Killer inhibitory receptor (CD158b) modulates the lytic activity of tumor-specific T lymphocytes infiltrating renal cell carcinomas

Tumor cells from 9 previously untreated patients with RCC were studied. The autologous tumor cell lines were derived from the primary tumor as described previously.31 Briefly, total cell suspensions were subjected to Ficoll-Hypaque centrifugation to eliminate the dead cells. Viable cells were cultured in specific culture medium to amplify tumor cells and TIL. For tumor cells, Dulbecco modified Eagle medium/Ham F12 medium supplemented with 10% fetal calf serum (FCS) and Ultroser G (Gibco BRL, Scotland) was used as complete medium. Confluent culture dishes were treated with EDTA and trypsin, and cells were subcultured 1 in 3.

The phenotype of TIL was analyzed by indirect 2-color fluorescence. Cells (2 × 10⁵) were first incubated for 30 minutes at 4°C with the unlabeled monoclonal antibodies (mAbs) HP3B1 (anti-CD94, IgG2a), EB6 (anti-p58.1, IgG1), or GL183 (anti-p58.2, IgG1), which were purchased from Immunotech (Marseille, France) and correspond to purified mAbs. Culture supernatants of the mAbs 11PB6 (anti-p58.1, IgG1), Z27 (anti-p70, IgG1), Q66 (anti-p140, IgG1), and Z199 (anti-NKG2-A, IgG2b) were also used. Cells were washed twice with phosphate-buffered saline (PBS) and then incubated for 20 minutes at 4°C with phosphatidyl ethanolamine (PE)–conjugated goat antimouse Ig. After a saturation step for 10 minutes with mouse Ig, cells were incubated with anti-CD8 fluorescein isothiocyanate–conjugated (FITC) mAb or anti-CD3 FITC mAb (Immunotech). Expression of activation markers was analyzed by using anti-CD69 PE and anti-CD25 FITC (Immunotech), anti-CD56, anti-CD28, and anti-CTLA-4 followed by GAMPE. In single-cell tumor suspensions, analysis was performed on 5000 gated TIL (appearing in FSC/SSC as small lymphocytes comparable to peripheral blood mononuclear cells (PBMC) after exclusion of dead cells and debris. Background levels were measured by using isotypic controls. Analysis was done on a fluorescence-activated cell sorter (Becton Dickinson, Pont de Claix, France) using Cell Quest software (Becton Dickinson). Compensation was set up with single stained samples.

TIL cell lines were obtained from mixed lymphocytes–tumor cell culture (MLTC). Briefly, from dissociated tumor cell suspensions, 3 × 10⁵ TIL/mL were grown for 2 weeks in medium (RPMI and 10% human AB serum) supplemented with 30 U/mL IL-2 (Roussel Uclaff, Romainville, France) in 24-well culture plates. The medium was renewed 3 times a week. After 2 weeks, the percentage of CD3⁺ T cells expressing p58 receptors was determined, and the p58⁺ T cells were immunoselected by using anti-p58.2 mAb (GL183) and immunomagnetic beads (Dynal, France). After immunoselection, p58⁺ and p58⁻ T-cell subsets were expanded separately in bulk cultures (3 × 10³ cells/well) on allogeneic feeder cells (10⁴cells/well) in the presence of autologous tumor cells (2 × 10³ cells/well). Then, p58⁺ T cells were cloned by limiting dilution 0.8 to 1 cell/well and using the feeder-cell conditions described above.

The cytolytic activity of T-cell lines against autologous tumor cells (patient DM), allogeneic renal tumor cell lines (LM, LR, MT, GF, and VM), and NK targets K562 and Daudi cells was measured in a 4-hour chromium 51 (⁵¹Cr)–release assay. Tumor cells were used in amounts of 2 × 10³ cells/well, and the ratio of effector to target (E:T) cells ranged from 20:1 to 2:1. The percentages of specific lysis were determined as described previously.31 In some experiments, mAbs 11PB6 (anti-p58.1), GL183 (anti-p58.2), B1.23.2 (anti-HLA-B or C, IgG2a), or UCHT1 (anti-CD3, IgG1) were added at predefined saturating concentrations at the beginning of the cytolytic assay. Data were expressed as the percentage of specific lysis at the indicated E:T ratio.

The cytolytic activity of T-cell lines was also assessed in a CD3-redirected lysis assay using P815 mastocytoma mouse cells. Briefly,⁵¹Cr-labeled P815 cells (2 × 10³) coated with anti-CD3ε (1μg/mL) were incubated with serial dilutions of T cells (E:T ratio ranging from 10:1 to 2:1). CD3-redirected lysis of labeled P815 cells was modulated by the presence of the anti-NK-R mAbs.

Usage of the TCRBV gene segment was determined by a semiquantitative polymerase chain reaction (PCR) analysis as described previously.32 Briefly, tumor samples (0.2-0.5 g) squeezed by a pulverizer (Spex 6700; Spex Industries, Edison, NJ) were resuspended in 6 mol/L guanidinium thiocyanate buffer. Total RNA was then purified by cesium chloride gradient centrifugation. For TIL and PBMC (5 × 10⁶ cells), total RNA was extracted with use of a modified guanidinium thiocyanate–phenol–chloroform method (Trizol; Eurobio, Les Ulis, France). Complementary DNA (cDNA) was prepared with a standard method using reverse transcriptase and an oligo-deoxythymidine primer (Invitrogen, Netherlands). cDNA amplification was done over 30 to 40 cycles with the 5′ sense primers specific for the 24 BV subfamilies and one 3′ antisense primer specific for BC. The intensity values for the different peaks in all BV subfamilies were added, and the percentages of each BV subfamily were evaluated and represented as histograms. Although PCR efficiency may vary from 1 oligonucleotide pair to another, this method was previously found to allow intersample comparison.33 

To study the overexpressed TCRBV transcripts further, a PCR-based method that determines size-distribution patterns of the V-D-J junction (CDR3) was used.33 Briefly, BV/BC-amplified products were copied in 1- to 5-cycle runoff reactions primed with fluorescence-labeled oligonucleotides (ABI fluorophore Fam) specific for a BC or the 13 BJ fragments. Runoff products were then subjected to electrophoresis on an automatic sequencer (ABI; Applied Biosystems, Foster City, CA) in the presence of fluorescent size markers and analyzed by automatic fluorescence quantification and size determination with use of a computer program (Genescan 672; Applied Biosystems).

Phenotypic analysis of the TIL population before in vitro culture was performed by using 2-color immunofluorescence analysis in 9 tumors. From 7 tumors containing a substantial CD3⁺ T-cell population, p58⁺ T cells were identified and corresponded to a marked subset of CD8⁺ T cells (range, 6% to 30%; Figure 1). From one of these tumors (patient DM: HLA A1, A2, B5, B8, and Cw7-Cw14), a permanent tumor cell line was established, and phenotypic, functional, and molecular analyses of p58⁺ T cells were done. Our data indicate that CD3⁺ represented 71% of the cells in the TIL gate, with 40% of CD4 and 38% of CD8 cells. A small subset of p58.2 cells representing 6% of the CD8⁺ T cells was detected (Figure2). Although the percentages of p58.2 cells were low, these cells constituted a well-defined CD8^brightT-cell subset, since the staining intensity of anti-p58.2 mAb (GL183) was high. With respect to the lectin-type NK receptors, CD94 was detected on 7.5% of CD8⁺ T cells and NKG2-A was detected on 6% of CD8⁺ T cells (Figure 2). In addition, 4-color analysis revealed that about 30% of p58⁺ T cells expressed CD94 but not NKG2-A (data not shown). NK receptors were not detected on CD4⁺ T cells.

Fourteen days after the MLTC was begun, the TIL cell line was phenotyped to detect CD3⁺ p58⁺ T cells. In these cultures, anti-p58.2 (GL183) stained 35% of the cells, indicating that p58⁺ cells were amplified during MLTC. From this culture, p58⁺ cells were immunoselected by using anti-p58.2 (GL183) and were expanded, as were p58⁻ T cells.

Comparative phenotypic analysis of both subsets was performed after a 10-day expansion. As shown in Figure 3, p58.2⁺ cells were CD3/CD8⁺ T cells, CD45RO⁺, CD28⁻, CTLA-4⁻, and 60% CD56^low. In addition, p58⁺ T cells were CD25⁺, CD69⁺ and expressed a TCR α/β when BMA031 mAb was used (data not shown). As expected from examination of the HLA-C alleles of patient DM that correspond to the HLA-C ligand of p58.2, the use of specific anti-p58.1 mAb 11PB6 confirmed the exclusive membrane expression of the p58.2 receptor. Expression of the other Ig-type NK-R was not detected (p70, p140), and 45% expressed CD94 but not NKG2-A (data not shown). On the other hand, p58⁻ cells also corresponded to CD3/CD8⁺, CD45RO⁺, CD25⁺, CD69⁺, TCR α/β⁺ T cells that were CD56⁻, CD28⁺, and CTLA-4^−/low (Figure 3) and did not express any Ig-type NK-R. A low-intensity expression of CD94 was detected on 15% of the cells, but NKG2-A was not detected. Thus, expression of CD56 and CD28 distinguished p58⁺ from p58⁻ T cells, whereas early activation markers CD25 and CD69 were expressed on both subsets.

When the lytic activity of p58⁺ and p58⁻ T cells was tested against autologous tumor cells, both T-cell subsets showed low levels of cytotoxic activity (Figure 4). In addition, p58⁺ T cells lysed K562 with a low efficiency and did not kill Daudi cells, suggesting that these cells do not have NK/LAK activity. On the other hand, p58⁻ T cells efficiently lysed NK cell targets K562 and NK-resistant Daudi cells, indicating that these cells mainly mediate non-major histocompatibility complex–restricted killing (eg, LAK activity) (Figure 4).

To refine the functional analysis, the p58⁺ T-cell line was further cloned by limiting dilution. A CTL clone (4C7) that showed a high proliferative capacity was derived. CTL clone 4C7 expressed p58.2 and CD56 and was CD28⁻. It had a low level of lytic activity against autologous tumor cells.

To assess whether the lytic activity of p58⁺ T cells is modulated by the p58 receptor, cytotoxic assays were performed in the presence of neutralizing anti-p58 mAbs. Lysis of autologous tumor cells by p58⁺ T cells and CTL clone 4C7 increased significantly when anti-p58.2 mAb was added during the cytotoxic assay, whereas the control, anti-p58.1 mAb 11PB6, had no effect (Figure5A). Preincubation of target cells with anti-HLA-B/C–specific mAb (BB1 23.2) induced a similar increase in lysis of tumor cells. Thus, the p58⁺ T-cell line and CTL clone exhibit marked lytic activity against tumor cells when interaction of p58 and HLA-C is blocked. Further evidence that p58⁺ T cells and CTL clone 4C7 mediated a TCR-specific lysis of tumor cells was that the addition of anti-HLA-B/C to counteract the p58 receptor in the presence of anti-CD3 mAb resulted in background-level lytic activity (Figure 5A). Moreover, the production of IFNγ by p58⁺ T cells stimulated by tumor cells increased significantly after blockage of interaction of HLA-C and p58 (data not shown). With regard to p58⁻ T cells, the marginal lysis observed was not modified when anti-p58.2 (GL183) was added to the test (Figure 5A), and anti-CD3 induced a partial inhibition of p58⁻ T-cell cytotoxicity (Figure 5A).

To confirm the presence of the modulator effect of the p58.2 receptor on the TCR-mediated signal, we performed CD3-redirected lysis experiments with FcγR⁺ murine P815 cells. Both p58⁺ and p58⁻ T cells, as well as CTL clone 4C7, mediated efficient CD3/TCR-triggered lysis of P815 cells but did not kill P815 in the absence of anti-CD3 mAb. Furthermore, in p58⁺ T cells, CD3-redirected lysis of P815 was almost completely abrogated in the presence of anti-p58.2 mAb, whereas anti-p58.1 had no significant effect and anti-p70 had no effect at all. Comparable results were obtained with CTL clone 4C7. In p58⁻ T cells, CD3-redirected lysis was not modified by anti-p58 mAb (Figure 5B). It is noteworthy that the 2 T-cell subsets displayed comparable levels of CD3-redirected lysis, indicating that the CD3 lytic machinery normally operates in p58⁺ T cells.

To analyze the specificity of lysis mediated by p58⁺ T cells, normal autologous cells (phytohemagglutinin (PHA)-blasts and Epstein-Barr virus (EBV) B cells), as well as allogeneic tumor cell lines, were used as targets. Autologous PHA-blasts cells and EBV-transformed B cells were not lysed, even in the presence of anti-p58.2 or anti-HLA-B/C mAbs, indicating that the antigen recognized by p58⁺ T cells is not expressed by activated T and B cells (Table 1).

Five allogeneic renal tumor cell lines (3 HLA-A1⁺ and 2 HLA-A2⁺) were tested. As shown on Table 1, p58⁺T cells lysed the 3 HLA-A1⁺ cell lines (LM, GF, and MT) in the presence of anti-p58.2 or anti-HLA-BC mAbs. Cell lysis of these allogeneic tumor cells was comparable to the lysis of autologous tumor cells, whereas the LR and VM tumor cell lines (HLA-A2⁺) were not lysed, even in the presence of anti-p58.2 mAb. In addition, normal renal cells from patient LM were susceptible to lysis by CTL clone 4C7. These data show that antigen expressed by DM tumor cells is present in other renal tumors and possibly expressed by normal renal cells and that the DM p58⁺ T cells exhibited HLA-A1–restricted lysis.

To assess the presence of p58⁺ T cells in vivo, TCR transcripts expressed by p58⁺ T cells were characterized and their in situ representativeness was analyzed. Analysis of usage of the BV gene fragment in the p58⁺ T-cell line revealed that the TCRBV repertoire was very restricted, with expression of 2 specificities: BV6 (43%) and BV8 (55%). On the other hand, the TCR repertoire of p58⁻ T cells was polyclonal (Figure6A and 6B). In tumor, although the TCR repertoire was diverse, BV6 and BV8 specificities corresponded to 7% and 11%, respectively, of the BV transcripts (Figure 6C). The repertoire in PBMC was polyclonal and different from that of tumor (Figure 6C).

We focused on the 2 TCR BV6 and BV8 specificities in the p58⁺ T-cell line to refine identification of clonal transcripts using the 13 different BJ primers. One dominant BV6 clone was detected, corresponding to a TCR BV6BJ2S1 rearrangement of 274 nt. The TCRBV8 transcripts contained 2 cDNA clones, 1 prominent rearrangement, TCR BV8JB2S1 of 214 nt, and 1 minor rearrangement, TCR BV8BJ2S7 of 242 nt (Figure 7).

The presence of these rearrangements in vivo was determined by CDR3 size analysis of BV6 and BV8 TCR transcripts in tumor and PBMC (Figure7). In tumor, the TCR VB6BJ2S1 (274 nt) and BV8BJ2S1 (214 nt) rearrangements were prominent, whereas the minor BV8BJ2S7 peak was not clearly detected. In PBMC obtained at the time of nephrectomy, TCR BV6 and VB8 transcripts displayed a bell-shaped pattern consistent with polyclonality, and the transcripts specific for p58⁺ were barely detectable. Interestingly, the cytotoxic CTL clone 4C7 expressed the TCR BV6J2S1 rearrangement of 274 nt prominently in situ, suggesting that the selective intratumor amplification of the TCR BV6BJ2S1 transcript corresponds to the dominant p58⁺ tumor-specific CTL. This molecular analysis of p58⁺ tumor-specific T cells confirms that such cells are amplified at the tumor site and constitute the main tumor-specific population infiltrating RCC.

The current studies emphasize that NK-R may be an important component of the regulation of the T-cell response in RCC. First, double-fluorescence analysis of uncultured TIL provided evidence that most renal tumors contain a substantial subset of p58⁺ T cells. Furthermore, the functional study indicated that p58⁺ TIL correspond to antigen-specific cytotoxic T cells, whose lytic and secretory activities were significantly inhibited by the p58 receptor. Exploiting the high-resolution properties of the analysis of the length of the CDR3 TCRBV transcripts and the fact that TCR expression is a stable and conserved feature of specific antigen recognition, we demonstrated expansion of the p58⁺ tumor-specific T-cell clones at the tumor site. On the other hand, p58⁻ TIL mainly mediated a non-specific lytic activity and exhibited a polyclonal TCR repertoire.

The tumor-specific T cells we obtained after a single in vitro stimulation by tumor cells (while avoiding culture skewing as much as possible) corresponded to few p58⁺ T-cell clones bearing an inhibitory p58 receptor and represented 5% to 10% of the CD8⁺ TIL subset. Interestingly, CTL clone 4C7 derived from a p58⁺ T-cell line had the same phenotypic and functional characteristics as the T-cell line and expressed the BV6BJ2S1 rearrangement that was prominent in the p58⁺ T cells and found to be amplified in the tumor in vivo. These results indicate that, in RCC, p58⁺ T cells correspond to tumor-specific T cells that are primed in vivo, proliferate, and display high lytic activity in vitro. A recent study using the sensitive tetramer technology provided evidence of the presence of CD94/NKG2-A⁺, Mela A/Mart1–specific CTL in the peripheral blood of a few patients with melanoma. In these patients, lysis of melanoma cells was inhibited by this inhibitory NK receptor.34 

Although there is no evidence that CD158b expression by itself provides antigen specificity, expression of this marker correlates in this model to high lytic activity, clonal TCRBV gene usage, and in situ expansion of the corresponding T cells. The expression of inhibitory NK receptors on tumor-specific T cells would explain the absence of tumor-specific activity in bulk TIL populations and the low frequency of CTL detected by conventional cytotoxicity assays of autologous tumor cells in RCC compared with that observed in melanoma. It might also explain the skewing usually observed between in situ–amplified and in vitro–derived T-cell subsets in RCC.6 

Few data are available on the mechanisms controlling the induction of these receptors on T cells. In contrast to the CD94/NKG2 receptor, which can be induced in vitro under particular T-cell stimulation conditions,35-37 Ig-type NK-R induction cannot yet be reproduced in vitro. In our model, expression of p58 was stable and was not modulated by either interferon-γ or interleukin 10 after T-cell stimulation with tumor cells (data not shown). In addition, few data implicating Ig-type NK-R in the recognition of solid tumors have been reported. Introduction of an inhibitory p70 receptor (KIR3DL) in a gp100 melanoma-specific, HLA-A2–restricted CTL resulted in inhibition of lysis of melanoma cells coexpressing HLA-A2 and HLA-Bw4 allotypes.38 A melanoma-specific p58⁺ CTL was obtained from PBMC stimulated by a metastatic cell line that had lost expression of the complete HLA haplotype. Although this CTL killed metastatic cells efficiently, its specific activity was inhibited by primary tumor cells expressing HLA-Cw7. This p58⁺ CTL recognized a new antigen, PRAME, which is expressed in a large range of melanomas, sarcomas, and myeloid leukemias and also detected in normal testicular, endometrial, and ovarian tissues.30 In the context of an immune response against tumor, expression or overexpression of normal tissue-specific antigen presented by tumor cells with altered HLA molecules in the presence of particular immunoregulatory cytokines may be responsible for amplification of KIR⁺ antigen-specific CTL.

In patients with RCC, who frequently present with slow-growing large tumor burdens, prolonged chronic exposure of T cells to antigen (overexpressed normal proteins) may lead to activation of potentially autoreactive T cells. Therefore, expression of NK-R may represent a mechanism involved in the down-regulation of a deleterious immune reaction. Alternatively, activation and expansion of p58⁺CTL may be a consequence of altered expression of HLA-C molecules at the surface of tumor cells. Indeed, like many tumors, RCC lose or have alterations of expression of the HLA molecules with tumor progression, and several mechanisms may be involved in these alterations.39,40 Furthermore, it has been shown that HLA-C molecules bind to the inhibitory NK receptor with extremely fast association and dissociation rates, and such kinetics may facilitate the rapid and precise immunosurveillance of cells with absent or diminished expression of HLA-I molecules.41 In this context, reduced levels of expression of HLA molecules have been shown to play a major role in the susceptibility of melanoma cells to NK-mediated lysis.42 The fact that in vitro–derived tumor cells were not lysed by p58⁺ T cells in the absence of anti-p58 mAb suggests that these cells expressed HLA-C. This finding supports the hypothesis suggesting that the nature of the antigen is the main mechanism activating p58⁺ T cells in RCC. The p58⁺ CTL correspond to high-affinity CTL requiring few interactions of TCR and HLA-I, and an overall (not allele specific) down-regulated expression of the HLA-I molecule probably affects the NK-R activity more than the TCR-mediated recognition. Thus, in RCC infiltrated by p58⁺ T cells, HLA-C expression may tune the lytic-activity level of the tumor-specific response.

The nature of the antigens recognized by p58⁺ T cells is still highly undetermined. It may be postulated that such antigens derived from self proteins recognized by autoreactive CTL that are unable to lyse normal cells expressing large amounts of HLA molecules but are efficiently activated by contact with cells with altered expression of HLA-I molecules. In our model, although the antigen recognized by p58⁺ CTL was not expressed by normal activated T and B cells, it may correspond to a tissue-specific antigen expressed by normal and tumoral renal cells. It is interesting that RCC do not express any of the tumor-specific and tumor-associated antigens identified so far by tumor-specific CTL in melanoma. In addition, the antigens recognized by classical CD8⁺ CTL in RCC that have been characterized correspond to private antigens and are encoded by mutated epitopes of the HLA-A2 allele,43hsp70-2,44 the open reading frame of carboxyl esterase,45 or the rarely expressed RAGEgene.46 Thus, the characterization of the antigens recognized by p58⁺ CTL may be of major interest.

In summary, we found that substantial subsets of TCR α/β CD8⁺ T cells that expressed NK receptors infiltrated RCC. Tumor-specific T-cell clones derived from this subset inhibited the lysis of renal tumor cells, suggesting that NK receptors may modulate T-cell activities in vivo. Overall, these results point to a new mechanism in the regulation of T-cell effector functions against human renal tumors that may have important implications for antitumor immunity in RCC.

We thank Yann Lécluse for immunofluorescence analyses and Dr Ryad Tamouza for HLA genotyping of several tumor cell lines.