Immunotherapeutic Response in Tumors Is Affected by Microenvironmental ROS

The tumor microenvironment (TME) is a complex ecosystem commonly including several cell types and structures. Prominent among these are carcinoma-associated fibroblasts (CAF), blood vessels, nerves, and various leukocyte populations. The TME plays an essential but still incompletely understood role in regulating the growth of tumors. The concept of immune surveillance as a biological mechanism to inhibit the growth of tumors was discussed by Ehrlich in the first decade of the 20th century. As our understanding of the immune system and the cancer microenvironment grew, an appreciation of the ability of tumors to evade immune surveillance also developed. Practical immunotherapeutic approaches to counter this immune evasion by disrupting immune checkpoints, notably programmed death-ligand 1 (PD-L1) and its receptor programmed cell death protein 1 (PD-1), predominantly using mAb-based inhibitors are now in routine clinical use and have been effective in some tumor types. However, many forms of cancer are resistant to these approaches, and even in tumor types that can respond well, there are frequent instances of therapy resistance. Therapeutic cancer vaccines are a promising strategy with more limited applications at present, and like checkpoint inhibitors, they require the presence of T cells to elicit tumor killing. The interactions of the tumor epithelium with the cells present in, and recruited to, the TME can positively and negatively regulate tumor growth and progression. In contrast to the tumor epithelium, the cells comprising the TME have generally been shown to be free from major genetic alterations. As such, they have been considered to be potential therapeutic targets. Here, Ford and colleagues suggest an approach of modifying a CAF characteristic [NADPH oxidase 4 (NOX4) expression] to regulate immune cell response and thus sensitize a tumor to immune therapy approaches (1).

In addition to immune surveillance, the adaptive immune system plays an important function in tumor suppression. Cytotoxic CD8⁺ T lymphocytes, which can release perforins to trigger the lysis of tumor cells, are important contributors to this protective mechanism. The presence of large numbers of CD8⁺ and CD4⁺ Th cells at a tumor site represents an “immunologically hot” phenotype likely to respond favorably to immune checkpoint inhibitor therapy. PD-1 is widely expressed in leukocytes including T cells, while its ligand PD-L1 is often expressed on cancer cells (2). Binding of PD-L1 to its receptor results in T-cell exhaustion and dysfunction, thus allowing the tumor to avoid CD8⁺ T-cell–mediated death. PD-1 and PD-L1 antagonists can release CD8⁺ cells in “hot” tumors. However, an effective tumor-killing response to checkpoint release requires both a large number of appropriate effector T cells and the ability of these CD8⁺ and CD4⁺ T cells to interact to mediate a response. Without the appropriate conditions, a checkpoint release will not result in clinically significant tumor cell lysis. The immune system is normally tightly controlled, because excessive action is potentially dangerous. For example, regulatory T cells (Treg) normally act to control disproportionate immune responses and maintain homeostasis in the immune system. In cancer, Tregs can support tumor growth as a consequence of their immunosuppressive actions on effector T cells. Large numbers of Tregs around a tumor are associated with a poor prognosis. Various other mechanisms can also disrupt the leukocytic infiltrate present at specific sites. Myeloid cells, notably CD11b+Gr1+ suppressor cells as well as M2 macrophages, found in tumors can also influence the leukocytic milieu suppressing T-cell infiltrate.

CAFs are an important microenvironmental constituent of many tumors. The term CAF is largely descriptive, indeed even the nature of normal fibroblasts, their origins, and functions are not well determined. There is no equivalent in fibroblast biology to the cell surface markers or understanding of lineage that define the various leukocyte populations. As such, fibroblasts are still largely described by what they are not. This grouping can include mesodermally derived cells of a variety of lineages. It is recognized that CAF are composed of a number of subpopulations, however, the number, origin, and function of these are not well-defined, especially in human tumors. CAF may include cells derived from resident mesenchymal lineages, senescent populations, and potentially the conversion of cells such as bone marrow–derived mesenchymal stem cells (recently reviewed by Sahai and colleagues; ref. 3). Experiments in mouse models of pancreatic cancer demonstrate some specific interactions between subpopulations of CAF and the immune system. For example, it has been shown that a broad-based myofibroblast suppression can decrease immune surveillance (4). Spatial location of the CAF subpopulations is likely also important because a model with a distinctive separation of subpopulations of CAF indicated that one such subpopulation secretes proinflammatory cytokines (5). However, a general understanding of the CAF subpopulations present in human tumors, let alone the assignment of specific functions to define spatial or gene expression clusters, is currently an area of cancer biology that requires further study.

Reactive oxygen species (ROS) include a number of radicals such as superoxide, hydrogen peroxide, and hydroxyl groups, which have become recognized as important molecules for both intra- and extracellular signaling. ROS can have dose-dependent effects with cytostatic levels associated with normal biological processes, while cytotoxic levels can result in the death of cells and the dysregulation of the immune system. The tumor-promoting effects of ROS seem to be tied to supraphysiologic but subtoxic levels of these radicals (6). ROS are generated by a number of mechanisms, notably including the NADPH oxidase (NOX) enzymes. Hypoxia can influence the function of CAF in tumors and indeed ROS can affect the activation phenotype of CAF (7). The carefully controlled production and neutralization of ROS are a crucial part of T-cell function (8). NOX4 is a component of the oxygen sensing mechanism and has been postulated to function in the kidney as an oxygen sensor that regulates the synthesis of erythropoietin in the renal cortex. NOX4, expressed in many cell types including epithelial, endothelial, and mesodermal, is a prodigious generator of ROS, resulting in a potential to generate oxidative stress that has been implicated in several nonmalignant diseases, such as pulmonary fibrosis and atherosclerosis. NOX4 regulates a range of cellular functions including differentiation and senescence and is expressed in many human tumors including bladder, colorectal, renal, and prostate cancers, where it has been linked to stromal activation and CAF-like behavior and phenotype (9). It is noteworthy that NOX4 is not the only NADPH oxidase expressed in tumors. NOX1, NOX2, and NOX5 are all found distributed through many of the most common human malignancies (10). Thus, there are multiple potential sources of ROS generation in tumors. There are, however, also a number of small-molecule inhibitors that have been developed, allowing clinical trials using some of these approaches to be a reasonable hope in the short to medium term.

In this issue, Ford and colleagues examine the role of tumor stromal cells in negatively regulating putative immune therapies (1). They applied xenograft models in immune-intact mice of three different tumor types (lung, colon, and breast carcinomas) grafted subcutaneously with or without fibroblasts either activated by TGFβ exposure or derived from murine tumors. The presence of activated fibroblasts or CAF resulted in the exclusion of CD8⁺ T cells from the grafts, a finding also seen in CAF-rich human head and neck tumors. This CD8 cell exclusion in the mouse models was tied to increased NOX4 expression in the activated fibroblasts and an associated increase in ROS. Suppression of NOX4 using either RNAi or a small-molecule inhibitor resulted in an increase in intratumoral CD8⁺ cells and increased susceptibility to both anticancer vaccination and anti-PD-1 therapeutic approaches in the mouse models.

Important questions remaining to be addressed include whether the CAF found in human tumors mirror the behavior of either endogenous mouse CAF or the more simplified model system of TGFβ-exposed fibroblasts used in some of the in vivo models described. The potential for multiple sources of ROS to offer confounding effects in human disease is also a meaningful consideration for therapy development. Human cancers will continue to prompt new questions but a greater understanding of the role of TME in regulating specific immune cells provides scope for developing new approaches and understanding. Work in this arena will be useful to broaden the use of checkpoint inhibitors and other immunotherapeutic approaches and/or to select the most appropriate patients and to personalize the therapies that they receive.

No potential conflicts of interest were disclosed.

This study was supported in part by NCI through P50 CA180995 SPORE in Prostate Cancer and by the Rob Brooks Fund for Prostate Cancer Care.