Recognition of a Ubiquitous Self Antigen by Prostate Cancer-Infiltrating CD8+ T Lymphocytes

See allHide authors and affiliations

Science  11 Jan 2008:
Vol. 319, Issue 5860, pp. 215-220
DOI: 10.1126/science.1148886


Substantial evidence exists that many tumors can be specifically recognized by CD8+ T lymphocytes. The definition of antigens targeted by these cells is paramount for the development of effective immunotherapeutic strategies for treating human cancers. In a screen for endogenous tumor-associated T cell responses in a primary mouse model of prostatic adenocarcinoma, we identified a naturally arising CD8+ T cell response that is reactive to a peptide derived from histone H4. Despite the ubiquitous nature of histones, T cell recognition of histone H4 peptide was specifically associated with the presence of prostate cancer in these mice. Thus, the repertoire of antigens recognized by tumor-infiltrating T cells is broader than previously thought and includes peptides derived from ubiquitous self antigens that are normally sequestered from immune detection.

Tlymphocytes that are reactive to antigens expressed by tumor cells have been shown to modulate cancer development in animal models and human cancer patients (1, 2). Thus, much effort has been devoted to the development of immunotherapeutic strategies aimed at inducing cancer regression by enhancing antitumor T cell responses (3). The identification of T cell tumor antigens typically relies on the in vitro culture of Tcell lines reactive to clonal tumor cell lines or to candidate peptide antigens (4). However, antigens defined in this manner may not provide a complete picture of the range of antigenic peptides in primary, heterogeneous tumors comprising a variety of cell types at different stages of differentiation. In this study, we set out to define a naturally arising T cell response to a tumor-associated antigen in an autochthonous mouse model of cancer.

Our approach was based on the direct detection of T cell populations that infiltrate primary tumors in vivo, thereby bypassing the need for in vitro T cell manipulation. To do this, we used a model in which male transgenic adenocarcinoma of mouse prostate (TRAMP) mice express simian virus 40 T antigen (Tag) under the control of a prostate-specific promoter, resulting in the development of spontaneous adenocarcinoma in the prostate by 14 to 20 weeks of age (5, 6). In a screen to identify endogenous T cell responses in the TRAMP model, we analyzed T cell receptor (TCR) repertoire diversity within the prostate infiltrate of late-stage tumor-bearing TRAMP+/+ mice (≥27 weeks of age, B6 genetic background) in order to identify reproducible T cell clonal expansions expressing conserved TCRs. We analyzed repertoire diversity using TCR CDR3 size spectratyping (7), in which the length distribution of the hypervariable complementarity-determining region 3 (CDR3) of a TCR chain is analyzed via a polymerase chain reaction (PCR)–based approach (8). Spectratyping analysis of the TCRβ chain variable region (Vβ) families 1 through 20 revealed the consistent overrepresentation of Vβ8.3 TCR transcripts with a conserved CDR3 length in the prostates of tumor-bearing TRAMP+/+ mice but not in the prostates of control B6 mice (Fig. 1A and fig. S1A). Sequencing of the prominent Vβ8.3 transcripts from several TRAMP+/+ mice identified conserved transcripts that preferentially encoded a CDR3 of 10 amino acids in length and a conserved amino acid sequence (fig. S2). Spectratyping analysis of prostate-infiltrating CD4+ and CD8+ T cells from tumor-bearing TRAMP+/+ mice revealed that T cells expressing the conserved Vβ8.3 TCRβ chain fall within the CD8+ T cell subset (Fig. 1B and fig. S1B), suggesting that these cells are major histocompatibility complex (MHC) class I–restricted. In an analogous manner, we used CDR3 size spectratyping and sequence analysis to identify restricted Vα2 and Vα18 TCRα chains that preferentially pair with the conserved TCRβ chain to form a functional TCRαβ heterodimer (fig. S3 and tables S2 and S3). Overall, the finding that CD8+ T cells infiltrating TRAMP+/+ prostate tumors express highly conserved TCRs suggested that a T cell response reactive to the same antigen was arising spontaneously in most tumor-bearing TRAMP+/+ mice.

Fig. 1.

CD8α+ T cells bearing Vβ8.3+ TCRs with a conserved CDR3 length are reproducibly overrepresented in the prostate of TRAMP+/+ mice. (A and B) TCR Vβ8.3 CDR3 size spectratyping analysis of TRAMP+/+ prostate. (A) cDNA from prostate tissue from 9 (21- to 36-week-old) B6 control mice, 10 (21-week-old) TRAMP+/+ mice, and 10 (27-week-old) TRAMP+/+ mice. (B) cDNA from CD4+ or CD8α+ T cells purified from the prostates of nine 27-week-old TRAMP+/+ mice by fluorescence-activated cell sorting. The “reference” spectra [bottom row in (A)] are derived from B6 female spleen samples. NP indicates no PCR product. (C) Flow cytometric analysis of thymocytes and splenocytes from B6 mice and Rag1–/– TCRαβ transgenic mice expressing the conserved Vα2+ and Vβ8.3+ TCR chains.

Having identified a conserved TCRαβ receptor expressed by T cells infiltrating TRAMP+/+ prostate tumors, we generated transgenic mice expressing the conserved Vα2+ TCRα and Vβ8.3+ TCRβ chains (Fig. 1C and fig. S4). When a TCRαβ transgenic line was crossed to the Rag1–/– (recombination activating gene 1) background, in which rearrangement of endogenous TCRs is precluded, mature peripheral T cells were selected to the CD8+ lineage (Fig. 1C). In addition, T cells in Rag1–/– TCRαβ transgenic mice exhibited no evidence of clonal deletion or co-receptor downregulation in the thymus or activation in the periphery (Fig. 1C and fig. S4), suggesting that there was no overt immunological self-reactivity in tumor-free TCRαβ transgenic mice.

To facilitate the characterization and identification of the antigen driving the observed T cell response, we generated an immortalized T cell hybridoma expressing the conserved TCRαβ (hereafter referred to as the “clonotypic hybridoma,” clone 6B1-6) from T cells from TCRαβ transgenic mice. In vitro, stimulation of the clonotypic hybridoma was not observed after direct culture of the hybridoma with a variety of tumor cell lines [including the TRAMP-C2 prostate cell line (9)] and primary immune cell types. Instead, stimulation was only observed when the clonotypic hybridoma was cultured with high-performance liquid chromatography (HPLC) fractions of crude cellular acid extracts together with MHC-expressing antigen-presenting cells (APCs). When crude extracts from the prostate of tumor-bearing TRAMP+/+ males were assayed in this manner for stimulation of the clonotypic hybridoma, a single peak of stimulatory activity was observed (Fig. 2A). Additionally, stimulatory activity was found in extracts from other organs (including liver, lung, spleen, and thymus) and bone marrow from TRAMP+/+ mice, as well as in organs from male and female B6 mice (Fig. 2A and fig. S5). T cell stimulation was only observed when APCs expressed the Kb class I MHC molecule, and was blocked by culture with antibodies to Kb (Fig. 2A), indicating that antigen recognition is Kb-restricted. Taken together, experiments that used crude extracts from mouse tissue indicate that the antigen recognized by clonotypic prostate-infiltrating T cells is derived from a self molecule that is present in many organs of both tumor-bearing and tumor-free mice.

Fig. 2.

T cells expressing the conserved TCRαβ recognize a histone H4–derived peptide. (A to C) Reactivity of the clonotypic T cell hybridoma 6B1-6. Extracts or proteins were boiled in 10% acetic acid and resolved by reversed-phase HPLC. Fractions were then dried and cultured with 6B1-6 and Kb-expressing APCs, and 6B1-6 stimulation was assayed. (A) Reactivity of 6B1-6 to crude extracts from prostate (PR), spleen (SP), and liver (LIV) from 27-week-old male TRAMP+/+ mice and liver from male (M) and female (F) B6 mice. Where indicated, a cocktail of antibodies to Kb (anti-Kb) or control antibodies (ctrl Ab) was added to the culture. (B) Reactivity of 6B1-6 to extracts of subcellular fractions. B16 melanoma cells were fractionated according to Wysocka et al. (16). The indicated subcellular fractions were isolated, acid-treated, processed, and assayed. (C) Reactivity of 6B1-6 to histone H4. A mixture of histone proteins or the indicated individual histones were acid-treated, processed, and assayed. (D) Stimulation of clonotypic T cells with histone H4–derived peptides. T cells expressing the conserved TCRαβ (either 6B1-6 or TCRαβ transgenic T cells) were cultured with peptide and APCs [either primary dendritic cells (DCs) or L-Kb cells]. The histone H4–derived peptides assayed and the control SIINFEKL peptide are indicated.

In addition to the crude extracts from primary mouse tissue, the stimulatory activity was present in crude extracts from all mouse cell lines assayed. Biochemical fractionation of cellular extracts from a tumor cell line revealed that the stimulatory activity is preferentially associated with the chromatin-enriched nuclear fraction (Fig. 2B). We therefore assayed the highly conserved histone proteins, an abundant constituent of chromatin, and found that preparations from a histone mixture and from purified histone H4 stimulated the clonotypic hybridoma (Fig. 2C). To identify the potential antigenic histone H4–derived peptides, we acid-treated, purified, and sequenced histone H4 using liquid chromatography–tandem mass spectrometry (LC-MS/MS) (7). A single 17-mer peptide (i.e., 17–amino acid peptide) of sequence VVYALKRQGRTLYGFGG (10) (identical to the C terminus of histone H4, residues 86 to 102) was identified (fig. S6). LC-MS/MS analysis of peptides present in the stimulatory fractions obtained from crude extracts (Fig. 2A) revealed the presence of the histone H4(86–102) 17-mer. Analysis of clonotypic T cell stimulation after culture with histone H4(86–102) or smaller truncated peptides in the presence of APCs confirmed that clonotypic T cells recognize histone H4(86–102) and revealed that the minimal core epitope for recognition is the heptamer H4(86–92): VVYALKR (Fig. 2D). The finding that clonotypic T cells recognize histone H4(86–92) presented by Kb is unexpected, given the preference of Kb for peptides that are eight amino acids in length and that contain a phenylalanine or tyrosine in the fifth residue (11). Consistent with this idea, histone H4 peptide binding to Kb, as measured by stabilization of Kb on the surface of RMA-S cells, was barely detectable (fig. S7), suggesting that H4 peptides bind Kb with low affinity. Having determined that clonotypic T cells are reactive to histone H4(86–92)–related peptides, we designated TCRαβ transgenic mice as “HRC” (histone H4–reactive TCR transgenic, complete expression).

In order to directly identify endogenous histone H4–reactive T cells in tumor-bearing TRAMP+/+ mice, we used fluorescent peptide/MHC tetramers in conjunction with flow cytometry. Because the low affinity of wild-type histone H4(86–92) VVYALKR binding to Kb precluded the generation of stable tetramers, we produced Kb tetramers bearing the analog peptide VVYAFKR, in which the leucine lying in the dominant Kb-binding anchor position is mutated to a phenylalanine. The VVYAFKR peptide exhibited increased affinity for Kb and was recognized by clonotypic T cells in vitro (fig. S8), and VVYAFKR/Kb tetramers (hereafter referred to as “H4/Kb tetramers”) stained HRC T cells efficiently (fig. S4). With the use of H4/Kb tetramers, histone H4–reactive T cells were not observed in the prostates of tumor-free B6 mice (n = 9 mice, Fig. 3A). In contrast, antigen-experienced CD44high CD8+ H4/Kb tetramer+ T cells were detected in the prostate infiltrate of most tumor-bearing TRAMP+/+ mice (Fig. 3A, with a median frequency of 0.15% of CD8+ T cells), with prominent H4/Kb tetramer+ populations in 3 out of 13 mice that were analyzed (≥0.2% of CD8+ T cells and ≥20 tetramer+ cells were recovered from the prostate). The frequency of histone H4–reactive cells in TRAMP+/+ prostate is comparable to that reported for T cells specific for tumor-associated antigens found commonly in the tumor infiltrate and metastatic lymph nodes of human melanoma patients (12). In addition to those identified in the prostate, antigen-experienced H4/Kb tetramer+ cells were also identified at detectable frequencies in the prostate-draining periaortic lymph nodes (7 out of 13 mice, with median frequency 0.029%) and the spleen (4 out of 13 mice, with median frequency 0.023%) of some TRAMP+/+ mice (Fig. 3A), including the 3 mice with prominent H4/Kb tetramer+ populations in the prostate.

Fig. 3.

T cell recognition of histone H4 in vivo. (A) Identification of endogenous histone H4–reactive T cells with peptide/MHC tetramers. T cells from ≥24-week-old TRAMP+/+ and B6 mice were stained with H4/Kb tetramer and antibodies to cell-surface markers. (Top) Representative samples from SP, periaortic lymph nodes (pLN), brachial lymph nodes (bLN), and PR. (Bottom) Summary of tetramer staining results, pooled from five experiments. To determine which TRAMP samples were “tetramer+,” we used data from B6 samples to establish a limit of detection (dashed horizontal bars), defined as the mean plus two SDs. For PR samples, solid and open circles denote samples in which ≥20 or ≤20 H4/Kb tetramer+ cells, respectively, were identified. (B) Division of HRC T cells transferred into TRAMP+/+ mice. CD45.1+ HRC T cells were labeled with CFSE, transferred into TRAMP+/+ and B6 mice, and analyzed 5 days (top) or 30 days (bottom) after transfer. Asterisks indicate statistically significant differences between TRAMP+/+ and B6 pLN values (P ≤ 0.0001 for day 5, P ≤ 0.006 for day 30). For 5 and 30 days after transfer, data are pooled from four and two experiments, respectively. Solid horizontal bars in (A) and (B) indicate median values.

To evaluate antigen recognition by histone H4–reactive T cells in late-stage prostate cancer, we analyzed the proliferation of HRC T cells after adoptive transfer into TRAMP+/+ mice. HRC T cells labeled with the dye carboxyfluorescein diacetate succinimidyl ester (CFSE) were transferred into tumor-bearing TRAMP+/+ mice and B6 control mice and were analyzed 5 or 30 days after transfer. At both time points, HRC T cells that had undergone division (i.e., with diluted CFSE) were observed in TRAMP+/+ mice but not in B6 controls (Fig. 3B). CFSE-diluted HRC Tcells were predominantly found in the tumor-draining periaortic lymph nodes (Fig. 3B), indicating that T cell antigen recognition in the lymphoid organs is mainly restricted to the lymph nodes draining the tumor site.

Together, tetramer analysis of endogenous T cells and adoptive T cell transfer studies indicate that, despite the ubiquitous nature of histone H4, MHC class I–restricted T cell recognition of histone H4 peptide is specifically associated with the presence of prostate cancer in TRAMP+/+ mice (13).

Next, we performed a series of experiments to analyze the ability of HRC T cells to traffic to the prostate and develop effector function after antigen recognition in tumor-bearing TRAMP+/+ mice. First, when analyzed 4 to 8 weeks after transfer, prostate infiltration of HRC T cells was observed in a proportion (10 out of 31 mice, ∼32%, with ≥20 HRC T cells recovered) of TRAMP+/+ recipients, with a median frequency of 0.27% of prostate-infiltrating CD8+ T cells (Fig. 4A). Second, when analyzed either 5 or 30 days after transfer, HRC T cells that had undergone division exhibited increased expression of the activation marker CD44 but generally produced little or no interferon-γ (IFN-γ) (Fig. 4B). Third, HRC T cells transferred into tumor-bearing TRAMP+/+ mice did not exhibit detectable cytolytic activity (Fig. 4C) and did not express the cytotoxic effector molecules granzyme B or perforin (fig. S11). Thus, after transfer into tumor-bearing TRAMP+/+ mice, HRC T cells proliferate and traffic to the prostate in a proportion of mice but typically fail to develop measurable effector function. These functional characteristics are analogous to those of tumor-reactive T cells found in tumor lesions of melanoma patients (12, 14, 15).

Fig. 4.

Trafficking and effector function of histone H4–reactive T cells. (A) Trafficking of HRC T cells to TRAMP+/+ prostate. CD45.1+ HRC T cells were transferred into ≥23-week-old TRAMP+/+ or B6 mice and analyzed 4 to 8 weeks after transfer. Solid and open circles denote samples in which ≥20 or ≤20 CD45.1+ cells, respectively, were identified. For three representative mice, flow plots are shown. (B) CD44 expression and IFN-γ production by HRC T cells. CFSE-labeled CD45.1+ HRC T cells that had been transferred 30 days earlier into 23-week-old TRAMP+/+ mice were re-stimulated in vitro and analyzed. The positive control indicates cells from B6 mice that had previously received in vitro–activated CD45.1+ HRC cells. (C) Lack of detectable cytolytic activity of HRC T cells. CD45.1+ HRC cells were injected into ≥24-week-old TRAMP+/+ or B6 mice. Fourteen days later, mixtures of target cells labeled with either VVYAFKR (H4) or control SIINFEKL [ovalbumin (OVA)] peptide were injected, and the recovery of targets was quantified. The positive control (OVA vax) indicates B6 mice that had been vaccinated against OVA peptide. (D) Slight reduction in PR mass and in genitourinary tract (GU) mass in TRAMP+/– HRV mice. Organ masses were determined for 26- to 28-week-old TRAMP+/– HRV (n = 106 mice) and TRAMP+/– (n = 119 mice) animals. Mice that died before the analysis point (n = 9 for TRAMP+/– HRV and n = 10 for TRAMP+/–) were not included in the analysis. Median masses were 902 mg (GU) and 95 mg (PR) for TRAMP+/– HRV mice and 1042 mg (GU) and 112 mg (PR) for TRAMP+/– mice. Horizontal bars indicate median values.

To gain insight into the contribution of histone H4–reactive T cells to antitumor immunity or to tolerance, we crossed TRAMP+/+ mice to a TCR transgenic line exhibiting variegated expression of the histone H4–reactive TCR (designated “HRV”), resulting in expression of the complete TCRαβ heterodimer on 10 to 20% of mature T cells (fig. S4). The resulting male offspring, either TRAMP+/– or TRAMP+/– HRV, were euthanized between 26 and 28 weeks of age, and the mass of the genitourinary tract and prostate was determined. Although there was no difference in survival to the 26- to 28-week end point (fig. S12), analysis revealed a slight, statistically significant reduction (∼10%) in prostate and genitourinary tract mass in TRAMP+/– HRV mice relative to that in TRAMP+/– controls. This finding suggests that, on average, a high precursor frequency of histone H4–reactive T cells results in decreased tumor burden in TRAMP mice. Despite this effect, the animals eventually die of prostate cancer, indicating that histone H4–reactive T cells are not by themselves protective in the absence of immune intervention. We are currently performing histological analyses of TRAMP prostate tumor pathology and immune infiltration in an effort to elucidate the mechanisms underlying the observed reduction in tumor burden in TRAMP+/– HRV mice.

On account of its abundant, ubiquitous nature, the histone H4 antigen described here does not fit into the major classes of tumor-associated antigens (4) and may represent a previously undefined type of antigen. Hypothetically, antigens of this type would be derived from ubiquitously expressed proteins that, under normal circumstances, are not efficiently processed in the MHC class I presentation pathway because of sequestration within the cell, either by compartmentalization within the nucleus or mitochondria or by segregation within supermolecular complexes. The CD8+ T cell repertoire would be largely ignorant of such antigens. However, as a result of distinct conditions within the tumor microenvironment, changes in antigen processing and presentation would induce MHC class I–restricted presentation of the antigen, allowing recognition by CD8+ T cells. In the case of histone H4, the stimuli leading to the induction of presentation may include necrosis, apoptosis, neovascularization, tissue remodeling, DNA damage, genomic instability, and imbalances in histone pools and nucleosome assembly.

The study of histone H4–reactive responses in TRAMP mice will serve as a valuable model for the tractable, longitudinal analysis of the T cell–mediated immune regulation of tumor development and will provide a preclinical model for evaluating the efficacy of immunotherapeutic strategies aimed at augmenting antitumor T cell responses.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S12

Tables S1 to S3


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article