The tumor microenvironment in renal cell cancer

Abstract

Purpose of review:

In addition to the provision of nutrients and growth factors that facilitate tumor cell proliferation and metastasis, the tumor microenvironment (MEV) restricts immune surveillance of tumor-associated antigens and limits the efficacy of immune checkpoint inhibitors (ICIs), tumor vaccines, and other immune therapies. This review will focus on the immunosuppressive mechanisms operative within the tumor MVE of renal cell carcinoma (RCC).

Recent findings:

Several of the immunosuppressive mechanisms within the tumor MEV have been identified and are potentially druggable. Clinical trials with agents that target several of these inhibitory pathways are currently underway.

Summary:

Although RCC is one of several tumor types responsive to ICIs, the effectiveness of these agents is likely to be limited by the various tumor-infiltrating bone marrow-derived myeloid cells that comprise the MEV. Several strategies to combat the recruitment of these cells into tumor tissue or to neutralize their immunosuppressive function have shown encouraging results in animal tumor models and clinical trials.

INTRODUCTION

The RCC MEV consists of various lymphoid and myeloid cells, myofibroblasts, and endothelial cells recruited from nearby tissue or derived from precursors that originate in the bone marrow. To a large extent, the MEV of a typical clear cell RCC (ccRCC) resembles that of other tumor types, but there are a few distinguishing features peculiar to RCC – notably the extent of the leukocytic infiltrate and the luxurious, VEGF-driven vasculature. The cells that constitute the RCC MEV not only promote the growth and metastasis of the tumor cells but, through a host of elaborate mechanisms, undermine the ability of the immune system to recognize and eradicate neoplastic cells.

CELLULAR COMPOSITION OF THE RCC MICROENVIRONMENT

Tumor-Infiltrating Lymphocytes:

Perhaps more than any other tumor type, ccRCC are infiltrated with T cells of various phenotype and function [1,2]. This is particularly the case with granzyme A- and perforin-expressing CD8⁺ cytotoxic T cells (CTL), which, according to an analysis carried out by Rooney et al, are more abundant in ccRCC than in any of 17 other tumor types examined [3]. Tregs and Th2 cells are correspondingly underrepresented in ccRCC [4]. Although these histologic findings might be predicted from the documented instances of “spontaneous” regression seen with RCC [5] and the well-known sensitivity of the disease to treatment with IL-2 [6] and immune checkpoint inhibitors [7–9], there are several aspects of ccRCC biology that predict that these tumors should not be especially immunogenic. The response to immune checkpoint inhibitors has been shown to correlate with the mutational burden of the tumor cell [10,11] and the number of mutations in a typical ccRCC is several orders of magnitude less than that observed in melanoma or NSCLC [12], suggesting that ccRCC should have limited expression of novel tumor-specific antigens and at best, a modest immune cell infiltrate. The molecular basis for the robust immune cell infiltration and response to immunotherapy observed with RCC is not entirely clear but may be due to the upregulation of genes associated with antigen processing and peptide loading onto HLA Class I molecules such as TAP1 and TAP2 [13] or antigen presentation such as HLA-A, B, and C, and β₂-microglobulin. Senbabaoglu et al recently compared the expression of these genes in various tumors with that of adjoining normal tissue and observed the greatest variance between the paired tissues in ccRCC tumors [4]. The mechanism by which the genetic aberration of ccRCC (functional loss of VHL) results in the up regulation of the antigen-presenting machinery and the role played by these genes in the regulation of the immune infiltrate remain to be defined.

The nature and extent of the tumor infiltrate has prognostic significance in RCC. In the aforementioned study by Senbabaoglu et al, the ccRCC subset with the greatest accumulation of T cells was found to have the poorest survival [4]. This negative correlation was attributed to the high levels of CTLA-4 and PD-1 expressed on these invading T cells and the possibility that their cytolytic function might be impaired by interaction with other cells within the tumor microenvironment. Infiltration by Th2 cells and Tregs was associated with negative outcomes. On the other hand, the number of infiltrating Th17 cells correlated with improved survival, as did the CD8⁺ T cell/Treg ratio.

Myeloid-derived suppressor cells (MDSC):

In addition to Tregs, tumors are infiltrated to a varying extent with a population of bone marrow- derived myeloid cells (myeloid-derived suppressor cells, MDSC) that serve to promote tumor angiogenesis, metastasis, and to undermine the antitumor effects of infiltrating CTLs. These cells are usually characterized as a subpopulation of activated neutrophils which express high levels of CD66b, CD11b, and VEGFR1 and low levels of the Fc receptor CD16 [14]. They are, however, actually quite heterogeneous and are often classified as neutrophilic, monocytic, or immature based on their morphology and surface phenotype [15,16]. The differentiation of these cells and their accumulation within tumor infiltrates are driven by a number of factors, including cytokines or chemokines produced by the tumor cells [17]. Conditioned media from RCC cell lines, for example, induces the differentiation of monocytes into MDSC [18]. These treated cells have the phenotype and several of the immunosuppressive functions of patient-derived MDSC. They produce nitric oxide and suppress the proliferation of autologous CD3⁺ T cells. All of these acquired features can be partially blocked by inhibitors of GM-CSF or STAT3, suggesting that the production of GM-CSF by the tumor cells is one of the means by which MDSC are induced and recruited by RCC. In a study focused specifically on RCC, the influx of MDSC was found to be correlated with levels of CCL2 (MCP-1), IL-17, and IL-18, all of which were abundantly expressed in tumor tissue and present in the blood of patients [19]. Najjar et al observed a correlation between the infiltration of the neutrophilic type of MDSC in RCC with the expression of IL-1β, IL-8, CXCL5, and Mip-1α [20]. RCC infiltration by the immature type of MDSC was found to correlate with tumor levels of IL-8 and CXCL5. In the murine RCC model Renca, the administration of antibodies against IL-1β or CXCR2 (the receptor for IL-8 and other pro-angiogenic chemokines) was shown to reduce MDSC accumulation, augment tumor infiltration by CD4⁺ and CD8⁺ T cells, and to retard tumor growth, suggesting that tumor cell production of various inflammatory cytokines is an important mechanism by which RCC orchestrate the tumor microenvironment. Finally, the release of the nuclear protein HMGB1 by damaged RCC tumor cells was found to contribute to the differentiation and accumulation of MDSC [21]. Administration of an anti-HMGB1 antibody inhibited tumor growth and prolonged survival in the Renca model. This beneficial effect was absent in mice independently depleted of MDSC by treatment with an anti-Gr-1 antibody.

Tumor-Associated Macrophages (TAMs):

RCC are extensively infiltrated with macrophages of various type [22,23]. These cells support tumor growth and metastatic predisposition through the production of proteases capable of remodeling the extracellular matrix. They also contribute to the general state of immunosuppression of the tumor microenvironment. Although the total number of (CD68⁺) macrophages infiltrating RCC appears to be only weakly predictive of tumor behavior, the abundance of tumor-promoting M2 (CD163⁺, CD204⁺, or CD206⁺) macrophages or the ratio of M2/M1 (CD11c⁺) cells present within the tumor infiltrate has consistently been shown to correlate with poor survival [22–26]. Chevrier et al have observed that the abundance of CD204⁺ (M2) macrophages within RCC correlates with the number of T cells expressing an exhausted or regulatory phenotype [23]. These M2 macrophages were found to express the PD1 ligands CD273 and CD274 and to produce the immunosuppressive cytokine IL-10. Fu et al have shown that these cells undermine the immune response to tumor antigens in part by producing IL-23, which induces Treg proliferation and activates their production of IL-10 and TGF-β [27]. Thus, the accumulation of M2 macrophages within the tumor infiltrate would be expected to contribute, along with Tregs and MDSC, to the immunosuppressive state of the RCC microenvironment.

RCC Microvasculature:

RCC is one of the most highly vascularized of all tumor types. The robust microcirculation of RCC is the result of the massive overproduction of vascular endothelial growth factor (VEGF) and other pro-angiogenic cytokines by the tumor cells and to a lesser extent, the supporting stroma. The generation of VEGF by RCC is in turn due to high levels of the transcription factors hypoxia-inducible factor-1 and −2 (HIF-1 and −2), which mediate the cellular response to hypoxia [28]. In RCC, the high levels of HIF-1 and −2 are the result of the loss of a functional VHL gene, the protein product of which is part of a multi-protein complex that ubiquitinates and destabilizes HIF-1 and −2 when they have undergone oxygen-dependent proline hydroxylation [29, 30]. The loss of VHL through mutation, deletion, or epigenetic silencing ensures that the two HIFs are allowed to accumulate in tumor cells regardless of the availability of oxygen.

Stabilization of HIF-1 and −2 through the loss of VHL is not the only factor contributing to the high levels of these transcription factors in RCC. Mathew et al have shown that RCC have low levels of the microRNAs miR-30–2-3p and miR-30a-3p compared to the normal renal parenchyma [31,32]. These microRNAs interfere with the translation of the HIF mRNAs and their absence from RCC cells allows for the HIF transcripts to be efficiently translated. This effect, coupled with the inherent stability of the HIF proteins due to the VHL loss, guarantees the abundance of these transcription factors in RCC cells. Another genetic quirk of RCC is the consistent loss of the glycolytic enzyme fructose-1,6-bisphosphatase (FBP1)[33]. This enzyme has recently been shown to bind to and inhibit the nuclear function of HIF-1 and −2 through a mechanism that is completely independent of its catalytic function. The absence of FBP1 therefore not only promotes glycolysis (the Warburg effect) but ensures the function of HIF-1 and −2 in RCC cells. The downstream consequence of the loss or down modulation of VHL, the miR-30 micro-RNAs, and FBP1 is to maintain high levels of HIF transcription factors and the expression of proangiogenic cytokines such as VEGF. This ensures a robust microcirculation within the tumor but also creates a unique vulnerability to the effects of VEGF receptor inhibition. Indeed, RCC was the first tumor type for which drugs that block VEGFR signaling were approved by the FDA as single agents [34–36].

The microvasculature within RCC consists of several types of microvessels that differ with respect to their degree of differentiation. Qian et al have analyzed the morphology and phenotype of the RCC endothelium and have identified differentiated vessels, which express both CD31 and CD34, and a less differentiated type of vessel in which the endothelial cells express only CD31 [37]. The latter tend to have small lumens and no pericyte coverage and are located primarily within the centers of tumor deposits. The more differentiated vessels are covered with pericytes and are more abundant in the tumor periphery. The undifferentiated vessels are particularly abundant in high grade RCC and their presence is associated with short patient survival [38].

Similar studies have been carried out by investigators in the Dvorak laboratory using an adenovirus engineered to express VEGF-A [39–42]. Implantation of this virus triggers an influx of vessels that resemble those found in RCC. Sitohy et al and Nagy et al have identified six distinct vessel types that develop at the site of virus-induced VEGF production, including “mother vessels”(MV) that emerge early from preexisting venules [39–41]. These mother vessels evolve into glomeruloid microvascular proliferations (GMP) and later into capillaries and microvascular malformations (VM). Subsequent remodeling gives rise to feeding arteries (FA) and draining veins (DV) similar to those found in RCC. The early MV and GMP express high levels of VEGFR-2 and are particularly sensitive to the inhibition of VEGF signaling from the administration of the VEGF trap aflibercept, for example. Late-appearing capillaries, VM, FA, and DV on the other hand ultimately lose their initial dependency on VEGF and are unresponsive to VEGF inhibition. This loss of dependency on VEGF, and the persistence of all but two of the microvessel types despite the inhibition of the growth factor that triggered the development of vessels in the first place, are thought to limit the efficacy of agents that block VEGFR signaling in the treatment of patients with RCC.

Wragg et al have compared the gene expression profiles of tumor-associated with nonmalignant endothelium from a variety of tumor types and have identified melanoma cell adhesion molecule (MCAM) and its extracellular matrix ligand laminin-α4 (LAMA4) as consistently overexpressed in the microvasculature of RCC [43]. These microarray data were confirmed by immunohistochemistry. MCAM expression is induced by exposure to VEGF and its expression in RCC was linked to poor patient survival. Given its specificity for the microvasculature of RCC, the authors suggest that MCAM might be therapeutically targeted by monoclonal antibodies as part of a treatment strategy for this disease.

MECHANISMS OF IMMUNOSUPPRESSION: STRATEGIES TO OVERCOME RESISTANCE TO IMMUNOTHERAPY

The tumor MEV limits the antitumor activity of invading T cells through a variety of mechanisms. Certainly one constraining factor is the microenvironmental depletion of key nutrients essential for optimal immune function such as glucose and glutamine, both of which are avidly taken up by proliferating tumor cells. Glutamine is essential for the expression of the T cell transcription factor Tbet and the differentiation of IFN-γ-producing TH1 cells and in its absence, CD4⁺ T cells differentiate into Foxp3⁺ Tregs [44]. Although inhibitors of the enzyme glutaminase such as CB-839 have multiple effects on tumor cell metabolism, oxidative stress, and DNA methylation that might contribute to their antitumor activity, their ability to increase glutamine levels in the MEV and reverse some aspects of immune dysfunction may be an important component of their overall antitumor effect [45].

The amino acid arginine is often lacking in the MEV extracellular fluid due to its uptake by MDSC and the secretion of arginase by these cells [14]. MDSC express one of the isoforms of the enzyme nitric oxide synthase (iNOS) that generates NO from arginine. This free radical directly nitrates proteins such as STAT1, rendering the T cells incapable of responding to antigen-presenting dendritic cells [46]. Several investigators have conjectured that NO production by tumor-infiltrating MDSC is one of the dominant mechanisms by which the tumor MEV undermines immune function.

RCC cells are generally unable to generate their own arginine due to a lack of the enzyme arginosuccinate synthetase-1 [47]. One might therefore expect that microenvironmental arginine depletion would limit RCC growth. The net effect of the arginase secretion, however, turns out to be tumor-promoting due to its inhibitory effect on immune function. In the absence of arginine, T cells accumulate uncharged tRNA, which activates the eIF-2α kinase GCN2, which in turn arrests protein synthesis. Certain elements of the T cell signaling apparatus – notably the TCR-ζ chain, the kinase p56^lck, and components of the NF-κB pathway – are particularly affected by eIF-2α phosphorylation, and the down modulation of these molecules profoundly inhibits T cell function [48]. Recently, inhibitors of arginase (e.g. INCB001158) have entered clinical trials as adjuncts to immune checkpoint inhibitors [49], and it is hoped that they might improve the efficacy of these agents.

Metabolism of the amino acid tryptophan by the enzyme indoleamine dioxygenase (IDO) not only depletes the MEV of this critical amino acid but results in the accumulation of metabolites (kynurenines) that are particularly toxic for T cells [50]. Although IDO has been shown to mediate immunosuppression in numerous animal tumor models, the extent to which the tryptophan-kynurenine metabolic pathway disrupts immune function in the MEV of human cancer remains unclear [51]. Another immunosuppressive compound, adenosine, is present in the MEV extracellular fluid as a result of the action of the ectonucleotidase CD73, which is expressed by many tumor cells. This nucleotide metabolite suppresses immune function through the engagement of its purinergic receptor [52], and studies are currently underway to determine if inhibition of this pathway is able to restore immune function and augment the efficacy of immune checkpoint inhibitors.

Several of the strategies to improve the efficacy of immune checkpoint inhibitors currently under investigation involve the combination of an anti-PD1 antibody with a drug designed to block the differentiation of MDSC or their trafficking to tumor sites [53–55]. These potentially important agents include the histone deacetylase inhibitor entinostat, which reduces arginase and NO production by MDSC [56], as well as STAT3 inhibitors, which block the differentiation and function of MDSC [57]. Drugs that selectively inhibit PI3K-γ, the PI3K isoform needed for MDSC to suppress immune function [58], are also being paired with anti-PD1 antibodies in early clinical trials. In a similar vein, agents that block the chemokine CSF-1 (e.g. cabiralizumab) have been combined with nivolumab in an effort to exclude MDSC from the tumor MEV in RCC and other malignancies [59].

Finally, it is worth mentioning the various VEGF- or VEGFR-targeted drugs that have recently been shown to augment the efficacy of anti-PD1 or –PDL1 antibodies in the treatment of RCC [60–62]. VEGF is known to interfere with dendritic cell function and to induce the expression of FasL on tumor endothelium [63], which should promote apoptosis in T cells in the act of extravasation. VEGFR blockade would be expected to reverse these effects. In addition, VEGF has been shown to reduce STAT3 signaling in MDSC and to limit the trafficking of these cells into tumor tissue [53,64,65]. It is therefore conceivable that the ability of these agents to enhance the efficacy of immune checkpoint inhibitors is at least partly mediated through an immunological mechanism involving MDSC within the RCC MEV independent of their anti-angiogenic effects.

CONCLUSION

ccRCC are extensively infiltrated with regulatory T cells, myeloid, and other cell types that are capable of suppressing immune function through mechanisms completely independent of PD1. These cells constitute a tumor MEV that limits the effectiveness of immune surveillance and the activity of immune checkpoint inhibitors. Several of the mechanisms by which these cells inhibit tumor eradication by CTL and NK cells have been characterized, and drugs designed to block their function are currently being tested as adjuncts to immune checkpoint inhibitors. It appears likely that the future management of ccRCC will entail the use of a combination of drugs with complementary effects on the immune system.

• Clear cell RCC is one of several cancer types shown to respond to immune checkpoint inhibitors.

• The ccRCC microenvironment is extensively infiltrated by Tregs and various bone-marrow-derived cells (MDSC, TAMs) that suppress the immune system independent of the interaction between PD1 and its ligands.

• Several mechanisms by which these cells interfere with immune surveillance - e.g. amino acid depletion, NO and kynurenine production – have been identified.

• Pharmacologic agents that reduce the trafficking of these cells into tumor infiltrates or block their immunosuppressive function have been developed and are in clinical trials as adjuncts to immune checkpoint inhibitors.