Development of Tumor-Infiltrating CD8⁺ T Cell Memory Precursor Effector Cells and Antimelanoma Memory Responses Are the Result of Vaccination and TGF-β Blockade during the Perioperative Period of Tumor Resection Free

Curative resection remains the intent for all surgical interventions for localized solid tumors; however, local and/or distant recurrence is not uncommon and is a significant cause of morbidity and mortality (1). A major goal of cancer immunotherapy is to develop effective CD8⁺ T cell memory responses with cancer vaccines that would provide protection against dissemination of localized tumors. The development of vaccines to prevent cancer recurrence is particularly relevant in the treatment of poorly immunogenic melanoma, for which we have failed to develop adjuvant treatment with a favorable therapeutic outcome. Although prophylactic vaccination can often elicit tumor-reactive CD8⁺ T cell responses (2), current immunotherapeutic options including vaccination against authentic tumor-associated Ags for the treatment of solid tumors have not consistently prevented recurrent disease (3).

Several barriers to the development of effective tumor vaccination strategies have been identified thus far, including: 1) immune tolerance to tumor or tumor-associated Ags, which are essentially nonmutated self-peptides capable of only weakly activating the immune system; and 2) the immune-suppressive (4) and CD8⁺ T cell-exhausting (5) environment of the tumor-bearing host. This immune suppression has been shown to be mediated by TGF-β, which is overexpressed by tumors and tumor-induced suppressor cells (i.e., CD4⁺ regulatory T cells [Tregs]) (6). In both cases, TGF-β is thought to play a major role in inhibiting immune responses (7–13).

Previous studies, including ours, have shown that selective substitution of amino acid residues that anchor the peptide to the MHC class I (MHC-I) molecule to form a higher affinity bond between the peptide and MHC-I molecule is sufficient to overcome T cell tolerance or ignorance to weakly immunogenic self-Ags. Presentation of the mutant peptide by APC results in the activation of naive CD8⁺ T cells, which are then capable of recognizing the wild-type peptide when presented by the MHC-I molecule on tumor cells. These modified peptides are termed heteroclitic peptides (4, 14–19). We have previously shown that in contrast to DNA vaccination with wild-type mouse melanoma-shared Ag tyrosinase-related protein 1 (TRP1), DNA vaccination with a selectively mutated TRP1 (containing 10 single aa substitutions for the generation of heteroclitic peptides and lacking N-glycosylation sites for augmented proteolytic processing), termed Trp1ee/ng or diet white magic for its ability to induce autoimmune vitiligo in black mice (20–22), induces strong CD8⁺ T cell responses against wild-type TRP1 peptides and is protective against melanoma growth when administered prior to a tumor challenge with the poorly immunogenic mouse melanoma B16 (23).

Recent advances in the characterization of memory formation have shown that Ag presentation results in the differentiation of T cell phenotypes destined to form a stable pool of Ag-specific memory T cells (24–27). After the initial effector response, subsequent to Ag presentation, the majority of effector T cells, which are short-lived effector cells (SLECs), defined as T-bet^hi, undergo apoptosis, whereas a smaller population of memory precursor effector cells (MPECs), defined as T-bet^lo, go on to form a stable pool of memory cells capable of mounting an effective immune response at a distal time point. CD8⁺ T cell memory cells are poised to respond more rapidly, strongly, and for extended duration than naive T cells and are thought to be resistant to tumor-induced suppression (27–29).

Based on the above observations, in the current study, we investigated the conditions and mechanisms that could result in the formation of tumor-reactive CD8⁺ T cells in tumor-bearing hosts that could effectively protect against a second melanoma challenge at a distal time point. Using a highly aggressive and immune-suppressive mouse melanoma model (B16), we demonstrate that treatment with Trp1ee/ng DNA vaccination combined with systemic TGF-β blockade in the perioperative period of local tumor resection results in effective protection from later melanoma challenge. Importantly, we report that protection conferred by perioperative immunotherapy correlated with the increased formation of CD8⁺ T cell MPECs expressing low levels of the transcription factor T-bet, as well as augmented proportion of tumor-infiltrating effector CD8⁺ T cells during the memory phase.

Pathogen-free C57BL/6 and B6.Cg-Thy1^a/Cy Tg(TcraTcrb)8Rest/J (pMel) male mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed at The University of Chicago animal facility in accordance with Institutional Animal Care and Use Committee guidelines. Tumor challenges were carried out with B16 or JBRH melanoma cells or Lewis lung carcinoma (LLCa) cells, as described previously (4, 30). The B16F10 mouse melanoma cell line was originally obtained from I. Fidler (M.D. Anderson Cancer Center, Houston, TX). JBRH melanoma was provided by P. Livingston (Memorial Sloan-Kettering Cancer Center, New York, NY). LLCa was acquired from American Type Culture Collection (Manassas, VA). For the primary tumor challenge, C57BL/6 mice (8–10 wk old) were inoculated intradermally in the right flank with 1 × 10⁵ live B16 cells. For the secondary tumor challenge, C57BL/6 mice were rested for 30 d after the last immunotherapy treatment and then reinoculated intradermally in the left flank with either 1 × 10⁵ live B16, JBRH, or LLCa cells. For therapeutic tumor challenge, C57BL/6 mice were reinoculated with 1 × 10⁵ live B16 cells the day prior to surgical resection of the primary tumor. Tumor diameter was measured three times per week. If maximum tumor diameter reached 10 mm or the tumor became ulcerated, the mice were euthanized. For adoptive cell transfer, naive CD8⁺ T cells from pMel mouse spleens were purified using a MagCellect kit (R&D Systems, Minneapolis, MN), and 1 × 10⁵ cells were transferred into B6 recipients via retro-orbital injection.

The primary B16 intradermal tumors were surgically excised on day 10 after tumor inoculation at the time of the second vaccination treatment, when tumor diameter had reached 3–5 mm. Mice were anesthetized with isoflurane, and the tumors were excised sharply with a 2-mm margin. Incisions were closed with steel staples. Mice in which the tumor was unable to be excised (<10%) were removed from the study.

For DNA vaccination, mice were anesthetized with isoflurane, and each mouse received four dermal administrations (total dose of 4 μg plasmid DNA per mouse) of modified TRP1ee/ng or hgp100 DNA-coated gold particles via gene gun biolistic transfection as previously described (4, 16, 18–20). Mice were vaccinated every 5 d for a total of three vaccinations. Peptide immunization was performed by footpad injection with TiterMax emulsion (CytRx, Norcross, GA) containing 10 μg optTRP1₄₅₅ peptide (TAPDNLGYM) or OVA₂₅₇ (SIINFEKL) in 10 μl total volume (16, 31).

Anti–TGF-β mAb (1D11.16.8) and anti-CD4 mAb (GK1.5) was obtained from BioXCell (West Lebanon, NH). Anti–TGF-β mAb (250 μg) or anti-CD4 mAb (250 μg) was administered via i.p. injection at the time of DNA vaccination treatments.

All mouse Abs against cell surface and intracellular markers were purchased from eBioscience (San Diego, CA), except CD3 allophycocyanin-Cy7 (BD Biosciences, San Jose, CA), T-bet Pacific Blue (Biolegend, San Diego, CA), and OptTRP1₄₅₅-tetramer PE (Fred Hutchinson Cancer Research Center, Seattle, WA). Cells from spleen and lymph node were depleted of RBCs using Ack Lysis buffer (BioWhittaker, Walkersville, MD), washed once with PBS (BioWhittaker), incubated with Abs to extracellular markers for 30 min at 4°C in the presence of Mouse Fc Block (BD Biosciences), washed in PBS, and fixed with 2% formaldehyde. For intracellular marker staining, cells were stained for extracellular markers as described previously, fixed and permeabilized using BD Cytofix/Cytoperm (BD Biosciences), washed with BD Perm/Wash buffer (BD Biosciences), stained in Perm/Wash Buffer (BD Biosciences) with Abs to intracellular markers for 30 min at 4°C, washed twice with BD Perm/Wash Buffer (BD Biosciences), and fixed in a solution of 50% v/v 4% formaldehyde and BD Perm/Wash Buffer (BD Biosciences). Cell staining data were acquired using an FACSCanto or LSRII flow cytometer (BD Biosciences) with FACSDiva software (BD Biosciences) and analyzed with FlowJo software (Tree Star, Ashland, OR). For cytokine production experiments, cells were stimulated for 6–18 h, as indicated, with 10 pg–1μg/ml wild-type TRP1₄₅₅ peptide (TAPDNLGYA) in the presence of 1:1000 GolgiStop (BD Biosciences), prior to intracellular staining.

Matrigel (BD Biosciences) was thawed on ice at 4°C and mixed with RPMI 1640 containing 2 × 10⁵ B16 melanoma cells at a 1:1 volume ratio. B16-matrigel mixture was injected intradermally (200 μl) into the left flank of each mouse under anesthesia. Seven days later, B16-matrigel plugs were harvested and dissociated using the Miltenyi GentleMACS Dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany; per manufacturer’s recommendation for B16 dissociation). Cells recovered were analyzed using flow cytometry.

Statistically significant differences in tumor-free survival between groups of mice were determined with log-rank analysis of Kaplan–Meier data. For dose-dependent responses, best-fit sigmoidal curves were compared pairwise by F test. For all other data, Student t test (two-tailed) was used to calculate the p value for differences in measurements between two groups. A p value of <0.05 was considered statistically significant.

We have previously shown that prophylactic immunization with a selectively mutated construct (Trp1ee/ng) protects mice against B16 mouse melanoma challenge, but is not curative when applied during progressive tumor growth (4, 20). We hypothesized that progressively growing tumors inhibit the development of CD8⁺ T cells induced by vaccination. Thus, we examined Ag-specific CD8⁺ T cell responses to optTRP1₄₅₅ or OVA₂₅₇ peptide vaccination in the presence of B16 tumors. Groups of C57BL/6 mice were intradermally challenged in the right flank with 1 × 10⁵ B16 melanoma cells 5 d prior to peptide vaccination. Five days after the peptide vaccination, CD8⁺ T cell responses were assessed by measuring IFN-γ production in CD8⁺CD44^hiCD62L^lo T cells restimulated with the wtTRP1₄₅₅ or OVA₂₅₇ peptide. Ag-specific IFN-γ production by effector CD8⁺ T cells induced by vaccination was significantly reduced in tumor-bearing versus tumor-free mice (Fig. 1A, 1B). To quantify effector CD8⁺ T cell responses after vaccination, we examined the percent of CD8⁺ effector T lymphocytes by optTRP1₄₅₅ tetramer staining via flow cytometry of tumor-draining inguinal lymph nodes cells from tumor-bearing mice and equivalent lymph nodes from tumor-free mice 5 d after optTRP1₄₅₅ peptide vaccination. The percentage of tetramer⁺ optTRP1₄₅₅-reactive CD8⁺ T cells was significantly reduced (Fig. 1C) in tumor-bearing mice compared with tumor-free mice. Tetramer specificity was assessed by staining CD8⁺ T cells from mice vaccinated with OVA₂₅₇ (Supplemental Fig. 1).

Tumor-induced immune suppression has been attributed to multiple mechanisms that tumors have acquired to escape immune destruction. Studies of suppressive mechanisms by the Ochoa and Gabrilovich laboratories (32) have demonstrated that tumors can modulate the expression of the CD3ζ molecule in CD8⁺ T cells or induce changes in Gr-1⁺ immature myeloid cells. Consequently, we tested the role of these mechanisms of tumor-induced suppression in our melanoma model and found indistinguishable changes in CD3ζ expression and immature myeloid cell formation in B16 melanoma-bearing versus tumor-free mice (data not shown). Previous studies have shown that TGF-β is a key mediator of tumor-induced immune suppression (6, 33, 34). To determine whether tumor-induced immune suppression was mediated by TGF-β in our melanoma model, naive B6 mice were inoculated intradermally with 1 × 10⁵ B16 cells at day −10. Recipient mice were then vaccinated with optTRP1₄₅₅ peptide in adjuvant TiterMax emulsion (19) at day 0 (Fig. 2A). Intracellular IFN-γ production by CD44^hiCD8⁺ T cells was assessed 5 d postvaccination. Some mice were treated additionally with TGF-β blockade and i.p. injections of 250 μg mAb 1D11.16.8 (35) at days −1 and 3 relative to vaccination. As shown in Fig. 2, suppression of Ag-specific responses was rescued by TGF-β blockade. To determine whether suppression in tumor-bearing hosts was cell mediated, effector CD8⁺ responses were assessed in optTRP1₄₅₅-vaccinated mice after the adoptive transfer of lymphocytes from tumor-bearing mice (Fig. 3A). Specifically, donor naive B6 mice were inoculated intradermally with 1 × 10⁵ B16 cells. Ten days post tumor inoculation, lymphocytes were harvested from tumor-draining lymph nodes and adoptively transferred into naive B6 mice. Recipient mice were then vaccinated with peptide optTRP1₄₅₅ in adjuvant TiterMax emulsion. Intracellular IFN-γ production was assessed 5 d after vaccination in CD44^hiCD8⁺ T cells. Control recipient mice underwent adoptive transfer of lymphocytes from naive mice prior to vaccination. The mean percentage of effector CD8⁺ cells producing IFN-γ was significantly lower (>50%) in vaccinated recipient mice after the adoptive transfer of cells from tumor-bearing mice compared with adoptive transfer from naive mice (Fig. 3B). However, when mice were treated with vaccination in combination with TGF-β blockade, with i.p. injections of 250 μg mAb 1D11.16.8 at days −1 and 3 relative to vaccination, the Ag-specific effector responses were rescued (Fig. 3B). To further identify the cells involved in tumor-induced suppression, a similar experimental design as shown in Fig. 3A was applied; however, in this experiment, purified CD4⁺ cells from tumor-draining lymph nodes were transferred into naive recipients. Vaccine-mediated IFN-γ responses were significantly reduced in mice that received CD4⁺ T cells purified from tumor-draining lymph nodes (Supplemental Fig. 2A). Based on these results, we tested whether vaccinated-mediated responses in tumor-bearing mice vaccinated with optTRP1₄₅₅ in adjuvant were restored by in vivo CD4 depletion (GK1.5). Although CD8⁺ T cells responses were suppressed in CD4-undepleted tumor-bearing hosts, CD4-depleted recipients were able to mount strong Ag-specific IFN-γ responses (Supplemental Fig. 2B). These data demonstrate that tumor-induced suppression is mediated by CD4⁺ T cells and is molecularly driven by TGF-β.

To characterize the course of tumor-induced immune suppression in relation to the timing of surgical excision, optTRP1₄₅₅ vaccine-induced CD8⁺ effector T cell responses were assessed at different time points after the removal of primary B16 tumor. B6 mice were inoculated with 1 × 10⁵ B16 melanoma cells in the right flank. Tumors were surgically excised 10 d after tumor inoculation at 3–5 mm² in diameter. Control tumor-free mice underwent mock surgery. Mice were vaccinated once with optTRP1₄₅₅ peptide at day 5 after surgical removal tumor. Draining inguinal lymph nodes were harvested at day 10 after surgical excision to assess CD8⁺ T cell responses against wild-type TRP1₄₅₅. The percentage of effector CD8⁺ T cells producing IFN-γ was comparatively suppressed after tumor resection compared with mock surgery mice (Fig. 4A). These data demonstrate that surgical resection alone is insufficient to reverse tumor-induced immune suppression of CD8⁺ T cells. However, IFN-γ production by effector CD8⁺ T cells returned to levels observed in tumor-free mice 5 wk post surgical removal of the tumor (Fig. 4B), demonstrating that the suppressive phenotype is lost over time.

Given the ability to reverse B16-mediated immune suppression with administration of TGF-β blockade, we hypothesized that perioperative DNA vaccination with TRP1ee/ng in combination with TGF-β blockade would result in effective protection from a secondary B16 tumor challenge at a distant time point. Mice were injected intradermally in the right flank with 1 × 10⁵ B16 melanoma (Fig. 5A). Tumor-bearing mice were treated with Trp1ee/ng DNA vaccination ± TGF-β blockade three times over the course of 15 d. At the second vaccination, when the tumors had reached ∼5 mm in diameter, they were surgically excised. After the third vaccination, mice were rested for 30 d and then rechallenged with 1 × 10⁵ B16 cells in the left flank. Tumor growth was measured three times per week. Approximately 55–60% of the mice treated with primary tumor excision, DNA vaccination, and TGF-β blockade remained free from tumor growth (Fig. 5B). In contrast, mice treated with primary tumor excision and TRP1ee/ng DNA vaccination and mice treated with primary tumor excision and no immunotherapy were not protected from tumor growth after the second B16 challenge and had ∼25% tumor-free survival (Fig. 5B). Furthermore, we observed that tumor-free mice treated with mock surgery and TRP1ee/ng DNA vaccination ± TGF-β blockade were also not protected from a B16 tumor challenge administered 30 d after the final immunotherapy treatment (Fig. 5B). All untreated mice undergoing B16 tumor challenge developed tumors within 21 d (Fig. 5B). Tumor-free mice treated with TGF-β blockade and tumor-bearing mice treated with TGF-β blockade and excision of primary tumor were not protected from a B16 challenge at a distant time point (data not shown). Importantly, vaccination ± TGF-β blockade, in the absence of a primary tumor burden, is not protective from a tumor challenge given at a distant time point. To test perioperative immunotherapy in a therapeutic model, B16-bearing mice treated with TGF-β blockade and vaccination in the perioperative period were challenged with a second B16 melanoma 1 d prior to resection of the primary B16 melanoma. Significant delays in tumor incidence (Fig. 6A) and prolonged survival (Fig. 6B) were observed in mice treated with TGF-β blockade and vaccination with TRP1ee/ng. No treatment effects were detected in nonimmunized mice (Fig. 6A, 6B).

To determine whether memory responses were directed against tumor-unique Ags (i.e., mutations), we treated mice as in Fig. 5; however, secondary tumor challenge was carried out using another immune-suppressive melanoma cell line, JBRH (Supplemental Fig. 3), which is unrelated to B16 (Fig. 7A). We observed a 40% tumor-free survival in mice treated with B16 primary tumor excision, DNA vaccination, and TGF-β blockade after a secondary tumor challenge with 1 × 10⁵ JBRH cells compared with no protection observed in untreated mice undergoing a tumor challenge with JBRH (Fig. 7B). Furthermore, tumor-free survival did not differ between mice undergoing tumor challenge with B16 and mice challenged with JBRH at the secondary time point (Fig. 7B), indicating that the memory responses involved in preventing the growth of a secondary tumor challenge are not limited to unique B16-Ags, but are rather extended to melanoma-shared Ags. To demonstrate that these protective memory responses were directed against melanoma, similarly treated mice were challenged at day 45 with LLCa. No protection was observed in these mice (data not shown). From these data, we conclude that administration of vaccination with TRP1ee/ng DNA in combination with TGF-β blockade during the period around surgical resection of the primary tumor results in a memory response capable of providing protection to an unrelated form of melanoma.

To determine whether perioperative immunotherapy with TRP1ee/ng vaccination and TGF-β blockade influences the preferential differentiation of the MPEC versus SLEC phenotypes, we compared MPEC/SLEC distribution in B16-matrigel tumor-infiltrating lymphocytes in mice treated with perioperative TRP1ee/ng vaccination ± TGF-β blockade and primary tumor excision (Fig. 8A). B16 tumor-bearing mice were vaccinated ± TGF-β blockade during the perioperative period of primary tumor excision. At the third vaccination, mice were reinoculated with 1 × 10⁵ B16 cells in an acellular matrix mixture, Matrigel (BD Biosciences). Seven days after the last treatment, B16-matrigel plugs were harvested, and CD44^hiCD62L^loCD8⁺ T cells were analyzed for T-bet expression. Tumor-infiltrating lymphocytes harvested from mice treated with surgical excision of the primary tumor, TRP1ee/ng vaccination, and TGF-β blockade exhibited significantly lower T-bet expression and thus increased MPEC formation compared with mice treated with vaccination and surgical excision (Fig. 8B). To demonstrate that responses shown in this study were Ag specific, we employed an experimental design in which Thy1.1-marked naive pMel CD8⁺ T cells were adoptively transferred into B6 mice 1 d prior to vaccination, and mice were vaccinated alternatively with hgp100. Consistent with decreased T-bet levels observed in endogenous CD8⁺ T cell responses (Fig. 8B), pMel CD8⁺ T cells from mice receiving hgp100 vaccination and TGF-β blockade also expressed a significant reduction in T-bet levels compared with mice receiving vaccination only (Supplemental Fig. 4). However, no changes in the number of pMel CD8⁺ T cells were observed (data not shown). Additionally, the number of pMel CD8⁺ T cells and the relatively low intracellular expression of T-bet by CD8⁺ T cells indicates that an increased proportion of MPECs at the effector phase may be involved in the memory protection observed with perioperative immunotherapy with TRP1ee/ng vaccination combined with TGF-β blockade. To test this possibility, mice were TRP1ee/ng vaccinated ± TGF-β blockade and rested for 30 d. Instead of receiving B16-matrigel at the effector phase (day 15), these mice received B16-matrigel only at the memory phase (day 43) (Fig. 8A). Consistent with increased MPEC formation at the effector phase, we observed that in the memory phase B16-matrigel plugs from mice treated with TRP1ee/ng vaccination and TGF-β blockade were infiltrated with a higher proportion of effector CD8⁺ T cells (CD44^hiCD62L^lo) compared with mice treated without TGF-β blockade (Fig. 8C). Mice that received a matrigel plug without B16 were unable to recruit detectable numbers of cells within the matrigel (data not shown).

Memory CD8⁺ T cells are poised to respond more rapidly, strongly, and for an extended duration than effector T cells and are resistant to tumor-induced suppression making them ideal immunotherapy candidates against local and/or distant recurrence. Previous studies have shown that CD8⁺ T cells are primed within the tumor environment; however, these cells become suppressed and/or exhausted (4, 5), thus preventing potent antitumor effector and memory responses. In this study, we demonstrate that immunotherapy, in the perioperative period of primary tumor resection, with DNA vaccination against an authentic tumor-shared Ag combined with reversal of tumor-induced suppression, results in the development of protective immune responses against tumor in the therapeutic and memory setting. By immunologically targeting wild-type Trp1 expressed on B16 through DNA vaccination (Trp1ee/ng) and systemic TGF-β blockade during the perioperative period of primary B16 resection, mice were protected against a coexistent and a secondary (memory-phase) B16 challenge. Importantly, this protection was extended to JBRH, an independently derived melanoma tumor cell line, that shares melanoma Ags with B16. This concurrent immunity against B16 and JBRH indicates that the responses observed in this study are not against B16-specific mutations. Importantly, in this study, we demonstrate that perioperative immunotherapy increases the formation of tumor-infiltrating and tumor-reactive CD8⁺ T cells expressing low levels of the transcription factor T-bet, defined as MPECs.

Tregs play a pivotal role in immunological suppression (36). Studies have shown that depletion of Tregs by CD25 or CD4 depletions prior to development of immune responses results in augmented immunological responses against tumors (30). However, CD25 is also expressed by activated CD8⁺ and CD4⁺ T cells; therefore, removal of Tregs by CD25 depletion can only be applied in the absence of an active immune response. In this study, we demonstrate that CD4⁺ T cells in tumor-bearing mice are cellular mediators of suppression. However, CD4 depletion results in the concurrent depletion of Th cells, which are necessary for optimal CD8⁺ T cell responses. Importantly, we also demonstrate that TGF-β is the pivotal suppressive mechanism; therefore, we reversed tumor-induced suppression by TGF-β blockade.

Under physiological conditions, regulation by TGF-β plays an important role in immune homeostasis. TGF-β inhibits T cell functions by blocking both proliferation and differentiation (34). In tumor-bearing hosts, TGF-β is produced by both tumor cells and tumor-induced suppressor cells (e.g., Tregs) (10, 34). Although studies in which TGF-β was blocked by either neutralizing Abs or a soluble TGF-β–RII-Fc (10, 34) have demonstrated increased antitumor CD8⁺ T cell responses induced by vaccination, in this study we show that TGF-β blockade alone is insufficient to permit the development of memory T cell responses. This was shown by the lack of protection against a memory-phase tumor challenge in mice that were treated with TGF-β blockade and/or vaccination with TRP1ee/ng in the effector phase. Importantly, whereas TGF-β blockade alone is insufficient, it is necessary for the development of protective memory responses against melanoma when delivered during vaccination in the perioperative period of primary tumor resection.

MPECs and SLECs have similar functional ability at the peak of the immune response; however, they greatly differ in their memory potential and survival (26, 37). A major challenge that remains is the identification of extrinsic and intrinsic signals that make SLECs susceptible to apoptosis and the signals that promote the survival and maintenance of the MPECs (38). Studies by Hand and Kaech (38) show that according to the amount of inflammation, a gradient of T-bet is created in which high T-bet expression induces SLECs and low expression promotes MPECs. We found that T-bet^lo tumor-specific pMel CD8⁺ T cells accumulate in tumor-bearing mice that have had primary tumors surgically resected and received perioperative therapy with TRP1ee/ng vaccination and TGF-β blockade.

Based on these observations, in our model, three elements are necessary for effective antitumor responses: 1) induction of CD8⁺ T cells by vaccination against an authentic tumor Ag; 2) prevention of tumor-induced immune-suppressive effects by TGF-β blockade; and 3) the presence of the tumor provides a potential source of immunologically shrouded but common melanoma-derived Ags. Together these elements result in the formation of MPEC and memory responses capable of protecting against an immediate and future tumor re-encounter. We show that conditions for an immunologically fertile environment are met when TGF-β blockade and vaccination are applied during the perioperative period of primary tumor resection. These findings address limitations of current CD8⁺ T cell immunotherapies against cancer by generating effective CD8⁺ T cell memory recall responses.

We thank Alan Houghton (Memorial Sloan-Kettering Cancer Center, New York, NY, for B16F10, originally obtained from I. Fidler [M.D. Anderson Cancer Center, Houston, TX], and JBRH originally provided by P. Livingston [Memorial Sloan-Kettering Cancer Center]). We also thank L.A.K. and J.J.K. for constructive discussions and the Flow Cytometry Facility at The University of Chicago for invaluable support.

The authors have no financial conflicts of interest.