Definition of the HLA-A2 restricted peptides recognized by human CD8⁺ effector T cells by flow-assisted sorting of the CD8⁺ CD45RA⁺ CD28⁻ T cell subpopulation

INTRODUCTION

Human CD8⁺ T cells can be segregated into naive (CD45RA⁺) and activated/memory (CD45RO⁺) T cells [1]. Recent studies have shown that CD8⁺ CD45RA⁺ T cells can be divided into naive and effector T cells based on CD28 expression [2]. True naive T cells are CD28⁺ and are present at high numbers at birth. This population decreases with age, while the CD45R0⁺ T cells subset increases. The antigen-experienced CD45RO⁺ T cells can convert back to CD45RA⁺ T cells which may represent effector T cells associated with high cytolytic capability, high IFN-γ and TNF-α production [2]. Thus, this T cell subset may represent an attractive source to define the fine specificity in ex vivo-sorted antigen-experienced T cells without the need for in vitro manipulation. This information will provide useful insights into which target epitopes are recognized in vivo in a T cell subset associated with high T cell effector functions. In addition, these data may be useful as a screening assay to define T cell epitopes targeted by terminally differentiated effector T cells.

Alternatively, with the advent of soluble tetramer MHC class I peptide complexes, a combination to screen for antigen-specific immune responses and T cell marker analysis is feasible [3,4]. However, it is desirable to link the enumeration of antigen-specific T cells with functional analysis, i.e. cytokine secretion, which is not achieved easily in tetramer-binding T cells. In addition, flow-cytometry-sorted CD8⁺ CD45RA⁺ CD28⁻ T cells allow screening of a high number of unrestricted peptide epitope candidates of the MHC class I allele if autologous antigen-presenting cells (e.g. EBV-transformed B cell lines) are used as recipient target cells.

In order to define whether peptide-specific T cells reside in the CD8⁺ CD45RA⁺ CD28⁻ T cell population in patients with chronic viral (e.g. human papilloma virus (HPV)) or bacterial (e.g. Mycobacterium tuberculosis) infection, we evaluated the frequency of CD8⁺ CD45RO⁺, CD8⁺ CD45RA⁺ CD28⁺ and CD8⁺ CD45RA⁺ CD28⁻ T cells in PBL from patients with either carcinoma in situ neoplasia (CIN) or invasive carcinoma and from patients with active pulmonary tuberculosis.

MATERIALS AND METHODS

RESULTS

We tested the distribution of CD45RA⁺ CD8⁺ for T cells in PBL from patients with HPV⁺ lesions or from patients with pulmonary tuberculosis (Table 3). The majority of patients with HPV⁺ lesions showed a higher number of CD8⁺ CD45RA⁺ T cells (ranging from 7·6 to 17·2% in PBL) compared to CD8⁺ CD45RO⁺ T cells (ranging from 2 to 6·5% in PBL, see Table 3). This distribution is not restricted to patients with cervical cancer lesions; it could also be detected in PBL obtained from patients with a chronic (myco-) bacterial infection, ranging from 16% to 22% in PBL compared to a lower number of CD8⁺ CD45RO⁺ T cells (4–10% in PBL). In general, the distribution of antigen-experienced (CD28⁻) T cells within the CD8⁺ CD45RA⁺ T cell subset is below 10% in healthy individuals [2], but ranged from 12 to 40% in patients with cervical cancer. Remarkably, a single patient with tuberculosis exhibited up to 92% CD28⁻ T cells in the CD8⁺ 45RA⁺ T cell subset (Table 3).

In order to test if this T cell pool would represent a source to probe for antigen-specific T cells, we sorted T cells based on CD8⁺ CD45⁺ CD28− expression (Fig. 1a,b) and tested for HLA-A2-restricted peptide recognition. CD8⁺ CD45RA⁺ CD28⁺ T cell populations from the patients listed in Tables 1 and 3 were used to characterize the pattern of T cell reactivity to defined peptide antigens presented by HLA-A2. After flow-assisted sorting, T cells were rested overnight in AIM-V medium containing 10% FCS and 50 ng rIL-7. Each of the CD8⁺ CD45RA⁺ CD28⁺ or CD28⁻ T cell population was tested for recognition of peptide epitopes provided by the HPV16 E7 protein (Table 2) as defined by IL-2, GM-CSF, IFN-γ and TNF-α production by ELISA (Fig. 2a,b). T cells from three of six patients with cervical cancer responded to the peptide epitopes as defined by either TNF, IL-2 or GM-CSF production. No IFN-γ secretion could be observed. Three of six patients showed TNF-α production in response to the HPV16 E7₁₁₋₁₉ peptide epitope YMLDLQPET (Fig. 2a). This response could be observed exclusively in the CD28⁻ T cell population. Two of six patients exhibited also GM-CSF and IL-2 production to HPV16 E7 epitopes. GM-CSF was associated with TNF-α production. In contrast, CD8⁺ CD45RA⁺ CD28⁻ T cells from patient CC-I also reacted to the HPV16 E7₇₃₋₈₁ epitope by IL-2 and to the E7₈₂₋₉₀ epitope by IL-2 and GM-CSF production. CD28⁻ T cells from patient CC-II reacted to the E7₈₂₋₉₀ epitope exclusively with IL-2 and to the E7₈₅₋₉₃ epitope with GM-CSF, but not with IL-2 or TNF-α production. It is of note that we could not detect T cell reactivity to HPV16-E7 peptides presented by surrogate HLA-A2⁺ antigen (T2)-presenting cells either in HPV16⁺ but HLA-A2⁻ patients, or alternatively in HLA-A2⁺ but HPV16⁻ patients (data not shown).

A similar situation was found to be true for T cells obtained from patients with tuberculosis, i.e. antigen-specific and HLA-A2 restricted T cells could be observed in the CD8⁺ CD45RA⁺ CD28⁻ T cell subpopulation (Fig. 2b), particularly against the M. tuberculosis-associated antigen 85a/b. Thus, peptide-specific T cell responses can be visualized easily if the CD8⁺ CD45RA⁺ CD28⁻ T cell subpopulation is used as the effector T cell population in peptide screening assays.

DISCUSSION

In contrast to elevated numbers of CD8⁺ CD45RA⁺ CD28⁻ observed in patients with a (chronic) HPV infection in our report, the increase of CD8⁺ T cells in acute HIV or EBV infection is related to CD8⁺ CD45RO⁺ CD28⁺ and CD45RO⁺ CD28⁻ T cells [10]. This T cell subset is also present in low numbers in healthy individuals. In addition, the expression of CD27, a member of the TNF receptor family, and CD28 are closely linked in this T cell subset, resulting in a low number of circulating CD27⁺ CD28⁻ or CD27⁻ CD28⁺ T cells [10]. The same was true for the patients in our study: CD28⁻ T cells did not express CD27 (Table 3). Data from a recent report suggest that CD8⁺ T cells may be grouped into distinct T cell subsets based on CD27 and CD28 expression [11]: naive T cells are CD27⁺ CD28⁺, and T cell differentiation leads to sequential down-regulation of CCR7, CD28 and finally CD27 with a concomitant up-regulation of T cell effector functions [10,12].

The elevation of the effector CD45RA⁺ CD28⁻ T cell subset in PBL does not appear to be associated with HPV infection, but rather with the presence of an ‘inflammatory lesion’, as a single patient (CC-VI, Table 1) tested negative for HPV (diagnosed with endometritis), but also presented elevated numbers of CD28⁻ (38%) of CD8⁺ CD45RA⁺ T cells. Thus, elevation of CD8⁺ CD45RA⁺ CD28⁻ T cells may reflect a redistribution of T cells in the peripheral circulation associated with a local inflammatory lesion. This was confirmed in the patient population with a chronic bacterial infection: CD45RA⁺ CD28⁻ effector T cells ranged from 8 to 92% in patients with pulmonary tuberculosis (Table 3). The number of CD8⁺ CD45RA⁺ CD28⁺ T cells in PBL from HLA-A2⁺ individuals was too low to allow for flow sorting, followed by testing for peptide-specific T cell reactivity. However, flow cytometry data obtained from HLA-A2⁺ healthy subjects using M. tuberculosis Ag85b peptide-loaded HLA-A2 tetramer complexes showed that M. tuberculosis Ag85b-reactive T cells are absent in healthy individuals (our unpublished data).

It is important to note that the evaluation of PBL for T cell activation rests on the presumption that the peripheral T cell pool is in steady exchange with the entire T cell pool. However, PBL represent only 2% of the entire T cell pool at any time: T cells may home to inflammatory lesions, or they may be present in secondary lymphoid organs [13,14]. Thus, different homing characteristics or apoptosis may skew the T cell pool, which is accessible by venipuncture. Indeed, T cells specific for tumour-associated antigens have been detected exclusively in tumour lesions, but not in peripheral blood [15]. A similar situation may be true for anti-HPV or antimycobacterial antigens: specific T cells may reside in cervical or pulmonary lesions, respectively. Indeed, the ex-vivo longitudinal analysis of M. tuberculosis Ag85b-reactive T cells using tetramer complexes revealed that the composition of the CD8⁺ T cell pool may be stable over time (e.g. in patient IV, Fig. 3b), or alternatively may undergo substantial alterations defined by an increase of memory and terminally differentiated effector T cells concomitant with the decline in naive or activated CD8⁺ T cell subsets (e.g. PBL from patient V; Fig. 3b). Remarkably, despite the different dynamics of CD8⁺ T cell subsets in PBL from patients with tuberculosis, the majority of antigen (tetramer)-reactive T cells resided in the CD8⁺ CD45RA⁺ CD28⁻ T cell population, except for a single time-point in PBL from patient IV: most M. tuberculosis Ag85b-specific T cells resided in the naive CD8⁺ CD45RA⁺ CD28⁺ T cell pool (Fig. 3c). Recent studies supported the notion that chronic (viral) infections lead to an enrichment of antigen-specific T cells in individual CD8⁺ T cell subsets associated with the nature of the antigen [11]. For instance, CMV-specific T cells appear to accumulate in CD8⁺ CD45RA⁺ CD28⁻ CCR7⁻ effector T cells; HIV-reactive T cells are predominantly in the ‘intermediate’ CD8⁺ CD45RA⁺ CD28⁻, CD27⁺ T cell pool and EBV-reactive T cells reside in the ‘early’ activated CD45RO⁺ T cell subset [11]. Thus, detection and sorting of antigen-specific T cells may have to be complemented with an evaluation of the composition of the peripheral T cell pool, based on CD45RA, CD28 or CCR7⁻ expression. These data may provide the functional basis for earlier observations that certain CD8⁺ T cell subsets represent a good source for expanding antiviral T cell responses [16,17].

Caution should be exercised if peptides are screened for T cell recognition. Only the HPV16-E7₁₁₋₁₉ peptide could induce TNF-α production; other peptides were successful only in inducing IL-2 or GM-CSF upon antigen exposure. Thus, screening for T cell reactivity is dependent on the cytokine readout, particularly if T cell reactivity is screened in an ex-vivo analysis in patients with cancer which may exhibit dysfunctional T cell responses due to T cell signalling defects [18]. T cell reactivity directed to HPV16-E7 peptide epitopes has been described in patients with cancer, or in healthy volunteers after in vitro restimulation. Full-length HPV16-E7 protein-pulsed autologous dendritic cells have been shown to expand MHC-class I-restricted and antigen-specific CD8⁺ T cells in patients with cervical cancer [19] or in healthy individuals [20]. These T cells secreted large amounts of IFN-γ which was not detectable in ex-vivo-harvested CD8⁺ CD45RA⁺ CD28⁻ T cells. IFN-γ-secreting T cells may reside in alternate T cell subsets, e.g. CD56⁺ [20] or CD16/CD8αα ⁺ [21]. T cells. Alternatively, IFN-production may be impaired in patients with chronic viral or bacterial infections due to decreased TCR zeta chain expression and impaired T cell signalling [22,23].

However, ex-vivo expanded T cells may not necessarily reflect the in-vivo situation: T cells directed against a defined peptide epitope may undergo apoptosis after antigenic stimulation and may not be detectable in functional T cell assays. Is there a difference pertaining to the pattern of reactivity in T cells which were expanded in vitro with HPV16 E7 epitopes and the peptides recognized by effector T cells in the current report? Human CTL induced by in-vitro stimulation with peptides (HPV16-E7 aa 11–20, 82–90, 86–93) were capable of lysing an HLA-A2⁺ HPV16⁺ cervical cancer cell line and therefore most probably present peptides which are also naturally processed and presented on human HPV16⁺ tumour cells [24]. In vitro stimulation with peptides in two of eight (25%) healthy donors yielded MHC class I-restricted and antigen-specific CD8⁺ T cells; three of six (50%) of freshly isolated CD8⁺ CD45RA⁺ CD28⁻ T cells reacted strongly with TNF-α production against the HPV16 E7₁₁₋₁₉ epitope (Fig. 2a). In contrast, in four of eight healthy donors (50%) CD8⁺ T cells could be expanded directed against the HPV16-E7₈₅₋₉₃ epitope [24]. However, we could not detect CD8⁺ T cells reactivity in naive or in memory T cells against this epitope. Thus, a dichotomy may exist if PBL are expanded in vitro against potential peptide epitopes. Alternatively, the HLA-A2 presented HPV epitopes, e.g. the HPV16 E7₈₅₋₉₃ peptide, may not represent the optimal antigenic stimulus, as recent data suggested that the HPV16 E7₈₆₋₉₃ peptide is superior compared to the E7₈₅₋₉₃ epitope [25] implemented in this report as the target antigen.

In conclusion, the CD8⁺ CD45RA⁺ CD28⁻ T cell subset appears to be identical to terminally differentiated CD8⁺ CD45RA⁺ T cells which stain negative for expression of the chemokine receptor CCR7 [7,26] (see Fig. 4). A recent report showed that CMV-specific (50% of all antigen-specific T cells) but not HIV-specific T cells (approximately 5% of all antigen-specific T cells) reside in this T cell subset [11,26]. We describe in this report that flow-guided sorting of this T cell subset is easy to achieve and represents an alternative source to define HLA-A2-restricted peptide recognition by highly effective T cells in the peripheral circulation without the need for in-vitro expansion. Not only a single defined T cell epitope, but also a mixture of potential antigenic candidate peptides may be screened by isolating CD8⁺ CD45⁺ CD28⁻ T cells. In addition, flow-sorted CD8⁺ T cell subsets may supplement the immunological relevance of the ELISPOT assay, which detects T cells that produce a cytokine in response to antigenic stimulation; the number of spots allows determination of the frequency of cytokine-secreting cells responding to a defined antigen (reviewed in [27]). The combination of flow-sorted T cell subsets and ELISPOT analysis would enable enumeration of antigen-specific T cells in T cell populations which differ in their homing characteristics and life span: this would allow clinical definition of valuable surrogate markers to measure antigen-specific T cell responses.