Antigen-specific Tregs control T cell responses against a limited repertoire of tumor antigens in patients with colorectal carcinoma

Spontaneous antitumor T cell responses in cancer patients are strongly controlled by Tregs, and increased numbers of tumor-infiltrating Tregs correlate with reduced survival. However, the tumor antigens recognized by Tregs in cancer patients and the impact of these cells on tumor-specific T cell responses have not been systematically characterized. Here we used a broad panel of long synthetic peptides of defined tumor antigens and normal tissue antigens to exploit a newly developed method to identify and compare ex vivo the antigen specificities of Tregs with those of effector/memory T cells in peripheral blood of colorectal cancer patients and healthy subjects. Tregs in tumor patients were highly specific for a distinct set of only a few tumor antigens, suggesting that Tregs exert T cell suppression in an antigen-selective manner. Tumor-specific effector T cells were detectable in the majority of colorectal cancer patients but not in healthy individuals. We detected differences in the repertoires of antigens recognized by Tregs and effector/memory T cells in the majority of colorectal cancer patients. In addition, only effector/memory T cell responses against antigens recognized by Tregs strongly increased after Treg depletion. The selection of antigens according to preexisting T cell responses may improve the efficacy of future immunotherapies for cancer and autoimmune disease.

Malignant transformation and cancer progression are immunologically relevant events in immunocompetent hosts (1). Molecular identification of human tumor-associated antigens (TAAs) in the last decade has led to the development of antigen-specific immunotherapy targeting these antigens. Despite promising results from animal studies, limited objective clinical responses were observed in cancer patients (2, 3). Recent advances in understanding the mechanisms underlying antigen presentation, notably the role of DCs, have led to the development of synthetic long peptides for vaccination purposes. Use of these peptides circumvents HLA restrictions (4), leads to more efficient peptide presentation (5), and minimizes induction of tolerance through antigen presentation on nonprofessional APCs, such as T and B cells (6, 7). Such peptides have been shown to elicit T cell responses against TAA in colorectal carcinoma (CRC) patients (8). However, several challenges remain to be overcome, including the insufficient antitumor responses due to immunosuppression driven by Tregs.

Tregs play a critical role in the maintenance of peripheral self tolerance. Naturally occurring CD4⁺CD25^hi Tregs are produced in the thymus (9) and express FoxP3, a transcriptional factor required for establishment and maintenance of Treg lineage identity and suppressor function (10, 11). Tregs accumulate at the tumor site, where they suppress the effector function of tumor antigen–specific T cells, resulting in tumor growth despite the presence of tumor antigen–specific T cells (12, 13). Increased densities of tumor-infiltrating FoxP3⁺ Tregs have been associated with poor prognosis in various solid tumors, including pancreatic (14), ovarian (12, 15), and hepatocellular carcinoma (16, 17). Depletion of Tregs results in enhanced antitumor immunity and tumor rejection in murine models (18) but may also result in the development of autoimmune diseases (19, 20). Consistent with the enhanced antitumor immunity observed in mice, depletion of Tregs in the peripheral blood of patients with CRC was recently shown to boost CD4⁺ T cell responses to TAAs (21).

Because of the superior suppressive capacities of antigen-specific Tregs over non-antigen-specific Tregs, as shown in mouse studies (22–24), it is critical to determine the TAA recognition patterns of Tregs in cancer patients (25). Previous studies demonstrated the presence of antigen-specific Tregs for restricted sets of antigens (26–28). However, thus far, there has been no comprehensive study comparing the actual distribution of Treg specificities for multiple tumor antigens in cancer patients with their impact on effector T cell (Teff) responses.

In this report, we demonstrate the presence of TAA-specific Tregs in CRC patients against multiple TAAs. Using a broad panel of long synthetic peptides of defined tumor antigens and normal tissue antigens, we here exploited a newly developed method to identify and compare ex vivo the antigen specificities of Tregs with those of memory/effector T cells (Tmems/Teffs). Depletion of Tregs led to increased recognition rates of TAAs as determined by IFN-γ ELISPOT analysis. The strongest increases in TAA recognition were seen for the same TAAs for which specific Tregs were frequently found in the patients. Strikingly, not all TAA-specific T cell responses were under the control of Tregs. The ability to determine Treg specificities in patients raises the possibility of choosing TAAs for vaccination purposes without further impairing tumor-specific immunity by inducing preexistent Tregs in patients.

TAA-specific Teff/Tmem responses are found in CRC patients. Nine TAAs frequently overexpressed in CRCs were selected for the study (Table 1). Additionally, 3 normal self antigens that have not been described to be tumor specific (collagen IV, proinsulin, colon-specific intestinal membrane A4 protein [A4]) were chosen as normal self antigens for comparison (Table 2). From these antigens, long peptide sequences (30–50 amino acids in length) were chosen that contained at least one of the characterized HLA-A*0201–restricted nonapeptides. The remaining sequence was defined using the SYFPEITHI database (29) in such a way that the predicted sequence was optimal for presentation on a wide range of HLA types.

Table 1

Tumor antigens and synthetic long polypeptides

Table 2

Normal self antigens and synthetic long polypeptides

Peripheral blood was taken from 170 patients before tumor resection. Median age of the tumor patients was 66 years (mean, 63.7 ± 12.0 years), ranging from 22 to 88 years. International Union Against Cancer (UICC) staging was available for 142 patients. Of these patients, 21 (15%) were classified as UICC stage I, 33 (23%) UICC stage II, 29 (20%) UICC stage III, and 59 (42%) UICC stage IV. The lymph node status was 0 in 68 patients (49%), 1 in 35 patients (25%), and 2 in 35 patients (25%). Hepatic metastases were detectable in 50% of the patients at the time of analysis.

Autologous DCs loaded with respective peptides were used for stimulation of ex vivo isolated purified peripheral blood T cells and tested in 40-hour IFN-γ ELISPOT assays. T cells stimulated similarly with irrelevant human IgG served for determination of the unspecific background. Primary data from one representative patient are shown in Figure 1A. Antigen-specific responses, defined by significantly increased spot numbers in triplicate wells of test antigen relative to negative control antigen, are indicated by asterisks. To verify that our findings on TAA-reactive T cell responses were not due to a decrease in IFN-γ spots in control wells, we compared total IFN-γ spots in responding and nonresponding patients. Low background spot numbers were not correlated with a higher probability of achieving positive results. Instead, positive results were characterized by significantly increased reactivity against TAAs (Figure 1B). These TAA-specific T cell responses were confined to the T cell compartment of CD45RO⁺ Tmems/Teffs since we did not detect any TAA-specific reactivity in the population of CD45RO^– naive T cells (Figure 1C). Frequencies of TAA-reactive T cells were calculated by subtracting mean background values from mean test values in case of significant antigen-specific responses (Figure 1D). The highest frequencies of tumor-specific T cells in responding patients were seen against autologous tumor lysate and the polypeptides derived from mucin-1 and heparanase. Figure 1E shows the percentage of CRC patients and healthy individuals in whom T cell responses against the respective TAAs or normal self antigens were measured. Interestingly, the recognition rates of the individual TAAs varied strongly. Whereas we found frequent responses against p53 (21.3%), carcinoembryonic antigen (CEA; 17.2%), HER2/neu (17.6%), and heparanase pp1 (16.1%), we did not find any telomerase-specific T cells in any of the tested patients. Most frequently, we detected T cell responses against autologous tumor lysate (29.6%). The response rates in healthy donors were significantly lower than in the CRC patients, and we did not find self antigen–specific T cell responses in CRC patients or healthy donors. Recognition of TAAs was very individual and mostly oligovalent, with responding patients on average recognizing 1.7 TAAs (Figure 1F). Only rarely did the patients recognize more than 3 TAAs. To assess the strength of spontaneous antitumor T cell responses, we compared the frequencies of TAA-reactive T cells in CRC patients and healthy donors (Figure 1G). TAA-specific T cells were observed at significantly higher frequencies in CRC patients than in healthy donors, showing the selective presence of TAA-specific Tmem/Teff responses in tumor patients.

Figure 1

T cell responses against TAAs in CRC patients and healthy donors. (A) Primary data from an exemplary IFN-γ ELISPOT assay. Peripheral blood T cells of 120 patients and 32 healthy donors were stimulated with polypeptides derived from negative control antigen (human IgG), normal self antigens, or the respective test tumor antigens. Data represent mean spot number of 3 wells per antigen ± SEM. **P < 0.05, spot numbers in test wells compared with pooled spot numbers of negative control antigen (2-tailed Student’s t test). TC, T cells. (B) Cumulative total IFN-γ spots from Treg-undepleted T cells for control antigen (IgG) in responding and nonresponding patients and responding and nonresponding TAAs. Mean ± SEM; *P < 0.05 (2-tailed Student’s t test); IgG, n = 21–28; TAAs, n = 29–127. (C) Primary data of an IFN-γ ELISPOT assay with CD45RO⁺ Tmems or CD45RO^– naive T cells. Mean ± SEM; asterisks indicate significant differences between those groups that are connected by horizontal lines. *P < 0.05. (D) Frequency of TAA-specific T cells from responding CRC patients for single TAAs. Mean ± SEM; asterisks indicate significant differences between those groups that are connected by horizontal lines; *P < 0.05; **P < 0.01; n = 2–8. (E) Proportions of CRC patients and healthy donors among all individuals tested that exerted significant T cell reactivity against TAAs from autologous tumor or against polypeptides derived from the respective tumor antigens. Data represent the percentage of patients or healthy donors with TAA-specific T cells; n = 10–36. (F) Number of different TAAs recognized in individual responding patients; n = 25. (G) Frequency of TAA-specific T cells from individual CRC patients and healthy donors (HD). Mean ± SEM. ***P < 0.0001; n = 238–269.

TAA-derived synthetic long peptides are recognized by CD4⁺ and CD8⁺ T cells. The polypeptide sequences used in our analysis contained sequences that have been shown to be presented on HLA-A*0201 MHC class I alleles. To test whether these polypeptides can also be presented on MHC class II complexes, we assessed the ability of CD4⁺ T cells to recognize the same tumor antigens. For this, T cells were separated into a CD4⁺ T cell fraction and a CD8⁺ T cell fraction using magnetic beads. Both fractions were then tested separately for their ability to secrete IFN-γ upon incubation with TAA-pulsed DCs in an ELISPOT assay. Primary data from 1 representative experiment are shown in Figure 2A. Response rates against TAAs and the frequencies of TAA-specific T cells were higher in the CD8⁺ cytotoxic T cell compartment as compared with the CD4⁺ T helper cell compartment for most antigens tested (Figure 2, B–D). However, CD4⁺ T cell responses were found for all antigens, except for EGFR, demonstrating that the length of the polypeptides is sufficient for presentation on MHC class II alleles and subsequent recognition by CD4⁺ T cells.

Figure 2

CD4⁺ and CD8⁺ T cell responses against TAAs in CRC patients. (A) Primary data of an exemplary IFN-γ ELISPOT assay with separated CD4⁺ and CD8⁺ T cell populations. Peripheral blood T cells of CRC patients were separated into a CD4⁺ and CD8⁺ T cell fraction and subsequently stimulated with TAA-derived polypeptides. Data represent mean spot number of 3 wells per antigen ± SEM. *P < 0.05, spot numbers in test wells compared with pooled spot numbers of negative control antigen (2-tailed Student’s t test). (B) Proportions of CRC patients among all individuals tested that exerted significant CD4⁺ or CD8⁺ T cell reactivity against polypeptides derived from the respective tumor antigens. Data represent the percentage of patients with TAA-specific T cells; n = 5–18. (C) Frequency of TAA-specific CD4⁺ and CD8⁺ T cells from individual CRC patients. Mean ± SEM. **P < 0.01; n = 99–110. (D) Comparison of TAA-specific CD4⁺ and CD8⁺ T cell reactivity in CRC patients. Data represent the percentage of tests where significant T cell reactivity against a TAA was measured (Fisher’s exact test); n = 99–110.

Treg depletion results in stronger TAA-specific Tmem/Teff function. To examine the impact of Tregs on the TAA-specific Tmem/Teff responses, we first looked at the presence of Tregs in the peripheral blood of CRC patients. CD4⁺CD25⁺Foxp3⁺ Tregs accounted for approximately 5% of total CD3⁺CD4⁺ cells (Figure 3, A and B). We then developed a protocol for the depletion of CD4⁺CD25⁺ Tregs from the blood. For this, purified CD3⁺ cells were subjected to CD4 followed by CD25 magnetic bead isolation. Depletion of Tregs resulted in the loss of Foxp3⁺ cells as determined by flow cytometry (Figure 3, C and D). Next, we compared TAA-specific T cell responses in Treg-undepleted T cell fractions with T cell responses in Treg-depleted T cell fractions (Figure 4A). As seen for the Treg-undepleted fraction, there was no correlation between low background spot numbers and a higher probability of positive test results in Treg-depleted fractions (Figure 4B). We found a significant increase in IFN-γ spots only for TAAs for which reactive T cells were found in comparison with nonresponding TAAs. For most tested TAAs, we saw an increase in the response rates of CRC patients after depletion of Tregs (Figure 4, C and D). Such an increase was not seen in healthy donors. However, some TAAs in CRC patients were not influenced by a depletion of Tregs and only showed minor increases in response rates (p53, survivin, MAGE-3) or no increase (heparanase). As with the Treg-undepleted T cells, no T cell responses against the tested normal self antigens in CRC patients were measured in the absence of Tregs. Only in healthy donors, we rarely found T cell responses against proinsulin after Treg depletion. In addition to increased response rates in the patients, we measured stronger TAA-specific responses after depletion of Tregs as determined by an increase in the frequencies of TAA-reactive T cells (Figure 4E). Furthermore, depletion of Tregs led to an increased fraction of responding patients that displayed reactivity against more than 3 TAAs, as compared with the Treg-undepleted T cells (data not shown). Responding patients on average recognized 2.1 TAAs after Treg depletion. Our analyses suggest that not all TAAs are under the control of Tregs. As the depletion of Tregs did not result in any increases in TAA recognition of more than 10% in healthy donors, we considered any change below this threshold as background. We therefore classified the TAAs into a group that showed in CRC patients a more than 10% increase in T cell response rate after Treg depletion (Treg-dependent TAAs) and a group with Treg-independent Tmem/Teff responses (less than 10% increase in response rate after Treg depletion). According to this classification, the increase in IFN-γ ELISPOT response rate was primarily found in the Treg-dependent Tmem/Teff response group (Figure 4F).

Figure 3

Presence of Tregs in CRC patients. (A) Exemplary flow cytometry of CD25⁺FoxP3⁺ Tregs gated on CD3⁺CD4⁺ T cells from peripheral blood of 1 CRC patient. (B) Percentage of CD4⁺CD25⁺FoxP3⁺ Tregs of CD4⁺ T cells in the peripheral blood of individual CRC patients. Mean ± SEM. n = 7. (C) Foxp3 expression in CD3⁺ cells before and after depletion of CD4⁺CD25⁺ Tregs. (D) Enrichment of CD4⁺CD25⁺FoxP3⁺ Tregs after isolation of CD4⁺CD25⁺ cells.

Figure 4

Influence of Tregs on TAA-specific T cell responses in CRC patients and healthy donors. (A) Primary data of an exemplary IFN-γ ELISPOT assay (CRC patient 1056) with T cell fractions before and after depletion of CD4⁺CD25⁺ Tregs. Mean ± SEM; horizontal lines indicate samples containing (black bars) or lacking (white bars) Tregs; *P < 0.05 (2-tailed Student’s t test). (B) Cumulative total IFN-γ spots from Treg-depleted T cells for control antigen (IgG) in responding and nonresponding patients and responding and nonresponding TAAs. Mean ± SEM; ***P < 0.0001 (2-tailed Student’s t test); IgG, n = 21–28; TAAs, n = 50–106. (C) Data represent the percentage change in tested individuals that exerted significant T cell reactivity against the tested TAAs after depletion of Tregs relative to the response rates before Treg depletion. *P < 0.05 (Fisher’s exact test); n = 10–34. (D) Comparison of TAA reactivity in CRC patients before and after depletion of Tregs. Data represent the percentage of tests where significant T cell reactivity against a TAA was measured. *P < 0.05 (Fisher’s exact test); n = 246–254. (E) Frequencies of TAA-specific Tmems before and after Treg depletion in individual CRC patients and healthy donors. Mean ± SEM; **P < 0.01; ***P < 0.001 (2-tailed Student’s t test); n = 209–269. (F) TAAs were classified into Treg-dependent Tmem responses (>10% change in response rate) or Treg-independent Tmem responses (<10% change in response rate). Mean ± SEM; asterisks indicate significant difference between the groups indicated by horizontal line; **P < 0.01. Telo, telomerase. (G) Patients were tested for the presence of HLA-A*0201 and grouped into an HLA-A*0201–positive and an HLA-A*0201–negative cohort. Depicted are the proportions of CRC patients for both groups among all individuals of the respective group tested that exerted significant T cell reactivity against tested polypeptides. **P < 0.01; NS, P > 0.05 (Fisher’s exact test); n = 93–189.

Recognition of long synthetic TAA-derived polypeptides is independent of HLA type. Since long synthetic polypeptides contain a large number of sequences that could potentially be presented on MHC complexes, we investigated the dependency of polypeptide recognition in patients on the HLA type of these patients. All polypeptides used contained at least 1 sequence that has been described to be efficiently presented by APCs on HLA-A*0201–positive cells. We therefore performed HLA typing for all patients and classified them accordingly in an HLA-A*0201–positive and an HLA-A*0201–negative cohort. Comparison of these groups revealed no difference in the ability to recognize TAAs, suggesting that the length of the peptides used is sufficient to allow antigen presentation independent of the patients’ HLA type (Figure 4G). In line with this, we did not see differences in antigen recognition between HLA-A*0201–positive and –negative cohorts for any of the respective antigens (data not shown).

TAA-specific Tregs are found in CRC patients but not in healthy donors. To investigate whether the differential influence of Tregs on TAA-specific T cell responses is due to an unequal distribution of Treg specificities, we developed an assay to investigate the TAA recognition capabilities of Tregs. As displayed in Figure 5A, we pulsed DCs with polypeptides derived from the TAAs or with irrelevant human IgG as control and added Tregs to these cells in order to selectively activate Tregs specifically recognizing a TAA sequence presented by the DCs. The proliferation of polyclonally activated Teffs was then suppressed in comparison to test wells with irrelevant antigens. As an additional control, we also used unpulsed DCs since all antigens tested are self antigens, including the control antigen IgG, and might therefore lead to an activation of Tregs and subsequent suppression of T cell proliferation. Primary data of 2 representative patients are shown in Figure 5, B and C. Antigen-specific Treg suppression, defined by significantly decreased proliferation in triplicate wells of test antigen compared with negative control antigen, is indicated by asterisks. TAA-specific Tregs were most prevalently observed in CRC patients for CEA, telomerase, HER2/neu, and MUC-1, whereas Tregs specific for survivin were only rarely detected (Figure 5D). Furthermore, we did not detect any Tregs specific for p53 or either one of the 2 heparanase-derived polypeptides. No TAA-specific Tregs were detected in healthy donors. Interestingly, we frequently detected Tregs specific for the normal self antigen proinsulin in both CRC patients and healthy donors.

Figure 5

Specificity of Tregs for TAAs in CRC patients and healthy donors. (A) Scheme for determination of Treg specificity. DCs were pulsed with TAA polypeptide or control antigen. Autologous Tregs were added. Tregs specific for an antigen presented by the DCs were selectively activated. Polyclonally activated Teffs were added to the DCs and Tregs. Tregs activated by their respective antigen then suppressed the proliferation of the Teffs, which was measured by [³H]thymidine incorporation. (B and C) Primary data of 2 representative Treg specificity experiments displaying selective suppression of proliferation in wells where DCs were pulsed with HER2/neu and CEA (B) or telomerase (C). DC, no antigen added. Data represent mean counts per minute of 3 wells per antigen ± SEM; *P < 0.05, counts in test wells compared with pooled counts of negative control antigen (2-tailed Student’s t test). (D) CRC patients and healthy donors were tested for the presence of TAA-specific Tregs. Data represent the percentage of individuals with Tregs specific for the respective antigen; CRC patients, n = 14–56; HD, n = 10–14.

High individuality of Treg specificities and TAA-specific Tmem/Teff responses. To examine whether the recognition of TAAs by Tregs directly correlates with TAA-specific spontaneous Tmem/Teff responses, we compared Treg specificity analyses with IFN-γ ELISPOT analyses in 12 patients (Figure 6). A proportion of patients displayed TAA-specific Tregs but no Tmem/Teff responses against any of the tested antigens (patients 1296, 1317, 1333). Specific Treg and Teff responses after depletion of Tregs against the same antigens were detected in patients 1314, 1341, 1299, and 1295. Furthermore, some patients displayed Treg specificities or TAA-reactive Teffs without noticeable correlation. Thus, Treg specificities and TAA-specific spontaneous Tmem/Teff responses were highly individual in CRC patients.

Figure 6

Individuality of Treg specificities and T cell responses against TAA-derived polypeptides. Displayed is the pattern of reactivity in Treg specificity analysis and IFN-γ ELISPOT analysis in 12 patients. For Treg specificity assays, data represent mean inhibition of proliferation of 3 wells per antigen compared with control antigen for antigens that exerted significant inhibition of proliferation. For IFN-γ ELISPOT analysis, data show the mean number of TAA-specific T cells of 3 wells per antigen for those antigens that showed significant difference in spot numbers compared with control antigen. n.d., not determined.

Treg specificities correlate with Treg influence in IFN-γ ELISPOT. For TAA-reactive T cell responses, we classified the tested TAAs based on the differential influence of Tregs (Figure 4F). We then applied the same classification on the Treg specificities that we measured in the peripheral blood of CRC patients. As shown in Figure 7A, 96% of the Treg specificities measured were against TAAs in which a strong influence of Tregs was also seen in the IFN-γ ELISPOT experiments, but only 4% of the specificities were in the TAA group in which no or only minor Treg influence was present in these experiments. The unequal recognition of TAAs by Tregs was not due to a general bias of T cells to recognize only the TAAs in the Treg-dependent TAA group. The TAA-reactive T cell responses were similar in the Treg-dependent TAA group and the Treg-independent TAA group (53% and 47%, respectively) (Figure 7B). From some patients, we isolated sufficient numbers of T cells and Tregs to evaluate simultaneously TAA-specific Tmem/Teff responses and Treg responses. While this group was too small for generalized conclusions, the data obtained confirm on an individual level our observation that the presence of TAA-specific Tregs controls the reactivity of T cells against the same antigens and that such Treg-mediated T cell inhibition is predominantly directed against the group of “Treg-dependent” antigens (MUC-1, Her-2, telomerase, CEA, and EGFR) but not against the group of “Treg-independent antigens” (heparanase, p53, survivin, or MAGE) (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI39608DS1).

Figure 7

Correlation of Treg specificities with Tregs’ influence on TAA-specific Tmem responses. (A) Distribution of Treg specificities to TAAs in Treg-dependent Tmem responses group and Treg-independent Tmem responses group based on IFN-γ ELISPOT analysis. Data represent the percentage of tests with TAA-specific Tregs. **P < 0.01 (Fisher’s exact test); n = 80–152. (B) Distribution of TAA-specific Tmem responses before depletion of Tregs in Treg-dependent Tmem responses group and Treg-independent Tmem responses group. Data represent the percentage of responding patients; n = 23.

This study provides what we believe to be the first detailed examination of the influence of Tregs on TAA-specific immune responses in a large cohort of cancer patients. Using a protocol developed by our group, we found TAA-specific Tregs for a broad range of antigens that have been previously described as relevant TAAs in CRC. The identification of tumor-specific Tregs did not require cloning of Tregs and was readily done from peripheral blood samples. Interestingly, we frequently found Tregs specific for the same antigens for which we saw strong increases in Teff responses after depletion of Tregs.

Preexisting T cell responses in the peripheral blood of cancer patients were found against all tested TAAs, except for telomerase, whereas no T cell responses against the normal self antigens were measured (Figure 1E). T cell reactivity was confined to the CD45RO⁺ population of Tmems and Teffs (Figure 1C). Therefore, it is possible that our approach of using total T cell populations in the tests underestimated to some degree the presence of tumor-reactive T cells, which might have consequences for our estimation of the impact of Tregs on these responses. However, the procedure of separating Tmems/Teffs from naive T cells is associated with a major loss of approximately 50% of cells. The limited amount of cells that could be obtained from a patient’s blood sample, the further cell loss caused by the procedure of Treg depletion, and the need to test in parallel T cell responses against various TAAs forced us to use total T cell populations instead of Tmem purification. In healthy donors, TAA-specific T cell responses were only rarely observed. Furthermore, the frequency of TAA-specific T cells as determined by IFN-γ ELISPOT analysis was massively increased in cancer patients as compared with healthy individuals (Figure 1G), demonstrating that the observed TAA responses are limited to cancer patients. T cell responses against TAAs were very individual and oligovalent.

The use of long synthetic polypeptides has been previously described to be advantageous over short peptides with exact MHC binding properties (4). In line with this, TAA-derived polypeptides were recognized by both CD4⁺ and CD8⁺ T cells (Figure 2). Interestingly, most TAA were recognized more frequently by CD8⁺ than CD4⁺ T cells. This is consistent with previous studies demonstrating that an increase in the length of the peptide used for vaccination resulted in more robust and effective CD8⁺ T cell responses than vaccination with a minimal T cell epitope (2, 3). In these studies, an increase in CD8⁺ T cell responses was also observed in the absence of CD4⁺ T cell help, as shown by significantly higher CD8⁺ T cell responses in MHC class II–knockout mice and in CD40-knockout mice when vaccinated with a long-peptide vaccine in contrast to a minimal MHC class I epitope.

Furthermore, the recognition of TAA-derived polypeptides was independent of the patients’ HLA type, demonstrating that the length of the polypeptides was sufficient to allow for efficient presentation on various HLA alleles (Figure 4G). Testing of TAA-specific T cell reactivities is therefore not confined to certain HLA alleles present only in a subset of patients. Furthermore, this lack of dependence on restricted HLA types might circumvent the potential immunogenic escape of cancer cells in a vaccination setting through downregulation of individual HLA alleles or the mutation of T cell epitopes.

On average, 5% of CD4⁺ T cells were Foxp3⁺ Tregs in CRC patients (Figure 3, A and B). Depletion of Tregs led to increased recognition rates of TAA by TAA-specific Tmems (Figure 2, C and D). Interestingly, Treg depletion did not lead to increased T cell recognition rates for some TAAs, such as p53 and heparanase, suggesting that Treg suppression is not equal for all tested antigens. Our data revealed that Tregs suppress preexisting TAA-specific immune responses in some patients. In line with this, Treg depletion did not result in increased response rates in healthy donors. Of note, no T cell responses were observed against normal self antigens in the absence of Tregs in CRC patients. Only rarely did we detect T cell responses against proinsulin in healthy individuals in the absence of Tregs. Classification of TAAs, depending on the influence of Treg depletion on T cell responses against these antigens, revealed that the increase in patient response rates were due to a specific set of TAAs, including MUC-1, EGFR, CEA, and HER2/neu.

Examination of Treg specificities in CRC patients demonstrated that TAA-specific Tregs are frequently present in these patients (Figure 5D). This is in line with the observation that in the presence of Tregs, Teff responses were dampened in some patients. Preexisting Tregs with specificities for CEA, telomerase, HER2/neu, MUC-1, and EGFR were prevalent. However, we did not find specific Tregs for p53 or heparanase in any of the tested patients. The discriminative recognition of TAAs by Tregs was not due to the fact that the polypeptides that were not or were only poorly recognized did not contain any sequences that can be presented on MHC class II complexes (Figure 2) or to an overall increased recognition of the TAAs for which specific Tregs are found (Figure 7, A and B). The recognition rate of TAAs by memory T cells was similar for both the TAAs that were recognized by Tregs and for those that were not. One explanation for this finding may be that the different environments in which the proteins are expressed may account for an unequal capacity to induce Tregs specific for a particular antigen. Heparanase is a proteolytic enzyme that degrades polymeric heparin sulfate molecules in the extracellular matrix and is primarily abundant at inflammatory sites (30). The microenvironment of these sites may inhibit the induction of Tregs against this protein. The tumor suppressor p53 is involved in the regulation of the cell cycle. p53 is inactive in normal cells and is bound to the protein MDM2, which prevents its action and promotes its degradation by acting as a ubiquitin ligase. The low abundance of this protein in normal cells may also lead to a weakened induction of Tregs against p53. The impact of the antigen-specific Treg-derived suppressive function on antitumor immunity is further emphasized by reports from mouse studies that CD4⁺CD25⁺ Tregs inhibited CD4⁺CD25^– conventional T cells with the same antigen specificity when cotransferred at similar numbers (31). The presence of preexisting Tregs in patients is of eminent importance when TAA-derived peptides are used in a vaccination setting that may induce or reactivate preexisting Tregs and might therefore further hamper an effective immune response, as was observed in a study by van der Burgh and colleagues (32).

Strikingly, we commonly found Tregs specific for proinsulin, a precursor of insulin produced by β-islet cells in the pancreas. These Tregs were present in CRC patients and in healthy individuals alike. Several reports have shown the presence of insulin-specific Teffs (33, 34) and that depletion of Tregs can lead to destruction of β-islet cells and development of a diabetes phenotype in mice (35). Our data indicate that Tregs with specificity for β-islet–specific proteins, such as proinsulin, are commonly present in the peripheral blood and may be involved in the prevention of autoimmune-mediated destruction of β-islet cells and sequential development of type 1 diabetes.

Although we found Tregs specific for the same antigens for which we measured increased memory T cell response rates after Treg depletion, the individual TAA-specific T cell responses and Treg specificities did not always correlate within single patients (Figure 5). Several factors may be responsible for this observed high individuality, including the possibility that the time frame of in vitro culture in the absence of Tregs may not be long enough to unmask all T cell responses that have been suppressed by Tregs. In addition, since the TAA-derived polypeptides used in this study represent only part of the total protein sequence, it is possible that we detected Treg responses against some epitopes within these sequences, while preexisting memory T cell responses against epitopes that are not present in the sequence may have remained undetected, and vice versa. TAA-specific memory T cell response rates and TAA recognition by Tregs against a chosen antigen may therefore increase when overlapping polypeptides are used that cover the full protein sequence. Therefore, the grouping of the TAAs into Treg-dependent Tmem/Teff and Treg-independent Tmem/Teff responses is not inalterable and may differ depending on the exact antigen sequence tested.

It is likely that other Treg-mediated effector functions, besides the suppression of Teff proliferation, add to a confinement of antitumor immune responses. Several suppression modalities have been recently shown to be involved in Treg function (36–42). Investigation of these mechanisms using the protocol described here might bear valuable information on Treg activity in cancer patients. However, the strong correlation between the effect of Treg depletion on Teff function (antigen-specific IFN-γ production) and the specificities found for Tregs suggests that the measured suppression of Teff proliferation by Tregs, as determined in our analysis, is a reliable readout for Treg involvement.

In conclusion, CRC patients develop multivalent and individual T cell responses against a broad variety of different CRC-associated TAAs. Some, but not all of these T cell responses are kept under the control of TAA-specific Tregs. TAA-specific Tregs are predominantly present in the blood of patients but are not detectable in healthy individuals. Differences between Tregs and Teffs in the repertoires of recognized antigens can be detected in a majority of patients. Using selected sets of TAAs for tumor vaccinations that induce optimal Teff responses but minimal Treg activity may improve the efficacy of vaccination protocols.

View Supplemental data

We thank the Dietmar Hopp Stiftung for support.

Address correspondence to: Philipp Beckhove, Translational Tumour Immunology Unit, The German Cancer Research Center, Heidelberg, INF 280, 69120 Heidelberg, Germany. Phone: 49-6221-423745; Fax: 49-6221-423702; E-mail: p.beckhove@dkfz.de.

Conflict of interest: The authors have declared that no conflict of interest exists.

Reference information: J. Clin. Invest.119:3311–3321 (2009). doi:10.1172/JCI39608.

See the related articles at Identification of heme oxygenase-1–specific regulatory CD8⁺ T cells in cancer patients and Identification of heme oxygenase-1–specific regulatory CD8⁺ T cells in cancer patients.

1. Dunn, G.P., et al. 2002. Cancer immunoediting: from immunosurveillance to tumor escape. Nat. Immunol. 3:991-998.
   View this article via: CrossRef PubMed Google Scholar
2. Banchereau, J., Palucka, A.K. 2005. Dendritic cells as therapeutic vaccines against cancer. Nat. Rev. Immunol. 5:296-306.
   View this article via: CrossRef PubMed Google Scholar
3. Tacken, P.J., et al. 2007. Dendritic-cell immunotherapy: from ex vivo loading to in vivo targeting. Nat. Rev. Immunol. 7:790-802.
   View this article via: CrossRef PubMed Google Scholar
4. Melief, C.J., van der Burg, S.H. 2008. Immunotherapy of established (pre)malignant disease by synthetic long peptide vaccines. Nat. Rev. Cancer. 8:351-360.
   View this article via: CrossRef PubMed Google Scholar
5. Faure, F., et al. 2009. Long-lasting cross-presentation of tumor antigen in human DC. Eur. J. Immunol. 39:380-390.
   View this article via: CrossRef PubMed Google Scholar
6. Bennett, S.R., et al. 1998. B cells directly tolerize CD8(+) T cells. J. Exp. Med. 188:1977-1983.
   View this article via: CrossRef PubMed Google Scholar
7. Bijker, M.S., et al. 2008. Superior induction of anti-tumor CTL immunity by extended peptide vaccines involves prolonged, DC-focused antigen presentation. Eur. J. Immunol. 38:1033-1042.
   View this article via: CrossRef PubMed Google Scholar
8. Speetjens, F.M., et al. 2009. Induction of p53-specific immunity by a p53 synthetic long peptide vaccine in patients treated for metastatic colorectal cancer. Clin. Cancer Res. 15:1086-1095.
   View this article via: CrossRef PubMed Google Scholar
9. Sakaguchi, S. 2004. Naturally arising CD4+ regulatory t cells for immunologic self-tolerance and negative control of immune responses. Annu. Rev. Immunol. 22:531-562.
   View this article via: CrossRef PubMed Google Scholar
10. Fontenot, J.D., Gavin, M.A., Rudensky, A.Y. 2003. Foxp3 programs the development and function of CD4+CD25+ regulatory T cells. Nat. Immunol. 4:330-336.
    View this article via: CrossRef PubMed Google Scholar
11. Hori, S., Nomura, T., Sakaguchi, S. 2003. Control of regulatory T cell development by the transcription factor Foxp3. Science. 299:1057-1061.
    View this article via: CrossRef PubMed Google Scholar
12. Curiel, T.J., et al. 2004. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat. Med. 10:942-949.
    View this article via: CrossRef PubMed Google Scholar
13. Nummer, D., et al. 2007. Role of tumor endothelium in CD4+ CD25+ regulatory T cell infiltration of human pancreatic carcinoma. J. Natl. Cancer Inst. 99:1188-1199.
    View this article via: CrossRef PubMed Google Scholar
14. Hiraoka, N., et al. 2006. Prevalence of FOXP3+ regulatory T cells increases during the progression of pancreatic ductal adenocarcinoma and its premalignant lesions. Clin. Cancer Res. 12:5423-5434.
    View this article via: CrossRef PubMed Google Scholar
15. Sato, E., et al. 2005. Intraepithelial CD8+ tumor-infiltrating lymphocytes and a high CD8+/regulatory T cell ratio are associated with favorable prognosis in ovarian cancer. Proc. Natl. Acad. Sci. U. S. A. 102:18538-18543.
    View this article via: CrossRef PubMed Google Scholar
16. Gao, Q., et al. 2007. Intratumoral balance of regulatory and cytotoxic T cells is associated with prognosis of hepatocellular carcinoma after resection. J. Clin. Oncol. 25:2586-2593.
    View this article via: CrossRef PubMed Google Scholar
17. Kobayashi, N., et al. 2007. FOXP3+ regulatory T cells affect the development and progression of hepatocarcinogenesis. Clin. Cancer Res. 13:902-911.
    View this article via: CrossRef PubMed Google Scholar
18. Needham, D.J., Lee, J.X., Beilharz, M.W. 2006. Intra-tumoural regulatory T cells: a potential new target in cancer immunotherapy. Biochem. Biophys. Res. Commun. 343:684-691.
    View this article via: CrossRef PubMed Google Scholar
19. Kim, J.M., Rasmussen, J.P., Rudensky, A.Y. 2007. Regulatory T cells prevent catastrophic autoimmunity throughout the lifespan of mice. Nat. Immunol. 8:191-197.
    View this article via: CrossRef PubMed Google Scholar
20. Sakaguchi, S., et al. 1995. Immunologic self-tolerance maintained by activated T cells expressing IL-2 receptor alpha-chains (CD25). Breakdown of a single mechanism of self-tolerance causes various autoimmune diseases. J. Immunol. 155:1151-1164.
    View this article via: PubMed Google Scholar
21. Clarke, S.L., et al. 2006. CD4+CD25+FOXP3+ regulatory T cells suppress anti-tumor immune responses in patients with colorectal cancer. PLoS ONE. 1:e129.
    View this article via: PubMed Google Scholar
22. Samy, E.T., et al. 2005. Continuous control of autoimmune disease by antigen-dependent polyclonal CD4+CD25+ regulatory T cells in the regional lymph node. J. Exp. Med. 202:771-781.
    View this article via: CrossRef PubMed Google Scholar
23. Tang, Q., et al. 2006. Visualizing regulatory T cell control of autoimmune responses in nonobese diabetic mice. Nat. Immunol. 7:83-92.
    View this article via: CrossRef PubMed Google Scholar
24. Tarbell, K.V., et al. 2004. CD25+ CD4+ T cells, expanded with dendritic cells presenting a single autoantigenic peptide, suppress autoimmune diabetes. J. Exp. Med. 199:1467-1477.
    View this article via: CrossRef PubMed Google Scholar
25. Zou, W. 2006. Regulatory T cells, tumour immunity and immunotherapy. Nat. Rev. Immunol. 6:295-307.
    View this article via: CrossRef PubMed Google Scholar
26. Vence, L., et al. 2007. Circulating tumor antigen-specific regulatory T cells in patients with metastatic melanoma. Proc. Natl. Acad. Sci. U. S. A. 104:20884-20889.
    View this article via: CrossRef PubMed Google Scholar
27. Wang, H.Y., et al. 2004. Tumor-specific human CD4+ regulatory T cells and their ligands: implications for immunotherapy. Immunity. 20:107-118.
    View this article via: CrossRef PubMed Google Scholar
28. Wang, H.Y., et al. 2005. Recognition of a new ARTC1 peptide ligand uniquely expressed in tumor cells by antigen-specific CD4+ regulatory T cells. J. Immunol. 174:2661-2670.
    View this article via: PubMed Google Scholar
29. Rammensee, H., et al. 1999. SYFPEITHI: database for MHC ligands and peptide motifs. Immunogenetics. 50:213-219.
    View this article via: CrossRef PubMed Google Scholar
30. McKenzie, E.A. 2007. Heparanase: a target for drug discovery in cancer and inflammation. Br. J. Pharmacol. 151:1-14.
    View this article via: CrossRef PubMed Google Scholar
31. Klein, L., Khazaie, K., von Boehmer, H. 2003. In vivo dynamics of antigen-specific regulatory T cells not predicted from behavior in vitro. Proc. Natl. Acad. Sci. U. S. A. 100:8886-8891.
    View this article via: CrossRef PubMed Google Scholar
32. Welters, M.J., et al. 2008. Induction of tumor-specific CD4+ and CD8+ T-cell immunity in cervical cancer patients by a human papillomavirus type 16 E6 and E7 long peptides vaccine. Clin. Cancer Res. 14:178-187.
    View this article via: CrossRef PubMed Google Scholar
33. Baker, C., et al. 2008. Human CD8 responses to a complete epitope set from preproinsulin: implications for approaches to epitope discovery. J. Clin. Immunol. 28:350-360.
    View this article via: CrossRef PubMed Google Scholar
34. Toma, A., et al. 2005. Recognition of a subregion of human proinsulin by class I-restricted T cells in type 1 diabetic patients. Proc. Natl. Acad. Sci. U. S. A. 102:10581-10586.
    View this article via: CrossRef PubMed Google Scholar
35. Sgouroudis, E., Piccirillo, C.A. 2009. Control of type 1 diabetes by CD4(+)Foxp3(+) regulatory T cells: lessons from mouse models and implications for human disease. Diabetes Metab. Res. Rev. 25:208-218.
    View this article via: CrossRef PubMed Google Scholar
36. Cao, X., et al. 2007. Granzyme B and perforin are important for regulatory T cell-mediated suppression of tumor clearance. Immunity. 27:635-646.
    View this article via: CrossRef PubMed Google Scholar
37. Collison, L.W., et al. 2007. The inhibitory cytokine IL-35 contributes to regulatory T-cell function. Nature. 450:566-569.
    View this article via: CrossRef PubMed Google Scholar
38. Deaglio, S., et al. 2007. Adenosine generation catalyzed by CD39 and CD73 expressed on regulatory T cells mediates immune suppression. J. Exp. Med. 204:1257-1265.
    View this article via: CrossRef PubMed Google Scholar
39. Kobie, J.J., et al. 2006. T regulatory and primed uncommitted CD4 T cells express CD73, which suppresses effector CD4 T cells by converting 5′-adenosine monophosphate to adenosine. J. Immunol. 177:6780-6786.
    View this article via: PubMed Google Scholar
40. Li, M.O., Wan, Y.Y., Flavell, R.A. 2007. T cell-produced transforming growth factor-beta1 controls T cell tolerance and regulates Th1- and Th17-cell differentiation. Immunity. 26:579-591.
    View this article via: CrossRef PubMed Google Scholar
41. Rubtsov, Y.P., et al. 2008. Regulatory T cell-derived interleukin-10 limits inflammation at environmental interfaces. Immunity. 28:546-558.
    View this article via: CrossRef PubMed Google Scholar
42. Takahashi, T., et al. 2000. Immunologic self-tolerance maintained by CD25(+)CD4(+) regulatory T cells constitutively expressing cytotoxic T lymphocyte-associated antigen 4. J. Exp. Med. 192:303-310.
    View this article via: CrossRef PubMed Google Scholar
43. Feuerer, M., et al. 2001. Therapy of human tumors in NOD/SCID mice with patient-derived reactivated memory T cells from bone marrow. Nat. Med. 7:452-458.
    View this article via: CrossRef PubMed Google Scholar
44. Brossart, P., et al. 1999. Identification of HLA-A2-restricted T-cell epitopes derived from the MUC1 tumor antigen for broadly applicable vaccine therapies. Blood. 93:4309-4317.
    View this article via: PubMed Google Scholar
45. Novellino, L., Castelli, C., Parmiani, G. 2005. A listing of human tumor antigens recognized by T cells: March 2004 update. Cancer Immunol. Immunother. 54:187-207.
    View this article via: CrossRef PubMed Google Scholar
46. Nagorsen, D., et al. 2003. Differences in T-cell immunity toward tumor-associated antigens in colorectal cancer and breast cancer patients. Int. J. Cancer. 105:221-225.
    View this article via: CrossRef PubMed Google Scholar
47. Schirle, M., et al. 2000. Identification of tumor-associated MHC class I ligands by a novel T cell-independent approach. Eur. J. Immunol. 30:2216-2225.
    View this article via: CrossRef PubMed Google Scholar
48. Ries, J., et al. 2005. Investigation of the expression of melanoma antigen-encoding genes (MAGE-A1 to -A6) in oral squamous cell carcinomas to determine potential targets for gene-based cancer immunotherapy. Int. J. Oncol. 26:817-824.
    View this article via: PubMed Google Scholar
49. Fisk, B., et al. 1995. Identification of an immunodominant peptide of HER-2/neu protooncogene recognized by ovarian tumor-specific cytotoxic T lymphocyte lines. J. Exp. Med. 181:2109-2117.
    View this article via: CrossRef PubMed Google Scholar
50. Sommerfeldt, N., et al. 2006. The shaping of a polyvalent and highly individual T-cell repertoire in the bone marrow of breast cancer patients. Cancer Res. 66:8258-8265.
    View this article via: CrossRef PubMed Google Scholar
51. Andersen, M.H., et al. 2001. Spontaneous cytotoxic T-cell responses against survivin-derived MHC class I-restricted T-cell epitopes in situ as well as ex vivo in cancer patients. Cancer Res. 61:5964-5968.
    View this article via: PubMed Google Scholar
52. Vonderheide, R.H. 2002. Telomerase as a universal tumor-associated antigen for cancer immunotherapy. Oncogene. 21:674-679.
    View this article via: CrossRef PubMed Google Scholar
53. Shomura, H., et al. 2004. Identification of epidermal growth factor receptor-derived peptides immunogenic for HLA-A2(+) cancer patients. Br. J. Cancer. 90:1563-1571.
    View this article via: CrossRef PubMed Google Scholar
54. Antonacopoulou, A.G., et al. 2008. EGFR, HER-2 and COX-2 levels in colorectal cancer. Histopathology. 53:698-706.
    View this article via: CrossRef PubMed Google Scholar
55. Albers, A.E., et al. 2005. Immune responses to p53 in patients with cancer: enrichment in tetramer+ p53 peptide-specific T cells and regulatory T cells at tumor sites. Cancer Immunol. Immunother. 54:1072-1081.
    View this article via: CrossRef PubMed Google Scholar
56. Naomoto, Y., et al. 2005. The role of heparanase in gastrointestinal cancer. Oncol. Rep. 14:3-8.
    View this article via: PubMed Google Scholar
57. Beckhove, P., et al. 2004. Specifically activated memory T cell subsets from cancer patients recognize and reject xenotransplanted autologous tumors. J. Clin. Invest. 114:67-76.
    View this article via: JCI PubMed Google Scholar
58. Barnea, E., et al. 2002. Analysis of endogenous peptides bound by soluble MHC class I molecules: a novel approach for identifying tumor-specific antigens. Eur. J. Immunol. 32:213-222.
    View this article via: CrossRef PubMed Google Scholar
59. Hausmann, S., et al. 1999. Peptide recognition by two HLA-A2/Tax11-19-specific T cell clones in relationship to their MHC/peptide/TCR crystal structures. J. Immunol. 162:5389-5397.
    View this article via: PubMed Google Scholar