The Balance Players of the Adaptive Immune System Mads Hald Andersen1,2

Published OnlineFirst February 13, 2018; DOI: 10.1158/0008-5472.CAN-17-3607 Cancer Perspective Research

Abstract

Equilibrium between immune activation and suppression may Vaccination against genetically stable cells with regular HLA be necessary to maintain immune homeostasis, because proin- expression is an attractive way to directly target immunosuppres- flammatory effector T cells (defined as antiregulatory T cells) sive cells in addition to attracting proinflammatory cells into the counteract the functions of regulatory immune cells. These self- tumor microenvironment. Importantly, vaccination toward IDO reactive T cells recognize human leukocyte antigen (HLA)–restrict- or PD-L1 to potentiate such T cells have proven safe, with minimal ed epitopes derived from proteins expressed by regulatory immune toxicity in the clinical phase I trials conducted thus far. cells such as IDO, PD-L1, PD-L2, or arginase. The activation of Significance: Autoreactive effector T cells that specifically rec- such proinflammatory effector T cells offers a novel way to directly ognize regulatory cells might be useful to harness for cancer target the tumor microenvironment, potentially giving them immunotherapy to target the immune-suppressive tumor micro- considerable clinical value, especially in patients with cancer. environment. Cancer Res; 1–4. 2018 AACR.

It was the recognition of suppressor T cells, now called regu- HLA-restricted derived epitopes, which are generated from intra- latory T cells (Treg), pioneered by Dr. Sakaguchi that highlighted cellular degraded antigens, are able to directly eliminate regula- the importance of regulatory cells in the maintenance of immu- tory immune cells (Fig. 1; ref. 3). In addition, anti-Tregs can boost nologic self-tolerance and immune homeostasis (1, 2). However, local immune activation by the secretion of effector cytokines. although a lot of attention has been given regulatory suppressive Thus, anti-Tregs may function as specific first responder helper T cells, new findings suggest that regulatory T cells may also have cells at the site of inflammation. It must be assumed that anti- effector capabilities. We recently reported that the immune system Tregs themselves are hampered by the suppressive effects of their has established a mechanism to counteract the variety of immune- targets. Thus, in immune regulatory networks, anti-Tregs may suppressive feedback signals: self-reactive, proinflammatory suppress the function of other regulatory immune cells and vice T cells that target immunosuppressive cells. Thus, we identified versa. Hence, under normal physiologic conditions equilibrium self-reactive T cells that recognize human leukocyte antigen between immune activation and suppression may be necessary to (HLA)–restricted epitopes derived from proteins, including indo- maintain immune homeostasis. The role of self-reactive effector leamine 2,3-dioxygenase (IDO; refs. 3–7) and PD-L1 (8–12), and suppressor cells in immune-regulatory networks may thus be expressed at inflammation sites in regulatory immune cells. miscellaneous. Professional antigen-presenting cells highly Because these T cells can directly react against regulatory immune express proteins such as PD-L1 and IDO, which are induced by cells, such cells were termed antiregulatory T cells (anti-Tregs; interferons expressed at inflammation sites. In fact, circulating ref. 13). Anti-Tregs may directly suppress the function of regula- IDO- or PD-L1 specific anti-Tregs are present in healthy donors, tory immune cells within immune regulatory networks as well as although detection was not as frequent as detection in patients assisting the adaptive immune response by secreting proinflam- with cancer (3, 6, 9, 10). We have further verified that interferons matory cytokines at inflammation sites. expand populations of IDO-specific T cells by demonstrating To keep the immune balance, regulatory immune cells, for that known IDO inducers (e.g., IFNg) lead to expansion of example, Tregs, different dendritic cell subtypes, myeloid-derived IDO-specific T cells among human PBMCs without additional suppressor cells, and M2 macrophages suppress or terminate stimulation (3). Likewise, we recently found that subcutaneous immune responses. This regulatory arm secures the unrespon- IFNg injections in C57 mice expand populations of PD-L1– siveness or tolerance to self-antigens. Regulatory immune cells specific T cells. If the mice were sacrificed 1 week after IFNg suppress immunity through a number of different cellular and injections, the spleens of the IFNg-treated mice showed a strong extracellular factors. In contrast, specific anti-Tregs recognizing PD-L1– specific T-cell response. Similarly, 3 days of treating C57 mice with the allergen DNFB led to an influx (or expansion) of PD-L1-specific T cells at the inflammation site (in prep.). Further- 1Center for Cancer Immune Therapy (CCIT), Department of Hematology, Copen- more, it has been described that the very common cytomegalo- 2 hagen University Hospital, Herlev, Denmark. Department of Immunology and virus (CMV) induces IDO expression in vivo, which has been Microbiology, University of Copenhagen, Copenhagen, Denmark. suggested to confer an advantage to CMV-infected cells, allowing Corresponding Author: Mads Hald Andersen, Herlev Hospital, Herlev Ringvej 75, them to escape T-cell responses (14). Notably, we found that the 2730 Herlev, Denmark. Phone: 453-868-2602; Fax: 454-453-0176; E-mail: presence of IDO-specific T-cell responses in the periphery was [email protected] associated with the CMV T-cell response (6). It has long been doi: 10.1158/0008-5472.CAN-17-3607 thought that self-reactive T cells harboring T-cell receptors 2018 American Association for Cancer Research. with high affinity to a target/human leukocyte antigen complex

www.aacrjournals.org OF1

Andersen

Figure 1. Regulatory immune cells (e.g., TAMs, MDSCs, Tregs) or cancer cells express high levels of certain proteins, for example, PD-L1, IDO, arginase, Foxp3 (red arrow), and are suppressing the function of immune effector cells in the microenvironment by several different means (red inhibitory arrow bar). These include the upregulation of inhibitory surface receptors and ligands, induction of distinct sets of metabolic enzymes or chemokines, and cytokines that recruit or modify regulatory immune cells. These molecules are transiently induced in normal tissues in response to inflammation or stress but are often hijacked by malignant cells and are constitutively expressed in various cancer tissues, where they create an immunosuppressive microenvironment and contribute to immune evasion of cancer. For example, arginase (ARG1 and ARG2) catalyzes the conversion of arginine. ARG1 is inducible in M2 macrophages, MDSCs, DCs, and granulocytes. ARG-dependent arginine depletion leads to downregulation of the TCRz chain and suppression of the proliferation of effector T cells and natural killer (NK) cells. Likewise, the expression of indoleamine 2,3-dioxygenase (IDO: IDO1 and IDO2) and tryptophan 2,3-dioxygenase (TDO) play a vital role in immune tolerance. Both IDO and TDO catalyze the degradation of the tryptophan to kynurenine. Tryptophan depletion and the accumulation of kynurenine metabolites lead to proliferative arrest of effector T cells as well as the induction and recruitment of Tregs as well as MDSCs. Furthermore, the interaction of checkpoint proteins programmed death-1 (PD-1) and its ligands PD-L1 and/or PD-L2 represent another example of an important immunosuppressive mechanism in the tumor microenvironment. Finally, induction of Treg differentiation and the recruitment, for example, by CCL22 expression, is another way in which malignant cells evade the host's immune system. Anti-Tregs are defined as specific proinflammatory T cells recognizing HLA-restricted derived epitopes, which are generated from proteins expressed by regulatory immune cells (blue arrow). Anti-Tregs react by releasing proinflammatory cytokines and/or granzymes (black arrow) in response to their cognate target cells. Hence, anti-Tregs exhibit cytotoxic activity against both target-expressing cancer cells, including melanoma, breast cancer, colon cancer, and AML as well as normal immune cells like myeloid cells and CD25hi FOXP3þ CD127 Tregs. Furthermore, anti-Tregs can indirectly augment both the effector functions as well as the proliferation of other effector cells. Hence, the activation of anti-Tregs boosts both antiviral immunity, for example, T-cell responses toward CMV or influenza antigens, as well as the response to cancer antigens like Mart-1. Under normal physiologic conditions, equilibrium between immune activation (anti-Tregs) and suppression (regulatory cells) may be necessary to maintain immune homeostasis.

undergo clonal deletion to maintain self-tolerance. However, Yu as metabolic enzymes (e.g., IDO) that restrain the antitumor and colleagues (15) recently demonstrated that clonal deletion activity of antitumor-specific T cells in the tumor microenviron- prunes the T-cell repertoire but does not eliminate self-reactive T- ment. In recent years, our growing knowledge of the factors cell clones. Hence, self-peptide–specific CD8 T cells were present responsible for protecting cancer cells from immune destruction in similar frequencies as those specific for non-self-antigens in the has led to the development of novel, immune-based, anticancer blood of healthy humans. These self-reactive T cells were signif- treatment modalities (16, 17). Indeed, impressive clinical icantly anergic compared with foreign-specific T cells; however, responses have been achieved by characterizing inhibitory T-cell they could be activated by strong activation signals. These self- pathways and targeting them with monoclonal antibodies against reactive T cells may escape thymic selection not only to participate specific membrane proteins (18). The activation of anti-Tregs, e.g., in the fight against pathogens directly but also to provide the by vaccination, may offer a novel very to directly target immune immune system with yet another layer of immune regulators as inhibitory pathways in the tumor microenvironment, modulate anti-Tregs that may contribute to immune homeostasis. Thus, immune regulation, and potentially altering tolerance to tumor both suppressive as well as effector regulatory cells participate in antigens. Thus, if successfully targeted, a therapeutic vaccination the complex network of cells that control the immune system and, approach to activate anti-Tregs can, like the other approaches that consequently, together function as the Balance Players of the target immune suppression (by checkpoint inhibition or small Adaptive Immune System (Fig. 1). molecule inhibitors that target immunosuppressive molecules), Many of the immune regulatory mechanisms considered help- contribute to antitumor immunity by relieving the immune ful in autoimmune settings are used by tumors to suppress suppression and thereby potentiating effective antitumor T-cell immune responses toward malignant cells in cancerous settings. responses. However, unlike other approaches, it could also lead to Hence, various immune-tolerance mechanisms are exploited by epitope spreading of the potential target cells and immunologic cancer cells to achieve immune escape, which becomes more memory (19), because anti-Tregs directly kill their target cells. pronounced with disease progression. Thus, cancer cells as well The first clinical testing of IDO vaccinations was performed as other regulatory immune cells [e.g., tumor-associated dendritic in patients with NSCLC (NCT01543464; ref. 20). Furthermore, cells and myeloid-derived suppressor cells (MDSC)] express two additional trials have recently started to evaluate the safety checkpoint inhibitors (e.g., PD-L1), inhibitory cytokines as well of a vaccine targeting PD-L1–specificTcellsinmultiple

OF2 Cancer Res; 2018 Cancer Research

T-Cell Responses against Regulatory Cells

myeloma (NCT03042793) and a combination vaccine that (29, 30). In a subset of patients with various cancer types, durable targets both IDO- and PD-L1–specific T cells with Nivolumab therapeutic responses can be generated via immune checkpoint in metastatic melanoma (NCT03047928). Finally, industry- blockade using antibodies against cytotoxic-T-lymphocyte– sponsored clinical trials in combination with anti-PD1 anti- associated protein 4 (CTLA4) or PD1/programmed cell death 1 bodies are being initiated in patients with NSCLC (www. ligand 1 (PD-L1). However, checkpoint inhibitors are only iobiotech.com). In all of these trials the vaccinations were efficient in a small fraction of patients with cancer. MDSCs and well tolerated by all patients with no severe toxicity. In the tumor-associated macrophages (TAM) play important roles in NSCLCstudy2patientsreceivedthevaccinationsfor5years tumor immune evasion, and their accumulation in the tumor bed (ESMO 2017, manuscript in preparation). Anti-Tregs are as restrict the accumulation of T cells within the vicinity of cancer described naturally present in vivo without vaccinations and cells. Therefore, these suppressive cells constitute a major reason both PD-L1 and IDO are induced by interferons as a counter- for the limited efficacy of checkpoint blockade in cancer treat- response to the inflammatory response. This provides a mech- ment. The novel understanding of anti-Tregs may lead to a anism that ensures immune homeostasis, which keep anti- translatable strategy for improving the efficacy of checkpoint Tregs in check—therefore, the risk of triggering autoimmune- blockade through the activation of specific T cells that react to related adverse events by vaccination is minimal. This is regulatory cells (including tumor cells) at the tumor site, thereby confirmed in pre-clinical models as mice vaccinated with IDO inducing local inflammation. We hypothesize that an anti-Treg– or PD-L1 have shown no sign of toxicity (unpublished data), activating vaccine would attract T cells into the tumor, thereby which supports the safety data from the clinical studies inducing Th1 inflammation, which would further induce PD-L1 conducted so far. expression in cancer and immune cells, generating targets more Because immune-suppressive cells might antagonize the susceptible to anti-PD1/PDL1 immunotherapy. Indeed, this was desired effects of therapeutic cancer vaccines, the addition of confirmed in pre-clinical models as anti-PD1 and IDO vaccina- anti-Treg antigens would consequently be easily implementable tion show synergy in both C57 and Balb-C mice (in preparation). and highly synergistic. Indeed, in a pre-clinical study, co-stimu- Combinatorial therapy with an anti-Treg–based vaccine and lation of anti–PD-L1 T cells augments T-cell response to a den- checkpoint blockade could be effective in a much wider popula- dritic cell (DC) vaccine (21). In addition to PD-L1 and IDO we tion of cancer patients. and others have identified pro-inflammatory self-reactive T cells The role of anti-Tregs in immune reactions supports a rationale that recognized HLA-restricted epitopes derived from other pro- for the targeting of these T cells in cancer immunotherapy. An teins normally expressed by regulatory immune cells—For exam- obvious risk for such an approach, potential long-term toxicity ple, tryptophan 2,3-dioxygenase (TDO; ref. 22), C-C motif che- due to vaccine-induced autoimmune mechanisms, appears to be mokine 22 (23), forkhead box P3 (Foxp3; refs. 24, 25) and more minimal, illustrated both in mouse in vivo studies and in the recently programmed death-ligand 2 (PD-L2; ref. 26), and Argi- human safety trials conducted thus far. Investigations to address nase (27). Especially, the latter is highly interesting to activate in a some of the most clinically relevant questions are ongoing in therapeutic setting, because arginase-expressing myeloid cells preclinical as well as clinical studies, especially elucidating the contributes to an immunosuppressive tumor microenvironment potential benefits of combination strategies with other therapeu- that prevents effector lymphocyte proliferation (28). Specific tic modalities. targeting of arginase-expressing myeloid cells (e.g., neutrophils, TAMs, and MDSC) could potentially induce T-cell infiltration at Disclosure of Potential Conflicts of Interest the tumor site. The fact that Th1 inflammation signals expand M.H. Andersen is a board member, scientific director, and has ownership IDO- and PD-L1–specific T cells suggests that combination of interest (including patents) in IO Biotech. arginase with IDO and/or PD-L1–based vaccines may work syn- ergistically. In this situation, arginase vaccination could induce Disclaimer Th1 inflammation at tumor sites where regulatory myeloid The funders did not have a role in the writing of the article or the decision to submit the article for publication. cells otherwise prevent infiltration of lymphocytes. This would, in turn, induce IDO and PD-L1, enabling further targeting by Acknowledgments fi PD-L1- and/or IDO-speci c T cells. The combination of vaccine This work was supported by the Danish Cancer Society, the Danish Council epitopes like arginase, PDL1, and IDO would thus be highly for Independent Research, and Herlev Hospital. beneficial and easy to implement in a clinical setting. Immune system suppression plays a major role in cancer Received November 20, 2017; revised December 12, 2017; accepted January progression, with major mechanisms of tumor immune escape 2, 2018; published OnlineFirst February 13, 2018.

References 1. Hori S, Nomura T, Sakaguchi S. Control of regulatory T cell development by 4. Sorensen RB, Kollgaard T, Andersen RS, van den Berg JH, Svane IM, thor the transcription factor Foxp3. Science 2003;299:1057–61. Straten P, et al. Spontaneous cytotoxic T-cell reactivity against indoleamine 2. Takahashi T, Tagami T, Yamazaki S, Uede T, Shimizu J, Sakaguchi N, et al. 2,3-dioxygenase-2. Cancer Res 2011;71:2038–44. Immunologic self-tolerance maintained by CD25(þ)CD4(þ) regulatory T 5. Sorensen RB, Berge-Hansen L, Junker N, Hansen CA, Hadrup SR, Schu- cells constitutively expressing cytotoxic T lymphocyte-associated antigen 4. macher TN, et al. The immune system strikes back: cellular immune J Exp Med 2000;192:303–10. responses against indoleamine 2,3-dioxygenase. PLoS ONE 2009;4:e6910. 3. Sorensen RB, Hadrup SR, Svane IM, Hjortso MC, thor Straten P, Andersen 6. Munir S, Larsen SK, Iversen TZ, Donia M, Klausen TW, Svane IM, et al. MH. Indoleamine 2,3-dioxygenase speciﬁc, cytotoxic T cells as immune Natural CD4(þ) T-cell responses against indoleamine 2,3-dioxygenase. regulators. Blood 2011;117:2200–10. PLoS ONE 2012;7:e34568.

www.aacrjournals.org Cancer Res; 2018 OF3

Andersen

7. Kollgaard T, Klausen TW, Idorn M, Holmgaard RB, Straten PT, Andersen 20. Iversen TZ, Engell-Noerregaard L, Ellebaek E, Andersen R, Larsen SK, MH. Association of a functional indoleamine 2,3-dioxygenase 2 genotype Bjoern J, et al. Long-lasting disease stabilization in the absence of with specific immune responses. Oncoimmunology 2012;1:441–7. toxicity in metastatic lung cancer patients vaccinated with an epitope 8. Ahmad SM, Larsen SK, Svane IM, Andersen MH. Harnessing PD-L1-specific derived from indoleamine 2,3 dioxygenase. Clin Cancer Res 2014;20: cytotoxic T cells for anti-leukemia immunotherapy to defeat mechanisms 221–32. of immune escape mediated by the PD-1 pathway. Leukemia 2014;28: 21. Ahmad SM, Martinenaite E, Hansen M, Junker N, Borch TH, Met O, et al. 236–8. PD-L1 peptide co-stimulation increases immunogenicity of a den- 9. Munir S, Andersen GH, Met O, Donia M, Frosig TM, Larsen SK, et al. HLA- dritic cell-based cancer vaccine. Oncoimmunology 2016;5:e1202391. restricted cytotoxic T cells that are specific for the immune checkpoint 22. Hjortso MC, Larsen SK, Kongsted P, Met O, Frosig TM, Andersen GH, et al. ligand PD-L1 occur with high frequency in cancer patients. Cancer Res Tryptophan 2,3-dioxygenase (TDO)-reactive T cells differ in their func- 2013;73:1674–776. tional characteristics in health and cancer. Oncoimmunology 2015;4: 10. Munir S, Andersen GH, Svane IM, Andersen MH. The immune checkpoint e968480. regulator PD-L1 is a specific target for naturally occurring CD4þ T cells. 23. Martinenaite E, Ahmad SM, Hansen M, Met O, Westergaard MW, Larsen SK, Oncoimmunology 2013;2:e23991. et al. CCL22-specific T cells: modulating the immunosuppressive tumor 11. Munir S, Andersen GH, Woetmann A, Odum N, Becker JC, Andersen MH. microenvironment. Oncoimmunology 2016;5:e1238541. Cutaneous T cell lymphoma cells are targets for immune checkpoint ligand 24. Nair S, Boczkowski D, Fassnacht M, Pisetsky D, Gilboa E. Vaccination PD-L1-specific, cytotoxic T cells. Leukemia 2013;27:2251–3. against the forkhead family transcription factor Foxp3 enhances tumor 12. Ahmad SM, Svane IM, Andersen MH. The stimulation of PD-L1-specific immunity. Cancer Res 2007;67:371–80. cytotoxic T lymphocytes can both directly and indirectly enhance antileu- 25. Larsen SK, Munir S, Woetmann A, Froesig TM, Odum N, Svane IM, et al. kemic immunity. Blood Cancer J 2014;4:230–3. Functional characterization of Foxp3-specific spontaneous immune 13. Andersen MH. Immune regulation by self-recognition: novel possibilities responses. Leukemia 2013;27:2332–40. for anticancer immunotherapy. J Natl Cancer Inst 2015;107:154. 26. Ahmad SM, Martinenaite E, Holmstrom MO, Jorgensen M, Met O, Nastasi 14. Raftery MJ, Wieland D, Gronewald S, Kraus AA, Giese T, Schonrich G. C, et al. The inhibitory checkpoint, PD-L2, is a target for effector T cells: Shaping phenotype, function, and survival of dendritic cells by cytomeg- novel possibilities for immune therapy. Oncoimmunolgy 2017;7: alovirus-encoded IL-10. J Immunol 2004;173:3383–91. e1390641. 15. Yu W, Jiang N, Ebert PJ, Kidd BA, Muller S, Lund PJ, et al. Clonal deletion 27. Martinenaite E, Mortensen RE, Hansen M, Holmstrom MO, Ahmad SM, prunes but does not eliminate self-specific alphabeta CD8(þ) T lympho- Met O, et al. Frequent spontaneous adaptive immune responses towards cytes. Immunity 2015;42:929–41. arginase. Oncoimmunology 2018;7:e1404215. https://doi.org/10.1080/ 16. Topalian SL, Hodi FS, Brahmer JR, Gettinger SN, Smith DC, McDermott DF, 2162402X.2017.1404215. et al. Safety, activity, and immune correlates of anti-PD-1 antibody in 28. LiuY,YuY,YangS,ZengB,ZhangZ,JiaoG,etal.Regulationofarginase cancer. N Engl J Med 2012;366:2443–53. I activity and expression by both PD-1 and CTLA-4 on the myeloid- 17. Pure E, Allison JP, Schreiber RD. Breaking down the barriers to cancer derived suppressor cells. Cancer Immunol Immunother 2009;58: immunotherapy. Nat Immunol 2005;6:1207–10. 687–97. 18. Tumeh PC, Harview CL, Yearley JH, Shintaku IP, Taylor EJ, Robert L, et al. 29. Munn DH, Mellor AL. Indoleamine 2,3-dioxygenase and tumor-induced PD-1 blockade induces responses by inhibiting adaptive immune resis- tolerance. J Clin Invest 2007;117:1147–54. tance. Nature 2014;515:568–71. 30. Maine CJ, Aziz NH, Chatterjee J, Hayford C, Brewig N, Whilding L, et al. 19. Gulley JL, Madan RA, Pachynski R, Mulders P, Sheikh NA, Trager J, et al. Programmed death ligand-1 over-expression correlates with malignancy Role of antigen spread and distinctive characteristics of immunotherapy in and contributes to immune regulation in ovarian cancer. Cancer Immunol cancer treatment. J Natl Cancer Inst 2017;109:2982600. Immunother 2014;63:215–24.

OF4 Cancer Res; 2018 Cancer Research

The Balance Players of the Adaptive Immune System

Mads Hald Andersen

Cancer Res Published OnlineFirst February 13, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-17-3607

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2018/02/13/0008-5472.CAN-17-3607. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.