Research ArticleImmunotherapy

Anti–CTLA-4 therapy broadens the melanoma-reactive CD8+ T cell response

See allHide authors and affiliations

Science Translational Medicine  17 Sep 2014:
Vol. 6, Issue 254, pp. 254ra128
DOI: 10.1126/scitranslmed.3008918

Abstract

Anti–CTLA-4 treatment improves the survival of patients with advanced-stage melanoma. However, although the anti–CTLA-4 antibody ipilimumab is now an approved treatment for patients with metastatic disease, it remains unknown by which mechanism it boosts tumor-specific T cell activity. In particular, it is unclear whether treatment amplifies previously induced T cell responses or whether it induces new tumor-specific T cell reactivities. Using a combination ultraviolet (UV)–induced peptide exchange and peptide–major histocompatibility complex (pMHC) combinatorial coding, we monitored immune reactivity against a panel of 145 melanoma-associated epitopes in a cohort of patients receiving anti–CTLA-4 treatment. Comparison of pre- and posttreatment T cell reactivities in peripheral blood mononuclear cell samples of 40 melanoma patients demonstrated that anti–CTLA-4 treatment induces a significant increase in the number of detectable melanoma-specific CD8 T cell responses (P = 0.0009). In striking contrast, the magnitude of both virus-specific and melanoma-specific T cell responses that were already detected before start of therapy remained unaltered by treatment (P = 0.74). The observation that anti–CTLA-4 treatment induces a significant number of newly detected T cell responses—but only infrequently boosts preexisting immune responses—provides strong evidence for anti–CTLA-4 therapy–enhanced T cell priming as a component of the clinical mode of action.

INTRODUCTION

In recent years, major advances have been made in the field of cancer immunotherapy, and the clinical relevance of tumor-reactive T cells is now beyond doubt. The evidence for the clinical activity of tumor-reactive T cells is particularly strong for melanoma, both from trials exploring the adoptive transfer of ex vivo expanded tumor-infiltrating lymphocytes (1, 2) and from trials exploring treatment with antibodies that target T cell checkpoint molecules such as CTLA-4 and PD-1 (35).

Of these promising new immunotherapeutic treatments, positive phase 3 clinical trial data are thus far only available for ipilimumab, an antibody that targets the T cell checkpoint molecule CTLA-4 (4, 6). In a first trial, treatment of melanoma patients with ipilimumab, either as single agent or coadministered with a gp100 peptide vaccine, was shown to improve overall survival relative to a gp100 peptide vaccine control arm (4). In a second trial, ipilimumab plus dacarbazine treatment was shown to improve overall survival relative to dacarbazine monotherapy (6). On the basis of these data, ipilimumab has been approved as first- and second-line therapy for patients with metastatic melanoma by both the U.S. Food and Drug Administration and European Medicines Agency.

A particularly striking feature of ipilimumab treatment is that although antibody administration is restricted to a 9-week time window, a subgroup of patients experience a prolonged clinical benefit, indicating that treatment can induce a durable alteration in the interaction between tumor cells and the T cell compartment.

CTLA-4 is expressed on both activated effector CD4 and CD8 T cells and on regulatory T cells (Tregs) and competes with CD28 for binding to B7.1 and B7.2 on antigen-presenting cells. CTLA-4 displays a substantially higher affinity for the B7 molecules, and the inhibitory signals that are generated through this receptor effectively counteract the costimulation provided via CD28 (7).

Strikingly, even though there is now substantial clinical experience with anti–CTLA-4 treatment (8), its mechanism of action has not been fully elucidated. Seminal work by Leach and co-workers provided the first evidence that anti–CTLA-4 can enhance antitumor activity in murine models (9), an observation that has since then been reproduced in a variety of murine cancer models, either as monotherapy or in combination with other immunotherapeutic strategies [reviewed in (10)]. A significant part of the activity of anti–CTLA-4 treatment was recapitulated in murine models in which only CTLA-4 on effector CD4 and CD8 T cells was the target of antibody blockade. However, maximal activity of anti–CTLA-4 treatment required the targeting of CTLA-4 on both effector cells and Tregs (11), suggesting multiple mechanisms of action. With respect to the relative role of CTLA-4 blockade on CD4 versus CD8 cells, antibody-mediated T cell depletion studies have demonstrated that anti–CTLA-4–mediated tumor control is in large part dependent on the presence of cytotoxic CD8 T cells (12) and in some, but not all, mouse models. This effect also requires the presence of CD4 cells, suggesting that anti–CTLA-4 treatment may in part work through increased CD4 T cell help (12, 13).

Finally, recent studies from a number of groups have provided evidence for an unexpected mechanism of action of anti–CTLA-4 in murine models. Specifically, work from Simpson and co-workers has demonstrated that anti–CTLA-4 administration leads to the specific depletion of Tregs within the tumor microenvironment but not lymph nodes, in keeping with the higher expression level of CTLA-4 on intratumoral Tregs. Together with the observation that the antitumor effect of anti–CTLA-4 treatment in this preclinical model is dependent on Fcγ receptor expression, these data argue in favor of an intratumoral mechanism of action of CTLA-4 blockade that involves the depletion of local Tregs (14). Additional support for this model comes from work by Bulliard et al. that also demonstrated Fc receptor γ (FcRγ)–dependent intratumoral Treg depletion upon anti–CTLA-4 treatment (15). Finally, the potential role of antibody-dependent cell-mediated cytotoxicity (ADCC) in the antitumor activity of anti–CTLA-4 is also supported by the observation that anti–CTLA-4 isotypes that are capable of inducing ADCC exhibit superior antitumor activity in preclinical models, relative to isotypes that lack this capacity (16).

Although there is now ample data on the effects of CTLA-4 blockade on T cell responses in murine models, complementary knowledge on the mechanism of anti–CTLA-4 therapy in humans is still largely lacking. An increase in the ratio of peripheral blood effector T cells over Tregs has been described by several groups (1719), with this increased ratio reflecting an increase in the number of activated effector T cells rather than a reduction in Treg. Likewise, an increase in the number of activated CD4 T cells at the tumor site has been demonstrated, using ICOS expression as a readout (19). Data on the effect of anti–CTLA-4 treatment on defined tumor-specific CD8 T cell responses are, however, still restricted to a small number of antigens. Specifically, for a limited number of patients, priming of T cell responses specific for NY-ESO-1 has been demonstrated after anti–CTLA-4 therapy (20), whereas this was not the case for T cell responses against MART-1, tyrosinase, and gp100 (21).

Here, we provide a systematic analysis of the effects of anti–CTLA-4 treatment on melanoma-specific CD8 T cell responses. We selected a cohort of patients who express the human leukocyte antigen-A*0201 (HLA-A*0201) allele to allow monitoring of T cell activity against the large collection of melanoma-associated shared epitopes that have been described for this allele. We use the resulting data to examine whether there is evidence for T cell priming as a mode of action of anti–CTLA-4 therapy or whether the effects of anti–CTLA-4 on the melanoma-specific CD8 T cell repertoire are more consistent with a primary activity at the tumor site. In the former case, appearance of new T cell reactivities upon treatment would be expected to be a dominant effect. In the latter case, preexisting and newly appearing melanoma-specific T cell responses would be expected to be influenced in a similar manner (Fig. 1A).

Fig. 1.

(A) Potential effects of anti–CTLA-4 treatment on the melanoma-reactive T cell repertoire. Top: Increase in the magnitude of the preexisting melanoma-reactive T cell response. Bottom: Broadening of the melanoma-reactive T cell response. (B) Reproducibility and sensitivity of pMHC multimer combinatorial coding technology. Data show the correlation between the frequency of antigen-specific T cell responses as detected in primary screens and in independent confirmation stains, using newly generated MHC multimer reagents labeled with distinct fluorochrome combinations.

RESULTS

Kinetics of anti–CTLA-4 induced alterations in melanoma-specific CD8+ T cell responses

To first assess whether melanoma-specific CD8 T cell responses in peripheral blood from patients with advanced melanoma were substantially influenced by anti–CTLA-4 treatment, we selected a sample set of 10 patients for which a large number of time points were available before, during, and after treatment. All samples were analyzed by combinatorial coding (22, 23), using a panel of peptide–major histocompatibility complex (pMHC) multimers loaded with 72 melanoma-associated epitopes (table S1). In addition, three viral epitopes, derived from Epstein-Barr virus (EBV), cytomegalovirus (CMV), and influenza A, were included to allow comparison with T cell reactivity against persistent and nonpersistent viruses.

With this strategy for the analysis of melanoma-specific T cell responses, antigen-specific T cell populations were detected as dual-fluorochrome–positive cell populations, an approach that substantially increases the sensitivity of antigen-specific T cell detection relative to single-marker–based analyses (22, 23). Antigen-specific T cell populations were identified using a standardized cutoff of 0.005% of total CD8 cells and ≥10 events. Furthermore, all T cell responses detected were validated in independent experiments, using a different fluorochrome code (see Materials and Methods). The robustness and sensitivity of this assay system are reflected by the high correlation (R2 = 0.9626, Spearman R test; n = 179) between these pairs of independent measurements (Fig. 1B). With a detection limit of 1 in 20,000 CD8 cells, which lies only slightly above estimates of human T cell precursor frequencies (24, 25), any antigen-specific naïve T cell pool that has undergone more than just a few cell divisions can be expected to be detected.

Analysis of peripheral blood samples of the first cohort of patients revealed the presence of at least one melanoma-specific CD8 T cell response in 9 of 10 patients (example given in Fig. 2; additional results shown in fig. S2). An increase in the number of detectable melanoma-specific CD8 T cell responses at week 12 after anti–CTLA-4 treatment was seen in 5 of 10 patients (median, 0.007% of total CD8+ cells; range, 0.005 to 0.419% of total CD8+ cells). These newly detected T cell responses generally became apparent within the first few weeks after start of therapy (median, 13.5 days from the start of therapy; range, 3 to 54 days), indicating that the effects of anti–CTLA-4 on the melanoma-reactive T cell repertoire can be rapid. Notably, within this small cohort, the magnitude of preexisting melanoma-specific T cell responses was not influenced by therapy and stable for up to 1 year (example of preexisting and newly detected T cell responses shown in Fig. 2).

Fig. 2. T cell responses detected pre- and post–anti–CTLA-4 treatment.

(A) Examples of T cell reactivities detected in one of the analyzed sample sets. Antigen-specific T cell populations are located in the diagonal, as the double-positive position, and indicated in black. Cells indicated in green are double positive in a color combination that includes only one of the two colors displayed in that particular plot. Numbers indicate the percentage of MHC multimer-positive CD8 cells of total CD8 cells. Gating strategy is provided in fig. S1. (B) Kinetics of T cell responses before, during, and after anti–CTLA-4 therapy. Top panel shows the course of preexisting melanoma and virus-specific CD8 T responses (defined as all responses detected at the first available time point). Bottom panel shows newly detected CD8 T cell responses (defined as responses that are first detected at any point in time after the first analyzed time point). In the example shown, a CD8 T cell response specific for the gp100YVL epitope is first detected 7 days after the first infusion of ipilimumab. X axis indicates time relative to start of therapy. Red dots indicate time points for blood collection.

Anti–CTLA-4 therapy broadens the melanoma-specific CD8 T cell repertoire

Having demonstrated that changes in melanoma-specific T cell responses upon anti–CTLA-4 therapy can be detected, we set out to analyze a larger sample set to describe the pattern by which CTLA-4 blockade influences melanoma-specific CD8 T cell reactivity. To this purpose, we compared CD8 T cell reactivity in samples from 30 additional HLA-A*0201 patients before and after anti–CTLA-4 treatment. Furthermore, to assess whether anti–CTLA-4 treatment influences T cell reactivity against different classes of melanoma-associated antigens in the same manner, the 72-epitope panel was extended to include all known shared melanoma-associated epitopes that are restricted to HLA-A*0201 (n = 145). Peripheral blood mononuclear cell (PBMC) samples from each time point were analyzed for reactivity with this extended panel of epitopes, and the magnitude and number of melanoma-specific T cell responses before and after therapy were compared.

Within the combined cohort of 40 patients, melanoma-reactive T cell responses were identified in 33 sample sets (Fig. 3). A total of 121 melanoma-specific T cell responses were detected, directed against 22 different epitopes, with a median magnitude of 0.023% of total CD8 cells (range, 0.005 to 6.214%). A large fraction of the previously described shared melanoma-associated epitopes is derived from so-called overexpressed antigens, and this is reflected in the composition of the epitope set of known shared HLA-A*0201 epitopes [of the 145 epitopes, overexpressed antigens, 45%; cancer/germ-line antigens, 27%; melanocyte differentiation antigens, 18%; and unclassified/mutated antigens, 10% (26, 27)]. Analysis of T cell responses in the cohort of 40 patients revealed that reactivity was only observed for a relatively small subset of previously described epitopes. In particular, T cell populations were only detected for 10% of the previously described epitopes derived from overexpressed antigens, whereas recognition of epitopes derived from cancer/germ-line and melanocyte differentiation antigens was more common (25% of epitopes each). Notably, the majority of the T cell responses detected against overexpressed antigens are specific for epitopes derived from alternative open reading frame or alternative splicing events, suggesting that within the class of overexpressed antigens, this subset may be of greatest relevance.

Fig. 3. Anti–CTLA-4 therapy selectively induces novel melanoma-reactive CD8 T cell responses.

Heat map summarizing all melanoma-reactive T cell responses detected within this sample set, with the color scale indicating response magnitude pre- and posttherapy. Color code: light green, 0.005 to 0.099%; dark green, 0.1 to 0.99%; orange, 1.00 to 4.99%; and red, >4.99% pMHC multimer+ CD8+ of total CD8+ cells. Only those epitopes against which reactivity was detected in at least one sample are shown.

To assess whether anti–CTLA-4 treatment commonly increases the magnitude of melanoma-specific T cell responses that already exist before onset of treatment, we determined the relative frequency of those antigen-specific T cell responses that were detected both before and after treatment. Alternately, to assess whether there is evidence for anti–CTLA-4–induced T cell response broadening, we determined how often the number of melanoma-specific T cell responses increased after treatment. Strikingly, the magnitude of preexisting T cell responses was in most patients completely unaltered upon ipilimumab treatment (median fold change of 1.04; P = 0.69, paired, two-tailed Student’s t test; n = 40; Fig. 4A). Likewise, preexisting CD8 T cell responses specific for epitopes from influenza A, CMV, or EBV were not influenced by ipilimumab administration (median fold change of 1.1; P = 0.74, paired, two-tailed Student’s t test; n = 96; Fig. 4B). In contrast, upon ipilimumab treatment, an increase in the number of melanoma-specific T cell responses was commonly observed (19 of 40 patients), with a decrease occurring in only 2 (P = 0.009, paired, two-tailed Student’s t test, comparing the number of responses before and after therapy per patient; n = 40) and with the number of virus-specific T cell populations remaining stable (P = 0.74, paired, two-tailed Student’s t test; n = 40). In many patients, only a single novel HLA-A*0201–restricted T cell response appeared upon anti–CTLA-4 treatment. However, in some patients, up to six novel melanoma-specific T cell responses that were restricted by HLA-A*0201 were induced within this short time window.

Fig. 4. Preexisting melanoma and virus-specific CD8 T cell responses are generally unaltered by anti–CTLA-4 treatment.

(A) The relative magnitude of 40 melanoma-associated T cell responses that were detected both before and after therapy is plotted. Median fold change is 1.04; P = 0.69, paired, two-sided Student’s t test. (B) The relative magnitude of 96 virus-specific T cell responses that were detected both before and after therapy is plotted. Median fold change is 1.1; P = 0.74, paired, two-sided Student’s t test. (C) The rate of T cell response broadening from time of diagnosis to start of anti–CTLA-4 treatment is compared to the rate of T cell response broadening in the 12-week pretherapy to posttherapy time window. P = 0.015, paired, two-sided Student’s t test; n = 20.

To further explore whether the observed broadening is directly linked to therapy or merely a reflection of a continuing broadening of melanoma-specific T cell reactivity during disease progression, we compared the kinetics with which novel T cell responses appear before start of anti–CTLA-4 therapy with the kinetics of broadening in the first 12 weeks after therapy. For 20 patients from whom we had data on time of diagnosis available, the rate with which new T cell responses appeared in the absence of anti–CTLA-4 treatment was calculated as the number of detected responses in the sample taken before start of treatment divided by time (months) from diagnosis (Fig. 4C). This rate of immunotherapy-unrelated T cell response broadening was compared to the rate of anti–CTLA-4 therapy–related T cell response broadening (that is, the number of newly detected responses in the posttherapy sample per time unit between pre- and postsample collection). This comparison demonstrates that anti–CTLA-4 treatment leads to a significant increase in the rate of T cell response broadening (pretherapy, 0.029 T cell responses/month; posttherapy, 0.321 T cell responses/month; P = 0.015, paired, two-tailed Student’s t test; n = 20).

Collectively, the above data reveal a specific signature of the effect of anti–CTLA-4 treatment on the melanoma-specific CD8 T cell repertoire in which novel T cell responses are seen in a substantial fraction of patients. To test whether these newly detected T cell responses could potentially contribute to tumor control, we isolated the pMHC multimer reactive cells of four T cell populations that were detected only after anti–CTLA-4 therapy, and cloned their T cell receptors (TCRs) by TCR gene capture (28). Subsequently, TCR-encoding retroviruses were generated and used to modify donor PBMC, and tumor reactivity of the resulting TCR-modified T cell populations was tested against a set of tumor lines expressing the relevant antigens. Notably, for three of the four TCRs tested (specific for the NY-ESO-1SLL, SSX-2KAS, and MAGE-C2ALK antigens), recognition of multiple antigen-expressing melanoma lines was observed (Fig. 5). These data demonstrate not only that anti–CTLA-4 treatment induces a broadening of the melanoma-specific CD8 T cell responses but also that these newly detected T cell reactivities have a demonstrable tumor recognition potential.

Fig. 5. Newly detected T cell responses after ipilimumab therapy show tumor reactivity.

TCRs isolated from melanoma-reactive T cell responses induced upon anti–CTLA-4 treatment were transduced into PBMC, and resulting cells were tested for recognition of tumor lines known to express the relevant targets, using CD107a expression as a measure of T cell activation. Experiments were conducted in duplicates, and individual values are depicted for two separate experiments for each TCR (data points separated on the x axis). The percentage of tumor cell–reactive cells was corrected for transduction efficiency.

DISCUSSION

In spite of intense clinical testing of anti–CTLA-4 therapy in particular in melanoma patients, little is known about the effects of this therapy on the tumor-reactive cytotoxic T cell repertoire in humans. Here, we carried out in total more than 11,000 individual analyses of antigen-specific T cell responses against all known shared HLA-A*0201–associated epitopes to address this issue.

We have focused this analysis on the HLA-A*0201 allele because the number of known melanoma-associated epitopes is by far the highest for this HLA class I allele (26). In spite of this focus on a single HLA allele, melanoma-specific CD8 T cell responses were detected in more than 80% of the sample sets analyzed, and on the basis of this observation, it seems plausible that T cell recognition of melanoma antigens through any of the available HLA class I alleles occurs in virtually all, if not all, melanoma patients. Most of the melanoma-specific T cell responses that are detected in this patient group are of a low magnitude, suggesting that it would be difficult to reveal them using less-sensitive assays. Nevertheless, it is apparent that, in some cases, the effects of ipilimumab on melanoma-specific CD8 T cell responses can be profound; in one patient, experiencing a partial clinical response (ongoing at 11 months), NY-ESO-1/LAGE-1–specific T cells increased from <0.005 to 6.2% of total CD8 cells in a 3-month period from start of treatment. Previous work has suggested that T cell recognition of the NY-ESO-1 antigen may be associated with tumor regression. In particular, a clinical trial with TCR gene–modified T cells targeting NY-ESO-1 has demonstrated that NY-ESO-1 can function as a tumor regression antigen (29). Furthermore, antibody responses against NY-ESO-1 have been shown to correlate with positive clinical outcome subsequent to anti–CTLA-4 therapy (20). On the basis of these earlier data from the Yuan and Robbins groups (20, 29) and the observation that the one patient in whom a profound NY-ESO-1–specific T cell response was induced by anti–CTLA-4 treatment experienced a prolonged partial response (overall response rate: partial response, 17.5%; complete response, 5%), it could be of interest to combine anti–CTLA-4 treatment with an NY-ESO-1 vaccine in patients with NY-ESO-1–positive tumors.

Here, we have assessed changes in melanoma-reactive T cell populations in the peripheral blood compartment. It has previously been reported that the expression of CTLA-4 is high on tumor-infiltrating T cells (30), and it would therefore be valuable to also examine changes in T cell frequencies at the tumor site. Conceivably, this may be achieved by comparison of pre- and posttherapy biopsy material. However, the high variability between antigen-specific T cell responses previously observed for different areas of the same tumor site will reduce the ability of such an approach to reveal differences (27). In addition to the analysis of T cell reactivity against shared antigens performed here, it will be useful to analyze the impact of ipilimumab on T cell responses against mutated antigens. The vast majority of mutations in human melanomas are considered private, complicating the analysis of large patient numbers; but with the recent developments in screening technologies, such analyses may become feasible in the coming years (31). Finally, by focusing on T cell reactivity toward shared antigens restricted by one of six possible HLA alleles in each patient, this study was not designed to assess the relationship between T cell activity and clinical response, and exploratory analysis indeed did not reveal any significant relationship between T cell response characteristics and clinical course.

The major finding of this study is that anti–CTLA-4 treatment leads to a significant broadening of the melanoma-specific CD8 T cell response. In contrast, the magnitude of preexisting T cell responses was, on average, not increased by therapy. Specifically, whereas nine of the preexisting CD8 T cell responses against shared melanoma antigens increased more than twofold upon therapy, a decrease of a similar magnitude was seen in seven cases. Recently, we demonstrated a fivefold increase in a T cell response against a patient-specific neoantigen in a patient treated with ipilimumab. Whether this represents one of the infrequent cases in which a preexisting T cell response increases in magnitude upon therapy or reflects a differential effect of anti–CTLA-4 therapy on T cell responses that are specific for neoantigens needs to be established in further work.

Recent work that has used TCR V-β CDR3 (complementarity determining region 3) sequencing has demonstrated that anti–CTLA-4 therapy leads to alterations within the T cell repertoire that can be detected within peripheral blood, and in one study, maintenance of high-frequency TCR clonotypes during anti–CTLA-4 treatment was associated with patient survival (32, 33). An advantage of the approach taken in these studies is that it provides a comprehensive overview of changes within the entire TCR repertoire. However, whether these changes also involve melanoma-specific CD8 T cell responses has not been established, making it difficult to compare the results from these analyses with the data obtained here.

Recent data obtained in mouse models have revealed that one mechanism of action of anti–CTLA-4 treatment is the intratumoral depletion of Tregs (14, 15), a mechanism that relies on the subsequent antitumor effect of local CD8 T cells. The binding affinity of ipilimumab to human FcγRIIIA (the human equivalent to the mouse FcγRIV) is similar to that of rituximab, an antibody for which there is substantial evidence for an FcR-mediated mode of action (14). Although direct data in humans are lacking, it is therefore plausible that the effects of ipilimumab in humans can also in part be mediated by FcR-dependent Treg depletion at the tumor site. The data obtained here provide direct evidence that, in addition to this postulated tumor site–restricted action, ipilimumab significantly broadens systemic CD8 T cell responses. This broadening of the melanoma-reactive T cell repertoire may provide the tumor-specific CD8 T cell reactivities that could benefit from the effects of ipilimuab treatment within the tumor microenvironment.

MATERIALS AND METHODS

Patient material

Peripheral blood samples were collected from patients with advanced melanoma treated with ipilimumab, in accordance with local guidelines and following signed informed consent. Patients were treated either in the expanded access program or after approval of ipilimumab, as standard second-line therapy. Samples were selected on the basis of four-digit genotyping for HLA-A*0201. PBMCs were isolated using standard Ficoll gradient centrifugation separation, according to local operating procedures. Cells were cryopreserved in liquid nitrogen, in fetal calf serum (FCS) with 10% dimethyl sulfoxide (DMSO), or in FCS supplemented with RPMI and 10% DMSO, for up to 3 years before analysis. Cells were thawed and recovered with DNAse (deoxyribonuclease) for 1 hour before counting.

Cell lines

For T cell reactivity assays, previously established patient-derived melanoma tumor cell lines were used. Tumors 1 to 5 were obtained in-house from freshly resected material, whereas tumor 6 was provided by I. M. Svane and P. thor Straten (Center for Cancer Immune Therapy, Herlev Hospital, Copenhagen). Tumors 1 to 4 were established as in vitro cell lines after xenotransplantation (single in vivo passage) of fresh primary material into non-obese diabetic/severe combined immunodeficient IL2Rg(null) mice. Tumors 5 and 6 were established directly in vitro. Cell lines were cultured in RPMI supplemented with FCS (10%), penicillin (100 IU/ml), and streptomycin (100 μg/ml) and passaged when 90% confluence was reached. The references for the tumor cell lines are as follows: tumor 1: M026.X1, tumor 2: M026R.X1, tumor 3: M029.X1, tumor 4: M002.X1, tumor 5: NKIRTIL006, and tumor 6: MM909.04.

pMHC multimer collections

The 72-epitope panel is given in table S1; the 145-epitope panel has been described previously (26, 27). To verify successful pMHC multimer panel generation and cell staining, a set of viral epitopes from influenza A, EBV, and CMV was included within each analysis. For all sample sets included, at least one virus-specific T cell response was detected.

Melanoma-associated peptides as well as ultraviolet (UV) cleavable peptides were synthesized in-house, as described previously (34). Recombinant HLA-A*0201 heavy chains and human β2m light chain were produced in Escherichia coli and isolated from inclusion bodies. MHC class I refolding reactions and purification by gel filtration high-performance liquid chromatography were performed as described previously (34). Specific pMHC complexes were generated by UV-induced ligand exchange in a 96-well format. In brief, pMHC complexes loaded with UV-sensitive peptide (100 μg/ml) were subjected to 366-nm UV light (CAMAG) for 1 hour at 4°C in the presence of rescue peptide (200 μM).

pMHC multimers were generated using a total of eight different fluorescent streptavidin (SA) conjugates (Invitrogen). For each 10 μl of pMHC monomer (100 μg/ml), the following amounts of SA conjugates were added: 1.5 μl of SA-QD605 (Q10101MP), 1.0 μl of SA-QD625 (A10196), 1.5 μl of SA-QD655 (Q10121MP), 1.5 μl of SA-QD705 (Q10161MP), 1.0 μl of SA-QD800 (Q10171MP), 1.1 μl of SA–PE (phycoerythrin) (1 mg/μl, SA1004-4), 1.1 μl of SA-Cy7-PE (1 mg/μl, SA1012), and 0.6 μl of SA–APC (allophycocyanin) (1 mg/μl, SA1005). For each pMHC monomer, conjugation was performed with two of these fluorochromes. Mixtures were incubated for 30 min on ice. NaN3 (0.02% w/v) and an excess of d-biotin (26.4 mM, Sigma) were added to block residual binding sites.

pMHC multimer combinatorial coding

For T cell staining, the following amounts of fluorescently labeled pMHC complexes were pooled together for combinatorial coding or used separately for confirmations: 1 μl of PE-pMHC, 2 μl of APC-pMHC, 3 μl of QD605-pMHC, 2 μl of QD625-pMHC, 2 μl of QD655-pMHC, 4 μl of QD705-pMHC, 4 μl of QD800-pMHC, and 3 μl of Cy7-PE-pMHC. Final staining volume was 128 μl, and cells were incubated at 37°C for 15 min. Subsequently, 2 μl of anti–CD8–FITC (fluorescein isothiocyanate) (BD, 345772), 1 μl of anti–CD4-AF700 (Invitrogen, MHCD0429), 1 μl of anti–CD14-AF700 (Invitrogen, MHCD1429), 1 μl of anti–CD16-AF700 (Invitrogen, MHCD1629), 3 μl of anti–CD19-AF700 (Invitrogen, MHCD1929), and 0.5 μl of LIVE/DEAD Fixable IR Dead Cell Stain Kit (Invitrogen, L10119) were added for 20-min incubation on ice. Before flow cytometric analysis, cells were washed twice.

Flow cytometry

Data acquisition was performed on an LSR II flow cytometer (Becton Dickinson) with FACSDiva software. The following 11-color instrument setting was used for combinatorial coding analyses: UV laser (355 nm): QD605, 595LP, 605/12; QD705, 685LP, 710/50; QD800, 750LP, 780/60. Violet laser (405 nm): QD625, 610LP, 625/20; QD655, 635LP, 655/8. Blue laser (488 nm): FITC, 505LP, 525/50. Yellow-green laser (561 nm): PE, 585/15; PE-Cy7, 750LP, 780/60. Red laser (640 nm): APC, 670/14; AF700, 685LP, 710/50; IRDye, 750LP, 780/60. To identify antigen-specific T cells, the following gating strategies were used: (i) selection of live (IRDye dim) single-cell lymphocytes [forward scatter (FSC)–W/H low, side scatter (SSC)–W/H low, FSC/SSC-A], (ii) selection of anti–CD8-FITC+ and “dump” (anti-CD4, anti-CD14, anti-CD16, and anti-CD19) negative cells, and (iii) selection of CD8+ T cells that were positive in two and only two MHC multimer channels. Cutoff values for the definition of positive responses were ≥0.005% of total CD8+ cells and ≥10 events. A minimum of 50,000 CD8+ T cells were acquired per sample. All responses were confirmed in independent experiments, using a different fluorochrome combination. Because of the sensitivity of the assay, magnitude of T cell responses is listed with four digits, and this does not reflect precision of detection. To monitor the reproducibility of the assay system, reference samples with four known T cell responses present at varying frequencies were included in each analysis.

TCR recognition of tumor lines

TCRs were isolated from four antigen-specific T cell populations that were first detected after therapy. In brief, antigen-specific T cells were identified by pMHC multimer staining and sorted as single cells. After a short in vitro expansion, antigen specificity was confirmed by pMHC multimer staining, DNA was isolated, and TCR sequences were determined by TCR gene capture (28). Tumor reactivity of the isolated TCRs was assessed after retroviral transduction of TCR sequences into healthy donor PBMCs using allogeneic tumor cell lines as target cells. After retroviral transduction of T cells, transduction efficiency was determined by pMHC multimer staining, and cells were cryopreserved after 2 weeks of expansion in complete medium (RPMI 1640 + 10% Human AB serum) supplemented with interleukin-2 (IL-2) (100 IU/ml) (Novartis) and IL-15 (5 ng/ml) (PeproTech). Cocultures were set up by addition of 2 × 105 effector cells to target cells in a 1:1 ratio in 96-well plates and were incubated for 5 hours at 37°C in the presence of anti–CD107a-PE antibody (BD Biosciences). After 1 hour, brefeldin A and monensin (BD Biosciences) were added, according to the manufacturer’s recommendations. After incubation, cells were washed and stained with anti–CD3-FITC (BD Biosciences) and anti–CD8–peridinin chlorophyll protein (PerCP)–Cy5.5 (BD Biosciences). Functional activity was analyzed by flow cytometry using a BD Calibur and FlowJo and was corrected for transduction efficiency.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/6/254/254ra128/DC1

Fig. S1. Overview of combinatorial encoding approach.

Fig. S2. Kinetics of T cell responses detected before and after anti–CTLA-4 therapy.

Table S1. Epitope sequences.

REFERENCES AND NOTES

  1. Acknowledgments: We thank A. Pfauth and F. van Diepen for flow cytometric support, M. Toebes for generation of essential MHC reagents, K. Kemper for four of the tumor cell lines, and I. M. Svane and P. thor Straten for one of the melanoma cell lines. Funding: This work was financially supported by Bristol-Myers Squibb (CA184–155) and by a Stand Up To Cancer—Cancer Research Institute Cancer Immunology Translational Cancer Research grant. Stand Up To Cancer is a program of the Entertainment Industry Foundation administered by the American Association for Cancer Research. Author contributions: P.K. performed, designed and interpreted experiments, and cowrote the manuscript; D.P., L.H., and C.L. performed experiments; S.K. provided patient information, C.O., D.J.-P., M.J.P.W., S.v.d.B., E.K., O.M., E.R., D.S., and C.B. provided patient material; J.B.H. and T.N.S. designed and interpreted experiments and cowrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article