Abstract
We performed T cell cloning experiments with a tumor-infiltrating lymphocyte subpopulation derived from a renal cell carcinoma tumor site (RCC-7) in which the TCR clonotypic repertoire had been analyzed in terms of TCRBV complementarity-determining region 3 size distribution. We report in this work the characterization of one of the five RCC-specific MHC class I-restricted CTL clones isolated in RCC-7. This TCRBV6J1S1 CTL recognized only the autologous RCC-7 tumor cell line in the context of HLA-A*0201, and the Ag is encoded by a mutated form of the hsp70-2 gene found in the tumor cells, but not in autologous PBLs nor in 47 other tumors. The identification of this gene was achieved by cotransfecting into COS cells a cDNA library of RCC-7 together with HLA-A*0201. Transfectants expressing the Ag were identified by their ability to stimulate TNF release by the CTL clone. The antigenic peptide is a decamer with a mutated residue at position 8. Half-maximal lysis was obtained with only 5 × 10−11 M of decapeptide in target sensitization assays compared with 5 × 10−8 M for the wild-type decapeptide. This difference in recognition was not related to difference in binding HLA-A*0201-presenting molecules, as assessed in an immunofluorescence-based peptide-binding assay using T2 cells. Constitutive hsp70 expression in various tumors suggests that this stress-induced protein may be recognized in situ by tumor-infiltrating lymphocytes. The finding in the tumor of a mutated form of the stress-induced hsp70-2 gene whose product is specifically recognized by TILs with high avidity is discussed in view of the present use of mycobacteria or heterologous heat-shock proteins as immunomodulators or as subunit vaccine candidates.
Stimulation of the immunologic rejection of tumors has been a longstanding goal of cancer research. This goal is based on the hypothesis that many tumors express Ags that can potentially serve as targets for their destruction by the immune system. It remains controversial to what extent spontaneous tumors can be recognized as foreign by the immune system in conventional immunization. A major objective is thus to identify immunogens and/or adjuvants able to successfully boost antitumor T cell responses. There is some evidence that such a goal may be achieved in patients. For instance, induction of an inflammatory response in both delayed-type hypersensitivity and vaccine sites in an immunized patient has been shown to be associated with the recruitment of only a few T cell clones, which may correspond in some instances to TILs3 already expanded in the tumor before immunization (1). Also, injection of immunodominant peptides from the gp100 melanoma-associated Ag has been shown to be associated with both successful immunization and objective cancer responses (2). Thus, in humans, tumor-associated Ags may be used to recruit or expand locally clonal tumor-specific T cells previously primed in vivo by relevant Ags. In vivo primed cells may be further amplified by IL-2 administration (2), because clonally expanded T cells are induced in the blood and tumors of patients receiving IL-2 (3). Nonetheless, new adjuvants or vehicles for Ag delivery are clearly needed to improve current immunization procedures and the induction of strong and long-lasting CTL responses.
It has been difficult to isolate and expand renal cell carcinoma (RCC)-specific CTL clones (4, 5). This may be, in part, because RCC-TILs are defective in the proliferative response (6). Nonetheless, there is some indication that RCC may be immunogenic in vivo because these tumors are often largely infiltrated by T cells, mainly composed of TCR α/β+ DR+ Th1-polarized lymphocytes (7), and because of the relatively high response rate (15–20%) to a number of immunotherapeutic protocols (8). Overall, the biological significance of the natural T cell-mediated antitumor immune response in RCCs remains ill-defined in contrast to melanomas, in which numerous specific CTLs and tumor-associated peptides recognized by the CTLs have been characterized.
Among a series of 12 RCC patients, we have previously identified a patient, termed RCC-7 (HLA-A2, -A29, -B44, -B51, -Cw15, -Cw16), with suggestive evidence of RCC Ag recognition at the tumor site, including in situ clonal expansion of specific T cell subpopulations and cytolytic activity of TILs against the autologous tumor cell line (7). In this work, we report the characterization of a RCC-specific MHC class I-restricted CTL clone from these TILs, the cloning of the Ag, and characterization of its expression. The Ag is encoded by a stress-induced mutated hsp70-2 gene, and this finding is discussed in view of the current emphasis being placed on HSPs as immunomodulators or as subunit vaccine candidates.
Materials and Methods
Cells and culture conditions
The autologous EBV cell line was derived following infection of PBMC of patient 7. The EBV-transformed cell line was maintained in RPMI 1640 (Life Technologies, Gaithersburg, MD) supplemented with 10% FCS. WEHI-164 clone 13, a mouse fibrosarcoma cell line sensitive to TNF, was kindly provided by Benoit Van den Eynde (Ludwig Institute for Cancer Research, Brussels, Belgium) and was maintained in suspension and cultured in RPMI 1640 (Seromed, Biochrom KG, Berlin, Germany) supplemented with l-glutamine, sodium pyruvate, antibiotics, and 5% FCS at a concentration of 0.01–0.05 × 106 cells/ml. The human mutant cell line CEM × 721.174.T2 (T2) (9) was maintained in RPMI 1640 medium supplemented with 10% FCS and was kindly provided by Pierre Langlade (Institut Pasteur, Paris, France). All cell cultures were kept in a water-saturated atmosphere with 5% CO2.
TILs from thawed single cell suspensions of tumors were seeded in six-well plates in RPMI 1640 containing penicillin, streptomycin, 1% l-glutamine, 1% sodium pyruvate, and 8% human AB serum (Institut Jacques Boy S.A., Reims, France), referred to as complete medium (7). TILs (i.e., TILs and tumor cells) were seeded at a final concentration of approximately 2 × 106 TILs/ml in complete medium supplemented with 10 IU/ml rIL-2, 50 IU/ml rIL-7 (Sanofi, Toulouse, France), and 10 IU/ml rIL-12 (Genetics Institute, Cambridge, MA) for 3 days. From the third day, TILs were fed using complete medium with 30 IU/ml rIL-2, 50 IU/ml rIL-7, and 10 IU/ml rIL-12. TIL cell lines were phenotyped and characterized for their cytotoxic activity after 14 and 21 days of stimulation.
After culturing for 3 wk, lymphocytes were cloned by limiting dilution. Cloning was performed from 600 to 0.6 cells/well in 96-well plates in RPMI containing 8% human AB serum supplemented with 30 IU/ml rIL-2 and 3% TCGF (Tull growth factor, an activated T cell conditioned medium), on a feeder layer including irradiated autologous tumor cells (1 × 104/well), allogeneic irradiated PBL (8 × 104/well), and irradiated EBV-transformed cells (2 × 104/well). Clones were fed three times per week with complete medium containing rIL-2 and TCGF.
mAbs, serologic reagents, and phenotyping analysis
Fluorescein (FITC)- or phycoerythrin-conjugated mAbs directed against TCR α/β, CD3 (Leu4), CD4 (Leu3a), CD8 (Leu8), CD80 (B7.1), and HLA-DR (L249) were purchased from Becton Dickinson (Mountain View, CA). CD56 (NKH1A) was purchased from Coultronics (Hialeah, FL). TILs were characterized by dual immunostaining performed by incubating cells for 30 min at 4°C with FITC or phycoerythrin mAbs. Flow cytometry analysis was performed on a FACScan (Becton Dickinson) cytometer using the Cellquest software. The ascites were W6.32 (anti-HLA-A/B/C), MA2.1 (anti-HLA-A2 and -B17), and B1.23.2 (anti-HLA-B/C), and were used for functional and immunofluorescence assays at predetermined saturating concentrations from 1/200 to 1/2000 final dilution.
Cytotoxicity assay
Cytotoxicity assays were performed using a standard 4-h chromium release assay, as previously described (7). Briefly, 2 × 103 51Cr-labeled target cells were incubated for 4 h at 37°C with effector cells at different E:T ratios in a final volume of 200 μl. For peptide recognition assays, 2 × 103 51Cr-labeled autologous EBV-transformed cells were incubated for 1 h at 37°C in 96-well microplates with various concentrations of peptide before adding CTL 11C2. For inhibition of lysis by mAbs, target cells were preincubated for 2 h in the presence of saturating mAb concentrations before subsequent addition of effector cells for an additional 4 h. At the end of the incubation, 40 μl of the supernatant was transferred in Lumaplate 96 solid scintillation plates (Packard Instruments, Meriden, CT) and, after overnight drying, counted in a Top Count beta counter (Packard Instruments).
CDR3 size analysis
CDR3 size analysis of TCRBV gene segments expressed by the CTL clones or found in blood or tumor fragments was performed as previously described (7). Briefly, total RNA was prepared using RNAB (Bioprobe Systems, Montreuil, France). First-strand cDNA was synthesized with oligo(dT) priming and reverse transcriptase (Invitrogen, Leek, The Netherlands). The procedure used for CDR3 size analysis includes independent reverse-transcriptase PCR amplifications (30 or 40 cycles) of TCRBV-BC fragments (10) from the cDNA, copying of the PCR products using nested fluorescent TCRBV or TCRBJ primers (11), and size determination of fluorescent runoff products by electrophoresis on an automated DNA sequencer (model 373A; Applied Biosystems, Foster City, CA) using the Immunoscope software (12). Since the 5′ and 3′ primer positions are fixed, variations in size in the runoff products are only due to differences in the length of CDR3, thus reflecting an imprecise V/D/J joining process. The peaks, at 3-nt intervals, correspond to in-frame transcripts. The graphs representing CDR3 size patterns were standardized at 100% for the highest peak, and data used to generate these graphs could be used to determine the intensity of each peak, expressed as fluorescence units (FU), and to evaluate the background. Dominant peaks were defined as high-intensity signals (e.g., from 20,000 to 100,000 FU), with a dramatic reduction in other CDR3 signals.
Cloning and expression of HLA molecules
HLA class I alleles were cloned using the PCR method described by Parham and coworkers (13), with slight modifications. Briefly, total RNA was prepared from RCC-7 cell line using RNAB (Bioprobe Systems). First-strand cDNA was synthesized with oligo(dT) priming and reverse transcriptase (Invitrogen). The cDNA was used as template in a 30-cycle PCR amplification with the following primers: 5P2-H (5′-GGGCAAGCTTGGACTCAGAATCTCCCCAGACGCCGAG-3′) and 3P2-X (5′-GCCCTCTAGATCTCAGTCCCTCACAAGGCAGCTGTC-3′), corresponding, respectively, to consensus sequences in the 5′ and 3′ untranslated regions of class I alleles. These primers are identical to the HLA-5P2 and HLA-3P2 primers described previously (13), except for the cloning sites: we replaced the SalI and HindIII sites of 5P2 and 3P2 with HindIII and XbaI sites, respectively. The PCR products were digested with HindIII and XbaI and ligated into plasmid pcDNA I (Invitrogen). The constructs were transfected into Escherichia coli MC1061/P3. Plasmid DNA was extracted from several colonies using QIAGEN columns (Qiagen, Chatsworth, CA). DNA sequencing was performed using the ABI PRISM Dye Terminator cycle sequencing ready reaction kit (Applied Biosystems) and an automated DNA sequencer. The sequences were compared with HLA class I nt sequences available from databanks.
Construction of the cDNA library
Poly(A)+ RNA was extracted from the RCC-7 cell line using mRNA isolation system (Fast Track kit 2.0; Invitrogen) following manufacturer’s instructions. First-strand cDNA was synthesized using AMV reverse transcriptase with an oligo(dT) primer containing a NotI site at its 5′ end. The RNA-cDNA hybrid created by first-strand synthesis was converted to double-stranded cDNA by DNA polymerase I in combination with RNase H and E. coli DNA ligase. The double-stranded cDNA was subsequently blunt ended with T4 DNA polymerase. BstXI adaptators were added, and the cDNA was sized on agarose gel. Size-selected cDNA (>800 nt) was ligated into BstXI/NotI cut expression vector pcDNA I and transformed into the appropriate E. coli strain (MC1061/P3). In screening experiments, plasmid DNA from pools of bacteria was prepared as follows: 100 or 200 colonies cultured on LB-agar plates (with 30 μg/ml ampicillin and 10 μg/ml tetracyclin) were seeded in 2 ml of LB medium for overnight incubation at 37°C. Plasmid DNA was extracted using the alkaline lysis method (14) and was resuspended in 30 μl of 10 mM Tris/EDTA 1 mM, pH 7.5, containing 20 μg/ml of RNase A. The concentration of plasmid DNA was then adjusted at 40 ng/μl.
Transfection of COS-7 cells and screening of transfectants
Transfection experiments were performed by the DEAE-dextran-chloroquine method (15). Briefly, 3 days before transfection, COS-7 cells were seeded in 96-microwell flat-bottom plates at 5 × 103 cells/well in 150 μl of RPMI containing 20% FCS. For transfection, medium was discarded and replaced by 30 μl of DEAE-dextran/DNA mixtures. These mixtures were prepared for duplicate transfections in V-bottom microwells by adding sequentially: 200 ng of plasmid DNA from the cDNA library, 200 ng of plasmid pcDNA I/HLA-A*0201, 25 μl 150 mM NaCl/10 mM Tris, pH 7.4 (referred as TBS buffer), and 35 μl TBS containing 1 mg/ml DEAE-dextran (Pharmacia Biotech Europe, Saclay, France). Cells were incubated with this mixture 30 min at room temperature, and 150 μl of DMEM supplemented with 10% decomplemented NuSerum (Becton Dickinson) and 100 mM choloroquine (Sigma-Aldrich Chimie SARL, Saint Quentin Fallavier, France) were added and cells were incubated 4 h at 37°C under 5% CO2. After incubation, medium was removed and cells were incubated 2 min in 1× PBS buffer containing 10% of dimethyl-sulfoxide solution. Cells were washed once in 1× PBS buffer and incubated with RPMI/10% FCS for 48 h. The medium was then discarded, cells were washed once with 1× PBS, and 3000 CTLs were added to the wells in 100 μl of RPMI/10% FCS. After 20 h, the supernatant was collected and its TNF content was determined by testing its cytotoxicity for WEHI-164 clone 13 in a MTT (3-[4, 5-dimethylthiozol]-2, 5-diphenyl-tetrazolium bromide; Sigma-Aldrich) colorimetric assay, as described (16). For inhibition of TNF secretion by mAbs, target cells were preincubated for 2 h in the presence of saturating mAb concentrations before subsequent addition of effector cells for additional 20 h.
Transfection of tumor cell lines and bioassay
Three days before transfection, HLA-A2+ RCC and melanoma cell lines were plated at 1 × 104 cells/well in 150 μl of RCC medium (7). Two hundred nanograms of DNA mixed with 1 μl of lipofectamine were added to 100 μl of Optimem-1 medium (Life Technologies) for 45 min at room temperature. The RCC medium was discarded and the DNA/lipofectamine mixture was then added to the cells and incubated for 5 h at 37°C, and then overnight after addition of 100 μl of RCC medium containing 20% FCS. The cells were then washed once with 1× PBS, and 3 × 103 CTL 11C2 cells/wells were added in 100 μl of RPMI/10% FCS. TNF content in supernatants was tested as described above.
PCR assay for isolation of full-length hsp70-2
Genomic DNA was extracted with DNAzol (Life Technologies). One microliter of the DNA was used in a PCR using Taq DNA polymerase (Perkin-Elmer, Norwalk, CT). The following primers were used: primer hsp70-2A, 5′-GGGCAAGCTTAGTCTCAGAGCGGAGCCCAC-3′ (nt −36 to −18, sense); primer hsp70-2B, 5′-GCCCTCTAGAGTCCCAACAGTCCACCTCAA-3′ (nt 1955–1974, antisense) containing HindIII and XbaI restriction cloning sites, respectively. PCR conditions were 98°C for 1 min, followed by 30 cycles of amplification (98°C for 15 s, 65°C for 1 min, 72°C for 2 min), with a final extension for 10 min at 72°C. The PCR product obtained was digested by HindIII and XbaI, purified by adsorption on glass beads (Geneclean), and then subcloned into the HindIII and XbaI sites of the pcDNA I expression vector for sequencing and cotransfection into COS-7 cells together with HLA-A*0201.
Identification of the minimal nucleotidic region coding for the antigenic peptide
The cDNA A18 was isolated from the cDNA library made in the pcDNA I expression vector. The plasmid was digested with SphI and XbaI before treatment with exonuclease III. To generate progressive deletions from the 3′ end of the cDNA A18, and thereby obtain a large number of truncated cDNA clones, we used the Exo Mung Bean Deletion Kit (Stratagene). After ligation, MC1061/P3 E. coli bacteria were transformed with the truncated cDNAs. Plasmid DNA was extracted from each clone, sequenced, and cotransfected into COS-7 cells together with HLA-A*0201.
Peptide synthesis and peptide-binding assay
In the screening assay, the peptides used were synthesized by the PepSet technology (Chiron Technologies, Suresnes, France). For functional assays, peptides were synthesized on solid phase using F-moc for transient NH2-terminal protection, and were purified by preparative HPLC. All of the peptides were >99% pure, as indicated by analytical HPLC. Lyophilized peptides were dissolved at 1 mM in DMSO and water and stored at −80°C.
For the binding assay, T2 cells (17) were cultured, 48 h before assay, in complete serum-free AIM-V medium (Life Technologies). A total of 106 cells was incubated at 26°C for 16 h in the same medium in 0.8% DMSO with or without peptides at a concentration of 20 μM. Then peptides (20 μM) were added again and cells were incubated at 37°C. At intervals of 30 min or 1 h, aliquots of cells were collected and the level of HLA-A2 expression was monitored with an anti-HLA-A2 mAb (MA2.1).
Isolation of RNA and Northern blot analysis
Cells were either maintained at 37°C or heat shocked at 42°C for 2 h before their collection by centrifugation. Total RNA was extracted by guanidinium isothiocyanate lysis and cesium chloride ultracentrifugation. Samples of total RNA (15 μg) were fractionated in 1% agarose-formaldehyde denaturing gel and transferred onto Hybond-N+ nylon membranes following the manufacturer’s instructions (Amersham France SA, Les Ulis, France). Northern blot was hybridized both with a hsp70-2-specific (nt 1955–2159) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probes. All probes were labeled with [α-32P]dCTP (3000 Ci mmol−1) using the Prime-IT II Random Primer Labeling Kit (Stratagene). Hybridation was performed at 45°C for 16 h with 106 cpm/ml of hsp70-2-specific probe and 105 cpm/ml of GAPDH probe. Membranes were washed twice with 2× SSC at room temperature, once for 45 min with 2× SSC/0.1% SDS at 62°C, and once at 62°C for 10 min with 0.1× SSC before autoradiography at −80°C between intensifying screens for 11 days.
Results
CTL clones specific for a RCC tumor
RCC-CTL clones isolated from RCC-7 TILs were obtained using either IL-2 or IL-2 + IL-7 + IL-12. Between days 20 and 30, the cultures that displayed significant lysis of the autologous tumor cell line RCC-7, but that did not lyse K562, were cloned by limiting dilution. Of the eight clones obtained in this manner, five were retained based on their distinct TCR phenotype (assessed by TCRBV CDR3 length distribution analysis) and MHC class I-restricted Ag recognition.
All five clones were cytotoxic CD4−CD8+TCR α/β+ cells and produced TNF when stimulated with autologous tumor cells. The characterization of one of these clones, termed 11C2, obtained with the IL-2 + IL-7 + IL-12 culture condition and which killed the RCC-7 tumor cell line, but not the autologous PHA blasts or EBV-transformed B cells (Fig. 1⇓A), is reported in the present study. Normal kidney cells were not tested because proximal tubule epithelial cultures, the normal counterpart of clear cell RCC, were not available for this patient. The cytotoxic activity of 11C2 was restricted by HLA-A2 molecules, as shown by the blocking effect of anti-class I mAb W6/32 and HLA-A2-specific MA2.1 mAb, while the HLA-B/C-specific B1.23.2 mAb had no effect (Fig. 1⇓A). The same Abs were tested for their inhibitory effect on TNF production and similar results were obtained (Fig. 1⇓B). It was concluded that HLA-A*0201 (specificity determined by sequencing the six RCC-7 HLA class I alleles) was most probably the presenting molecule for CTL 11C2. Finally, CTL 11C2 was unable to kill any one of the six HLA-A2+ allogeneic RCC cell lines tested (data not shown), suggesting that the target Ag is rarely expressed in tumor cell lines.
Recognition of autologous RCC-7 cell line by TIL 11C2. A, Specific lytic activity. Cytotoxicity of the 11C2 CTL clone toward the autologous RCC (RCC-7), PHA blasts, or EBV-transformed B cell line was tested in a standard chromium-release assay at different E:T ratios. Inhibition of cytolytic activity of 11C2 was tested after preincubation of the cells for 2 h with the indicated anti-HLA class I mAb at a predetermined saturating concentration. B, TNF secretion. Three thousand 11C2 cells were incubated with 2 × 104 RCC-7 cells. The amount of TNF was measured after 20 h of culture, by testing toxicity of the supernatants for TNF-sensitive WEHI-164 clone 13 cells. Inhibition of TNF secretion was tested after preincubation of 11C2 cells for 2 h with the indicated anti-HLA class I mAb.
In the RCC-7 tumor (7), as well as in other RCC tumors analyzed in our laboratory (18), in situ clonal expansion of TILs has been reported by analyzing TCRBV CDR3 size distribution. For instance, one of the five other CTL clones, termed 2A11, displays a unique TCRBV-BC pattern in terms of CDR3 length (a dominant 188 nt TCRBV1 peak, data not shown) that is similar to the one already observed in tumor fragments (7). When refining the analysis using 13 TCRBJ primers, a unique signal, TCRBV1J1S6 (149 nt with a CASNSDRGIGSPLHF amino acid junctional sequence), was observed, the latter being also dominant (i.e., a single peak with 83,166 FU) in the tumor (Fig. 2⇓). To confirm that the 2A11 TCR clonotype was dominant at the tumor site, we cloned TCRBV1-BC PCR products derived from a tumor fragment and found that of 21 colonies analyzed by sequencing, 11 (52%) corresponded to the 2A11 junctional region. The TCRBV1-BC CDR3 profile was found to be polyclonal in blood and, by randomly sequencing the corresponding PCR products, we did not find any transcript with the 2A11 junctional region (data not shown). Thus, the tumor-specific 2A11 CTL may be clonally amplified specifically at the tumor site (i.e., not in blood). A similar analysis was performed for CTL 11C2 (a 286 nt TCRBV6J1S1 signal), but no clear-cut evidence for clonal expansion at the tumor site was found using TCRBV6-BC primers or a more refined analysis with the TCRBV6-BJ1S1 primers (Fig. 2⇓).
The 11C2 CTL clone does not appear to be amplified at the tumor site. RNA from a tumor fragment and from 2A11 and 11C2 CTL clones was reverse transcribed and amplified for 40 cycles using a TCRBV1 or TCRBV6 and a BC primer. Amplified DNA was copied for five cycles in a runoff reaction using a fluorescent TCRBJ primer (13 BJ primers tested, BJ1S1-BJ1S7, BJ2S1-BJ2S6). The amplified products were analyzed on an automated sequencer. The profiles obtained show the size in nt (x-axis) and fluorescence intensity (y-axis) of the different amplified products. The absolute FU values obtained for dominant peaks are shown.
A cDNA coding for the Ag recognized by CTL 11C2
To identify the gene coding for the Ag recognized by 11C2, we used a genetic approach involving the transfection of a cDNA library into COS 7 cells, together with the cDNA coding for the HLA-A2-presenting molecules (19). We used a cDNA library constructed in pcDNA I expression vector using RNA extracted from RCC-7 cell line. This library was divided into 400 pools of 200 recombinant plasmids, and each pool was cotransfected in duplicate in COS-7 cells together with the autologous HLA-A*0201 pcDNA I construct. COS-7 cells were then tested for their ability to stimulate the production of TNF by clone 11C2. After 48 h, cotransfected COS-7 cells were incubated overnight with 11C2, and concentration of TNF in supernatant was determined by its cytotoxic effect on WEHI cells. The amount of TNF in supernatants was below 5 pg/ml, except for two pairs of higher duplicates (40 and 45 pg/ml). A second screening was performed by cotransfecting COS cells with 100 pools of 20 recombinant plasmids extracted from the positive duplicates. Finally, a third screening led us to the isolation of two cDNA clones that transferred the expression of the Ag in HLA-A*0201+ COS-7 cells. The results obtained with the A18 cDNA clone are shown in Fig. 3⇓A. The sequence of the longest cDNA, A18, was 1.9 kb long with a 100% homology to nt 577 to 2876 of the hsp70-2 cDNA (+1 is the hsp70-2 translation initiation site) (20), except for a mutation at position 877 (an adenine instead of the thymine), leading to a phenylalanine (F) to isoleucine (I) replacement at position 293 in the deduced protein sequence.
CTL 11C2 recognizes a mutated form of hsp70-2. A, Stimulation of CTL 11C2 by COS-7 cells transiently cotransfected with autologous HLA-A*0201 cDNA and either cDNA A18 or the full-length wt or mutated (∗) hsp70-2 cDNA. CTL 11C2 was added 48 h after cotransfection, and the TNF content of the supernatants was estimated 20 h later by testing its toxicity on WEHI-164 clone 13 cells. Control stimulator cells included the RCC-7 cell line as a positive control and COS-7 cells not transfected or transfected with HLA-A*0201 cDNA alone as negative controls. B, Stimulation of CTL 11C2 by allogeneic HLA-A2+ tumor cell lines transiently transfected with the full-length wt or mutated (∗) hsp70-2 cDNA. CTL 11C2 was added 24 h after transfection of a allogeneic RCC (RCC-5) or melanoma (MELA-1) cell line, and TNF content of the supernatants was estimated 20 h later by testing its toxicity on WEHI-164 clone 13 cells. Control stimulator cells included the RCC-7 cell line as a positive control and untransfected RCC-5 and MELA-1 cell lines as negative controls.
To identify the full-length hsp70-2 DNA and to check out whether the mutation is only found at the tumor site (or in other unrelated tumors), we performed a PCR (hsp70-2 is an intronless gene) on DNA extracted from the RCC-7 cell line as well as from the autologous EBV-transformed B cell line and PHA blasts. A unique ∼2-kb PCR product was obtained in each case corresponding to nt −36/+1974 and subcloned in pcDNA I vector for sequencing and expression. In the tumor fragment, four of seven DNA clones obtained included the mutation (no loss of heterozygosity), while in the EBV-transformed cell line or PHA blasts, none of the 14 DNA sequences analyzed display the mutation. Thus, the mutation was present only on one chromosome in the tumor cells and absent in the normal cells of the patient. These results were confirmed by SSCP (single-strand conformation polymorphism) analysis of RCC-7 DNA versus the autologous EBV-transformed B cell line DNA (data not shown). Interestingly, the SSCP pattern obtained with RCC-7 DNA was not found in a large panel of tumor DNA (including 9 head and neck squamous cell carcinomas, 15 breast carcinomas, 14 neuroblastomas, and 9 RCC), indicating that this particular mutation is rare in tumor hsp70 genes (data not shown).
Both the mutated and wt forms of full-length hsp70-2 were recognized when cotransfected with the HLA-A2 allele in COS-7 cells (Fig. 3⇑A). The possibility that the high-level replication of hsp70-2/pcDNA I in COS-7 cells generated an artificial CTL target that would not be produced under moderate expression conditions was then tested. We found that the mutated hsp70-2 and, to some extent, the wt form conferred recognition by CTL 11C2 in TNF release assays when transfected into allogeneic RCC or melanoma cells expressing HLA-A2.1 (Fig. 3⇑B). Overall, the mutation at amino acid position 293 seems to be specifically recognized by CTL 11C2, given that a clear-cut difference was observed between recognition of the mutated versus wt hsp70 products in three different experiments.
Identification of the antigenic peptide
To delimit the minimal nucleotidic region coding for the antigenic peptide, various truncated cDNA were obtained from cDNA clone A18 using exonuclease III to generate progressive deletions from the 3′ end of the cDNA (Fig. 4⇓). These cDNA fragments were cotransfected into COS-7 cells together with the autologous HLA-A*0201 cDNA. A minimal nucleotidic coding region was located between nt 730 and 944. Truncation of the region carrying the unique tumor-specific mutation abolished recognition by CTL 11C2 (Fig. 4⇓). We then searched for peptides bearing the HLA-A*0201-binding motif in this region, and among the 18 peptides tested, only 2, namely nonapeptide SLFEGIDIY (amino acids 286–294) and decapeptide SLFEGIDIYT (amino acids 286–295), with the isoleucine mutant residue at position 8, were recognized. Half-maximal lysis was obtained with only 5 × 10−11 M of the decapeptide compared with 5 × 10−7 M of nonapeptide (Fig. 5⇓). CTL 11C2 did also recognize the wt decapeptide 286–295 (SLFEGIDFYT) with a half-maximal lysis of 5 × 10−8 M, but not the wt nonapeptide 286–294 (Fig. 5⇓).
Location of hsp70-2 epitope region recognized by CTL 11C2 (A). The full-length hsp70-2 cDNA is schematically represented in white and black. 5′UT and 3′UT correspond to 5′ and 3′ untranslated regions, respectively. The coding region (in black) begins with the translation initiation site (ATG codon) and corresponds to the nt +1. The various truncated clones obtained from cDNA A18 are represented in grey. The A18 cDNA begins at nt 577 of the coding region. The mutated nt is marked by an asterisk (position 877). B, Stimulation of CTL 11C2 by COS-7 cells transiently cotransfected with the autologous HLA-A*0201 cDNA and each of the different truncated A18 cDNAs. Transfected cells were incubated for 24 h with 3000 CTL 11C2, and the amount of TNF was measured 20 h later. Control stimulator cells included COS-7 cells not transfected or transfected with A18 cDNA alone as negative controls and cotransfected by HLA-A*0201 and A18 cDNA as positive control.
Lysis of autologous EBV-transformed cell line incubated with hsp70-2-encoded peptides by CTL 11C2. Two thousand 51Cr-labeled EBV-transformed cells were incubated for 1 h in the presence of the indicated hsp70-2 peptides at various concentrations. CTL 11C2 were then added at an E:T ratio of 35:1. Chromium release was measured after 4 h. The asterisks indicate the mutated amino acid.
Binding of hsp70-2 peptides to HLA-A2
HLA-A2-binding peptide Ags are known to up-regulate expression of HLA-A2 molecules on T2 cells (17). The ability of hsp70-2 mutated or wt decapeptide Ag 286–295 was compared with that of the nonapeptide 286–294. The binding of both decapeptides was stable over time (till 4 h, not shown) at 20 μM (Fig. 6⇓), with no difference between the mutant and the wt forms at concentrations leading to up-regulation of HLA-A2 molecules (20, 2, and 0.2 μM). The hsp70-2 nonapeptides were less efficient, but their binding was still comparable with the HLA-A2 MART-127–35 peptide. As expected, no effect was seen for the control HLA-B7 peptide.
Induction of HLA-A2 expression on T2 cells by hsp70-2 peptide Ags. T2 cells were incubated at 26°C for 16 h in serum-free medium containing 0.8% DMSO with or without peptides at a concentration of 20 μM. Then peptides were added again, and cells were incubated at 37°C. At intervals of 30 min or 1 h, aliquots of cells were collected and change in HLA-A2 expression was monitored by flow-cytometric analysis with an anti-HLA-A2 mAb (MA2.1). Amino acid sequences of peptides are given; the mutated amino acid is marked by an asterisk.
Northern blot analysis
A hsp70-2 locus-specific probe, including the 3′ untranslated region, was used to examine the expression of the MHC-linked hsp70-2 gene (20). This probe does not hybridize to hsp70-1, which, in contrast to hsp70-2, is constitutively expressed at low levels. A 2.4-kb mRNA was observed from the heat-shocked RCC-7 and autologous EBV-transformed cell lines (Fig. 7⇓). Low level expression was also observed in the RCC-7 tumor (Fig. 7⇓, lane 1) and in the RCC-7 tumor cell line (lane 2), but not in the EBV-transformed B cell line (lane 3). Similar low level expression was also observed in other tumors, including melanoma, neuroblastoma, colon adenocarcinoma, and bladder tumor fragments (not shown). Thus, expression of hsp70-2, which is strictly stress induced (in contrast to hsp70-1), is commonly found in human tumors.
Northern blot analysis. Total cytoplasmic RNA (15 μg), derived from RCC-7 tumor (lane 1), RCC-7 cell line (lane 2), and autologous EBV-transformed cell line (lane 3) either maintained at 37°C (C) or heat shocked for 2 h at 42°C (HS) before harvesting, was fractionated on 1% agarose-formaldehyde denaturing gel and transferred onto Hybond-N+ nylon membrane. Northern blot was hybridized both with a hsp70-2-specific fragment and GAPDH cDNA as probes. The position of migration of the 28S and 18S RNA is shown. The hsp70-2 ∼2.4-kb transcript is indicated by an arrow above both the 18S RNA and the strong GAPDH hybridizing band.
Discussion
In the case of melanomas, many investigators have described the isolation of tumor-specific CTLs (21, 22), and such CTLs have been shown to play a role in tumor immunosurveillance (23). A number of Ags recognized by the CTLs have been characterized, and some of them are encoded by genes such as MAGE-1, MAGE-3, BAGE, and GAGE (24), which are not expressed in RCC. Melanoma-reactive T cells have also been shown to recognize the products of rare mutations such as the ones found in the β-catenin (25), MUM1 (26), and CDK-4 (27) genes. The immunogenicity of RCC in vivo may be much less than that of melanoma. This would explain why the frequency of detectable CTLs in conventional cytotoxicity assays against autologous tumor cells is demonstrably lower in RCCs than in melanomas. Nonetheless, evidence for the existence of common tumor-specific Ags in RCCs has been reported (5), and the recognition pattern of the four other CTL clones derived from RCC-7 (data not shown) is in line with this observation. The two Ags recognized by CTLs in human RCCs that have been characterized to date are encoded by a mutated HLA-A2 allele (28) or the rarely expressed (2% in RCC) RAGE gene (29).
In the TCR structure, the V-(D)-J junctional site, or CDR3, is thought to play a major role in determining fine specificity for Ags and to interact directly with antigenic peptides (30). Ag-driven T cell expansion in vivo would lead to the finding of recurrent TCR transcripts with identical junctional regions. Indeed, T cell clonal expansion has been demonstrated in the tumoral and peritumoral areas of solid tumors, such as melanomas (23, 31), RCCs (18, 32) including RCC-7 (7), head and neck cancers (33), and neuroblastomas (34). In most instances, the nature of the T cell epitopes recognized is not known, but the multiplicity of the T cell clonotypes observed in some patients suggests that multiple T cell epitopes are involved in triggering these immune reactions. Evidence for immune reactivity in the RCC-7 tumor was further supported by the finding of five different TILs that killed the autologous tumor cell line in vitro, while no CTL clone was isolated from 11 other tumors in similar culture conditions. Some of these TILs are clonally amplified in vivo to the extent of being detectable by FACS analysis, such as the 2A11 cell subpopulation that represents about 3% of TCR α/β+ TILs in RCC-7 (as determined with the B237.2 TCRBV1-specific mAb) (7). When analyzing freshly dissociated TILs in this tumor, more than 80% of them are activated CD8+DR+LAG-3+ cells. In addition, a polarized Th1-type response was observed (i.e., secretion of IL-2 and IFN-γ, and not of IL-4), following short-term in vitro activation of these TILs on anti-CD3-coated plates (7). Overall, the RCC-7 tumor cells with an acquired mutation of the hsp70-2 gene appear to be quite immunogenic in this patient.
The major hsp70 are encoded by a duplicated locus (hsp70-1, hsp70-2) located in the MHC region, 92 kb telomeric to the C2 gene (20). This segment of the MHC has been proposed to be termed the class IV region since it includes at least seven genes implicated to some degree in inflammation and in stress responses (35). The two intronless genes (hsp70-1 and hsp70-2) encoded an identical protein product of 641 amino acids. There are some sequence differences in the promoter regions, suggesting that they may be differentially regulated in response to stress factors. Hsp70-1 is constitutively expressed at low levels, while hsp70-2 is not. There is also complete divergence in the 3′ untranslated region that may confer different regulation on the two mRNA species. Using a hsp70-2-specific probe, we showed that elevated levels of a 2.4-kb mRNA are detected in cells following heat shock. In addition, this hsp70-2 probe detected low levels of these 2.4-kb transcripts in constitutive RNA from RCC tumor cell lines as well as from different frozen surgical tumor samples. The hsp70-2 promoter is apparently sensitive to various stress factors usually found in solid tumors (e.g., hypoxemia), leading to higher expression of HSPs in some tumors. The reason that allogeneic HLA-A2+ RCC cell lines that expressed low levels of hsp70-2 mRNA are not killed by CTL 11C2 may be related to the differences observed in target sensitization assays between the mutated and wt decapeptide 286–295 (5 × 10−11 M versus 5 × 10−8 M of peptide for half-maximal lysis). Since the point mutation at position 293 does not occur in two of the six allogeneic HLA-A2+ cell lines (as determined by SSCP analysis) tested in cytotoxicity assays, we speculate that HLA-A2 loading of a wt decapeptide 286–295 in these target cells was not enough to induce killing by CTL 11C2. In contrast, overexpression of the wt hsp70-2 cDNA in COS-7 cells induces TNF secretion by CTL 11C2, and transfection of allogeneic HLA-A2+ tumor cell lines with wt hsp70-2 cDNA also leads to some TNF secretion. The latter results implicate that the wt peptide loaded intracellularly onto MHC class I molecules in these transfected cells is recognized by CTL 11C2. TCR engagement by wt gene products in similar experimental conditions has, in contrast, not been observed in the case of another mutated gene, CDK-4 (27). Accordingly, recognition of stress-induced overexpressed hsp70 epitopes present on tumor cells by specific TILs may occur in vivo. Thus, the mutated hsp70 decapeptide described herein may be useful in breaking tolerance to the corresponding wt epitope in HLA-A2+ cancer patients.
Members of the hsp70 protein family are molecular chaperones that mediate protein de novo folding, translocation, and multimer assembly. Hsp70-2 has been identified as a putative susceptibility locus in organ-specific autoimmune diseases (36, 37). In addition, the use of a hsp70-2 allele-specific PCR (38) has shown that homozygosity was significantly associated with lymphoma (relative risk or RR = 18) and with breast carcinoma (RR = 16) (39). It is interesting to note that our RCC patient is homozygous for this allele. Frequency of this homozygosity is only 0.02 in control subjects, while it is 0.25 in patient groups (39). Altogether, hsp70-2 may have a potential role in cancer pathogenesis by participating in the regulation of antitumor immunity such as acting as a chaperone molecule for immunogenic tumor-associated peptides, but also in regulatory processes such as cell cycling. Indeed, hsp70-2 appears to be a molecular chaperone for CDC2 and is required for CDC2/cyclin B1 complex formation (40). Destruction of this complex would prevent development of the CDC2 kinase activity required to trigger G2/M-phase transition. Spermatogenic cells synthesize hsp70-2 during meiosis, and in hsp70-2−/− knockout mice infertility is associated with a failure of meiosis and a dramatic increase in spermatocyte apoptosis (41). The possibility that a mutated hsp70-2 chaperone has a dominant effect in tumor cells in triggering the G2/M phase transition during the mitotic cell cycle could not be excluded at the present time. Interestingly, among more than 100 RCC tumors studied in the laboratory, RCC-7 is the most aggressive one, with a rapid doubling interval in vitro and a high growth rate in SCID/nu mice. Preliminary experiments testing either the immortalizing or the transforming capacity of the mutated versus wt hsp70-2 cDNA in recipient cells have not led, however, to any direct evidence that the mutation plays a role in the oncogenic process.
HSPs are supposed to be nonpolymorphic molecules that do not differ in their primary structure among normal tissues and cancers or among normal and virus-infected cells. It has thus been reported that the immunizing ability of HSP preparations is related to the association of HSP molecules with peptides generated in the cells from which the HSPs are isolated. Indeed, HSP-peptide complexes can be generated in vitro, and the biological activity of these complexes is comparable with that of HSP-peptide complexes generated in vivo (42). Hsp70 has also been used with CTL epitopes to induce efficient protective antiviral immunity in addition to the generation of peptide-specific CTLs, and therefore represents an alternative to adjuvant and DNA vectors for the delivery of CTL epitopes to APCs (43). These observations demonstrate that HSPs are efficient CD8+ T cell response-eliciting adjuvants, but our results indicate that, in some instances, HSP-derived peptides are directly immunogenic. To our knowledge, the present work is the first example of mutated hsp70 in human disease.
HSP-specific human T cells are present with very high frequency at birth (44), and the immune system is supposed to be educated via diverse mechanisms such as peripheral tolerance, anergy, or apoptotic elimination so that the development of autoimmune diseases with an implication of HSP-specific T cells (e.g., diabetes, rheumatoid arthritis) is limited. HSPs are among the major targets of the immune response in mammals to bacterial, fungal, and parasitic pathogens. For instance, Mycobacterium tuberculosis hsp70 is an especially powerful Ag containing multiple B and T cell epitopes. The powerful immunologic features of HSPs have led to their experimental use as immunomodulators and as subunit vaccine candidates. In particular, hsp70 can stimulate in the absence of adjuvants strong and long-lasting immune responses against molecules that have been covalently attached to the HSPs (45, 46). It is interesting to note that HSPs are highly conserved proteins between divergent species and still represent in mammals preferred targets for immune recognition of complex organisms such as bacteria. In other words, much of the recognition of complex organisms by T cells occurs via conserved proteins (e.g., HSPs) and not via species-specific epitopes (for review, see 47 . Any deviation, such as a point mutation on these conserved structures, will induce a dramatic increase in immunogenicity. For instance, the mutated hsp70-2 decamer described in the present work is recognized with high avidity by TILs, while not differing in affinity for this HLA-A2-binding pocket compared with the wt decamer. Higher expression of HSPs in some tumors may act as a danger “flag” and permits HSP-directed attack induced by immunizations with appropriate bacterial preparations or heterologous HSPs. Thus, we propose that mutated human HSPs might be used in place of mycobacterial preparations to induce strong Th1/Tc1 immunogenicity and cytotoxic responses to HSPs and be more effective than bacillus Calmette-Guérin as tumor immunotherapy.
Acknowledgments
We thank Dr. Van Den Eynde for his valuable comments and help in optimizing the COS cell transfection procedures, Dr. Jean Benard for providing a panel of tumor DNA, and Dr. Graham A. W. Rook for helpful discussion.
Footnotes
↵1 This work was funded by a research grant from Ligue Nationale Contre le Cancer and Ligue Contre le Cancer, Comité des Hauts-de-Seine.
↵2 Address correspondence and reprint requests to Dr. Frédéric Triebel, Laboratoire d’Immunologie Cellulaire, Institut Gustave Roussy, 39, rue Camille Desmoulins, 94805 Villejuif Cedex, France. E-mail address: ftriebel{at}igr.fr
↵3 Abbreviations used in this paper: TIL, tumor-infiltrating lymphocytes; CDR3, complementarity-determining region 3; FU, fluorescence unit; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HSP, heat-shock protein; nt, nucleotide; RCC, renal cell carcinoma; SSCP, single-strand conformation polymorphism; wt, wild-type.
- Received June 29, 1998.
- Accepted October 8, 1998.
- Copyright © 1999 by The American Association of Immunologists
References
In this issue
