Generation of RAGE‐1 and MAGE‐9 peptide‐specific cytotoxic T‐Lymphocyte lines for transfer in patients with renal cell carcinoma

Renal cell carcinomas (RCCs) are supposed to be immunogenic, and several clinical trials of immunotherapy using tumor lysate‐pulsed dendritic cells (DCs) have been performed. We report on the generation of RAGE‐1 and MAGE‐9 peptide‐specific CTL lines. RAGE‐1 and MAGE‐9 are expressed in 56% and 38% of RCCs. Seven MAGE‐9‐ and 13 RAGE‐1‐derived peptides were found to be immunogenic in the context of the HLA‐A*0201 MHC. CTLs were generated by coculture with peptide‐pulsed, activated B cells, which were easily generated in great quantities and displayed functional activity for a prolonged period of time. MAGE‐9 and RAGE‐1 peptide‐specific CTL lines were strictly peptide‐specific and displayed high cytotoxic activity not only against peptide‐loaded T2 cells but also against HLA‐A*0201‐positive RCC lines, which naturally express MAGE‐9, RAGE‐1 or both. Thus, B cells are well suited as APCs for the generation of large numbers of tumor peptide‐specific CTLs for adoptive transfer. MAGE‐9 as well as RAGE‐1 may well provide suitable targets for immunotherapy of RCC. © 2005 Wiley‐Liss, Inc.


Flow cytometry
For FACS analysis, 3 3 10 5 cells were stained according to routine procedures. For cytokine and intracellular CD152 expression, cells were fixed and permeabilized in advance. Fluorescence was determined using a FACStar (Becton Dickinson).

Peptide selection and stability assay
HLA-A*0201 binding peptides of MAGE-9 and RAGE-1 were selected according to epitope prediction using SYFPEITHI and BIMAS. 66,67 Functional assays were performed with those peptides that ranked highest in both predictions. As internal control, the influenza pMP GILGFVFTL and an abundant self-peptide of the P68 protein, YLLPAIVHI, 68 were used. Binding stability was evaluated using TAP-deficient T2 cells. T2 cells (1 3 10 6 ) were cultured overnight with 5 mg/ml b 2 -microglobulin and 50 mg/ml peptide. Cells were washed and stained with anti-HLA-A2.1. The fluorescence index was calculated as (mean fluorescence intensity in the presence of peptides 2 mean fluorescence intensity in the absence of peptides)/mean fluorescence intensity in the absence of peptides. 69,70 Binding stability of the peptides is shown in Table I. All peptides in which binding had been predicted by SYFPEITHI and BIMAS displayed a fluorescence index >0.3, which is supposed to be sufficient for antigen presentation. 71 Therefore, all peptides were included in further analysis.

Generation of APCs and long-term culture of B cells
For the generation of DCs, PBMCs were isolated by Ficoll-Hypaque centrifugation. Cells were washed, resuspended in RPMI supplemented with 5% autologous serum and incubated for 2 hr at 37°C, 5% CO 2 . Nonadherent cells were washed off, and adherent cells were incubated in RPMI-1640 supplemented with 5% autologous serum, 150 U/ml rhGM-CSF, 50 U/ml rhIL-4 and 50 U/ml rhIFN-g (Strathman, Hannover, Germany). Medium was exchanged after 3-4 days, and cells were maintained for 10 days. Maturation of DCs was evaluated by veiled appearance and fluorescence analysis, which revealed staining with anti-MHCI and anti-HLA-A*0201 in >90%, anti-MHCII in 80%, anti-CD14 in <10%, anti-CD40 in >50% and anti-CD80 and anti-CD86 in >60%. B-cell lines were generated from adherent cells after Ficoll Hypaque centrifugation of PBMCs. Cells were resuspended in Iscove's medium supplemented with 5% autologous serum, 100 U/ml rhIL-4 and 5.5 3 10 -7 M cyclosporin A. Cells (2 3 10 6 ) were seeded on irradiated (60 Gy) NIH3T3-CD154 cells (3 3 10 5 ) in 6-well plates. B cells were transferred to new NIH3T3-CD154coated plates every 3-5 days. After 21 days, cyclosporin A was omitted from the culture medium. Starting after 15 days, growth and maturation of CD154/CD40-activated B cells was controlled weekly by cell counting and flow cytometry. CD154/CD40-activated B cells were used for the generation of T-cell lines after 6 weeks of culture.

Proliferation assay
DCs were loaded for 3 hr at 37°C, 5% CO 2 with 10 mg/ml peptide. Cells were washed and mixed with adherent cell-depleted PBMCs at a ratio of 1:50. DCs (2 3 10 3 ) and T cells (1 3 10 5 ) were seeded in triplicate in 96-well plates and cultured for 5 days, adding 10 mCi 3 H-thymidine during the last 12-16 hr of culture. Plates were harvested with an automatic harvester, and 3 H-thymidine incorporation was evaluated in a b-counter.

IFN-c ELISpot assay
Nitrocellulose plates (96-well) were coated with 50 ml anti-IFNg (15 mg/ml). After overnight incubation at 4°C, plates were washed and free binding sites were blocked by incubation in PBS/ 1% BSA. T cells (3 3 10 4 /well) and peptide (4 ml of a 10 mg/ml solution) were added and incubated for 16-24 hr at 37°C. Plates were washed 6 times with PBS/0.05% Tween-20, and 100 ml biotinylated anti-hIFN-g (1:1,000 dilution) were added for 4 hr at room temperature. Plates were washed 4 times with PBS/0.05% Tween-20 and incubated with streptavidin-coupled alkaline phosphatase (1:1,000 dilution) for 2 hr. After washing, IFN-g-secreting cells were visualized by addition of 50 ml NBT/BCIP for 1 hr at room temperature. Spots were counted with an ELISPOT reader.  The fluorescence index defines the stability of HLA-A*0201 expression in peptide-loaded T2 cells, as described in Material and methods. The fluorescence index was derived from 4 hr after loading vs. time point 0.
Cytotoxicity assay 51 Cr-labeled target cells were incubated in triplicate in 96-well plates with effector T cells at ratios of (T:E) 1:3-100. Cells were incubated for 4 hr (T2 cells) and 8 hr (RCC lines) at 37°C. Plates were centrifuged, and aliquots of the supernatants were harvested and counted in a g-counter. Percent cytotoxicity was calculated as 100 3 (cpm test sample 2 cpm spontaneous release)/(cpm total release 2 cpm spontaneous release). Spontaneous release was in the range 8-15% (T2 cells) and 12-19% (tumor lines).

RT-PCR
mRNA was isolated with the Oligotex mRNA purification system (Qiagen, Hilden, Germany) using the manufacturer's protocol. cDNA was synthesized using 1 mg of polyA 1 RNA, MuMLV reverse transcriptase and oligo-dT. First-strand cDNA was subjected to RT-PCR amplification. Using polyA 1 RNA-derived cDNA, the following oligonucleotides were used for amplification: MAGE-9 (

Statistical evaluation
Statistical evaluation of in vitro assays was done by Student's t-test; significance of antigen expression was calculated by the sign test, significance of antigen coexpression by the concordance index and significance of differences in antigen expression by the McNemar test.

MAGE-9 and RAGE-1 expression in human RCC
We previously identified MAGE-9 expression in RCC by SSH. 52 Because only a limited number of immunotherapeutic targets for RCC are known, it became of interest to evaluate the frequency of MAGE-9 expression in RCC and its possible suitability as an immunogenic target. The expression profile of RAGE-1 in RCC is well known, and there are several reports on its immunogenicity. Therefore, RAGE-1 was used as a reference antigen throughout.
MAGE-9 and RAGE-1 expression was evaluated in 34 RCCs by RT-PCR. Due to a high degree of homology of different family members, primers were designed to amplify only small DNA fragments (197 and 150 bp for MAGE-9 and RAGE-1, respectively). As particularly with MAGE-9 additional bands became frequently visible, each RT-PCR product was sequenced to verify expression of MAGE-9 and RAGE-1. RAGE-1 was detected in 19 of 34 RCCs and MAGE-9 in 13 of 34 RCCs. G250, for comparison, was expressed in 71% of RCCs (Fig. 1, Table II). The concordance index of RAGE-1 and MAGE-9 expression was 0.08; i.e., coexpression was incidental. Also, the frequency of RAGE-1 and MAGE-9 expression did not differ significantly (sign test). There was no correlation between MAGE-9 or RAGE-1 expression and tumor stage (data not shown) or tumor grade (McNemar test). We could also not detect a correlation between MAGE-9 or RAGE-1 expression and histologic subtype. However, this accounts only for clear cell vs. non-clear cell RCC because 80% of the tissues were clear cell RCC and the number of, e.g., RCC tissues of the granular or oxyphil type was too small to allow for statistical evaluation.
Though RAGE-1 and MAGE-9 were less frequently expressed than G250, with expression rates of 56% (RAGE-1) and 38% (MAGE-9), respectively, evaluation of their potential as target structures for immunotherapy in RCC appeared justified.
Screening of HLA-A*0201 binding RAGE-1 and MAGE-9 peptides for induction of T-cell proliferation RAGE-1 and MAGE-9 peptides, which had been shown to bind to HLA-A*0201 (Table I), were used for DC loading and screened for induction of T-cell proliferation upon coculture. DC and T cells were derived from 4 healthy donors. All assays were done in triplicate and repeated 3-5 times. pMP served as positive control. Proliferation indices >2.0 were taken as indicating a response. The 4 donors' PBMC proliferation indices to pMP varied between 3.1 6 0.22 to 5.2 6 0.45. The proliferative response of the individual donors' PBMCs toward 7 MAGE-9 and 13 RAGE-1 peptides was rather constant, but the 4 donors' PBMCs displayed different reactivity profiles; i.e., not all donors' PBMCs reacted with the same peptides. Thus, proliferation indices of the 4 donors' PBMC toward peptide pRC were 2.41 6 0.24, 1.55 6 0.17, 2.76 6 0.68 and 2.89 6 0.52, respectively. Because no peptide has induced consistently a particularly strong response, T-cell lines were generated by concomitant stimulation with 5 peptides, which was considered to increase the likelihood of generating CTL lines (see below).

Generation of MAGE-9 and RAGE-1 peptide-specific T-cell lines using CD154/CD40-activated B cells as APCs
Though DCs are known to be potent APCs, due to their short lifetime after activation, repeated collection of PBMCs or bone marrow is required if patients should be vaccinated with tumor APCs or treated with in vitro activated T cells. Therefore, we considered the possibility of generating T-cell lines using activated B cells as APCs. B-cell lines were generated as described in Material and methods. For 28 days, B cells grew slowly, frequently encountering a crisis after 4 weeks. Thereafter, B cells grew exponentially, as demonstrated for B cells from 3 donors in Figure 2a. They were kept in culture for over 12 months. After 4-6 weeks, cultures contained only B cells (CD19 1 ) and B cells expressed MHC II, CD80, CD86 and CD40 at a high level (Fig. 2b). Expression of the costimulatory molecules remained unchanged throughout the culture period (data not shown).  (43) T cells of 5 donors were cultured with 4 peptide mixtures containing either M38-M42 or M43, M44, RA, RB and RC or RD-RH. As exemplified for one donor (Fig. 3a), T-cell expansion started after 4-5 weeks and exponential growth was observed after 7-8 weeks. Neither this donor's nor the other 4 donors' T cells proliferated in response to the peptide mix RI-RM. It should be mentioned that T-cell cultures which did not start to proliferate within 4-5 weeks died off after 8 weeks. T cells expressed continuously CD25, CD28, CD69 and CD152 at a high level; and a considerable percentage expressed IFN-g. CD8 1 cells were dominant already after 4-5 weeks. CD8 was expressed by >95% of cells after 10 weeks (Fig. 3b).
Thus, CD154/CD40-activated B cells are an easy-to-establish and long-lasting source of APCs that appears well suited for the generation of CTL lines.
Specificity and cytotoxic activity of RAGE-1 and MAGE-9 peptide-specific T-cell lines Evaluation of IFN-g secretion provided evidence that each Tcell line responded selectively to one or 2 individual peptides; i.e., the number of spots in response to the peptide mix corresponded to the number of spots observed with one of the peptides, whereas a background response was seen with the remaining peptides. As demonstrated for the T-cell lines shown in Figure 3, one culture contained T cells responding to pM39 and weakly to pM40, one culture specifically responded to pRC and one contained a mixture of cells responding either to pRE or pRG (Fig. 4a). Peptide selectivity was confirmed by evaluating specific lysis of peptide-loaded (10 mg/ml) T2 cells. The 3 T-cell lines lysed only T2 cells loaded with those peptides that induced IFN-g secretion but did not lyse T2 cells loaded with an irrelevant peptide or T2 cells loaded with a peptide that had been presented during culture but had not induced IFN-g secretion (Fig. 4b). We never succeeded in establishing a CTL line that did not secrete IFN-g. However, roughly 25% of IFN-g-secreting lines did not display cytotoxic activity (data not shown).
Though lysis of peptide-loaded T2 cells provided good evidence for functional activity of the T-cell lines, it was important to know whether these T cells would recognize the peptides naturally presented by the tumor cells. Eight HLA-A*0201 1 RCC lines were tested for RAGE-1 and MAGE-9 expression. Three of these lines,   (Table III).
In fact, the 3 T-cell lines lysed RCCs expressing the relevant tumor-associated antigen with close to 100% lysis observed at an E:T ratio of 100:1. T-cell lines did not lyse the NK target K562 or the HLA-A*0201 1 colon adenocarcinoma line SW707, which does not express RAGE-1 or MAGE-9. The MAGE-9 peptide-specific T-cell line did not lyse the KTCTL-30 line, which expresses RAGE-1 but not MAGE-9 (Fig. 5a).
The peptide specificity of these T-cell lines, which were generated by coculture with activated B cells loaded with a mix of 5 peptides, was further supported by cold target inhibition, using a 10-fold excess of unlabeled, peptide-loaded T2 cells. Inhibition of lysis of peptide-loaded T2 cells was only observed when T2 cells were loaded with the corresponding peptide. Accordingly, lysis of the KTCTL-30 line by the RAGE-1-specific T-cell line T2 could be inhibited by cold target T2 cells loaded with peptide RC but not by cold target T2 cells loaded with peptide M39 or peptide RE (data not shown).
To confirm the peptide specificity of the T-cell lines, T cells were restimulated with activated B cells loaded with individual peptides. T cells only survived when restimulated with the ''relevant'' peptide. Those T-cell lines were maintained in culture for over 3 months. After freezing and rethawing, the T-cell lines were restimulated and rescreened for their cytotoxic potential and peptide specificity. The T-cell lines maintained the characteristic fea-tures of CTLs (data not shown) and displayed high cytotoxic activity against the RCC lines expressing the target antigen, but not against the NK target or against the target antigen-negative, HLA-A*0201 1 or antigen-negative, HLA-A*0201 2 tumor line (Fig. 5b). Furthermore, inhibition of lysis of RCC lines by a 10-fold excess of cold target peptide-loaded T2 cells was strictly peptide-specific; i.e., lysis of KTCTL-53 by the pM39-specific CTL line could only be inhibited by pM39-loaded T2 cells and lysis by the pRC-specific CTL line could only be inhibited by pRC-loaded T2 cells. Lysis by the pRE-specific CTL line was also selectively inhibited by pRE-loaded T2 cells (Fig. 5c). Although all 3 CTL lines were inhibited exclusively by T2 cells loaded with the relevant peptide, inhibition by T2-pM39 and by T2-pRE was less efficient than inhibition by T2-pRC at an E:T ratio of 100:1. To confirm that the CTL lines recognize exclusively the respective peptides on the RCC lines, CTLs were titrated, keeping the ratio of RCC target to peptide-loaded T2 cold target cells constant at 1:20. At lower E:T ratios, cytotoxic activity of the pM39-(MAGE-9) and pRE-(RAGE-1) specific CTL lines could be completely inhibited by peptide-loaded T2 cells (Fig. 5d).
Thus, T cell lines generated with peptide-loaded B cells are capable of lysing efficiently peptide-loaded T2 cells as well as tumor cells presenting the naturally processed peptides.

Discussion
RCC has long been suggested to be an immunogenic tumor, 1,[7][8][9] and several clinical studies using tumor lysate-loaded or RNA/ DNA-infected DCs or DCs fused with tumor cells have confirmed induction of an immune response and improved survival  Figure 3 were tested for IFN-g secretion in response to the individual and pooled peptides used for activation. Peptide p68 served as control. Line 1 strongly responded to peptide pM39 and weakly to pM40 but not to pM38, pM40, pM41 and pM42; line 2 responded exclusively to pRC; and line 3 responded to pRE and weakly to pRG. As all 3 lines were derived from the same donor and reacted with all the peptides, the finding provides clear evidence for oligoclonal T-cell expansion.  Expression was evaluated by flow cytometry. Negative controls were adjusted to a mean fluorescence intensity of 10.

MAGE-9 expression
MAGE-9 is expressed in 38% of RCCs, the frequency of MAGE-9 expression not differing significantly from the frequency of RAGE-1 expression, which was observed in 56% of our pool of RCCs. There was no evidence that MAGE-9 and RAGE-1 expression are linked. Expression of MAGE-9 or RAGE-1 was independent of RCC stage and grade. Although this can only be taken as indirect evidence, it nonetheless points toward both molecules not intrinsically inducing a strong immune response that might provoke tumor escape mechanisms. [74][75][76][77][78] The majority of our collection of RCC tissues were clear cell RCC. Thus, there were too few samples of mixed, granular or oxyphil RCC tissues to define whether they express RAGE-1 or MAGE-9 at a significantly higher or lower frequency than clear cell RCC. However, we can state that clear cell RCCs do not express RAGE-1 and MAGE-9 at a significantly different frequency compared to non-clear cell type RCC. This is different for G250, which is expressed predominantly in clear cell RCC. 79 Thus, particularly for patients with non-clear cell type RCC, MAGE-9 may offer a target for vaccination. Finally, MAGE-9 was detectable in one normal kidney tissue sample by RT-PCR but not by Northern blot. As normal tissue was recovered from the same kidney as the tumor tissue, the most likely explanation is that macroscopically normal tissue contained few tumor cells.

MAGE-9 immunogenicity
As MAGE-9 is expressed at a significant frequency in RCC, the question arose as to its immunogenicity. Seven MAGE-9 and 13 RAGE-1 peptides with predicted HLA-A*0201 binding displayed sufficient stability (fluorescence index >0.3) to allow prolonged T-cell receptor binding. The proliferative response of individual healthy donors' PBMCs toward peptide-loaded DCs was very consistent but varied considerably between donors. The variability in response between the different donors could well be due to peptide binding to additional non-HLA-A*0201 alleles. In this context, it also should be mentioned that PBMC donors were not tested for homozygosity of the HLA-A*0201 allele. Peptide binding to additional MHC alleles possibly could explain why some peptides with suboptimal binding to HLA-A*0201 sufficed for the generation of a CTL line as, e.g., peptide pRE with a relatively low stability index of 0.493 for HLA-A*0201. The increased cytotoxic activity of CTL lines toward RCC lines compared to peptide-loaded T2 cells would also argue for a contribution of non-HLA-A*0201bound peptides. Irrespective of the possibility of peptide presentation by additional HLA alleles, peptide binding stability and the proliferative responses confirmed MAGE-9 and RAGE-1 as possible immunotherapeutic targets.
CD154-activated B cells have been described repeatedly as efficient APCs in tumor vaccination and may be well suited for the generation of large numbers of CTLs for adoptive transfer. 65,[80][81][82][83] Therefore, we explored the efficacy of CD154-activated B cells in the generation of MAGE-9 and RAGE-1 peptide-specific CTL lines. Although generation of an activated B-cell line takes more time than generation of DCs, activated B cells have some major advantages. First, large numbers of APCs can be generated from a small volume of peripheral blood. Second, CD154-activated B cells can be maintained in culture for a long period of time. In our hands, functional activity of B-cell lines, which were established for the presentation of MAGE-9, RAGE-1 and P68 peptides more than 12 months ago, is still unimpaired; and there has been no reduction in proliferative activity. These features are most valua- ble for cancer patients as a single blood sample can be sufficient for repeated applications of APCs in vaccination protocols or for repeated generation of CTLs in transfer regimen. Finally, it is less expensive to generate and maintain B cells than DCs. 65 Our protocol for the generation of T-cell lines followed routine procedures and does not require a detailed discussion. However, 2 features are worth mentioning. First, IFN-g secretion by itself may not be sufficient to estimate the generation of CTL. Although peptide-specific IFN-g secretion was observed already after 3-4 weeks of culture, roughly 25% of IFN-g-secreting T-cell cultures never developed cytotoxic activity. However, IFN-g expression is an essential prerequisite; cultures not secreting IFN-g did not develop cytotoxic activity. Also, the CTL lines lysed peptide-loaded T2 cells and RCC lines, which naturally present those peptides, with equal efficacy. As outlined by Tatsumi et al., 84 who observed natural HLA-DR4 MAGE-6 peptide presentation in RCC and malignant melanoma, the observation of natural peptide presentation could well provide a useful basis for selecting vaccine candidate peptides as well as for immune monitoring. Second, the T-cell lines likely were mono-or oligoclonal. Although we have not analyzed the T-cell receptor, it was surprising to see that against a mixture of peptides T cells only responded to one or 2, although the donor's T cells had been shown to respond in short-term cultures against all peptides. The monoclonality/oligoclonality was further strengthened by the observation that after cloning by limiting dilution with seeding one T cell/well, all arising clones displayed the same specificity as the starting T-cell line (data not shown). The fact that CTL lines could be maintained without a loss in cytotoxic activity upon repeated restimulation by the ''target peptide'' only but not by restimulation with peptides where the original line had not responded also argues for mono-/oligoclonality of the CTL lines. The apparent selection pressure for T cells with strong avidity is also supported by the high lytic activity of these cells, by the absence of any lymphokine-activated killer cell activity and by the efficacy of cold target inhibition. We did not observe expansion/ activation of CD4 1 T-helper or cytotoxic cells. This is likely due to the HLA-A*0201-selective peptide presentation.
We want to discuss briefly a third aspect of our culture system. We actually did not expect to obtain monoclonal/oligoclonal T-cell lines after stimulation with a mixture of 5 peptides, which had all been shown to induce a proliferative response after short-term activation. We speculate that only the ''clone'' with the highest avidity survived. As already mentioned, the lymphocyte donor was not tested for HLA-A*0201 homozygosity and the peptide which induced CTL activation may have been presented in addition by a non-HLA-A*0201 allele. Such an advantage in peptide presentation would have been missed by the stability assay, where peptides were presented by T2 cells. Irrespective of this hypothetical explanation, our finding strongly argues for offering whole protein, RNA or DNA for vaccination, if aiming for multiclonal CTLs. In view of the high cytotoxic activity of the described T-cell lines, we would argue that it may not be necessary to establish T-cell clones, which takes a longer time than establishing T-cell lines. An additional argument for the transfer of T-cell lines is our observation that these highly cytolytic lines could be maintained and expanded for a long period without a loss in lytic activity and the expected survival and homing advantage upon in vivo transfer. 85 Taken together, MAGE-9 and RAGE-1 peptide-specific CTLs displayed high cytolytic activity against MAGE-9/RAGE-1expressing RCC lines and large numbers of CTLs could be gained from a single blood sample by repeated stimulation on autologous CD154-activated B cells. Thus, MAGE-9 and RAGE-1 may be well included in the list of therapeutic targets in RCC. Also, the repeated adoptive transfer of CTLs may be greatly facilitated by using B cells as APCs and may be recommended at least as a mode of reinsurance in case the patient's health no longer allows repeated provision of DC progenitors.