Inhibition of arginase by CB-1158 blocks myeloid cell-mediated immune suppression in the tumor microenvironment

Myeloid cells in the tumor microenvironment (TME) are associated with poor prognosis across multiple types of cancer, including lung, colorectal, and breast [1]. Tumor-infiltrating myeloid cells contribute to an immunosuppressive TME through multiple mechanisms, constraining anti-tumor immunity and hindering immunotherapy [2, 3]. Agents that aim to block myeloid cell-mediated immunosuppression are currently in pre-clinical and clinical development, however there are no approved therapies specifically directed against tumor-associated myeloid cells [4, 5].

The major populations of tumor infiltrating myeloid cells include tumor-associated macrophages (TAMs), myeloid-derived suppressor cells (MDSCs), and granulocytes [2, 6, 7]. A feature common to all of these immunosuppressive cells is their expression of the enzyme arginase 1 (Arg1) [8]. Arg1 catalyzes hydrolysis of the amino acid L-arginine to produce urea and L-ornithine, thereby depleting extracellular L-arginine [9]. T cells are auxotrophic for L-arginine, requiring the amino acid for the rapid and successive rounds of proliferation that follow T cell antigen receptor (TCR)-dependent activation of effector cells [10–12]. In some inflammatory settings, myeloid-mediated arginine depletion is essential for suppressing excessive T cell proliferation [13]. Blocking Arg1 activity in the context of cancer could therefore shift the balance of L-arginine metabolism to favor lymphocyte proliferation. Indeed, in murine studies, injection of the arginase inhibitor nor-NOHA or genetic disruption of Arg1 in the myeloid compartment resulted in reduced tumor growth, indicating that Arg1 is pro-tumorigenic [14, 15]. Thus, pharmacological inhibition of Arg1 is a compelling therapeutic strategy for the treatment of cancer.

Here we describe CB-1158, a potent and orally-bioavailable small-molecule inhibitor of Arg1. In T cell co-cultures, CB-1158 reversed myeloid cell-mediated immunosuppression and restored T cell proliferation. In murine syngeneic tumor models, CB-1158 shifted the tumor immune landscape toward a pro-inflammatory TME, resulting in tumor growth inhibition. CB-1158 augmented the efficacy of other anti-cancer agents, including gemcitabine, antibodies to immune checkpoints, adoptive T cell therapy, and adoptive NK cell therapy, to inhibit tumor growth. The therapeutic potential of targeting Arg1 was further supported in a screen of cancer patient samples that revealed an abundance of Arg1-expressing myeloid cells in tumors and high amounts of Arg1 in plasma. CB-1158 is currently in clinical trials for patients with solid tumor malignancies (NCT02903914).

CB-1158 was next tested for the ability to inhibit native Arg1 in lysates of human granulocytes, peripheral blood erythrocytes, and primary hepatocytes [22–24]. We found that CB-1158 inhibited native arginase in lysates with similar potency to that observed for recombinant arginases (Table 1). In addition, Arg1 protein and activity have been reported to be elevated in the plasma of cancer patients compared to healthy donors [25, 26] and we observed inhibition of arginase activity by CB-1158 in plasma isolated from cancer patients (Table 1).

To determine if CB-1158 can restore lymphocyte proliferation in the context of immunosuppressive arginase-expressing myeloid cells, we first confirmed that lymphocytes require exogenous L-arginine to proliferate. Purified T cells or NK cells were stimulated with anti-CD3/anti-CD28 or IL-2, respectively, in the presence or absence of L-arginine in the media. Proliferation of human and murine T cells (Fig. 1a, left) and NK cells (Fig. 1a, right) only occurred in media that contained L-arginine, as expected [10, 11, 28, 29].

To determine if arginase activity is necessary for myeloid cell-mediated suppression of T cell proliferation, T cell proliferation was assayed in co-culture with human myeloid cells in the presence or absence of CB-1158. Granulocytes are an abundant myeloid cell recruited from peripheral blood to sites of wound healing, infection [30], and the TME [25, 31]. Arg1 is stored in cytoplasmic granules and upon activation, which occurs spontaneously in vitro, granulocytes release active Arg1 into the extracellular milieu [10]. Purified human granulocytes from a healthy donor consumed L-arginine from the media (Fig. 1b). When activated granulocytes were co-cultured with autologous T cells, T cell proliferation was inhibited (Fig. 1c). The addition of CB-1158 blocked depletion of L-arginine from the media (Fig. 1d, left) and restored T cell proliferation to 90% of the proliferation observed for T cells without granulocytes (Fig. 1d, right). These results show that L-arginine depletion by Arg1 is necessary for granulocyte-mediated suppression of T cell proliferation in vitro, and that CB-1158 reverses this immunosuppression.

We next tested if CB-1158 could block T cell suppression conferred by myeloid cells derived from cancer patients. G-MDSC purified from the peripheral blood of a patient with lung cancer or granulocytes purified from a head and neck cancer patient were cultured for 48 h and the conditioned media were then used in T cell proliferation assays. T cells from healthy donors were used in lieu of the cancer patient donors’ T cells due to the limited amount of sample available from individual patients. Cancer patient-derived G-MDSCs or granulocytes reduced the amount of L-arginine in the media, and arginine depletion was blocked by CB-1158 (Fig. 1e and f, left panels). T cell proliferation was inhibited in the media conditioned by G-MDSC or granulocytes, and proliferation was restored to 99% or 79%, respectively, of control levels by the addition of CB-1158 (Fig. 1e and f, right panels). CB-1158 also restored secretion into the media of the T cell inflammatory cytokines interferon-γ and granzyme-B (Fig. 1g). Together these data demonstrate that inhibition of arginase by CB-1158 blocks myeloid cell-mediated immunosuppression, rescuing T cell proliferation and cytokine secretion.

To determine if arginase inhibition by CB-1158 could be tested for anti-tumor efficacy in mouse models of cancer, pharmacokinetic and pharmacodynamic studies were performed in tumor-bearing mice. Following a single dose of CB-1158 or BID dosing for 5 days, dose-dependent exposure of CB-1158 was observed in both plasma and tumors (Fig. 2a and b). Oral dosing of CB-1158 in tumor-bearing mice also raised the amount of L-arginine in plasma and tumors, indicating an on-target pharmacodynamic effect of CB-1158 (Figs. 2a and b). Importantly, CB-1158 was well-tolerated at doses of 100 mg/kg twice daily for 23 days, with no significant clinical observations or impact on body weight (Fig. 2c).

CB-1158 was tested in multiple syngeneic murine models of cancer. Blocking arginase with CB-1158 significantly inhibited the growth of CT26, LLC, B16, and 4T1 tumors (Fig. 3a). Confirmation that CB-1158 targets Arg1 in vivo was assessed with a genetically altered mouse strain containing a conditional disruption of Arg1 in the myeloid lineage [14, 19, 32]. LLC cells injected into Arg ^flox/flox ; Tie2-Cre ⁺ mice (Fig. 3b, indicated as ARG1^ΔM) grew smaller tumors that were similar in size to CB-1158-treated Arg ^flox/flox ; Tie2-Cre ⁻ mice (Fig. 3b, indicated as ARG1^WT), and CB-1158 treatment of Arg ^flox/flox ; Tie2-Cre ⁺ animals conferred no further reduction in tumor growth, consistent with specific on-target inhibition of Arg1 by CB-1158 (Fig. 3b). Together, these results provide evidence that Arg1 activity promotes tumor growth and that elimination of myeloid cell Arg1 expression or pharmacological blockade of arginase by CB-1158 limits tumor growth in vivo.

A series of experiments was performed to address the in vivo mechanism of action of CB-1158. First, we confirmed that CB-1158 is not directly cytotoxic to murine cancer cell lines by assaying the growth of CT26, LLC, B16, and 4T1 cell lines in the presence of a dose-titration of CB-1158. No growth inhibition of murine cancer cell lines was observed in the presence of 1 mM CB-1158 (Fig. 4a). Next, to test if the mechanism of in vivo efficacy of CB-1158 is immune cell-mediated, CB-1158 was administered to LLC tumor-bearing SCID mice. The efficacy of CB-1158 was abrogated in the SCID background (Fig. 4b), indicating CB-1158 requires an intact immune system to inhibit tumor growth. To confirm the immune cell-mediated mechanism of action of CB-1158, tumor growth was assessed in mice lacking immune cell subsets. Depletion of either CD8⁺ cells or NK cells in the B16 (Fig. 4c) and CT26 (Fig. 4d) tumor models blocked the efficacy of CB-1158, indicating that both CD8⁺ cells and NK cells are required for the full anti-tumor effect of CB-1158 in these models. In the LLC tumor model, depletion of NK cells resulted in loss of efficacy, while depletion of CD8⁺ cells had a smaller effect (Fig. 4e). Collectively, these results indicate that the anti-tumor effect of CB-1158 is mediated by the immune system and requires cytotoxic lymphocytes.

To further investigate the immune cell-mediated mechanism of action of CB-1158, flow cytometry was performed on tumors from animals treated with vehicle or CB-1158 and changes in specific immune cell populations were quantified. In each of the models where tumor growth was inhibited by CB-1158, we observed statistically significant changes in immune cell populations indicative of an increase in inflammation in the TME (Fig. 5). In the CT26 (Fig. 5a) and B16 (Fig. 5b) models, CB-1158 treatment resulted in an increase in tumor-infiltrating activated CD8⁺ CD25⁺ cytotoxic T cells compared to vehicle-treated controls. In the LLC model, CB-1158 treatment caused an increase in CD8⁺ T cells in the tumor, as well as a decrease in CD68⁺ macrophages, compared to vehicle-treated controls (Fig. 5c). In the 4T1 model, where CB-1158 had a modest effect on tumor growth, we nevertheless observed changes in immune cell populations consistent with increased inflammation in the tumor: an increase in CD3⁺ T cells, an increase in NK cells, and a decrease in myeloid cells in the tumors of CB-1158-treated animals compared to vehicle-treated controls (Fig. 5d). Consistent with increased inflammation in the tumors of CB-1158-treated animals compared to vehicle controls, CB-1158 was linked to increases in transcripts of interferon-responsive genes (Fig. 5e) and increases in inflammatory cytokines (Fig. 5f). Taken together, our data indicate that the molecular mechanism of tumor growth inhibition by CB-1158 is immune cell-mediated and results from an increase in inflammation in the tumor microenvironment that includes an increase in cytotoxic lymphocytes and a decrease in myeloid cells.

In growing tumors, effective immunity may be blocked by more than one suppressive mechanism, including expression of immune checkpoint proteins and depletion of essential nutrients from the TME [33–35]. We reasoned that combining CB-1158 with other immune-modulating agents might further reduce tumor growth and tested this idea with four different combinations. First, CT26 tumor-bearing mice were treated with CB-1158 in combination with the checkpoint blockade therapy anti-PD-L1. While tumor growth was reduced with CB-1158 or anti-PD-L1 as monotherapies, tumor growth inhibition was enhanced by combining the two agents (Fig. 6a, left). Complete tumor regression was achieved in three out of ten mice treated with anti-PD-L1 alone, whereas the combination therapy resulted in six complete responses (Fig. 6a, center), with 90% survival at study day 46 for the combination group compared to 30% survival for the anti-PD-L1 single agent group (Fig. 6a, right). Combination efficacy with checkpoint blockade therapy was further explored in the orthotopic 4T1 tumor model, where the addition of CB-1158 to the doublet regimen of anti-CTLA-4 and anti-PD-1 resulted in a significant reduction in growth of the primary tumor as well as a decrease in the number of lung metastases (Fig. 6b). Next, we used two different cellular immunotherapies in combination with CB-1158. B16 tumor-bearing mice treated with CB-1158 in combination with T cells specific for the PMEL tumor antigen exhibited significantly reduced tumor growth compared to mice treated with either single agent (Fig. 6c). Of note, the significant tumor growth inhibition in mice treated with the combination of CB-1158 plus T cells is consistent with the striking survival benefit that was reported for T cell therapy in mice deleted for myeloid Arg1 [32]. We also used NK cell therapy in a lung metastasis model. CT26 tumor-bearing mice treated with CB-1158 in combination with NK cells had significantly fewer lung metastases compared to control groups (Fig. 6d). These results indicate that CB-1158 may be an attractive combination agent with other immuno therapies.

Some standard-of-care chemotherapeutics have been observed to modulate the immune infiltrate [36, 37]. While the primary mechanism of action of gemcitabine, a nucleoside analog, is considered to be inhibition of DNA synthesis, it has also been observed to reduce MDSCs [38–40]. Therefore, we reasoned that gemcitabine may augment the activity of CB-1158. Indeed, CB-1158 treatment in combination with gemcitabine resulted in a significant increase in tumor growth inhibition in the CT26, LLC, and 4T1 tumor models (Fig. 7a). We investigated the mechanism of action of the combination of CB-1158 and gemcitabine treatment in the CT26 tumor model by performing flow cytometry on these tumors. Gemcitabine treatment resulted in a significant decrease in CD11b⁺ myeloid cells in the tumors (Fig. 7b). We note that although gemcitabine can be toxic to immune cells, both lymphocytes and myeloid cells remained present in gemcitabine-treated animals with the dose regimen that was used. Interestingly, NK cells and inflammatory (M1-type, CD80⁺) macrophages were significantly increased, while immunosuppressive (M2-type, CD206⁺) macrophages were significantly reduced in tumors of animals treated with the combination of CB-1158 and gemcitabine, indicating an increase in inflammation and a decrease in immune suppression in the tumor microenvironment (Fig. 7b).

To investigate which tumor types may be more likely to respond to arginase inhibition, we interrogated human tumor microarrays by immunohistochemistry for Arg1 protein expression. The Arg1 antibody was validated using normal liver tissue as a positive control and normal tonsil tissue as a negative control (Fig. 8b). We found an abundance of Arg1⁺ infiltrating immune cells in multiple tumor types, with especially high numbers in tumors of the lung, gastrointestinal tract, and bladder (Fig. 8a and d). Tumor cells were largely negative for Arg1 staining with the exception of HCC (Fig. 8c). Tumor tissue microarrays were also stained by multi-parameter immunofluorescence for Arg1 and other immune cell markers, and we found that Arg1 was more frequently associated with the granulocytic marker CD15 than with the macrophage marker CD68 (Fig. 8e), and in some cases, striking co-localization between Arg1 and CD15 was observed (Fig. 8f). These data confirm Arg1 expression in multiple tumor types.

In addition to tumor expression, Arg1 protein and activity have been observed in the peripheral blood and have been reported to be higher in the plasma of some cancer patients compared to healthy volunteers [25]. To investigate whether the amount of Arg1 is higher in cancer patients than in healthy donors, Arg1 protein was measured in the plasma of 31 healthy donors and 76 cancer patients across 13 different histologies (see Methods). Arg1 was significantly higher for cancer patients compared to the healthy volunteers (Fig. 8g). Peripheral blood L-arginine has also been reported to be lower in cancer patients [25, 26]. Plasma L-arginine was measured for 20 healthy volunteers and 26 cancer patients across 7 different histologies. L-arginine was significantly lower for the cancer patients compared to the healthy individuals (Fig. 8h). These results suggest that cancer patients may experience immune suppression that is associated with higher circulating Arg1 and lower amounts of L-arginine compared to healthy individuals, and that inhibiting circulating Arg1 and raising plasma L-arginine with CB-1158 could confer an immune benefit in the context of cancer.

We reasoned that raising arginine levels would be immune-stimulatory in the context of cancer for the following reasons: first, cytotoxic lymphocytes require exogenous arginine for proliferation in response to in vitro stimulation [10, 11, 28, 29]; second, many cancer patients are immunosuppressed and have lower plasma arginine compared to healthy individuals [25, 26]; and lastly, activated immunosuppressive myeloid cells consume arginine and compete with other arginine auxotrophs such as cytotoxic lymphocytes in the TME for this amino acid [8, 12]. Thus, raising arginine in cancer patients could be critical for the immune system to mount an effective anti-tumor response.

Arginine depletion by myeloid cells is primarily mediated by the enzymes Arg1 and NOS [8]. While there is conflicting data regarding the role of NOS in inflammation [32, 41], Arg1 activity in myeloid cells has been shown to be immunosuppressive and pro-tumorigenic: first, Arg1-expressing myeloid cells consume arginine from the media and suppress T cell activity in co-culture, and importantly, T cell proliferation can be restored either by restoring arginine to the media or by the addition of an arginase inhibitor, showing that arginase activity is necessary for the observed immune suppression (Fig. 1) [10, 11, 25, 42, 43]. Secondly, genetic ablation of Arg1 in the myeloid compartment of tumor-bearing mice was shown to reduce tumor growth, indicative of a pro-tumorigenic and immune suppressive role for Arg1 in vivo [13, 14, 19, 32]. Thirdly, blocking arginase activity pharmacologically with nor-NOHA [15] or CB-1158 (this report) reduces tumor growth. The present study additionally demonstrates that Arg1 inhibition with CB-1158 raises tumor and plasma arginine and increases inflammation in the TME. Together these data argue that arginase activity is immunosuppressive and provide validation for arginase as a cancer immunotherapy drug target. A second arginase isoform, Arg2 is a constitutively-expressed mitochondrial matrix protein found at low levels in many tissues and at high levels in kidney and intestine [21]. Since Arg2 activity also affects plasma L-arginine levels, Arg2 could also be a regulator of immune function. In support of this notion, Arg2 has been reported to promote maternal-fetal immune tolerance [44].

A potential concern with systemic arginase inhibition is interrupting its function within the urea cycle, an important pathway in the liver for detoxifying ammonia generated from the breakdown of amino acids. In spite of this concern, pharmacological arginase inhibition has been well-tolerated in several animal studies, including one study involving a rat model of hypertension in which nor-NOHA was injected over a period of 10 weeks [45], as well as multi-day studies of nor-NOHA in mice [15]. In addition, we have observed that twice-daily oral dosing of CB-1158 was well-tolerated in mice for at least 40 days. Lack of apparent hepatic toxicity may be explained by several observations. First, CB-1158 does not readily enter cells, exhibiting IC₅₀ values for intracellular arginase in the HepG2 and K562 cell lines that are two orders of magnitude higher than for soluble arginases in cell lysates (Tables 1 and 2). Secondly, the subcellular localization and regulation of urea cycle Arg1 may protect it from pharmacological inhibition. In hepatocytes, urea cycle Arg1 is tightly associated at the mitochondria in a multi-enzyme complex, and studies using semi-permeabilized cells and radiolabeled substrate have demonstrated that tight channeling of product and substrate occurs among successive enzymes of the urea cycle: argininosuccinate synthase, argininosuccinate lyase, and arginase [46]. Thus, hepatic Arg1 may be less accessible to CB-1158 compared to cytoplasmic or extracellular arginase in plasma, tumors, and inflamed tissues. When we tested CB-1158 for the ability to inhibit arginase in intact primary hepatocytes (Table 2), we used media containing ornithine, not arginine, and to our knowledge, urea generated under these conditions should require a complete urea cycle. Therefore, this assay may be a measure of the Arg1 activity that is exclusively associated with the urea cycle. However we note that the low potency of CB-1158 on primary hepatocytes could also be due to the inability of CB-1158 to penetrate the cell membrane.

Arginine supplementation has been investigated as a potential therapy for cancer patients and the clinical results suggest that raising arginine levels may be beneficial. In one study of 18 colorectal cancer patients undergoing tumor resection, histopathologic analysis of biopsies revealed that supplementation of arginine prior to surgery resulted in an increase in CD16⁺ and CD56⁺ NK cells infiltrating the tumors [47]. In another study of 96 breast cancer patients, a significant increase in pathological response was observed in patients with small tumors receiving arginine compared to placebo [48]. These data indicate that supplemental arginine may have an immune-stimulatory and anti-tumor effect in cancer patients, and suggest that raising systemic arginine by pharmacological arginase inhibition would be similarly beneficial. However, major limitations to therapeutic oral arginine supplementation include severe gastrointestinal distress and extensive metabolism of arginine by the intestinal mucosa [9, 49–51]. CB-1158 treatment has the potential to both augment and maintain arginine levels in patients thereby avoiding the gastrointestinal distress and arginine oscillations inherent to dietary arginine supplementation.

Antibodies to T cell checkpoint proteins CTLA-4, PD-1, and PD-L1 have resulted in durable clinical responses for some cancer patients, but many patients do not respond [52, 53], suggesting that overcoming additional immunosuppressive mechanisms will be necessary to reactivate anti-tumor immunity in resistant patients. The cellular and molecular basis of resistance to checkpoint blockade is an area of intense investigation. Biomarker studies examining responsive and resistant patients point to a suppressive TME as one possibility to explain resistance. Patients with a higher baseline level of T cells infiltrating the tumor (T cell-inflamed tumors) compared to non-responsive tumors (cold tumors) were more likely to respond to anti-CTLA-4 or anti-PD-1 therapies [54], and patients resistant to anti-PD-1 therapy had higher baseline levels of tumor-infiltrating MDSCs compared to responders [55]. Furthermore, pre-clinical studies have shown that the depletion of essential extracellular metabolites in the TME, such as glucose and amino acids, can block T cell effector function [56, 57]. Together, these results suggest that immunosuppressive myeloid cells and depletion of essential metabolites may give rise to cold tumors that are resistant to checkpoint blockade and therefore targeting these myeloid cells and metabolic regulatory pathways, in combination with checkpoint antibodies, could restore inflammation and increase patient response rates. In support of this hypothesis, epacadostat, which inhibits the enzyme IDO-1 and prevents depletion of the essential amino acid tryptophan from the TME, is exhibiting striking results in combination with anti-PD-1 therapy in patients with melanoma, lung cancer, RCC, or head and neck carcinoma [58–62], arguing that targeting T cell checkpoints and immunosuppression by the TME is an efficacious combination.

In this report, we have shown that inhibition of arginase with CB-1158, a potent, selective, and orally-bioavailable small molecule, reverses suppression of T cell proliferation by blocking arginine depletion. CB-1158 reduced tumor growth in mouse models of cancer by increasing inflammation in the TME, confirming an immune-modulatory mechanism of action in vivo. In addition, CB-1158 treatment of tumor-bearing mice enhanced the efficacy of checkpoint blockade, adoptive T cell therapy, adoptive NK cell therapy, and gemcitabine treatment. A survey of human tumor microarrays and cancer patient plasma confirmed Arg1 expression is elevated in human cancer. These findings provide strong rationale for the clinical development of CB-1158 as an anti-cancer therapeutic agent. A Phase 1 study (NCT02903914) has been initiated to test the anti-tumor activity of CB-1158, as a monotherapy and in combination with anti-PD-1, in order to add to the armamentarium of agents that activate the immune system to fight cancer.