Exploiting T cell co-inhibition to delay autoimmune disease recurrence

Type 1 diabetes (T1D) results from T cell-mediated destruction of insulin-producing pancreatic β cells. Individuals with long-term disease are at risk of developing life-threatening complications. β cell replacement is a therapy for T1D but is limited by recurrent autoreactive T cell targeted β cell death. Thus, β cells better equipped to inhibit local T cell responses may survive longer in autoimmune recipients. Programmed-death 1 (PD-1) signaling through its ligand PD-L1 inhibits T cells, and may serve as a prominent defense in T1D. Using flow cytometric analysis, in the absence of T cells in NOD.RAG−/− mice we do not detect β cell PD-L1 expression. However, with T cells, we observed an increased proportion of β cells expressing PD-L1 in female non-obese diabetic (NOD) mice which had not developed diabetes. In addition, the majority of remaining live β cells at diabetes onset in NOD mice continue to express high levels of PD-L1. These three situations suggest that islet β cells may increase PD-L1 expression as a last line of defense to limit infiltrating T cell mediated destruction. To manipulate β cell PD-L1 expression prior to transplantation, we screened a panel of diabetes-related cytokines and found that IFN-γ enhances β cell PD-L1 expression. Unfortunately, islet transplant survival was not prolonged, which we hypothesized was due to enhanced MHC class I expression, facilitating CD8+ T cell-mediated killing. We therefore de-coupled PD-L1 from enhanced MHC I expression. Using this approach, enforced β cell PD-L1 expression delays disease recurrence. These data support our hypothesis that β cells expressing T cell co-inhibitory molecules, like PD-L1, can locally inhibit autoreactive T cells which may prevent transplant destruction.

Abstract PR05: Mechanisms of tumor response and resistance to radiation and dual checkpoint blockade in mice and patients

Immune checkpoint inhibitors result in impressive clinical responses but optimal results will require combination with each other and other therapies. This raises fundamental questions about mechanisms of non-redundancy and resistance. Here, we report major tumor regressions in a subset of patients with metastatic melanoma treated with an anti-CTLA4 antibody (anti-CTLA4) and radiation (RT) on a phase one clinical trial and reproduced this effect in mouse models. Although combined treatment improved responses in irradiated and unirradiated tumors, resistance was common. Computational analysis of genome-wide and immune profiles of mice revealed resistance was due to T cell exhaustion driven by adaptive resistance and prolonged interferon-gamma exposure, resulting in STAT1-mediated upregulation of PD-L1 on melanoma cells and tumor macrophages. Accordingly, optimal response in melanoma and other cancer types requires RT, anti-CTLA4, and anti-PD-L1/PD-1. Anti-CTLA4 predominantly inhibits T regulatory cells, and macrophage depletion and/or PD-L1 blockade reverses T cell exhaustion. RT promotes the infiltration of intratumoral antigen-specific CD8 T cells and enhances the diversity of the T cell receptor (TCR) repertoire. RT with dual checkpoint blockade shapes the TCR repertoire of the expanded peripheral clones in a manner consistent with antigen-driven selection. Similar to results from mice, patients on our clinical trial with tumors showing high PD-L1 did not respond to RT + anti-CTLA4, demonstrated persistent T cell exhaustion, and rapidly progressed. In contrast, patients with low PD-L1 on melanoma cells or macrophages had markedly improved survival, with the best survival observed among those patients with low PD-L1 on both cell types. Thus, our results suggest that 1) RT can enhance response to anti-CTLA4 when the TCR and/or antigen repertoire are sub-optimal, 2) upregulation of PD-L1 through STAT1-mediated adaptive resistance mechanisms inhibits response to anti-CTLA4-based therapy unless PD-L1/PD-1 is blocked, and 3) the combination of RT, anti-CTLA4, and anti-PD-L1 promotes response and immunity through distinct mechanisms. Finally, although PD-L1 was a dominant resistance mechanism in our models, PD-L1-independent resistance mechanisms were also evident. The next generation of clinical trials based on these findings are underway.

Citation Format: Christina Twyman-Saint Victor, Andrew Rech, Joseph Benci, Amit Maity, Ramesh Rengan, Kristen Pauken, Erietta Stelekati, Bihui Xu, Hannah Dada, Pamela Odorizzi, Ramin Herati, Ravi Amaravadi, Lynn Schuchter, Hemant Ishwaran, Rosemarie Mick, Daniel Pryma, Xiaowei Xu, Michael Feldman, Tara Gangadhar, Steve Hahn, John Wherry, Robert Vonderheide, Andy Minn. Mechanisms of tumor response and resistance to radiation and dual checkpoint blockade in mice and patients. [abstract]. In: Proceedings of the Fourth AACR International Conference on Frontiers in Basic Cancer Research; 2015 Oct 23-26; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2016;76(3 Suppl):Abstract nr PR05.

Abstract A036: Anti-PD1 therapy and CD8 T cell invigoration in metastatic melanoma

Antibodies targeting the PD-1 pathway can reverse T cell exhaustion resulting in re-invigoration of immune responses. Anti-PD1 therapies such as pembrolizumab have been shown to be efficacious across a broad range of human cancers including melanoma with clinical responses correlating with immune changes in the tumor microenvironment. However, our understanding of the immune mechanism of anti-PD1 therapy in humans remains incomplete. Specifically, there is relatively little information about changes in the differentiation, activation and reversal of exhaustion in peripheral blood CD8 T cells and correlation with clinical outcome. We collected peripheral blood at serial time points before and after pembrolizumab therapy on 39 consecutive patients with Stage IV melanoma and analyzed changes in T cell subsets and differentiation using 16 parameter flow cytometry. Here we show that pembrolizumab treatment results in increases in the peripheral blood CD8/Treg ratio, as well as invigoration of CD8 T cells. Focusing on key CD8 T cell subsets expressing combinations of T-bet, Eomes, PD-1 and other inhibitory receptors has allowed greater focus on populations responding to re-invigoration by pembrolizumab treatment. This invigoration can be demonstrated by upregulation of granzyme B (GzmB) and Ki67 in key CD8 T cell subsets. An early increase in GzmB+Ki67+ cells translates into a later increase in GzmB+ cells suggesting an early wave of proliferation giving rise to a pool of re-invigorated CD8 T cells. T cell subsets that express markers of exhaustion including T-bet, Eomes, and inhibitory receptors may represent populations that are invigorated by anti-PD1 therapy. These changes may also correlate with clinical outcomes and therefore may be useful as a biomarker of response. These results show that T cell responses to pembrolizumab treatment can be tracked in an easily accessible peripheral blood compartment during therapy. The kinetics of immune response in key CD8 T cell subsets may inform us of optimal treatment duration as well as the nature and timing of cancer response to therapy.

Citation Format: Alexander Huang, Wei Xu, Shannon Harmon, Felix Quagliarello, Ramin Herati, Kristen Pauken, Bertram Bengsch, Lynn Schuchter, Ravi Amaravadi, Suzanne McGettigan, Tara Gangadhar, John Wherry. Anti-PD1 therapy and CD8 T cell invigoration in metastatic melanoma. [abstract]. In: Proceedings of the CRI-CIMT-EATI-AACR Inaugural International Cancer Immunotherapy Conference: Translating Science into Survival; September 16-19, 2015; New York, NY. Philadelphia (PA): AACR; Cancer Immunol Res 2016;4(1 Suppl):Abstract nr A036.

Identification of self-specific CD4+ T cell subsets resistant to PD-1 blockade (BA11P.130)

The inhibitory receptor programmed death (PD)-1 and its ligand PD-L1 regulate T cell function to limit autoimmunity. PD-1 is highly expressed on exhausted T cells, limiting their antiviral or antitumor activity. Even though PD-1 pathway blockade has gained momentum in cancer treatment, it is unclear how PD-1/PD-L1 blockade impacts different T cell subsets. We examined the effects of PD-1 blockade on self-reactive CD4+ T cells in mouse models of varying autoimmune susceptibilities. We used insulin-peptide/MHC Class II tetramers to track endogenous insulin-specific CD4+ T cells in diabetes-prone (NOD), and diabetes-resistant (NOR and B6.g7) mice. PD-1 was expressed at lower levels on insulin-specific cells in diabetes-resistant mice compared to cells from NOD mice. Moreover, PD-1 pathway blockade accelerated autoimmunity in all NOD mice and 30% of NOR mice, but B6.g7 mice remained protected from disease. By transferring congenically marked islet-specific CD4+ T cells into pre-diabetic NOD mice, we asked whether anti-PD-L1 reinvigorates anergic cells. Surprisingly, anergic cells remained functionally blunted compared to effector cells after anti-PD-L1, suggesting that established anergic cells were not dependent on PD-1 signaling to maintain their unresponsive state. This work highlights how the differentiation status of a T cell may predetermine its susceptibility to PD-1 blockade, and influence autoimmunity or antitumor responses that ensue.

Antigen specific tolerance and the prevention of autoimmune type 1 diabetes occurs through indirect antigen presentation and requires programmed death ligand 1 (THER5P.918)

Antigen-coupled ethylene carbodiimide-fixed cells is an attractive therapeutic platform for treating human patients with type 1 diabetes since these cells can prevent and reverse diabetes in the non-obese diabetic mouse model of diabetes. However, the mechanisms contributing to the induction of T cell tolerance in vivo following treatment are unclear. Therefore, we used the BDC2.5 T cell transfer model to study how p31-coupled cells inhibit T cell function in vivo. Our results demonstrate that indirect antigen presentation was critical for tolerance induction in vivo. We further demonstrate that PD-L1 on the host antigen presenting cells is required for tolerance induction. Importantly, using two-photon microscopy we show that BDC2.5 T cells arrest on CD11c+ dendritic cells during the induction of tolerance, suggesting that these cells are responsible for presenting coupled cell antigen and tolerance induction. To increase the feasibility of translating antigen coupled cells to the clinic, we demonstrate that whole blood preparations and frozen cells are sufficient for tolerance induction, a significant concern for donor coupled cell sources when considering translation to human patients. These data further our understanding of how antigen-coupled cells induce tolerance and how this tolerance protocol could be altered to increase clinical feasibility.

Development of vaccination approaches that reverse leukemia-induced tolerance and reduce leukemic burden (TUM9P.1020)

Current therapies induce only transient responses in patients with BCR-ABL+ B Acute Lymphoblastic Leukemia. Thus augmenting immune responses to the BCR-ABL fusion peptide could be therapeutically beneficial. The fusion between BCR and ABL encodes a neo-antigen (called BAp) that stimulates an immune response, but this response does not eliminate leukemia. To characterize the immune response to BAp we examined BCR-ABL+ leukemic cells in mice. Principal Component Analysis demonstrated that a CD4+ T cell response driven by high CD40 and low PDL1 on leukemic cells inversely correlated with leukemic burden. Using BAp:I-Ab-specific tetramers we tracked BCR-ABL-specific CD4+ T cells in response to leukemia. Naive mice had a small population of BAp:I-Ab-specific T cells. Transfer of live leukemia resulted in modest expansion (55-fold) of BAp:I-Ab-specific T cells, which were largely diverted into FOXP3+ Tregs. Depletion of Tregs led to increased expansion of BAp:I-Ab specific T cells and protected mice from leukemia. Likewise, acute viral vaccination in the presence of BAp (but not infection with Listeria monocytogenes expressing BAp) induced expansion of Ly6C+ BAp:I-Ab-specific T cells, prevented Treg induction, and significantly reduced leukemic burden. Thus, BCR-ABL-specific T cells exist in non-leukemic mice but are tolerized in the presence of leukemia; this tolerance can be broken by Treg ablation or specific vaccination approaches, leading to protection against leukemia.

Resident memory CD8 T cells are master orchestrators of protective innate and adaptive immune responses (MUC7P.765)

CD8 T cells eliminate intracellular infections through direct cytolysis, making them ideal candidates for vaccine design. Recent studies have shown that resident memory CD8 (TRM) are critical in mediating early local pathogen clearance after re-infection. However, given the numerical paucity of CD8 TRM relative to the total number of potential target cells, it is unclear how CD8 TRM could rapidly clear pathogen strictly through direct cytolysis. We therefore hypothesized that T cell activation precipitated local polyfunctional cytokines, which were important mediators of protection after local re-infection. Here, we demonstrate that after local antigen re-encounter, CD8 TRM reactivation within the female reproductive tract (FRT) elicited secretion of multiple cytokines, which induce rapid functional changes in multiple leukocytes. During this recall response, local secretion of IL-2 and IL-15 upregulated granzyme B in local lymphocytes, TNFα matured tissue dendritic cells, and IFNγ recruited circulating memory CD8 T cells and B cells to the FRT. Thus, CD8 TRM cells employ multiple polyfunctional cytokines to orchestrate diverse and parallel responses to infection, modulating both circulating and local leukocytes. Future studies will continue to dissect apart the diverse mechanisms CD8 TRM use to clear pathogen. Vaccines should therefore aim to establish CD8 TRM cells for efficacious adaptive and innate immune responses to pathogen.

Indirect presentation of antigen-coupled cells mediates specific CD4+ T cell tolerance by PD-L1 for the prevention of type 1 diabetes (BA14P.207)

Antigen coupled to the surface of cells using ethylene carbodiimide is a potent inducer of tolerance in multiple models of autoimmunity including Type 1 Diabetes (T1D). Yet, the mechanisms contributing to the induction of T cell tolerance in vivo during T1D are unclear. We used the BDC2.5 T cell receptor transgenic T cell transfer model to study how p31-coupled cells inhibit T cell function in vivo. Antigen-coupled cells that were either deficient in MHC class II or MHC-mismatched induced tolerance similarly to MHC class II-sufficient matched cells, suggesting that indirect presentation of coupled cell antigens is required for tolerance induction in vivo. We also show that BDC2.5 T cells predominantly encounter coupled cell antigen that has been ingested by host antigen presenting cells, and that the inhibitory receptor PD-L1 is required on the host but not the coupled cells to induce tolerance, further highlighting an indirect mechanism of tolerance induction. These data add to our current understanding of how antigen-coupled cells induce tolerance and aid in the translation of this therapeutic into the clinic.

Requirement for PD-1 in the induction and maintenance of tolerance differs depending on tolerogenic stimulus (BA8P.118)

The inhibitory receptor Programed Death (PD)-1 interacting with PD-L1 is important for tolerance since blocking this pathway accelerates autoimmunity. However, the mechanisms by which PD-1/PD-L1 contribute to tolerance remain unclear. In this study we tracked endogenous insulin-specific CD4 T cells in mice that are susceptible (NOD) or resistant (B6.g7) to Type 1 Diabetes (T1D). Despite the presence of insulin-reactive CD4 T cells in both strains, anti-PD-L1 precipitated T1D in NOD but not B6.g7 mice. Accelerated T1D in NOD mice correlated with increased numbers of insulin-specific CD4 T cells in the spleen, pancreatic LN, and pancreas. The number of cells increased in the pancreatic LN of B6.g7, but not the spleen or pancreas, suggesting that loss of PD-1 signaling was insufficient to promote trafficking to the pancreas in these mice. Unexpectedly, the breakdown of self tolerance in NOD mice did not rely on PD-1 signals as anergic cells did not change with PD-1 blockade. Rather, PD-1 was more important for regulating effector T cells. This contrasts data using antigen-coupled ECDI-fixed cells to induce anergy in potently activated effector cells during T1D in NOD, where PD-1/PD-L1 ligation is required for both the induction and maintenance of tolerance. This work reveals the role PD-1 signaling has for different contexts of tolerance, and has important therapeutic implications for use of PD-1 inhibitors in patients.

Role of the PD-1 pathway in regulating islet-reactive CD4 T cells during autoimmune diabetes (123.39)

Engagement of the inhibitory receptor PD-1 with its ligand PD-L1 is critical for preventing Type 1 diabetes in non-obese diabetic (NOD) mice. The mechanism by which PD-1 signaling limits self-reactive CD4 T cells is not well understood. To investigate this we transferred a low number of self-reactive BDC2.5 T cell receptor transgenic CD4+ T cells into prediabetic NOD mice. These cells became activated due to the presence of endogenous antigen. Following activation we administered anti-PD-L1 antibody to block PD-1 signaling and determined the affect on the BDC2.5 cells. PD-1 blockade resulted in an increase in BDC2.5 T cells in the spleen, pancreatic lymph node, and pancreas. This increase correlated with an increase in proliferation in the secondary lymphoid organs measured by CFSE dye dilution. However, it is unclear whether the increase in the pancreas was due to increased proliferation or enhanced recruitment. To investigate the requirement for PD-1 and/or PD-L1 on effector CD4+ T cells, we utilized genetically deficient BDC2.5 T cells. We determined that PD-1 on the BDC2.5 T cell was required to suppress proliferation and recruitment to the pancreas, whereas PD-L1 on the BDC2.5 T cell was dispensable. These data suggest that PD-1 is required on the autoreactive CD4+ T cells and under normal conditions will suppress proliferation in the secondary lymphoid organs and limit recruitment to the pancreas in a cell intrinsic manner.

Programmed death-1 promotes tolerance by blocking the T cell receptor-induced stop signal. (167.26)

Programmed death-1 (PD-1) and CTLA-4 are inhibitory receptors expressed on activated T cells that are critical for preventing autoimmunity. However, the biological context in which PD-1 and CTLA-4 control T cell tolerance is not fully understood. In this study we used an antigen specific diabetes model and T cell dynamic motility during the induction and maintenance of tolerance in vivo by two-photon imaging. Using this system we demonstrate that unlike naive or activated islet antigen-specific T cells, tolerized islet antigen-specific T cells moved freely and did not swarm around antigen-bearing dendritic cells (DCs) in pancreatic lymph nodes. However, blocking interactions between PD-1 and PD-L1 altered the dynamic T cell motility and caused the breakdown in peripheral tolerance resulting in autoimmune diabetes. Neutralizing CTLA-4, on the other hand, did not impact tolerized T cell motility in vivo or clinical disease. The ability of PD-1 to influence the T-cell stop signal was antigen dependent, since T cell motility and T cell-DC contact times were unaffected by PD-L1 neutralization in non-draining (inguinal) LNs. These data provide in vivo evidence that PD-1 -PD-L1 interactions are critical for the maintenance of peripheral tolerance and prevention of autoimmunity. Further, the effects of PD-1 are fundamentally distinct from those of CTLA-4 and suggest unique mechanisms of action for these two inhibitory receptors.