Intratumoral IL-12 Gene Therapy Results in the Crosspriming of Tc1 Cells Reactive Against Tumor-associated Stromal Antigens

Abstract

HLA-A2 transgenic mice bearing established HLA-A2^neg B16 melanomas were effectively treated by intratumoral (i.t.) injection of syngeneic dendritic cells (DCs) transduced to express high levels of interleukin (IL)-12, resulting in CD8⁺ T cell-dependent antitumor protection. In this model, HLA-A2-restricted CD8⁺ T cells do not directly recognize tumor cells and therapeutic benefit was associated with the crosspriming of HLA-A2-restricted type-1 CD8⁺ T cells reactive against antigens expressed by stromal cells [i.e., pericytes and vascular endothelial cells (VEC)]. IL-12 gene therapy-induced CD8⁺ T cells directly recognized HLA-A2⁺ pericytes and VEC flow-sorted from B16 tumor lesions based on interferon (IFN)-γ secretion and translocation of the lytic granule-associated molecule CD107 to the T cell surface after coculture with these target cells. In contrast, these CD8⁺ T effector cells failed to recognize pericytes/VEC isolated from the kidneys of tumor-bearing HHD mice. The tumor-associated stromal antigen (TASA)-derived peptides studied are evolutionarily conserved and could be recognized by CD8⁺ T cells harvested from the blood of HLA-A2⁺ normal donors or melanoma patients after in vitro stimulation. These TASA and their derivative peptides may prove useful in vaccine formulations against solid cancers, as well as, in the immune monitoring of HLA-A2⁺ cancer patients receiving therapeutic interventions, such as IL-12 gene therapy.

Introduction

T cell-mediated antitumor immunity plays a role in regulating tumor growth, placing selective pressure on the antigenically heterogeneous cancer cell population throughout disease progression.^1,2,3 To date, most tumor-associated antigens (TAAs) recognized by T cells have proven to be nonmutated, “self” antigens that may be quantitatively overexpressed by tumor cells of one or more histologic types.⁴ Clinical trials implementing vaccines and immunotherapies targeting such antigens have exhibited success in promoting increased numbers of specific CD4⁺ and/or CD8⁺ T cell populations in the peripheral blood of patients, but they have only rarely demonstrated therapeutic efficacy in the advanced disease setting based on RECIST criteria.^5,6 Although transient objective clinical responses have been reported in some instances, responding patients may relapse with progressor tumors that fail to express elements of the major histocompatibility complex (MHC) antigen-presenting machinery and/or treatment-targeted antigens.^{2,3,4,5,6,7,8}

The modest success of current therapeutic vaccines targeting TAA suggests that alternate target antigens might instead be considered for integration into treatment designs in order to improve the efficacy of such approaches. In particular, a selection of antigens that are both crucial to tumor growth and survival, but which cannot be readily disposed of in the face of immune attack/selection (i.e., the oncogenes HPV-E6/E7 in cervical carcinoma,⁹ etc.) would appear most prudent. As an alternative to developing immune-based strategies against dominant oncogenes, serious consideration should be given to the targeting of antigens that are expressed not by tumor cells themselves, but rather by cells comprising the tumor-associated stroma; i.e. (myo)fibroblasts, vascular cells (including endothelial cells and their supportive mural cells, aka pericytes) and an array of infiltrating inflammatory cells.^10,11 Treatment-induced, immune-mediated disruption of the tumor “soil” would be expected to inhibit tumor growth and/or promote disease resolution.¹²

In this context, we investigated whether the crosspriming of CD8⁺ T cells reactive against tumor-associated stromal antigen (TASA) is a general paradigm for effective immunotherapy. We have previously shown that intratumoral (i.t.) delivery of syngenic dendritic cells (DCs) engineered to secrete interleukin (IL)-12p70 (i.e., DC.IL12) results in potent CD8⁺ T cell-mediated immunity against CMS4 sarcomas in Balb/c mice.¹³ Protection in this model was at least partially due to therapy-induced crosspriming of type-1 CD8⁺ T (i.e., Tc1) cells reactive against an H-2L^d-presented peptide derived from the β-hemoglobin (HBB) protein expressed by pericytes and/or vascular endothelial cells (VEC) within the sarcoma microenvironment.¹³ In the current report, we now show that delivery of DC.IL12 into B16 (HLA-A2^neg) melanomas established in HLA-A2 Tg (HHD) mice results in the induction of protective HLA-A2-restricted CD8⁺ T cells recognizing tumor-associated HLA-A2⁺ pericytes and VEC, as well as, an array of HLA-A2-presented TASA-derived peptide epitopes. Murine TASA-derived peptide epitopes share sequence identity with their human homologues, and human HLA-A2⁺ normal donors and melanoma patients displayed anti-TASA CD8⁺ T cell responses after in vitro sensitization (IVS). These data support the therapeutic targeting of TASA (via i.t. cytokine gene therapy or specific vaccination) as a potential means to treat vascularized solid tumors (including melanomas) that may be refractory to TAA-based therapeutics based on MHC/TAA expression heterogeneity and the progressive selection of immune escape variants.

Results

Analysis of TASA expression in the TME

We have previously reported that CD8⁺ T cells responses against peptides derived from the murine HBB or EphA2 proteins inhibit the establishment and progression of HBB^neg or EphA2^neg tumor cells, respectively, in syngenic wild-type hosts in vivo.^13,14 The “antitumor” efficacy of these Tc1 cells appeared to be due to their targeting of HBB⁺ pericytes and/or EphA2⁺ VEC within the tumor microenvironment (TME).^13,14 Based on these data, as well as, recent reports by other groups,^15,16,17,18 we hypothesized that TASA might serve as “universal” targets allowing for CD8⁺ T cell-mediated restricted growth of solid vascularized tumors. Among the known TASA expressed by pericytes and/or activated VEC,^{13,14,15,16,17,18,19,20,21,22,23,24} we selected an initial panel of 12 antigens for evaluation in the current studies (Table 1).

Table 1. TASA-derived peptides evaluated in this study: Summary of in vitro results.

To validate that the chosen TASA were indeed expressed in situ by stromal cells in the TME, we performed immunohistochemistry analyses using specific pAbs on tissue sections isolated from day 14 (HLA-A2^neg) B16 melanomas growing progressively in untreated HLA-A2 Tg (HHD) mice. Using immunofluorescence microscopy, we determined coexpression patterns of specific stromal target antigens with NG2⁺ pericytes and/or CD31⁺ VEC within the TME. The resulting images are depicted in Figure 1a, with a summary of cellular protein expression profiles provided in Table 2. Based on these imaging analyses, we assigned the DLK1, HBB, NG2, PDGFRβ, RGS5, and VEGFR2 antigens as predominantly tumor pericyte-associated, and the EphA2 and TEM1 antigens as predominantly tumor VEC-associated. The NRP1, NRP2, PSMA, and VEGFR1 antigens appeared to be expressed by multiple cell types including pericytes, VEC, and alternate stromal cells and/or tumor cells within the progressive B16 TME. To further corroborate TASA expression by NG2⁺ pericytes, CD31⁺ VEC, or H-2K^b+ tumor cells within the TME, these cell populations were flow-sorted from enzymatically digested B16 tumors resected from untreated recipient HHD mice. To gauge potential overexpression of TASA in tumor versus normal tissues, pericytes and VEC were also flow-sorted from single-cell digests of tumor-uninvolved kidneys harvested from these same animals. Reverse transcriptase (RT)-PCR analyses were then performed on complementary DNA isolated from each of these sorted cell populations. Quality control analyses supported the expression of NG2 transcripts only in pericytes, CD31 transcripts only in VEC and gp100 transcripts only in B16 cells (Figure 1b). These analyses also support: (i) tumor pericyte expression of all TASA transcripts with the exceptions of EphA2 and PSMA; (ii) tumor VEC expression of transcripts for DLK1, EphA2, HBB, PSMA, TEM1, VEGFR1, and VEGFR2; (iii) B16 expression of transcripts for NRP1, PDGFRβ, VEGFR1, and VEGFR2; (iv) higher levels of DLK1, EphA2, HBB, NRP1, NRP2, PDGFRβ, RGS5, TEM1, VEGFR1, and VEGFR2 transcript expression in tumor- versus normal kidney-derived stromal cells; and (v) comparable or greater levels of NG2, PSMA, and CD31 transcript expression in normal kidney- versus tumor-derived stromal cells (Figure 1b).

Figure 1.

Expression of tumor-associated stromal antigen (TASA) in the established B16 TME. (a) B16 melanoma cells were injected subcutaneously (s.c.) in the right flank of female HHD mice and allowed to establish/progress for 14 days. Animals were then euthanized, with tumors resected, fixed, sectioned, and analyzed for expression of the indicated antigens using specific Abs and fluorescence microscopy as outlined in Materials and Methods section. Specific pAb against NG2 (green), the indicated antigen of interest (red), and CD31 (blue) were used to distinguish preferential antigen expression in tumor-associated stromal pericytes, vascular endothelial cells (VEC), alternate stromal cells and/or tumor cells. (b) B16 melanoma cells, as well as, flow-sorted (PDGFRβ⁺, CD31^neg) pericytes and (PDGFRβ^neg, CD31⁺) VEC isolated from day 19 established B16 tumors and tumor-uninvolved kidneys were analyzed for expression of target gene product mRNAs using reverese transcriptase-PCR (RT-PCR) as described in Materials and Methods section. All data are reflective of three independent experiments performed for each tumor type.

Table 2. Cells expressing TASA in the B16 TME.

Selection of TASA peptides for immunologic analyses

Of the selected TASA, HLA-A2-presented epitopes recognized by CD8⁺ T cells have been previously reported for human EphA2, NG2, PSMA, RGS5, VEGFR1, and VEGFR2 (Table 1). Notably, these defined human epitopes share 100% sequence identity with their murine homologues. To identify novel HLA-A2-presented epitopes in the alternate six selected TASA, a prediction algorithm (http://www.bimas.cit.nih.gov/molbio/hla_bind/) was applied to each protein, and nonameric (9-mer) and/or decameric (10-mer) peptides were preferentially chosen for synthesis and corollary analyses based on two priority criteria: (i) a high algorithm predicted binding score to the HLA-A2.1 class I molecule, and (ii) identity in the human versus murine peptide sequences. This latter restriction was adopted for translational purposes; i.e., to insure that specific therapy-induced T cell responses would need to break operational tolerance in HLA-A2 Tg (HHD) recipient mice in order to provide antitumor protection (i.e., as would also need to occur for protection in HLA-A2⁺ patients with solid cancers). After selection, we showed that each of the chosen synthetic peptides was competent (to a varying degree) to bind and stabilize HLA-A2 complexes expressed by T2 cells (Supplementary Figure S1), a prerequisite to their ability to be presented to specific, HLA-A2-restricted CD8⁺ T cells.

Delivery of DC.IL12 into HLA-A2neg B16 tumors promotes the crosspriming of CD8⁺ T cells reactive against tumor pericytes, VEC and an array of TASA-derived peptide epitopes in HHD mice

In order to assess the potential in vivo relevance of therapy-induced CD8⁺ T cell responses against these TASA in the tumor setting, we strategically built upon our earlier work which resulted in the identification of HBB as tumor pericyte-associated antigen.¹³ In that study, i.t. delivery of syngenic DC adenovirally engineered to produce mIL-12p70 (i.e., DC.IL12), but not control DC, was competent to break operational tolerance against HBB and to yield protective immunity.¹³ As a consequence, in the current studies, DC.IL12 were prepared and injected directly into subcutaneous (HLA-A2^neg) B16 melanomas growing progressively in HLA-A2 Tg HHD mice on days 7 and 14 post-tumor inoculation. On day 19 post-tumor inoculation, the mice were euthanized and CD8⁺ splenic T cells were analyzed for their ability to secrete interferon (IFN)-γ in response to stimulation with TASA-derived peptides presented by the HLA-A2⁺ T2 cell line.

I.t. delivery of DC.IL12 resulted in dramatically reduced tumor growth (Figure 2a; P < 0.05 versus versus DC.ψ5-treated or untreated controls after day 11). Furthermore, splenic CD8⁺ T cells isolated from the DC.IL12 (but not DC.ψ5)-treated cohort of animals directly recognized HLA-A2⁺ pericytes and VEC flow sorted from single-cell digests of B16 tumors (but not kidneys isolated from these same tumor-bearing animals) or HLA-A2^neg B16 tumor cells (Figure 2b,c and Supplementary Figure S2). Tc1 recognition of tumor-derived pericytes and VEC was completely blocked in the presence of the anti-HLA-A2 mAb BB7.2 (but not an anti-MHC II mAb L243), supporting the HLA-A2-restricted nature of T cell reactivity. Splenic CD8⁺ T cells from DC.IL12- (but not control DC-) treated animals also responded against an array of TASA-derived peptides when presented by HLA-A2⁺ T2 cells in vitro (Figure 2c). The ability of these murine (HHD) CD8⁺ T cells to recognize TASA-derived peptides in the context of the human T2 cell line suggests these Tc1 effector cells exhibit moderate-to-high avidity for specific epitopes, since the murine CD8 coreceptor interacts inefficiently with the human HLA class I α3 domain²⁵ expressed by T2 cells.

Figure 2.

Induction of CD8⁺ T cells reactive against tumor-associated stromal antigen (TASA) after intratumoral delivery of DC.IL12. (a) HLA-A2^neg B16 melanoma cells were injected subcutaneously (s.c.) in the right flank of female HLA-A2 Tg (HHD) mice and allowed to establish for 7 days. On day 7, mice were randomized into three groups (n = 5 mice each) receiving no treatment, i.t. injection of syngenic dendritic cell (DC) that were previously infected with recombinant adenovirus encoding mIL-12p70, or DC infected with control (empty) adenovirus (i.e., DC.ψ5). Animals were retreated using the same therapy on day 14 post-tumor inoculation. In replicate cohorts of animals receiving DC.IL12 therapy, depleting mAbs against CD4 or CD8 were provided beginning on day 6 post-tumor inoculation as described in Materials and Methods section. Tumor sizes were assessed every 3–4 days and are reported as mean ± SD in mm². *P < 0.05 versus control or DC.ψ5-treated mice on days ≥14. (b) On day 19 post-tumor inoculation, the mice were euthanized and CD8⁺ splenocytes isolated by magnetic bead cell sorting (MACS) and cultured with PDGFRβ⁺CD31^negH-2K^b(neg) pericytes or PDGFRβ^negCD31⁺H-2K^b(neg) vascular endothelial cells (VEC) sorted by flow cytometry as described in Materials and Methods section. After coculture in the absence or presence of anti-HLA-A2 mAb BB7.2 or anti-major histocompatibility complex (MHC) class II mAb L243 (10 µg/well) for 48 hours at 37 °C, cell-free supernatants were analyzed for mIFN-γ content by specific enzyme-linked immunosorbent assay (ELISA). Data are mean ± SD for triplicate determinations, and are representative of two independent experiments performed. *P < 0.05 versus kidney cells (pericytes or VEC) and tumor pericytes/VEC in the presence of anti-HLA-A2 mAb BB7.2. (c) On day 19 post-tumor inoculation, the mice were euthanized and splenocytes and stimulated for 5 days with stromal peptides as outlined in the Materials and Methods section. On day 5, MACS-isolated CD8⁺ splenocytes were cocultured with HLA-A2⁺ T2 cells loaded with the indicated TASA-derived peptides or HLA-A2^neg B16 tumor cells. After a 48-hour culture period, cell-free supernatants were analyzed for mIFN-γ concentration by specific ELISA. Data are mean ± SD for triplicate ELISA determinations. *P < 0.05 versus FluM1 control peptide responses. All presented data are representative of three independent experiments performed.

We next analyzed the impact of therapy on the ability of CD8⁺ tumor-infiltrating lymphocytes (TIL) freshly isolated from day 17 tumors to recognize flow-sorted pericytes and VEC, as well as, TASA peptides presented by T2 cells. Using both intracellular IFN-γ staining (Figure 3a) and CD107 translocation (Figure 3b; i.e., a measure of effector T cell degranulation associated with perforin/granzyme-dependent lysis;²⁶) assays, we observed that 3–12% of CD8⁺ TIL isolated from animals treated with DC.IL12 mediated effector Tc1 responses against tumor (but not kidney)-derived pericytes and VEC. Similar frequencies of CD8⁺ TIL from the DC.IL12-treated cohort of mice recognized TASA peptides presented by T2 cells (Figure 3a,b). The ability of target cells to elicit effector responses from CD8⁺ TIL isolated from DC.IL12-treated mice was blocked by anti-HLA-A2 (but not anticlass II) mAb and these T cells display only background reactivity against HLA-A2^neg B16 tumor cells (Supplementary Figure S3). In contrast, the frequency of TASA-specific CD8⁺ TIL isolated from untreated or DC.ψ5-treated melanoma was lower (versus DC.IL12 treatment) in all functional analyses performed (Figures 2c, 3a,b, and Supplementary Figure S3).

Figure 3.

CD8⁺ tumor-infiltrating lymphocytes (TIL) from DC.IL12-treated mice are enriched in effector cells reactive against tumor pericytes and/or vascular endothelial cells (VEC), as well as tumor-associated stromal antigen (TASA) peptides. B16 tumor-bearing mice were treated as described in Figure 2. On day 17 post-tumor inoculation, CD8⁺ TIL were isolated from all cohorts of mice, and pericytes and VEC were isolated from the tumors and kidneys of untreated mice as described in the Materials and Methods section. Freshly sorted CD8⁺ TIL were then cocultured with pericytes, VEC, or T2 cells ± TASA peptides (1 µmol/l each of all peptides in Table 1 with the exception of NRP2- or PSMA-derived peptides) for 4–5 hours, before responder CD8⁺ T cells were analyzed for intracellular expression of (a) IFN-γ or cell-surface expression of (b) CD107a/b by flow cytometry. Inset numbers reflect the percentage of CD8⁺ T cells expressing intracellular interferon (IFN)-γ or cell surface CD107a/b. Data are from one representative experiment of two performed.

CD8⁺ T cells from HLA-A2⁺ normal donors or HLA-A2⁺ melanoma patients recognize TASA-derived peptides in vitro

To assess whether the TASA-derived peptides identified in our HHD tumor model were also capable of being recognized by human CD8⁺ T cells, we performed IVS using T cells isolated from the peripheral blood of HLA-A2⁺ donors or HLA-A2⁺ patients with melanoma. DC were pulsed with peptides derived from a given TASA for 4 hours at 37 °C, then washed and used as stimulator cells for autologous CD8⁺ T cells. In cases where more than one peptide existed for a given protein, we pulsed DC with an equimolar (10 µmol/l) mixture of each peptide. Based on our past experience using IVS protocols to elicit specific T cell responses against TAA (where the precursor frequency was far lower in normal donors versus cancer patients; ref. 27, we applied two rounds of IVS using TASA for normal donors and a single-round of IVS using TASA for melanoma patients. Using this approach, we observed that HLA-A2⁺ normal donors (Figure 4; Tables 1 and 3) and melanoma patients (Figure 4; Tables 1 and 3) were each capable of recognizing many of the TASA-derived peptides.

Figure 4.

In vitro immunogenicity of tumor-associated stromal antigen (TASA)-derived peptides in HLA-A2⁺ normal donors and patients with melanoma. The indicated peptides were pulsed onto autologous dendritic cells (DC) and used to prime and boost CD8⁺ T cells isolated from the peripheral blood of eight normal HLA-A2⁺ donors or ten HLA-A2⁺ patients with melanoma as described in the Materials and Methods section. Seven days after the primary in vitro sensitization (IVS) (melanoma patients) or a secondary IVS boost (normal donors), T cells were analyzed for their reactivity against HLA-A2⁺ T2 cells pulsed with the relevant peptide versus the negative control HIV-nef_190–198 peptide. After 24 hours of coculture, cell-free supernatants were analyzed for levels of secreted IFN-γ using a commercial enzyme-linked immunosorbent assay (ELISA). Data are reported in Bar and Whisker plots, with P values provided for paired pre- versus post-IVS data from normal donors and patients. In addition, we noted P < 0.05 for MEL-post versus ND-post for the following peptides: DLK1 (309), NG2 (770), NG2 (2238), PDGFRβ (891), and RGS5 (5).

Table 3. Normal donor and melanoma patient demographics and responsiveness to TASA.

Discussion

The major finding of the current report is that protective CD8⁺ T cells induced as a consequence of effective i.t. DC.IL12 therapy recognize both tumor-associated stromal cells (i.e., flow-sorted pericytes and VEC) and naturally processed and HLA-A2-presented peptides derived from TASA. CD8⁺ T cell recognition of pericytes and VEC was tissue (tumor) specific, since therapy-induced CD8⁺ T cells did not recognize these same cell populations sorted from tumor-uninvolved “normal” kidneys. Based on our RT-PCR analyses, such differential Tc1 recognition of tumor stromal cells may be directly related to the higher levels of TASA transcripts (and possibly protein) expressed by pericytes/VEC isolated from the TME versus the kidney. As expected, in our HHD recipient mouse model system, protective HLA-A2-restricted Tc1 cells failed to recognize HLA-A2^neg B16 tumor cells, even though CD8⁺ T cells appeared to be crossprimed against HLA-A2-presented B16 melanoma-associated antigens such as MART-1 and gp100 (Figure 5), presumably via crosspresentation mediated by HLA-A2⁺ APCs emigrating from the TME.¹¹

Figure 5.

Splenic CD8⁺ T cells from HHD mice effectively treated with DC.IL12 gene therapy develop HLA-A2-restricted responses against melanoma-associated antigens. HHD mice bearing day 7 HLA-A2^neg (MART-1⁺, gp100⁺) B16 melanomas were left untreated or they were treated with intratumoral (i.t.) injection of control dendritic cell (DC) (DC.ψ5) or DC.IL12 as described in Figure 2. On day 19 post-tumor inoculation (i.e., 5 days after receiving the second injection of DC), CD8⁺ spleen cells were isolated and analyzed for reactivity against the hMART-1_26–35 and h/mgp100_209–217 peptide epitopes presented by the HLA-A2⁺ T2 cell line. After 48 hours coculture of T cells and Ag-loaded T2 cells, cell-free supernatants were harvested and analyzed for interferon (IFN)-γ content by specific enzyme-linked immunosorbent assay (ELISA). *P < 0.05 versus T2 only control.

Overall, protective immunity in our model was associated with polyspecific Tc1 responses against at least one peptide epitope derived from 11 of 12 TASA evaluated (with NG2 being the lone exclusion). This does not preclude protective CD8⁺ T cell responsiveness against NG2-derived peptides, given the recent findings reported by Maciag et al.¹⁵ In their study, a recombinant Listeria monocytogenes vaccine was found to activate CD8⁺ T cells reactive against the HLA-A2-presented NG2_2238–2246 epitope in HLA-A2/K^b mice, in association with slowed tumor progression and inhibition of NG2⁺ pericyte frequencies in the TME. Our inability to observe a similar finding could relate to the comparative subdominance of the NG2 epitope among the many TASA being coordinately responded against under IL-12 gene therapy conditions in our B16 model system. Alternatively, or additionally, the functional avidity of murine anti-NG2 CD8⁺ T cells invoked in our model may be low, thereby disallowing recognition of target cells that are intrinsically NG2⁺ (i.e., pericytes) or pulsed with NG2-derived peptides (i.e., T2 cells). Furthermore, the time point at which we have investigated Tc1 responses (5–7 days after the second i.t. injection of DC.IL12) may have missed an anti-NG2 Tc1 response if this were to evolve later in the therapeutic process based on sequential rounds of therapy-induced crosspriming and the progressive broadening of the protective T cell repertoire. Prospective longitudinal analyses of CD8⁺ T cell responses against TASA should allow us to discriminate whether (and to what extent) specific T cell reactivity against a given TASA-derived peptide epitope fluctuates over time in effectively treated, long-term survivor animals.

Since the peptides analyzed in the current study are nonmutated and were chosen to be evolutionarily conserved sequences in humans and mice, i.t. DC.IL12 gene therapy must be capable of breaking operational tolerance in the T cell repertoire reactive against these “self” antigens/epitopes. Indeed, we observed that HHD mice and HLA-A2⁺ normal human donors, as well as, HLA-A2⁺ patients with melanoma exhibited detectable CD8⁺ T cell responses against (at least one peptide selected from) the vast majority of TASA evaluated. Although previous reports have characterized human CD8⁺ T cell responses against HLA-A2-presented peptides derived from the TASAs EphA2, NG2, PSMA, RGS5, VEGFR1, and VEGFR2,^{15,16,17,21,23,27} the current HHD model studies support the natural (tumor stromal cell) presentation of novel HLA-A2-presented epitopes derived from the DLK1, HBB, NRP1, PDGFRβ, and TEM1 gene products in situ.

Even though our study of HLA-A2⁺ normal donor and melanoma patient responses is small in size, at a minimum, these data suggest that melanoma patients and normal donors are capable of mounting type-1 CD8⁺ T cell responses against many “self” TASA when appropriately stimulated (as a proof-of-principle for the development of future vaccines targeting such antigens). Although somewhat an unfair comparison given the application of two IVS cycles for normal donors and one IVS cycle for melanoma patients, poststimulation CD8⁺ T cell responses between these cohorts were statistically different for only a minor subset of TASA peptides (i.e., DLK1 (309), NG2 (770), NG2 (2238), PDGFRβ (891), and RGS5 (5); Figure 4 legend). Such differences could reflect the differential presence of specific CD8⁺ memory T cells in the peripheral blood of melanoma patients, however, this is highly speculative. Prospective analyses will need to include segregation of CD8⁺ responder T cells into “naive” versus “memory” populations at experiment outset to discrimate whether in vivo priming of such T cells has occurred in patients, and the integration of polyfunctional effector analyses to discern the likely “clinical importance” of such CD8⁺ T cells in the cancer setting.²⁸

Furthermore, given the diversity of prior therapies received by the evaluated melanoma patients and variance in their current disease status (i.e., no evidence of disease versus active disease; Table 3), in addition to the small sample size, it is impossible to correlate T cell responsiveness to TASA with clinical outcome at the current time. Such information can only be determined in the context of longitudinal immunomonitoring studies applied to prospective randomized therapeutic trials for patients with melanoma. In this regard, our data argue for the translational utility of TASA peptides in the context of active vaccination protocols and/or clinical trials implementing immunotherapeutic/antiangiogenic approaches (including IL-12p70 gene therapy, tyrosine kinase inhibitors or VEGFR antagonists) for the treatment of solid cancers, such as melanoma. While one could readily envision the development of phase I/II TASA peptide-based vaccines, it may ultimately be most attractive to consider recombinant TASA protein- or gene-based formulations. These latter agents would presumably have the capacity to promote polyspecific, polyfunctional T cell-mediated immunity in HLA-heterogeneous cancer patients (thereby obviating the need to restrict accrual to a given HLA allotype; i.e., HLA-A2). Such approaches would also allow one to concomitantly elicit TASA-specific type-1 CD4⁺ T cell-mediated immunity that may prove directly reactive against HLA class II⁺ pericytes or VEC in the proinflammatory TME and/or support optimal Tc1 functionality/durability in cancer patients.²⁹ We are currently in the process of defining novel Th epitopes from the TASA evaluated in the current paper that may prove useful in vaccine formulation and the immune monitoring of patients.

Treatment-associated vascular “normalization,” which has been reported to be a preferred clinical outcome in successful cancer therapies,^{30,31,32,33,34} could also be the direct result of the CD8⁺ T cell-mediated death/regulation of VEC or pericytes that are required to sustain VEC within the TME in vivo. In such a scenario, tumor-associated pericytes/VEC could be induced to undergo either apoptosis or granzyme/perforin-mediated lysis by effector Tc1 cells.^35,36,37 Alternatively or additionally, the intimate communication between pericytes and VEC (i.e., via platelet-derived growth factor, vascular endothelial growth factor, transforming growth factor-β1, etc.) could be disrupted by IFN-γ, tumor necrosis factor-α, and/or additional factors secreted by Tc1 in response to cognate Ag presented by tumor pericytes or VEC, or crosspresented by tumor-associated APCs.^38,39 Such a pathway could be reinforced in an autocrine manner based on the expected upregulation of pericyte/VEC MHC class I expression by IFN-γ,⁴⁰ allowing for improved CD8⁺ T cell recognition of these target cells. Furthermore, it remains formally possible that IL-12 gene therapy-induced inhibition of vasculogenesis in the TME may be related to systemic T cell-mediated targeting and eradication of circulating TASA⁺ pericyte and/or VEC progenitors that could otherwise have been recruited into, and co-opted to become components of the tumor vascular bed.⁴¹ We are currently investigating the in vivo relevance of these various mechanisms of action and target cell populations in our therapeutic B16 model.

Beyond the predicted direct suppression of tumor growth by treatment-induced, TASA-specific CD8⁺ T cells, these strategies would be presumed to reduce tumor interstitial pressure, thereby enhancing the “deliverability” of systemic therapeutic agents (such as chemotherapeutic drugs, therapeutic mAbs or even therapeutic T cells themselves) into the TME. Additionally, such treatments would be expected to promote loco-regional tumor cell death (necrosis and/or apoptosis), providing an enriched source of tumor antigen in vivo that may allow for secondary waves of crosspriming and the “spreading” of the antitumor T cell repertoire.^42,43,44,45 This diversification in the specificity of protective T cells would theoretically allow for enhanced therapeutic efficacy and more durable T cell-mediated protection against tumor recurrence or the progression of micro-metastatic disease.^43,44,45

Ultimately, however, the clinical success of therapies/vaccines that evoke anti-TASA Tc1 responses will depend on whether such T effector cells are differentially recruited into, and sustained within, the TME versus normal vascularized tissues. In this regard, recent reports studying peptide- or DNA-based vaccines targeting the TASA EphA2, NG2, or VEGFR2 have all exhibited some degree of antitumor effectiveness with little or no off-target disruption of the normal vasculature,¹³ the cutaneous wound-healing process^15,16,17 or normal fertility, gestational period, or litter size in treated mice.¹⁵ Thus far, we have not observed any acute behavioral or physical manifestations of toxicity in HHD mice cured of B16 tumors as a result of DC.IL12 therapy. Furthermore, our in vitro analyses suggest that therapeutic anti-TASA Tc1 cells do not recognize normal tissue-derived pericytes or VEC. Nevertheless, given our belief that “epitope spreading” [a phenomenon classically associated with the development of chronic autoimmune diseases; ref. 46, underlies effective IL-12 gene therapy, we will continue to assess the health/performance status of treated animals for any signs of evolving autoimmune pathology (i.e., vasculitis, encephalopathy, retinopathy in the case of the current focus on TASA)].

In conclusion, our data suggest that therapies that promote CD8⁺ T cell targeting of tumor-associated TASA (such as DLK1, EphA2, HBB, NRP1, RGS5, and TEM1) may be meritorious for translation into the clinic for the treatment of patients with solid, vascularized tumors, such as melanoma. In particular, vaccines based on such TASA are of compelling interest and may prove effective in the treatment of a broad range of cancers regardless of the immunophenotype (MHC and tumor antigen) status of the patient's tumor cells in vivo.

Materials and Methods

Mice. HHD mice were obtained from Dr François A. Lemonnier through Dr Pravin T.P. Kaumaya (The Ohio State University, Columbus, OH). HHD mice fail to express H-2^b class I molecules, with their cells instead expressing an HLA-A*0201-hβ2 microglobulin single-chain (HHD) gene product.⁴⁷ Ag-specific CD8⁺ T cell responses in HHD mice recapitulate those observed in HLA-A2⁺ human donors.⁴⁷ Female 6–8-week-old mice were used in all experiments and were handled in accordance with an Institutional Animal Care and Use Committee-approved protocol. HLA-A2 expression on peripheral blood cells isolated from HHD mice via tail venipuncture was confirmed by coordinate positive staining as assessed by flow cytometry using two monoclonal antibodies (mAbs) MA2.1 (reactive against HLA-A2 and HLA-B17) and BB7.2 (reactive against HLA-A2 and HLA-Aw69) (both mAbs from the American Type Culture Collection, Manassas, VA).

Cell lines and culture. B16 is an HLA-A2^neg, mMART-1⁺, mgp100⁺ melanoma cell line (syngenic to the H-2^b background of HHD mice) and has been described previously.¹⁴ The T2 cell line is an HLA-A2⁺, TAP-deficient human T-cell/B-cell hybridoma.²⁷ Cell lines were free of mycoplasma contamination and were maintained in CM [RPMI 1640 supplemented with 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, 100 µg/ml streptomycin, and 10 mmol/l -glutamine (all reagents from Life Technologies, Grand Island, NY)] in a humidified incubator at 5% CO₂ and 37 °C.

RT-PCR. RT-PCR was performed using the following primer pairs (Supplementary Table S1). Cycling times and temperatures were as follows: initial denaturation at 94 °C for 2 minutes (1 cycle), denaturation at 94 °C for 30 seconds, annealing at 60 °C for 30 seconds and elongation at 72 °C for 1 minute (30 cycles), final extension at 72 °C for 5 minutes (1 cycle). PCR products were identified by image analysis software for gel documentation (LabWorks 4.6 Software; UVP, Upland, CA) following electrophoresis on 1.2% agarose gels and staining with ethidium bromide (Sigma-Aldrich, St Louis, MO).

Fluorescence imaging of tumor sections. Tumor tissue samples were prepared and sectioned as previously reported.¹³ For analysis of T cell subsets, sections were incubated with rabbit anti-mouse NG2 (Millipore, Bedford, MA) along with alexa488-conjugated anti-CD4 or -CD8β antibodies or matching isotype controls (all from BD Biosciences, San Jose, CA) for 1 hour. After washing with 0.5% bovine serum albumin in phosphate-buffered saline (PBS), sections were stained with donkey anti-rabbit Ig cy5 (Jackson ImmunoResearch, West Grove, PA) secondary pAb for 1 hour at room temperature. For analysis of CD31 versus NG2, sections were first incubated with rat anti-mouse CD31 (BD Biosciences) and rabbit anti-mouse NG2 (Millipore) Abs for 1 hour at room temperature and then washed. Sections were then treated with donkey anti-rat Ig cy3 and donkey anti-rabbit Ig cy5 (both from Jackson ImmunoResearch) Abs for 1 hour and washed. For the analysis of target antigens in B16 tumor lesions, all sections received dilutions of rat anti-mouse CD31 (BD Biosciences) and guinea-pig anti-mouse NG2 (kindly provided by Dr Bill Stallcup, The Burnham Institute for Medical Research, La Jolla, CA; ref. 48) Abs. In addition, each slide received a pAb reactive against a given TASA: rabbit anti-mouse pAb for DLK1 (R&D Systems, Minneapolis, MN), EphA2 (Santa Cruz Biotech., San Diego, CA), PSMA (Thermo Fisher Scientific, Rockford, IL), RGS5 (Sigma-Aldrich), VEGFR1 (Thermo Fisher Scientific) or goat anti-mouse pAb for HBB (Santa Cruz), NRP1 (R&D Systems), NRP2 (R&D Systems), PDGFRβ (R&D Systems), VEGFR2 (Abcam, Cambridge, MA). Sections were then again washed five times with 0.5% bovine serum albumin (in PBS), before a 1-hour incubation with dilutions of a mixture of secondary antibodies: (i) donkey anti-rat cy5 pAb, (ii) donkey anti-guinea-pig DyLight 488 pAb, and (iii) either donkey anti-rabbit cy3 pAb or donkey anti-goat cy3 pAb depending on the species of antibody directed against the TASA target (all secondary antibodies were purchased from Jackson ImmunoResearch). After secondary Ab staining, sections were then washed with three washes of PBS, coverslipped with gelvatol mounting media (made in-house) and stored at 4 °C until imaging using an Olympus Fluoview 500 Confocal microscope (Olympus America, Center Valley, PA).

Synthetic peptides. Peptides (Table 1) were synthesized by 9-fluorenylmethoxycarbonyl (Fmoc) chemistry by the University of Pittsburgh Cancer Institute's Peptide Synthesis Facility (a shared resource). Peptides were >96% pure based on high-performance liquid chromatography profile and mass spectrometric analysis performed by the University of Pittsburgh Cancer Institute's Protein Sequencing Facility (a shared resource).

Generation of HHD bone marrow-derived DCs and DC.IL12. DC were generated from bone marrow precursors isolated from the tibias/femurs of mice using in vitro cultures containing 1,000 U/ml recombinant murine granulocyte/macrophage colony-stimulating factor and 1,000 U/ml rmIL-4 (both from Peprotech, Rocky Hill, NJ), as previously described.¹³ The Ad.mIL-12p70 and Ad.ψ5 (empty) recombinant adenoviral vectors were produced and provided by the University of Pittsburgh Cancer Institute's Vector Core Facility (a shared resource), as reported previously.^13,42 Five million (day 5 cultured) DCs were infected at an multiplicity of infection = 50 with Ad.mIL-12p70 or the control, empty vector Ad.ψ5. While control DC produced <62.5 pg IL-12p70/ml/48 hour/10⁶ cells, DC.IL12 cells produced 1–10 ng IL-12p70/ml/48 hour/10⁶ cells (data not shown and refs. 13,42).

i.t. DC.IL12 therapy. B16 melanoma cells (1 × 10⁵) were injected subcutaneously in the right flank of HHD mice and allowed to establish for 7 days. Mice were then randomized into cohorts of five animals, with each cohort exhibiting an approximate mean tumor size of 30–50 mm². On days 7 and 14, tumor-bearing mice were untreated or treated with i.t. injections of 1 × 10⁶ adenovirus-infected DCs (DC.ψ5 or DC.IL12) in a total volume of 50 µl PBS. Tumor size was then assessed every 3–4 days and recorded in mm², determined as the product of orthogonal measurements taken using vernier calipers. In some experiments, as indicated, in vivo antibody depletions (on days 6, 13, and 20 post-tumor injection) of CD4⁺ T cells or CD8⁺ T cells were performed as previously described.¹³ Data are reported as mean tumor area ± SD. On day 17–19 post-tumor inoculation, CD8⁺ splenocytes and TIL were magnetic bead cell sorting-isolated from three mice/cohort, with cells pooled and assessed for reactivity against peptide epitopes or cell targets (pericytes, VEC, tumor cells) as described below.

Evaluation of murine CD8⁺ T cell responses in vitro. To analyze Ag-specific responses, spleens and TIL were harvested (from two mice/group) 3–5 days after the second i.t. injection of control DC or DC.IL12 (i.e., day 17–19 after tumor inoculation). Splenic lymphocytes were restimulated in vitro for 5 days with irradiated (2.5 Gy) naive peptide-pulsed HHD splenocytes at a stimulator:responder cell ratio of 1:1. Responder CD8⁺ T cells were then isolated using magnetic bead cell sorting (Miltenyi Biotec, Auburn, CA) and analyzed for reactivity against unpulsed or peptide-pulsed T2 cells, as indicated. To analyze T cell response to stromal cell targets and tumor cells, untreated HHD mice bearing established day 17–19 B16 tumors were sacrificed and tumors and kidneys removed. Tissues were then minced manually and enzymatically digested as described by Crisan et al.⁴⁹ using collagenases IA, II, and IV (Sigma-Aldrich) and DNAse I (Sigma-Aldrich) for 30 minutes at 37 °C, with gentle shaking. Cells were then being passed through a 70-micron cell strainer (BD Biosciences), washed with PBS, and single-cell suspensions stained with anti-mouse CD31 FITC (BD Biosciences), anti-mouse CD140b (PDGFRβ) PE (eBioscience, San Jose, CA), and anti-mouse H-2K^b APC (BD Biosciences). After washing with PBS, cells were sorted into enriched populations containing pericytes (PDGFRβ⁺CD31^negH-2K^b(neg)) or VEC (PDGFRβ^negCD31⁺H-2K^b(neg)) using a multicolor fluorescence-activated cell sorter (FACSAria; BD Biosciences). In all cases, cells were >95% pure for the stated phenotype. CD8⁺ T cells (10⁵) were then cocultured with 10⁴ pericytes or VEC in U-bottom 96-well plates (Sigma-Aldrich). To verify HLA-A2 restricted recognition of target cells by CD8⁺ T cells, 10 µg of anti-HLA-A2 mAb BB7.2 or control anti-HLA-class II mAb L243 (both from ATCC) were added to replicate coculture wells. Forty-eight hours after initiating splenic CD8⁺ T cell cocultures, cell-free supernatants were collected and analyzed for mIFN-γ content using a commercial enzyme-linked immunosorbent assay (BD Biosciences) with a lower limit of detection of 31.3 pg/ml. Data are reported as the mean ± SD of triplicate determinations. Alternatively, freshly sorted CD8⁺ TIL were cocultured with pericytes, VEC, T2 cells (±peptides) or B16 tumor cells at a T cell-to-target cell ratio of 3:2 for 4–5 hours at 37 °C and analyzed for intracellular levels of IFN-γ or cell-surface expression of CD107a/b using specific mAbs (APC-labeled anti-mouse CD8α from eBioscience; PE-labeled rat anti-mouse IFN-γ and FITC-labeled rat anti-mouse CD107a/b from BD Biosciences) and flow cytometry using the manufacturer's suggested protocol and ref. 26, respectively.

In vitro assessment of human CD8⁺ T cell responses against TASA- or TAA-derived peptides. Peripheral blood mononuclear cells were obtained by venipuncture or leukapheresis from HLA-A2⁺ normal donors or HLA-A2⁺ melanoma patients with written consent under institutional review board-approved protocols (Table 3). CD8⁺ T cells were then isolated by magnetic bead cell sorting (Miltenyi Biotec) and either not stimulated or stimulated with autologous, TASA peptide-pulsed DC as previously described.²⁷ Normal donor T cells were stimulated with TASA peptide-pulsed DC twice on a weekly schedule, with responder T cells harvested for analysis of their specificity 5 days after the booster stimulation (i.e., day 12 of T cell-DC coculture). Melanoma patient CD8⁺ T cells were analyzed after a single round of stimulation with TASA peptide-pulsed, autologous DC (i.e., day 5 of T cell-DC coculture) as indicated in text. For DC-based stimulations, DC were pulsed with an equimolar (1 µmol/l each) pool of the TASA peptides (Table 1) for 4 hours at 37 °C at 5% CO₂ tension. These antigen-loaded DC were then used to stimulate autologous CD8⁺ T cells at a T cell-to-DC ratio of 10:1 to generate a bulk population of responder T cells. T cells were maintained in IMDM media supplemented with 10% human AB serum, 100 U/ml penicillin, 100 mg/ml streptomycin, 10 mmol/l -glutamine and MEM nonessential amino acids (all reagents from Invitrogen, except human AB serum that was purchased from Sigma-Aldrich, Carlsbad, CA). Responder CD8⁺ T cells were analyzed for reactivity against control (HLA-A2⁺) T2 cells or T2 cells pulsed with individual TASA or TAA peptides (1 µmol/l for 4 hours at 37 °C⁵⁰) at a CD8⁺ T cell-to-T2 cell ratio of 5:1 for 24 hours. Harvested cell-free supernatants were consequently assessed for hIFN-γ content using a specific enzyme-linked immunosorbent assay (BD Biosciences) with a lower detection limit of 4.7 pg/ml.

Statistical analysis. Student's two-sided t-test and one-way analysis of variance were used to test for overall differences between groups (StatMate III; ATMS, Tokyo, Japan), with a P value <0.05 taken as significant.

SUPPLEMENTARY MATERIAL Figure S1. TASA-derived peptides bind to HLA-A2 to a variable degree based on the T2 class I stabilization assay. Peptide stabilization of HLA-A2 complexes on the T2 cell line by synthetic peptides was assessed as previously described [50]. FluM1_58-66 (GILGFVFTL) was used as a positive HLA-A2 binding control peptide [27]. Overlays of fluorescence histograms are provided for each peptide over a 1-10000 nM dose range, as indicated. Evidence for productive stabilization of HLA-A2 complexes is supported by a shift in staining intensity to the right vs. the no peptide control. Negative control (HLA-A3/A11-binding) HIV-nef_73-82 peptide [27] failed to promote enhanced HLA-A2 stabilization on T2 cells (data not shown). Data are from 1 representative experiment of 3 independent assays performed. Figure S2. CD8⁺ T cells isolated from B16-bearing HHD mice left untreated or treated with DC. ψ5 fail to recognize tumor-associated pericytes/VEC. CD8⁺ T cells were MACS-isolated from the spleens of tumor-bearing animals that were left untreated (Control) or that were treated with i.t. delivered DC.ψ5, as outlined in Fig. 2B. These T cells were then cultured with flow-sorted tumor- or kidney-derived pericytes or VEC +/- blocking anti-HLA-A2 (BB7.2) or class II (L243) mAbs as described in Materials and Methods. Cell-free supernatant was harvested after 24h incubation at 37^oC and analyzed using a specific IFN-γ ELISA. Representative data is presented from 1 of 2 independent experiments performed. Figure S3. CD8⁺ TIL isolated from B16-bearing HHD mice treated with DC.IL12 recognize tumor-associated pericytes in an HLA-A2-restricted manner, and fail to recognize HLA-A2^neg B16 tumor cells. TIL were isolated from the day 17 melanomas of mice (treated as indicated) and analyzed for reactivity against flow-sorted tumor pericytes as described in Fig. 3 for intracellular IFN-γ or cell surface expression of translocated CD107 using flow cytometry. To assess MHC-restriction in T cell recognition of tumor pericytes, 10 μg of anti-HLA-A2 mAb BB7.2 or anti-pan class II mAb L243 were added to cultures during the 4-5h co-incubation period prior to flow cytometry-based analysis. Inset numbers reflect the percentage of CD8⁺ T cells exhibiting positive response to tumor pericytes or B16 melanoma cells. Data derive from 1 representative experiment of 2 independent experiments performed. Table S1. RT-PCR primers used in this study.

Supplementary Material

Figure S1.

TASA-derived peptides bind to HLA-A2 to a variable degree based on the T2 class I stabilization assay. Peptide stabilization of HLA-A2 complexes on the T2 cell line by synthetic peptides was assessed as previously described [50]. FluM1_58-66 (GILGFVFTL) was used as a positive HLA-A2 binding control peptide [27]. Overlays of fluorescence histograms are provided for each peptide over a 1-10000 nM dose range, as indicated. Evidence for productive stabilization of HLA-A2 complexes is supported by a shift in staining intensity to the right vs. the no peptide control. Negative control (HLA-A3/A11-binding) HIV-nef_73-82 peptide [27] failed to promote enhanced HLA-A2 stabilization on T2 cells (data not shown). Data are from 1 representative experiment of 3 independent assays performed.

Figure S2.

CD8⁺ T cells isolated from B16-bearing HHD mice left untreated or treated with DC. ψ5 fail to recognize tumor-associated pericytes/VEC. CD8⁺ T cells were MACS-isolated from the spleens of tumor-bearing animals that were left untreated (Control) or that were treated with i.t. delivered DC.ψ5, as outlined in Fig. 2B. These T cells were then cultured with flow-sorted tumor- or kidney-derived pericytes or VEC +/- blocking anti-HLA-A2 (BB7.2) or class II (L243) mAbs as described in Materials and Methods. Cell-free supernatant was harvested after 24h incubation at 37^oC and analyzed using a specific IFN-γ ELISA. Representative data is presented from 1 of 2 independent experiments performed.

Figure S3.

CD8⁺ TIL isolated from B16-bearing HHD mice treated with DC.IL12 recognize tumor-associated pericytes in an HLA-A2-restricted manner, and fail to recognize HLA-A2^neg B16 tumor cells. TIL were isolated from the day 17 melanomas of mice (treated as indicated) and analyzed for reactivity against flow-sorted tumor pericytes as described in Fig. 3 for intracellular IFN-γ or cell surface expression of translocated CD107 using flow cytometry. To assess MHC-restriction in T cell recognition of tumor pericytes, 10 μg of anti-HLA-A2 mAb BB7.2 or anti-pan class II mAb L243 were added to cultures during the 4-5h co-incubation period prior to flow cytometry-based analysis. Inset numbers reflect the percentage of CD8⁺ T cells exhibiting positive response to tumor pericytes or B16 melanoma cells. Data derive from 1 representative experiment of 2 independent experiments performed.

Table S1.

RT-PCR primers used in this study.