ther 1996; 7: 1955–1963. This work was financially supported by the Italian 18 Salgaller ML, Marincola FM, Cormier JN, Rosenberg SA. Association for Cancer Research (AIRC), by Istituto Immunization against epitopes in the human melanoma antigen Superiore di Sanita` Special Program on Gene Therapy gp100 following patient immunization with synthetic peptides. and by CNR. CZ is supported by a fellowship from the Cancer Res 1996; 56: 4749–4757. Italian Federation for Cancer Research (FIRC). We thank 19 Cormier JN et al. Enhancement of cellular immunity in mela- Dr Enzo Bronte (Universita` di Padova) and Dr Nick noma patients immunized with a peptide from MART-1/Melan Restifo (NIH, Bethesda) for providing reagents. The A. Cancer J Sc Am 1997; 3: 37–44. excellent technical expertise of Mariella Parenza and 20 Saito S et al. Immunotherapy of bladder cancer with cytokine gene-modified tumor vaccines. Cancer Res 1994; 54: 3516–3520. Ivano Arioli is gratefully acknowledged. We thank 21 Brichard V et al. Individual differences in the orientation of the Grazia Barp for editorial help. We also thank Dr Chiara cytolytic T cell response against mouse tumor P815. Eur J Immu- Castelli and Dr Pier Luigi Lollini for critical reading of nol 1995; 25: 664–671. the manuscript. 22 Vanderlugt CJ, Miller SD. Epitope spreading. Curr Opin Immunol 1996; 8: 831–836. 23 Van Waes C et al. Immunodominance deters the response to References other tumor antigens thereby favouring escape: prevention by 1 Boon T, Coulie PG, Van den Eynde B. Tumor antigens recog- vaccination with tumor variants selected with cloned cytolytic nized by T cells. Immunol Today 1997; 18: 267–268. cells in vitro. Tissue Antigens 1996; 47: 399–407. 2 Rosenberg SA. Cancer vaccines based on the identification of 24 Maeurer MJ et al. Tumor escape from immune recognition; genes encoding cancer regression antigens. Immunol Today 1997; lethal recurrent melanoma in a patient associated with down- 18: 175–185. regulation of the peptide transporter protein TAP-1 and loss of 3 Mitchell MS. Active specific immunotherapy of melanoma. Br expression of the immunodominant MART-1/Melan-A antigen. Med Bull 1995; 51: 631–646. J Clin Invest 1996; 98: 1633–1641. 4 Morton DL et al. Prolongation of survival in metastatic mela- 25 Seung S, Urban JL, Schreiber H. A tumor escape variant that noma after active specific immunotherapy with a new poly- has lost one major histocompatibility complex class I restriction valent melanoma vaccine. Ann Surg 1992; 216: 463–482. element induces specific CD8+ T cells to an antigen that no 5 Parmiani G, Colombo MP, Melani C, Arienti F. Cytokine gene longer serves as a target. J Exp Med 1993; 178: 933–940. CD8 reactivity and clinical response to tumor vaccines M Rodolfo et al 872 26 Weber M et al. Expression of the Mage-1 tumor antigen is up- 33 Chang AE et al. Adoptive immunotherapy with vaccine-primed regulated by the demethylating agent 5-aza-2Ј-deoxycytidine. lymph node cells secondarily activated with anti-CD3 and Cancer Res 1994; 54: 1766–1771. interleukin-2. J Clin Oncol 1997; 15: 796–807. 27 De Smet M et al. The activation of human gene MAGE-1 in 34 Grabstein KH et al. Induction of macrophage tumoricidal tumor cells is correlated with genome-wide demethylation. Proc activity by granulocyte–macrophage colony-stimulating factor. Natl Acad Sci USA 1996; 93: 7149–7153. Science 1986; 232: 506–508. 28 Habers K et al. DNA methylation and gene expression: endogen- 36 Thomassen MJ et al. Induction of cytokine messenger RNA and ous retroviral genome becomes infectious after molecular clon- secretion in alveolar macrophages and blood monocytes from ing. Proc Natl Acad Sci USA 1981; 78: 7609–7613. patients with lung cancer receiving granulocyte–macrophage 29 Tahara H et al. Effective eradication of established murine colony-stimulating factor therapy. Cancer Res 1991; 51: 857–862. tumors with IL-12 therapy using a polycistronic retroviral vec- 36 Zilocchi C et al. IFN␥-independent rejection of IL-12-transduced tor. J Immunol 1995; 154: 6466–6474. carcinoma cells requires CD4+ T cells and GM-CSF. J Exp Med 30 Hock H et al. Mechanisms of rejection induced by tumor cell- 1998; 188: 133–143. targeted gene transfer of interleukin-2, interleukin-4, interleu- 37 Eberl G et al. Immunodominance of cytotoxic T lymphocyte epi- kin-7, tumor necrosis factor ␣, or ␥. Proc Natl Acad topes coinjected in vivo and modulation by interleukin-12. Eur Sci USA 1993; 90: 2774–2778. J Immunol 1996; 26: 2709–2716. 31 Goedegebuure PS et al. Reactivity of murine tumour-infiltrating 38 Miotti S et al. Characterization of human ovarian carcinoma- lymphocytes with solid-phase anti-CD3 : in vitro cyto- associated antigens defined by novel monoclonal antibodies kine production is associated with in vivo efficacy. Surg Oncol with tumor-restricted specificity. Int J Cancer 1987; 39: 297–303. 1994; 3: 79–85. 39 Langhorne J. Limiting dilution analysis of precursors of cyto- 32 Schwartzentruber DJ et al. In vitro predictors of therapeutic toxic T lymphocytes. In: Lefkovits I, Pernis B (eds). Immunolog- response in melanoma patients receiving tumor-infiltrating lym- ical Methods. Vol 2. Academic Press: New York, 1981, pp 221– phocytes and interleukin-2. J Clin Oncol 1994; 12: 1475–1508. 231.