Apoptosis-Regulated Low-Avidity Cancer-Specific CD8⁺ T Cells Can Be Rescued to Eliminate HER2/neu–Expressing Tumors by Costimulatory Agonists in Tolerized Mice Free

Cancer vaccines that activate cytotoxic CD8⁺ T cells are emerging as a promising treatment for patients with cancer (1). Vaccines are most effective when they induce a large high-avidity T-cell repertoire with populations specific for multiple epitopes of relevant tumor antigens (2). However, one barrier to effective immunization is the predominant establishment of lower avidity and therefore less potent T cells instead of the high-avidity T cells typically found in non–tumor-bearing hosts. This scarcity of high-avidity T cells is due to clonal deletion during thymic selection, from which the low-avidity T cells are capable of escaping (3, 4). This results in a population of low-avidity T cells available in the periphery to mount a potential antitumor response (3, 4). Although thymic deletion is an important mechanism to protect the host from the activation of self-reactive T cells, the remaining low-avidity T cells have a higher threshold of activation and are greatly impaired by the peripheral tolerance mechanisms (5–7).

A second barrier to effective activation of both high- and low-avidity T cells is the presence of many cell types within the tumor microenvironment, including the tumor cells themselves, which inactivate vaccine-induced tumor-targeted T cells. Both high- and low-avidity T cells are regulated by multiple peripheral tolerance mechanisms, including suppression by regulatory T cells (Treg) and myeloid-derived suppressor cells (MDSC), and intrinsic T-cell tolerance mechanisms such as anergy that result in increased expression of coinhibitory molecules including cytotoxic T-lymphocyte antigen-4 (CTLA-4) and programmed cell death receptor-1 (PD-1; refs. 6, 8–10). Low-avidity T cells are also subject to peripheral deletion triggered by the chronic presence of small amounts of antigen, but can either escape deletion when no antigen is present or can be protected against deletion through activation or anergy if they receive a strong enough T-cell receptor (TCR) signal (6). Although high- and low-avidity T cells may be regulated by similar mechanisms, emerging evidence suggests that there are also signaling pathways that predominantly regulate low- versus high-avidity populations within the T-cell repertoire (8, 11–15). Agents are being developed that can modulate a number of these Treg pathways, leading to enhanced antitumor activity in preclinical models and in patients with cancer (11, 12, 16–19). The first such agent, ipilimumab, was recently approved for use by the U.S. Food and Drug Administration (FDA) for its ability to activate T cells by downregulating the T-cell inhibitory signal CTLA-4 in patients with metastatic melanoma (20). It is still not clear which signaling pathways regulate low- versus high-avidity T cells and whether it is possible to convert a low-avidity T cell into one that functions with high-avidity potency. Thus, understanding the activation signaling differences between high- and low-avidity T cells will facilitate the discovery of additional drugs that activate a more complete T-cell repertoire.

We previously reported the development of high- and low-avidity CD8⁺ T-cell populations specific for the same HER2/neu–expressing (21) immunodominant epitope, RNEU_420–429 (22). These TCR transgenic mice were used to identify a subpopulation of Tregs that inhibit high-avidity T cells from functioning within the tumor. In addition, immunomodulating doses of cyclophosphamide (23) given with a neu-targeted, granulocyte macrophage colony—stimulating factor (GM-CSF)–secreting vaccine inhibit these Tregs and allow for complete eradication of neu-expressing mammary tumors by adoptively transferred high-avidity T cells (24). In contrast, adoptively transferred low-avidity T cells fail to traffic into and eradicate the tumor under the same conditions (22). Using these two TCR transgenic T-cell populations, we show for the first time that low-avidity T cells have increased expression of the proapoptotic proteins TNFRSF10B (DR5), FasL, and CD24; and the expression of these proteins correlates with reduced T-cell function and increased T-cell death. In addition, the presence of FasL-expressing low-avidity T cells causes death of high-avidity CD8⁺ T cells when cocultured in vitro. These studies establish that blocking activation-induced cell death (AICD) with TNF receptor (TNFR) agonists (anti-OX40 and 41BB antibodies) allows low-avidity T cells to traffic into the tumor and enhance tumor clearance. Therefore, these data identify the death receptors as a new mechanism responsible for regulating low-avidity CD8⁺ T-cell populations, providing new opportunities for developing another class of immunomodulating agents that can enhance the activity of cancer vaccines.

HER2/neu (neu-N) transgenic mice were purchased from The Jackson Laboratory [FVB-Tg(MMTV-Erbb2)NK1Mul/J], and bred and housed at the Johns Hopkins animal facility. Experiments were performed using female mice aged between 6 and 12 weeks old and protocols approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine (Baltimore, MD). High- and low-avidity TCR transgenic mice were generated from RNEU_420–429–specific T-cell clones as described previously (22). Dilutional tetramer staining was used to verify the avidity of each mouse before use.

The NT2.5 tumor cell line, 3T3neuGM vaccine, 3T3GM mock vaccine, and T2-D^q T-cell target cell line were created and maintained as previously reported (25). GM-CSF was verified by ELISA, and neu levels were verified by flow cytometry biannually (25).

The immunodominant RNEU_420–429 (PDSLRDLSVF) peptide and LCMV NP_118–126 (RPQASGVYM) negative control peptide were synthesized at >95% purity in the Johns Hopkins Biosynthesis and Sequencing Facility. The anti-OX40 antibody was produced from the OX86 hybridoma, a gift from the Weinberg laboratory. The hybridoma was grown in protein-free hybridoma media (Gibco) and purified over a protein G column (BD Pharmingen). Purified rat immunoglobulin G (IgG) was used as an irrelevant control (The Jackson Laboratory). Antibodies for flow cytometry were anti-DR5-PE, anti-CD24-APC, anti-Annexin V-Pacific Blue, anti-Thy1.2-PerCP, anti-Thy1.2-Pacific Blue, anti-CD8-FITC, anti-Bcl-2-APC (BioLegend), anti-FasL-PerCP-efluor710 (eBioscience), anti-Annexin V-PE, anti-Thy1.2-FITC, anti-IFN-γ-PE, anti-TNF-α-APC, and purified anti-Fas (BD Pharmingen), anti-Survivin-PE (Cell Signaling Technology), purified anti-CD24, and polyclonal rabbit IgG (Santa Cruz Biotechnology).

Microarrays were performed at the Johns Hopkins Deep Sequencing and Microarray Core Facility using the NuGen amplification system and an Affymetrix Exon 1.0 ST array [Gene Expression Omnibus (GEO) accession: GSE54020]. Cells were adoptively transferred into tumor-bearing, Cy-treated, vaccinated neu-N mice before being extracted from the tumor-draining node on day 3. RNA was extracted using the Qiagen's RNeasy Mini Kit.

A total of 1 × 10⁶ NT2.5 tumor cells were injected subcutaneously into the right mammary fat pad. One week after the tumor injection, mice received 100 mg/kg Cy intraperitoneally. Twenty-four hours later, mice received three subcutaneous injections of the 3T3neuGM vaccine (1 × 10⁶ cells/injection) into the left upper limb and both lower limbs. One day after the vaccination, 6 × 10⁶ high- or low-avidity CD8⁺ T cells were adoptively transferred into the mice. T cells were isolated from splenocytes of high- and low-avidity TCR transgenic mice using a Dynal mouse CD8-negative isolation kit (Invitrogen). Proliferation was assessed using the CellTrace CFSE Cell Proliferation Kit (Invitrogen). Of note, 300 μg of OX40 or 41BB antibody (R&D Systems) was given intraperitoneally with adoptive transfer 3 and 7 days after vaccination. Of note, 300 μg of CD4⁺ T cell-depletion antibody (BioLegend) was given intraperitoneally 3 days before adoptive transfer. All experiments described in this article were repeated at least twice with a minimum of 3 mice per group.

Three days after adoptive transfer, intracellular staining was performed on cells from tumor-draining lymph nodes using the mouse intracellular staining kit (BD Biosciences). A total of 1 × 10⁶ extracted cells were incubated with 2 × 10⁵ peptide-pulsed T2-D^q cells for 5 hours at 37°C with Brefeldin A (BD Biosciences) as described by Weiss and colleagues (22). Extracellular surface staining was performed overnight in fluorescence-activated cell sorting (FACS) buffer. Samples were read on an LSR-II flow cytometer and analyzed using FACSDiva software (BD Biosciences). Tumor-infiltrating lymphocytes (TIL) were analyzed by surface staining 5 days after adoptive transfer. Tumors were mashed and digested using 1 mg/mL collagenase (Gibco) and 25 μg/mL hyaluronidase (Sigma) before being stained and analyzed by flow cytometry.

A total of 5 × 10⁴ NT2.5 cells were injected subcutaneously into the right mammary fat pad 2 days before treatment with Cy, vaccination, and adoptive transfer as described above. Of note, 300 μg of OX40 or IgG was given intraperitoneally 3 and 7 days after vaccination. Tumor length and width were measured every 5 days. Mice were sacrificed when tumor size exceeded 1 cm².

T cells were isolated as described above from tumor-draining lymph nodes on day 3 and stained for surface markers before staining with Annexin V and 7-AAD (7-amino-actinomycin D) as described in BD Biosciences' Apoptosis Detection Kit. Cross-linking apoptosis studies were performed in vitro on splenocytes from high- and low-avidity TCR transgenic mice by adding 0.1 μg/mL purified Fas antibody, 500 ng/mL CD24 antibody, or 500 ng/mL IgG to 5 × 10⁵ cells/mL in a 96-well plate incubated at 37°C for 3 hours with T2-D^q cells pulsed with 10 ng of peptide. Following incubation, T cells were washed and stained as described previously.

Following lysis of the red blood cells using ACK buffer (ammonium-chloride-potassium buffer; Gibco), splenocytes from high-avidity TCR transgenic mice were mixed with CD8⁺ isolated low-avidity T cells at a ratio of 1:4 before incubating with peptide-pulsed (20 μg) T2-D^q cells at 37°C for 24 hours. Apoptosis staining was performed as described above using Vβ4 TCR staining to differentiate the high-avidity T cells from the Vβ2 low-avidity T cells.

Student t tests (paired and unpaired) were performed using GraphPad Prism software. Differences were considered statistically significant if a value of P < 0.05 was found.

We previously reported that vaccinated, Cy-treated low-avidity RNEU_420–429–specific TCR transgenic T cells adoptively transferred into a tolerant environment do not clear tumor, whereas similarly treated high-avidity T cells also specific for RNEU_420–429 are capable of clearing large burdens of tumor (22). Low-avidity T cells also do not secrete effector cytokines characteristic of fully activated T cells nor do they traffic to the tumor (22). To explain this disparity in functionality, we adoptively transferred naïve low- and high-avidity TCR transgenic T cells into neu-N transgenic mice to identify differences in protein expression between high- and low-avidity T cells in a tolerant environment. CD8⁺ T cells were adoptively transferred into tumor-bearing neu-N mice treated with Cy plus vaccine and isolated from the tumor-draining lymph nodes of the neu-N mice 3 days after adoptive transfer (22). mRNA was extracted from these cells to perform microarray analysis of gene expression (Supplementary Fig. S1). Ingenuity analysis of microarray data identified cell death pathways significantly increased in low- versus high-avidity T cells. Flow cytometry studies further confirmed that the prodeath proteins DR5, FasL, and CD24 are expressed at higher levels on low-avidity cells than high-avidity cells (Fig. 1A–C). These initial studies led to the hypothesis that higher expression of prodeath proteins on low-avidity T cells may explain the difference in function between low- and high-avidity T cells in a tolerant tumor microenvironment.

These studies demonstrate a correlation between the expression of DR5, FasL, and CD24 on T cells and avidity. Next, we evaluated whether increased expression of death pathway proteins also correlates with reduced function of low-avidity CD8⁺ T cells. Low- and high-avidity T cells, adoptively transferred into tumor-bearing neu-N mice treated with low-dose Cy and vaccine, were evaluated for cytokine secretion and trafficking into the tumor. The principal cytokine produced by activated RNEU_420–429–specific CD8⁺ T cells is IFN-γ, and low-avidity RNEU_420–429–specific CD8⁺ T cells produced much less IFN-γ than high-avidity T cells when adoptively transferred into a tolerant environment (22). To determine the effect of DR5, FasL, and CD24 on the function of low-avidity T cells, we first compared the expression of these proteins with IFN-γ secretion. Low-avidity T cells that express DR5, FasL, and CD24 secrete even less IFN-γ than low-avidity T cells that do not express these proteins (Fig. 2A). T-cell trafficking into tumors is another important function of activated cancer-specific T cells. As mentioned before, low-avidity T cells do not traffic into the tumor even in mice treated with Cy plus vaccine, whereas high-avidity T cells do traffic into and eradicate large burdens of neu-expressing tumors in mice treated with Cy plus vaccine. We therefore evaluated whether there is differential expression of death-associated proteins on high-avidity T cells that traffic into the tumor compared with those that do not. Analysis of these high-avidity TILs showed significantly lower expression of DR5, FasL, and CD24 on cells that trafficked into the tumor when compared with cells that remained in the tumor-draining lymph nodes (Fig. 2B). These proteins are also more highly expressed on low-avidity T cells activated in vivo or in vitro than in naïve cells (Supplementary Fig. S2). Annexin V and 7-AAD staining confirmed that DR5, FasL, and CD24 protein expression is upregulated on apoptosing T cells (Fig. 2C). The finding that T cells expressing DR5, FasL, and CD24 secrete less IFN-γ and are less likely to traffic into tumors indicates that T cells expressing these death receptor proteins are less functional as antitumor effector cells than cells that do not express these proteins.

To determine whether low-avidity T cells expressing these death receptors are more susceptible to cell death, as suggested by our initial studies, we used an anti-Fas receptor antibody. A total of 1 × 10⁴ naïve high- and low-avidity T cells from TCR transgenic mice were plated with 0.1 μg of anti-Fas receptor antibody for 3 hours. Death was determined by staining for 7-AAD and Annexin V. Low-avidity T cells demonstrated increased cell death following Fas receptor binding when compared with high-avidity T cells further confirming that low-avidity T cells are more susceptible to Fas-induced cell death than high-avidity T cells (Fig. 3A). CD24-mediated low-avidity T-cell death was also verified by antibody cross-linking of CD24 (Fig. 3B).

Because vaccination induces a polyclonal T-cell repertoire with a range of avidities specific for a tumor antigen, we wanted to address whether low-avidity T cells could negatively affect the life span of high-avidity T cells. We conducted an in vitro experiment to determine whether apoptosis would increase in high-avidity T cells when mixed with low-avidity T cells. High-avidity T cells were stimulated with T2-D^q cells pulsed with RNEU_420–429 peptide, with and without low-avidity T cells. We found that apoptosis does increase in high-avidity T cells when stimulated in the presence of low-avidity T cells (Fig. 3C). In addition, we found that blocking the Fas/FasL interaction on high-avidity T cells with a FasL-blocking antibody prevented the increase in high-avidity T-cell apoptosis. This indicates that low-avidity T cells cause death of high-avidity T cells in a Fas-dependent manner. These studies demonstrate that not only are low-avidity T cells more susceptible to death themselves but they are also able to induce cell death in other tumor-specific T-cell populations.

Next, we wanted to address whether low-avidity T cells would become more functional in clearing tumor if they were able to survive longer. An agonistic OX40 antibody was used because of the known role of OX40 in preventing AICD. Tumor-bearing Cy-and-vaccine–treated mice were treated with anti-OX40 antibody or rat IgG on the day of adoptive transfer. Intracellular staining of low-avidity T cells taken from the tumor-draining nodes of anti-OX40 antibody-treated mice on day 3 showed a significant increase in IFN-γ secretion over IgG-treated mice (Fig. 4A). Because anti-OX40 antibody induces function in low-avidity T cells, we next tested whether OX40 treatment had the ability to facilitate low-avidity T-cell trafficking into the tumors of neu-N mice and facilitate tumor rejection. Low-avidity T cells adoptively transferred into tumor-bearing mice treated with Cy, vaccine, and anti-OX40 monoclonal antibody (mAb) did traffic into the tumor, whereas low-avidity T cells adoptively transferred into tumor-bearing mice treated with Cy, vaccine, and IgG failed to traffic into the tumors (Fig. 4B). Low-avidity T cells from the tumor-draining nodes of mice treated with anti-OX40 antibody also had decreased expression of DR5, CD24, and FasL (Fig. 4C). The percentage of dead cells was also significantly lower in the anti-OX40–treated group in comparison with the IgG-treated group (Fig. 4D). Furthermore, increased cytokine activity and trafficking function exhibited by low-avidity T cells treated with Cy+vaccine+anti-OX40 antibody resulted in increased tumor-free survival when compared with mice treated with either Cy+vaccine+IgG, anti-OX40+vaccine alone without Cy, or anti-OX40+Cy+mock vaccine (Fig. 4E). Anti-OX40+vaccine+Cy–treated mice were also able to eradicate 10% of the established tumors (Supplementary Fig. S3).

Two mechanisms by which treatment with anti-OX40 antibody prevents cell death are increased expression of the antiapoptotic protein Bcl-2 and increased expression of the antiapoptotic protein survivin (Fig. 4F). Survivin expression was significantly increased in anti-OX40+Cy+vaccine–treated mice on days 3 and 5, whereas Bcl-2 expression was significantly increased on days 5 and 7. The increase in function exhibited by low-avidity T cells treated with Cy+vaccine+anti-OX40, although significantly increased over mice not receiving OX40 antibody, is not as functional as Cy+vaccine–treated high-avidity T cells (Fig. 4G). However, this increase in IFN-γ secretion, trafficking, and tumor-free survival demonstrates that preventing AICD can enhance the function of low-avidity T cells.

Peripheral deletion occurs when T cells are exposed to a weak TCR signal without the proper costimulation (7). T cells that do not encounter antigen survive without being activated and are nonfunctional, whereas T cells that receive a weak TCR signal are susceptible to peripheral deletion and T cells that receive a strong TCR signal are activated. We confirmed this finding by comparing cell survival between untreated, vaccinated, and OX40-treated mice (Supplementary Fig. S4). Because low-avidity T cells are more likely to be present in the periphery of tumor-bearing mice than high-avidity T cells but were identified as being more susceptible to cell death, we wanted to determine whether the amount of antigen exposure was contributing to the death of low-avidity T cells. To test this possibility, we compared untreated mice with Cy+vaccine–treated tumor-bearing mice where the vaccine provides an antigen source for low-avidity T cells. We found that the percentage of apoptosing T cells was increased in the spleen of tumor-bearing treated mice when compared with non–tumor-bearing untreated mice (Fig. 5A). We also found that giving the Cy+vaccine–treated neu-N mice anti-OX40 antibody reduced the amount of low-avidity T-cell death to levels equivalent to those observed in untreated non–tumor-bearing mice (Fig. 5A). These data suggest that low-avidity T-cell apoptosis is facilitated by the presence of T cell–specific antigens in the context of a T cell–alerting vaccine. In vitro, low-avidity T cells incubated with T2 cells pulsed with increasing amounts of peptide demonstrated that increased peptide stimulation led to increased low-avidity T-cell death (Fig. 5B). We also found that low-avidity T-cell death increases upon proliferation (likely a consequence of the T cell–alerting vaccine), with increased proliferation leading to increased cell death (Fig. 5C). This proliferating population of T cells most likely to apoptose is also the population with the highest expression of proapoptotic proteins and potential to secrete IFN-γ (Fig. 5D and E); both are likely consequences of the vaccine's attempt to activate these T cells. These studies demonstrate that low-avidity T cells are susceptible to apoptosis after vaccine activation, and that the terminally divided group of low-avidity T cells most likely to be functionally active is also the group most likely to express proapoptotic proteins and to be susceptible to cell death. OX40 treatment can rescue this low-avidity T-cell population from cell death, allowing these T cells to sustain functional activity.

To further evaluate the mechanism by which anti-OX40 agonistic antibody augments low-avidity T-cell function, we evaluated whether CD4⁺ T cells are necessary for anti-OX40 activity. A CD4-depletion antibody was used to deplete CD4⁺ T cells in vitro before assessing CD8⁺ low-avidity T-cell function. CD8⁺ low-avidity T cells were isolated from mice treated with Cy+vaccine+anti-OX40 antibody or control antibody as shown in Fig. 4A. CD4⁺ T-cell depletion was verified by flow cytometry (data not shown). CD4⁺ depletion did not affect IFN-γ secretion by low-avidity CD8⁺ T cells (Fig. 6A). To confirm that anti-OX40 treatment led to increased function of low-avidity T cells through increased costimulation of the TNFR (death receptor) pathway, agonistic 41BB-specific costimulation was also tested as it is also a member of this family of death receptor agonists. Cy-treated and vaccinated neu-N mice given anti-41BB also showed increased low-avidity T-cell function as measured by IFN-γ secretion and tumor trafficking (Fig. 6B and C). Anti-41BB was also able to eradicate established tumors (Supplementary Fig. S3). These results demonstrate that the increased rate of death and decreased function seen by low-avidity T cells can be overcome with strong costimulation through the TNFR pathway when a Treg-depleting agent is given with a T cell–activating vaccine.

Our data support four novel findings that link the control of tumor-specific low-avidity T-cell activity to the apoptosis pathway. First, this study shows that low-avidity T cells are nonfunctional and have limited survival in tumor-bearing hosts. Second, nonfunctional low-avidity T cells have increased expression of the proapoptotic proteins DR5, CD24, and FasL that are associated with reduced T-cell survival. Third, expression of these proapoptotic proteins by low-avidity T cells also induces cell death in neighboring high-avidity T cells. Fourth, costimulation of low-avidity T cells with TNFR (death receptor) family member agonists prevents low-avidity T-cell death and enhances low-avidity T-cell function and trafficking into tumors.

We recently reported that adoptively transferred high-avidity T cells require both an antigen-specific vaccination and Treg inhibitors to achieve long-term eradication of progressing murine tumors in neu-N mice (22). That study also reported that low-avidity T cells were ineffective under these conditions. This article demonstrates that this lack of function can be overcome with TNFR agonists, which has important implications for the treatment of patients with cancer.

The TNFR agonists used in this study to improve the function of low-avidity T cells were agonistic anti-OX40 and -41BB antibodies. OX40 is a costimulatory molecule that is able to protect cells from AICD (27–30). Both OX40 and 41BB have been shown in neu-N mice to be able to elicit an antitumor response from endogenous T cells when combined with vaccination of pulsed dendritic cells (31). Hombach and colleagues showed that chimeric antigen receptor (CAR)–engineered CCR7-negative T cells are susceptible to AICD and that combined CD28 and OX40 stimulation rescued the T cells, enabling them to provide a more efficient antitumor response (32). This is in direct agreement with our findings that OX40 agonists can allow nontrafficking tumor-specific T cells to traffic to the tumor. It has also been shown that OX40 can help T cells overcome tolerance and reverse anergy (33, 34). Our previously published work demonstrated an improved antitumor response in neu-N mice when an OX40 agonist is combined with vaccine treatment (33). That study looked at the endogenous population of CD8⁺ T cells as opposed to using adoptively transferred T cells of known avidities as we did in this work. In addition to OX40 agonists, 41BB agonists have also been shown to increase T-cell function and survival (35–38). Hernandez-Chacon and colleagues demonstrated that 41BB agonists could protect melanoma TILs from AICD (36). Although this study again is unrelated to avidity, it supports our study results demonstrating that T cells trafficking into the tumor benefit from 41BB agonists. Eliminating established tumors with OX40 and 41BB was performed to determine the mechanism that increases the function of low-avidity T cells and not to compare the efficacy of 41BB agonists with OX40 agonists. The eventual reestablishment of tumors is not surprising as only one dose was given. Although many studies show the benefits of TNFR agonists on tumor-specific T cells, our study is the first to elucidate a specific mechanism showing a direct effect on low-avidity antigen-specific T cells.

41BB and OX40 antibody treatments have been shown to increase the expression of the antiapoptotic protein Bcl-2 (36, 39). OX40 has also been shown to increase the expression of another antiapoptotic protein, survivin, resulting in T-cell proliferation and expansion (30). These findings are important because they demonstrate the link between the TNFR agonists and the mechanism these death receptors are using to prevent death. As in those studies, our study found that Bcl-2 and survivin expression were increased on low-avidity T cells with OX40 mAb treatment, overcoming the effects of the proapoptotic proteins highly expressed on neu-specific low-avidity T cells. Our data also show that OX40 is not affecting CD8⁺ T cells indirectly through its effects on CD4⁺ T cells, but rather, alters the CD8⁺ T-cell function and trafficking directly through the death receptor pathways.

The discovery that TNFR agonists enable low-avidity T cells to effectively eradicate tumor was a direct result of the major novel finding in this study that low-avidity T cells are ineffective at tumor trafficking and killing due to their early cell death upon activation. This finding is significant because the first step in engaging low-avidity T-cell populations in eradicating tumors is to identify the mechanism(s) that cause them to be nonfunctional. In this study, low-avidity T cells were found to be specifically vulnerable to cell death due to the increased expression of three proapoptotic proteins: DR5, CD24, and FasL. Our study found that TNFR agonists reduce the expression of these proapoptotic proteins. Because of early upregulation of the death receptors upon T-cell activation in our mouse model, we were unable to alter the function of low-avidity T cells with drugs targeting FasL and DR5 alone in Cy+vaccine–treated mice or in conjunction with OX40 treatment (data not shown). Published work has shown that mAb treatment of breast cancer with a DR5 receptor agonist results in decreased tumor burden, which was increased further with additional use of anti-Erbb-2 mAb (40). Therefore, systemic blockage of DR5, although possibly protective of tumor-specific T cells, may inhibit death of tumor cells. Although these death proteins do affect the survival of the T cells, targeting them specifically was ineffective as an anticancer therapeutic intervention. Blocking antibodies against CD24 led to non-CD8⁺ T cell–dependent death of the mice due to possible anaphylaxis when given with vaccine (data not shown). CD24 has also been shown to be involved in apoptosis in B cells and tumor cells as well as T cells, indicating that had CD24-blocking antibody not been lethal in those conditions it may have had side effects counterproductive to tumor elimination (41–44).

The increased expression of FasL on low-avidity T cells that was found by our study is noteworthy because death by ligation of Fas/FasL is the main pathway for AICD, the mechanism for peripheral deletion (45–47). In addition to causing autonomous death of the T-cell itself, FasL expression also causes death in Fas-expressing neighboring T cells (48). Thus, another novel finding by our study was that not only were low-avidity T cells more susceptible to cell death, but the presence of low-avidity tumor-specific T cells led to Fas-dependent cell death of high-avidity tumor-specific T cells. This finding indicates that low-avidity T cells could potentially be regulating high-avidity T cells and be an additional factor contributing to the suppressive environment causing a lack of tumor immunogenicity. These data also suggest that agents that only aim to activate high-avidity T cells may not achieve optimal activity if low-avidity T cells are not also activated. Thus, understanding the different mechanisms of T-cell suppression regulating subpopulations of effector T cells should allow the optimal design of the most effective immunotherapies.

Also important to the clarity of these findings is to demonstrate that low-avidity T cells are dying upon exposure to antigen. We found that low-avidity T cells have an increased rate of death when given Cy+vaccine versus no treatment, in tumor-bearing or nontumor-bearing mice, and that low-avidity T cells have increased death upon exposure to increased amounts of peptide. These data confirm and expand the results of Redmond and colleagues demonstrating increased numbers of surviving T cells when adoptively transferred without peptide transfer (6, 7). Interestingly, our data show that this death is in the population of divided cells, which may express proapoptotic proteins or secrete IFN-γ. These results are significant because they suggest that tumor-specific low-avidity T cells are dying upon exposure to antigen and initial activation signals, which has not been reported. In fact, one study reported that low-avidity CD4⁺ GAD-specific autoreactive T cells were actually less susceptible to AICD (49), but that study was performed under completely different activation conditions than ours.

In conclusion, these studies establish that low-avidity T cells are not inherently ineffective, and that blocking AICD with costimulatory signals through the TNFR (death receptor) pathway enables low-avidity T-cell activation in the presence of a simultaneous antigen-specific signal. In addition, low-avidity T cells committed to AICD may adversely affect the function and survival of other antigen-specific T-cell populations not typically under the control of AICD. These findings support combinatorial immunotherapies that alter the function of multiple effector T-cell populations to achieve optimal immunotherapy for cancer.

E.M. Jaffee has provided expert testimony for Aduro Biotech. No potential conflicts of interest were disclosed by the other authors.

Conception and design: C. M. Black, T.D. Armstrong, E.M. Jaffee

Development of methodology: C.M. Black, T.D. Armstrong, E.M. Jaffee

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.M. Black, T.D. Armstrong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.M. Black, E.M. Jaffee

Writing, review, and/or revision of the manuscript: C.M. Black, T.D. Armstrong, E.M. Jaffee

Study supervision: T.D. Armstrong, E.M. Jaffee

The authors thank the Johns Hopkins Flow Cytometry Core including Ada Tam and Lee Blosser for their technical assistance on this article. The authors also thank the Johns Hopkins Deep Sequencing and Microarray Core including Conover Talbot and Haiping Hao. Dr. Jaffee is the first recipient of the Dana and Albert “Cubby” Chair in Oncology.

This study was supported by R01CA122081 NIH/NCI grant, P50CA062924 NIH/NCI Spore grant in Gastrointestinal Cancer, and P30CA006973 Cancer Center Grant

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.