Diversity and recognition efficiency of T cell responses to cancer

Abstract

Background: Melanoma patients vaccinated with tumor-associated antigens frequently develop measurable peptide-specific CD8+ T cell responses; however, such responses often do not confer clinical benefit. Understanding why vaccine-elicited responses are beneficial in some patients but not in others will be important to improve targeted cancer immunotherapies.

Methods and findings: We analyzed peptide-specific CD8+ T cell responses in detail, by generating and characterizing over 200 cytotoxic T lymphocyte clones derived from T cell responses to heteroclitic peptide vaccination, and compared these responses to endogenous anti-tumor T cell responses elicited naturally (a heteroclitic peptide is a modification of a native peptide sequence involving substitution of an amino acid at an anchor residue to enhance the immunogenicity of the peptide). We found that vaccine-elicited T cells are diverse in T cell receptor variable chain beta expression and exhibit a different recognition profile for heteroclitic versus native peptide. In particular, vaccine-elicited T cells respond to native peptide with predominantly low recognition efficiency--a measure of the sensitivity of a T cell to different cognate peptide concentrations for stimulation--and, as a result, are inefficient in tumor lysis. In contrast, endogenous tumor-associated-antigen-specific T cells show a predominantly high recognition efficiency for native peptide and efficiently lyse tumor targets.

Conclusions: These results suggest that factors that shape the peptide-specific T cell repertoire after vaccination may be different from those that affect the endogenous response. Furthermore, our findings suggest that current heteroclitic peptide vaccination protocols drive expansion of peptide-specific T cells with a diverse range of recognition efficiencies, a significant proportion of which are unable to respond to melanoma cells. Therefore, it is critical that the recognition efficiency of vaccine-elicited T cells be measured, with the goal of advancing those modalities that elicit T cells with the greatest potential of tumor reactivity.

Figure 4. High RE Recognition of Native G209n but Not G209–2M Peptide Correlates with Efficiency in Tumor Cell Lysis

CTL clones 476.105 and 132.1 were assayed for lysis of T2 cells pulsed with 10-fold dilutions of (A) native or (C) heteroclitc peptide at concentrations ranging from 100 fg/ml to 100 ng/ml. (B) Lysis of Malme-3M melanoma cells by 476.105 and 132.1 CTLs. All assays were performed in triplicate, and each clone was assayed twice. Error bars reflect variation between two separate assays.

Figure 5. Endogenous T-Cell Responses Have Higher RE Than Vaccine-Elicited Responses

CTL clones representing different tetramer-positive populations in each patient expressing different VB were assayed for lysis of T2 cells pulsed with various dilutions of G209n, G209–2M, M27, or M26 peptides in ⁵¹Chromium release cytotoxicity assays as described in Figure 4 legend. A RE score was attributed to each clone equal to the negative log₁₀ of the peptide concentration that resulted in 40% lysis of peptide-pulsed T2 cells. (A and B) RE scores for both (A) MART-specific and (B) gp100-specific clones from all patients were correlated with efficiency in lysing melanoma cells. Correlation coefficients were 0.66 for MART-specific clones and 0.81 for gp100-specific clones. (C–F) Comparison of RE scores for endogenous (patients 461 and 132) and vaccine-induced (patients 517, 520, 422 and 476) responses. (C and D) RE analysis with native peptides (C) M27 and (D) G209n. Mean RE (weighted) for each response is indicated with horizontal bars. Weighted means were based on all clones, not only those assayed, and were estimated by summing the RE of each analyzed clone multiplied by the number of total clones expressing the same VB, in each patient. Weighted means were as follows: patient 517, 5.7; patient 520, 7.0; patient 461, 7.9; patient 422, 9.7; patient 476, 9.9; and patient 132, 11.2. One-tailed T-tests demonstrated that endogenous responses had significantly higher RE than vaccine-induced responses: patient 461 versus patient 517, p = 1.8 × 10⁻⁵; patient 461 versus patient 520, p = 1.1 × 10⁻³; patient 132 versus patient 422, p = 6 × 10⁻⁶; and patient 132 versus patient 476, p = 4.3 × 10⁻⁴. (E and F) RE analysis with heteroclitic peptides (E) M26 and (F) G209–2M. Weighted means were as follows: patient 517, 10.6; patient 520, 11.1; patient 461, 11.2; patient 422, 10.5; patient 476, 11.6; and patient 132, 11.3.

Figure 1. Melanoma Patient Samples Selected for Analysis of RE for Melanoma Cells

(A) Six patients with T cell responses reactive with for M26 or G209–2M tetramers were selected for analysis. PBMCs from each patient were stained with PE-conjugated peptide–MHC tetramers, G209–2M-tet PE or M26-tet PE, and co-stained with anti-CD8 fluorescein isothiocyanate and anti-CD14, -CD19, and -CD4 Cy5PE. The plots shown are gated for CD8+, CD14−, CD19−, and CD4− cells. Tetramer-positive cells are boxed and estimated for percent of total CD8+ cells: patient 422, 2.5%; patient 476, 0.31%; patient 132, 0.22%; patient 517, 0.23%; patient 520, 0.12%, and patient 461, 0.50%. (B) Microcytotoxicity ⁵¹Chromium release assay with tetramer-positive cells isolated by FACS from the CD8+ PBMC population from patient 422. Isolated cells were assayed for lysis of T2 cells treated with relevant or irrelevant peptide, or mel526 melanoma cells. Sorted cells were combined with 250 target cells at 13:1 E:T ratios for 4 h, and supernatants were assayed for percent specific release of radiolabel.

Figure 2. Endogenous T Cell Responses Are More Efficient in Melanoma Lysis Than Vaccine-Elicited Responses

Cells from 87 clonal CTL lines were assayed for lysis of melanoma cells mel526, Malme-3M, and A375 in ⁵¹Chromium release cytotoxicity assays. Mel526 and Malme-3M are HLA-A2.1+ and express both gp100 and MART-1. A375 cells are HLA-A2.1+ but do not express either gp100 or MART-1 and served as a negative control. T2 cells treated with 1 μg/ml G209–2M or M26 peptides served as controls for antigen-specific lysis. The CTL clones assayed were selected to represent different tetramer-positive subsets expressing different VB. Dominating tetramer-positive populations in each patient were represented with two or more clones. Each CTL clone was assayed in triplicate wells, and the data displayed are averages of two different experiments. Clones from the same patient expressing similar VB while exhibiting different lysis potential were viewed as separate subsets. Each assay was performed at 10:1 E:T ratio as detailed in Methods. The height of each bar represents percent specific lysis, while the width represents the relative size of the tetramer-positive subpopulations (defined by VB expression) in each patient. Population size was defined as the percent of clones from each patient expressing the same VB. Error bars show standard deviation between two experiments within each clone and/or between different clones where more than one clone was analyzed.

Figure 3. Most CTL Clones Isolated from Endogenous Responses Are Efficient in Tumor Cell Lysis

CTL clones derived from each patient were classified as “efficient” (greater than 40%), “intermediate” (between 10% and 40%), or “low/no” (less than 10%) in lysis of melanoma cells based on data displayed in Figure 2. Each bar represents the portion of total clones from each patient with “efficient,” “intermediate,” or “low/no” melanoma lysis potential.