Arginase 1/2 Inhibitor OATD-02: From Discovery to First-in-man Setup in Cancer Immunotherapy Free

This article is featured in Highlights of This Issue, p. 805

The clinical success of therapies that block immune checkpoint pathways (PD-1/PD-L1, CTLA4) and reactivate antitumor immunity focused research efforts on the discoveries of other immunotherapeutic agents that may enhance the clinical benefits of checkpoint inhibitors (1). Arginases (ARG1 and ARG2) are enzymes that hydrolyze semiessential amino acid l-arginine to ornithine and urea (2). The appropriate concentration of l-arginine plays an important role in the full immune response in mammals. l-Arginine deficiency downregulates the expression of the CD3ζ chain, a critical signaling element of the T-cell receptor, thereby impairing the T-cell function. Moreover, depletion of l-arginine leads to inhibition of IFNγ production (3). Arginase is mainly produced by myeloid-derived suppressor cells that are highly enriched in the tumor-bearing state, but also by tumor-associated macrophages, cancer-associated fibroblasts (CAFs), neutrophils, and some tumor cells themselves (Supplementary Fig. S1; refs. 2, 4–9). In addition to ARG1, several scientific reports have recently addressed the important immunosuppressive nature of ARG2. ARG2 has been reported to cause intracellular immunosuppression of CD8⁺ cytotoxic T cells, preventing from the formation of self-renewing memory T cells, as well as promoting cancer cell proliferation (10). In activated T cells, ARG2 degrades l-arginine leading to T-cell suppression (11). On the other hand, ARG2 has also been shown to enhance the metabolic fitness of tumor-supporting regulatory T cells (Treg; ref. 12). ARG2 was found also in Tregs from normal skin and its expression increased in metastatic melanoma suggesting that low intracellular l-arginine concentrations may facilitate the development of these tumor-supporting cells (2, 13, 14).

The clinical relevance of arginase inhibitors in cancer immunotherapy is under investigation in ongoing phase I/II trials with the use of numidargistat (Calithera Biosciences/Incyte Corporation; CB-1158, INCB001158, CAS 2095732-06-0, UNII: IFD73D535A, ClinicalTrials.gov Identifier: NCT02903914, NCT03314935; ref. 15). Early data from phase I studies in colorectal carcinoma show that numidargistat is moderately active in monotherapy and when combined with pembrolizumab (16). INCB1158 was well tolerated and inhibited plasma arginase activity, inducing increased plasma arginine levels at all doses tested. The published up-to-date results confirmed partial response and stable disease in 27% of patients observed in the monotherapy arm. When combined with pembrolizumab, partial response was observed in 37% of patients, while the overall response rate increased from 3% in monotherapy to 7% in the combo arm (vs. 0%–1% reported for checkpoint inhibitors) and the progression-free survival rate at 6 months doubled compared with pembrolizumab alone. However, numidargistat was shown to exert low intracellular activity (IC₅₀ = 32–139 μmol/L) and showed a low volume of distribution and short half-life (∼6 hours) so it has to be taken twice a day (16–18). Notably, it was designed as an extracellular ARG1 inhibitor, not capable of targeting intracellular ARG2 (16, 17).

Because numidargistat cannot fully counteract arginase immunosuppression, it has been tempting to search for more potent ARG1/2 inhibitors with intracellular activity and improved pharmacokinetic/pharmacodynamic properties to potentially target a broader range of malignancies. Several other compounds with proven efficacy in tumor syngeneic animal models are in the nonclinical stage of development. However, to the best of our knowledge, they were also designed exclusively to inhibit extracellular arginase (19–22). Importantly, inhibition of arginase must be appropriately balanced between overcoming immunosuppression in tumor microenvironment (TME) and on-target hepatotoxicity (2). Ideally, an arginase inhibitor for clinical development should inhibit extracellular arginases in TME and intracellular arginases in immune and immunosuppressive cells, as well as tumor cells, without affecting the hepatocytic ARG1. It could be achieved either by the selective distribution of a compound to TME thereby avoiding a hepatic pass or by selective cellular uptake or efflux through the specific cells.

Herein, we present the discovery and nonclinical development of OATD-02, the only dual arginase inhibitor with unique pharmacokinetic properties, that is, moderate to high volume of distribution, low clearance, and the ability to target intracellular ARG2 (thanks to adequate cell membrane penetration). OATD-02 is expected to provide therapeutic benefits by inhibiting arginases in cancer, with an anticipated low risk of systemic toxicity (Supplementary Fig. S1).

OATD-02 was obtained in a 5% overall yield starting from the simple cyclohexanone via asymmetric vinylation, electrophilic carboxylation, Ugi reaction, reduction-reductive amination, hydroboration, and hydrolysis (Supplementary Fig. S2; refs. 23, 24). Detailed synthetic preparation of OATD-02 and compounds 8–16 are described in Supplementary Materials and Methods. Additional compounds within Supplementary Figures S3 to S14 and Supplementary Tables S1 to S3 can be found in patent US10391077B2.

Recombinant enzymes (ARG1 or ARG2) were produced at Molecure SA, biosynthesized in E. coli expression system, purified by fast protein liquid chromatography, and stored at –80°C in the storage buffer (20 mmol/L Tris pH 8.0, 100 mmol/L NaCl, 10 mmol/L dithiothreitol, 10% glycerol). The enzymatic reaction was carried out at 37°C for 1 hour in the assay buffer (100 mmol/L sodium phosphate buffer, 130 mmol/L NaCl, pH 7.4) with the addition of 1 mg/mL BSA in the presence of 200 μmol/L MnCl₂ (enzyme cofactor), 10 mmol/L (for ARG1) or 20 mmol/L (for ARG2) of l-arginine hydrochloride (enzyme substrate) and serial dilutions of the test compounds. The inhibition of the arginase activity was determined by colorimetric detection of produced urea, as previously described with minor changes. Briefly, 150 μL of freshly prepared developing reagent [2 mmol/L o-phthaldialdehyde, 2 mmol/L N-(1-naphthyl)ethylenediamine dihydrochloride, 50 mmol/L boric acid, 1 mol/L sulfuric acid, 0.03% Brij-35] was added to each well incubated for 20 minutes at room temperature. Next, the absorbance at 515 nm was measured. Normalized values were fitted to a four-parameter equation using GraphPad Prism, and IC₅₀ values were determined.

The binding kinetic studies were evaluated by surface plasmon resonance (SPR). The initial buffer of the protein sample was exchanged to immobilization buffer (10 mmol/L NaHPO₄ pH 6.7, 10 mmol/L KCl) by triple diafiltration at 12°C on Amicon Ultra 0.5 mL Centrifugal Filter. Then 64 μL of ARG1 at a concentration of 200 μg/mL was injected (at 5 μL/minute) over the CM5 sensor surface activated with EDC/NHS (according to Cytiva protocol) resulting in an immobilization level of 8,374.1 RU. Interactions were assayed in 50 mmol/L NaHPO₄ pH 7.4, 150 mmol/L KCl, and 0.05% Tween-20 in a single-cycle kinetics mode with a flow rate of 30 μL/minute. The association steps were 5 minutes and the dissociation step 30 minutes with no regeneration step. Analyte (OATD-02) was injected at concentrations: 9.75, 37.5, 150, 600, and 2,400 nmol/L.

All cell lines and primary cell cultures were confirmed at the last used passage to be Mycoplasma-free (Mycoalert Detection Kit, Lonza).

The intracellular activity of arginase inhibitors was determined using murine M2-polarized macrophages. Bone marrow cells were collected from femur bones of C57BL/6 male mice (Charles River Laboratories). The cells suspended in DMEM (Thermo Fisher Scientific) supplemented with 5% FBS and 50 ng/mL MCSF (PeproTech) were poured through a 40-μm cell strainer and seeded into Petri dishes at 1 × 10⁶ cells/mL for 2 days (37°C, 5% CO₂). Next, nonadherent cells were removed and fresh medium was added (DMEM, 5% FCS, 50 ng/mL MCSF). The next day, the cells were dissociated using the CellStripper Dissociation Reagent (Corning) and resuspended in the fresh DMEM containing 5% FCS, 50 ng/mL MCSF, 20 ng/mL TGFβ, and 30 ng/mL IL4 (Biomibo). The cells were seeded into 96-well plates (64 × 10³ cells/100 μL/well). Afterward, the medium was replaced with 80 μL of OptiMEM (Thermo Fisher Scientific) and 10 μL of serial dilutions of the test compounds. Next, 10 μL of the mixture of l-arginine hydrochloride (final concentration 5 mmol/L) and MnCl₂ (final concentration 5 μmol/L) were added. Following 24 hours incubation (37°C, 5% CO₂), the generated urea was detected by mixing 50 μL of the supernatant with 75 μL of the developing reagent (as described in the above section). Normalized absorbance values (515 nm) were fitted to a four-parameter equation using GraphPad Prism and IC₅₀ values were determined. For the liver arginase inhibition assay, human cryopreserved hepatocytes (catalog No. HMCPQC, Thermo Fisher Scientific/Gibco) were thawed in the recovery medium (No. CM7000, Thermo Fisher Scientific/Gibco) supplemented with the Primary Hepatocyte Thawing and Plating Supplements (No. CM3000, Thermo Fisher Scientific/Gibco) and seeded onto collagen-coated 96-well plates at 9 × 10⁴ cells per well. After 6 hours incubation at 37°C, the medium was replaced with a prewarmed William's E Medium (No. A1217601, Thermo Fisher Scientific/Gibco) supplemented with the Primary Hepatocyte Maintenance Supplements (No. CM4000, Thermo Fisher Scientific/Gibco). The next day, the medium was replaced with 80 μL of the OptiMem Reduced Serum Medium (No. 11058021, Thermo Fisher Scientific/Gibco). Serial dilutions of the tested compound were added at the volume of 10 μL. Then, a mixture of the substrate (l-arginine, 5 mmol/L final concentration) and MnCl₂ (0.5 μmol/L final concentration) was added at the same volume. The cells were incubated for 24 hours and then the supernatants were collected. The urea detection was performed by mixing 50 μL of the supernatant with 75 μL of the developing reagent (as described in the above section). Normalized absorbance values (515 nm) were fitted to a four-parameter equation using GraphPad Prism (RRID:SCR_002798) and IC₅₀ values were determined.

CHO-K1 cells (chinese hamster ovary; ATCC catalog No. CCL-61, RRID:CVCL_0214) transiently transfected with the pcDNA3.1 vector (Invitrogen, catalog No. V79020) carrying human ARG1 or ARG2 cDNA were used to assess the cellular activity of the tested arginase inhibitors. The cell culture was maintained in Ham's F12 medium supplemented with 10% FBS. The transfection was achieved using Lipofectamine 2000 according to the instructions of manufacturer. Two days following the transfection, Ham's F12 medium was changed to OptiMEM (reduced serum medium) with an addition of l-arginine (5 mmol/L), manganese (II) chloride (150 μmol/L), and the tested compounds (dilution series). The level of urea released to the culture medium was detected in the culture supernatants. The colorimetrically detectable product was developed as described by Jung and colleagues (Clin. Chem., 1975, 21: 1136–1140) with minor changes. The enzyme activity curves were plotted using GraphPad Prism 7.0 (RRID:SCR_002798) and the IC₅₀ values were determined.

All in vivo experiments were conducted at Molecure SA. All animal experimental procedures were approved by the first Local Ethics Committee in Warsaw (Poland). For the syngeneic B16F10 model, 8-week-old BALB/c females (BALB/c, AnNCrl, Charles River Laboratories) were injected subcutaneously in the right flank with 5 × 10⁵ B16F10 cells (ATCC, catalog No. CRL-6475, RRID:CVCL_0159) in PBS.OATD-02 was dosed by oral gavage twice per day at 50 mg/kg body weight starting 1 day after the tumor implantation (n = 12). Control groups received vehicle (saline; n = 12). When the tumors became measurable, their volumes were determined three times a week by a digital caliper using the formula: width (mm)² × length (mm) × 0.5 (Supplementary Fig. S17). The levels of OATD-02 or l-arginine were determined in plasma samples or tumor homogenates using LC/MS-MS. Sample preparation was based on protein precipitation using acetonitrile. The resulting supernatants were analyzed using hydrophilic interaction liquid chromatography with subsequent mass spectrometry detection based on optimized Multiple reaction monitoring (MRM) transitions. Matrix-matched calibration and quality control samples were prepared for the quantification of OATD-02. Because of the endogenous character of l-arginine, a surrogate matrix was used for its detection. All statistical analyses were performed using GraphPad Prism (RRID:SCR_002798). Datasets with normal distribution were described by mean with SEM. Otherwise, the median value was shown and the U Mann–Whitney or the Kruskal–Wallis tests were applied for intergroup comparisons. Differences at a P value less than 0.05 were considered statistically significant. Tumor growth inhibition (TGI) index was calculated by the formula: (1 − A/B) × 100%, where A is the tumor volume in the treated group and B is the volume in the control group.

See the Supplementary Materials and Methods for more details about the experimental materials and methods, figures and tables (Supplementary Figs. S2–S23; Supplementary Tables S1–S11).

All data are available in the main text or the Supplementary Materials and Methods or from the corresponding author upon reasonable request. The structure factors and atomic coordinates for ARG1–OATD-02 complex structure were deposited in Protein Data Bank under accession no. 8AUP.

To design an oral arginase inhibitor that will be able to target both intracellular and extracellular ARG1 and ARG2 in the TME, we followed the following principles: the molecule must (i) be a nonselective arginase inhibitor, (ii) penetrate the cellular membrane of specific target cells (immune cells, including specifically suppressor cells, as well as tumor cells), (iii) be absorbed from the intestine, and (iv) demonstrate high volume of distribution and low clearance to achieve and maintain an adequate concentration of the compound in TME. There is no information in the literature on arginase inhibitors that combine high enzymatic activity (measured as the IC₅₀ < 100 nmol/L) and high intracellular activity with a high volume of distribution and low clearance that enable targeting the intracellular ARG2. Because the known arginase inhibitors are transition-state analogs of the hydrolysis of the very polar guanidinium group in arginine, they are therefore very polar amino acids (21). It is an extremely difficult medicinal chemistry effort to combine the above principles in one molecule as evidenced by previously described ARG inhibitors (21, 25). In this regard, we focused on lowering of molecular flexibility of 2-(S)-amino-6-boronohexanoic acid (ABH) structure 1 (known arginase inhibitor) to force an appropriate conformation and limit rotatable bonds by incorporation of cyclic moiety (Fig. 1; ref. 26).

Known cyclic conformationally restricted boronic acid ARG1 inhibitors include peptide ABH derivatives bearing heterocyclic rings formed at C1-C2 like numidargistat 2 or compounds recently developed by Merck and AstraZeneca (Fig. 1, compounds 3 to 4; refs. 17, 19–21). These structures, however, stabilize an amino acid moiety conformation leaving propylene-/butylene-boronic acid residue quite flexible and thus potentially exposing the compound to limited intestinal absorption, and off-target binding. In our approach, we decided to link C1 with C3 in the carbon chain to obtain a less flexible cyclic construct carrying ethyleneboronic acid instead of propylene or butylene while avoiding peptide bonds. We explored cyclobutane 5, cyclopentane 6, cycloheptane 7, and cyclohexane 8, and found that cyclohexane-1,3-constrained ABH analog 8 showed similar potency in ARG1 inhibition to the linear parent compound (Fig. 1; Supplementary Tables S1–S3). The relatively bulky cyclohexane ring may also protect the closely located B(OH)₂ group from its metabolism and may generally reduce the recognition of the compound by some off-target proteins as there are no cyclohexane-derived proteogenic amino acids in the human body.

We then focused on decorating the cyclohexane ring with additional substituents at positions 5 and 6 to improve potency by stabilizing the conformation of the cyclohexyl ring and generating additional interactions with aspartic acid residues present in arginases (23, 27). The 6-substituted derivatives turned out to be more active inhibitors (Table 1; Supplementary Table S3).

Insertion of methyl substituent, amine, hydroxyl, or fluorine resulted in an increased potency. Stabilization of the conformation of cyclohexane by insertion of methyl substituent in 8 significantly increased activity (1.2 vs. 0.37 μmol/L). Alcohol 11 and fluorine-derivative 12 as a hydrogen bond donor/acceptor improved the potency up to 0.15 μmol/L. It was found that elongation of the chain connecting amino groups with cyclohexane at the C6 position (13 vs. 10) is of key importance for the activity, with optimal distance being the methylene group (the potency increased up to 20-fold; Table 1; Supplementary Tables S1–S3). Incorporation of basic amines can also allow the compound to potentially bind to acidic components of cellular membranes (e.g., phospholipids), and it may enhance the volume of distribution required for intracellular activity in the TME (28). Primary amine 13, despite its excellent enzymatic activity against both ARG1 and ARG2, was found to be a very weak inhibitor in a cellular model. We used bone marrow–derived macrophages (BMDMs) polarized toward M2 immunosuppressing phenotype as a surrogate of the myeloid suppressor cells infiltrating the TME (29). In arginase-expressing BMDMs, 13 inhibited intracellular mARG with an IC₅₀ of 60 μmol/L. In addition, 13 was not orally available in rats (Table 1; Supplementary Table S5). To reduce the number of HB donors and their negative effects on membrane permeability, intestinal absorption, and clearance, we decided to investigate small secondary and tertiary amines. The secondary amine (methyleneaminomethyl derivative 14) was more active than 13 in an enzymatic assay and showed 3.5-fold higher potency in the BMDM model (still only moderate activity). However, 14 suffered from low plasma exposure after intragastric administration. Tertiary amines (dimethylamine 15 and pyrrolidine 16), which are not HB donors were as potent as 14 toward human recombinant AGR1/2 (IC₅₀ = 20/39 and 16/35 nmol/L, for hARG1 and hARG2, respectively) and inhibited intracellular mouse ARG1 with submicromolar activity (IC₅₀ = 0.91 and 0.37 μmol/L, for 15 and 16, respectively). Importantly, a significant difference in plasma exposure after oral administrations of 15 and 16 was found. It is known that low-molecular-weight polar hydrophilic compounds are absorbed by the intestine via a paracellular route, achieving moderate to good bioavailability after oral administration (30–31). In this context, we selected 15 as a compound with the most balanced structural and physicochemical properties and the best pharmacokinetic profile (Table 1) and focused on its further characterization.

The dose–response curves and IC₅₀ of OATD-02 were evaluated in vitro using biochemical assays with recombinant human arginases (hARG1 and hARG2) and mouse and rat ARG1 (IC₅₀ = 20, 39, 39, and 28 nmol/L toward hARG1, hARG2, mARG1, and rARG1, respectively; Table 2). In addition, OATD-02 inhibitory potential versus hARG1 was determined by assessing the inhibitory constant K_i.

The parameters of the competitive inhibition of hARG1 by OATD-02 were determined as shown in Fig. 2A. The Michaelis–Menten constant (K_m), describing the enzyme–substrate interaction in the absence of an inhibitor, was 9.8 mmol/L l-arginine. Next, similar Michaelis–Menten kinetics of hARG1 was determined at increasing OATD-02 concentrations. Data were successfully fitted to the competitive inhibition model where increasing inhibitor concentration did not change enzyme concentration at a half-maximal rate of metabolism but elevated its K_m value (apparent affinity decrease). The K_i for hARG1 and OATD-02 was estimated at 12.6 nmol/L (Fig. 2A; Table 2).

The binding kinetics was evaluated by SPR (Fig. 2B). OATD-02 binds ARG1 with high affinity (∼1 × 10⁻⁸ M), and relatively slow binding (8.1×10³ M⁻¹s⁻¹) is compensated by very slow dissociation (∼8.7×10⁻⁵ s⁻¹). Such binding kinetic results in a relatively high stability of the complex (residence time τ > 3 hours; half-life > 2 hours; Fig. 2B; Table 2).

The crystal structure of the hARG1–OATD-02 complex, solved at a resolution of 2.2 Å, reveals atomic details of inhibitor binding (Fig. 2C and D; Supplementary Fig. S15). The inhibitor is bound at an active site of the enzyme with hydroxyboronate anion interacting with two Mn²⁺ ions coordinated by active site residues His101, Asp124, His126, Asp128, Asp232, and Asp234. The cyclohexane ring and the ethylene group between the ring and hydroxyboronate anion are stabilized by van der Waals interactions with the side chains of His126 and His141, which sandwich the ethylene linker. The carboxylic group located in the cyclohexane ring forms a hydrogen bond with Ser137 and water-mediated interactions with side chains of Thr135, Asn139, and His141 and the backbone of Asn139. The primary amine group in the ring forms a hydrogen bond with Asp183 and water-mediated interactions with the backbone of Gly142 and the side chain of Glu186. The dimethylamine group forms a hydrogen bond with Asp183 and is stabilized by van der Waals interaction with Thr136.

OATD-02 demonstrated high inhibition of intracellular arginases in several cell lines. It inhibits mouse ARG in BMDM cells with IC₅₀ = 912.9 nmol/L and human ARG2 in transfected CHO-K1 cells with IC₅₀ = 171.6 nmol/L (Fig. 2E). Importantly, OATD-02 inhibits hARG1 in human primary hepatocytes at a much higher micromolar concentration (IC₅₀ = ∼13 μmol/L) compared with the above cell lines, suggesting that the anti-immunosuppressive effect may be achievable at concentrations that do not affect the urea cycle thus minimizing a potential on-target liver toxicity (Fig. 2E; Table 2).

OATD-02 showed favorable pharmacokinetic characteristics in nonclinical species evaluated after a single-dose intravenous or oral administration (Table 2; Supplementary Tables S6 and S7). The oral bioavailability increased from mouse through rat to dog (13%, 30%, and 61%, respectively), also achieving the highest exposure in dogs (Fig. 3A; Table 2).

Pharmacokinetic analysis showed a low plasma clearance (CL_p) and moderate to high volumes of distribution resulting in long-terminal plasma half-lives. The pharmacokinetic profile suggests that the fast tissue distribution and then slow elimination may be prevented by active or paracellular renal reabsorption. OATD-02 generated a “delayed,” but very strong and prolonged pharmacodynamic effect, measured as a fourfold increase in plasma l-arginine concentration at the PD steady-state in rats at an oral dose of 5 mg/kg, which sustained even 1 week after the end of a treatment (Fig. 3B). OATD-02, in comparison with numidargistat, demonstrated a superior pharmacodynamic activity in rats—the effect on plasma arginine level after 1 mg/kg of OATD-02 (intravenous) dose was comparable with 100 mg/kg of numidargistat (i.v.). Of note is the significant increase of pharmacodynamic activity of OATD-02 at 3 mg/kg dose versus 1 mg/kg (and vs. 100 mg/kg of numidargistat; Supplementary Fig. S16).

Anticancer effect of OATD-02 was previously demonstrated in several mouse cancer models (8, 32–34). Here, we present the nonclinical efficacy of OATD-02 in a mouse melanoma model B16F10 (Fig. 3C). The B16F10 model represents a syngeneic murine model offering an intact immune system and TME for testing immunomodulatory agents. The model is known to be rich in CAFs expressing immunosuppressive ARG2 (2, 14). OATD-02 significantly inhibited tumor growth (TGI 46%) when administered orally at the dose of 50 mg/kg twice daily for 15 days (Fig. 3C; Supplementary Figs. S17 and S18). The observed anticancer efficacy is similar to that obtained for numidargistat but with the use of a dose that is twice as high (50 mg/kg of OATD-02 vs. 100 mg/kg of numidargistat) despite much higher oral bioavailability of numidargistat in mice (17). No adverse effects were observed during the treatment (Supplementary Fig. S18). A significant increase in l-arginine levels was revealed after treatment with OATD-02 (∼threefold change vs. control; Fig. 3D). Two hours after the last dosing, OATD-02 had achieved a level of approximately 1.4 μg/mL in plasma and 2.6 μg/g in tumor indicating good TME penetration and accumulation. This has been achieved by appropriate volume of distribution and drug-target residence time, as tumor protein binding was found to be negligible. In vitro, OATD-02 has no cytostatic and antiproliferative effects on B16F10 cells. Incubation of these cells with OATD-02 concentrations of 0.05 to 100 μmol/L for 72 hours did not show an effect on viability, indicating no direct cytotoxic effects and highlighting the apparent immune-modulating properties of OATD-02 (Supplementary Fig. S19).

Evaluation of OATD-02 in comprehensive toxicology studies revealed that the compound is well tolerated and did not cause any major adverse effects in rats and dogs at 5 and 9 mg/kg doses, respectively. In 4-week toxicity good laboratory practice (GLP) studies with a 4-week recovery period, toxicokinetics, potential target organs, reversibility of the effects, and pharmacodynamic biomarkers were assessed at doses ranging from 5 to 15 mg/kg/day in rats and from 1 to 9 mg/kg/day in dogs. No observed adverse effect level (NOAEL) was established at 5 mg/kg/day for rats and 9 mg/kg/day for dogs. As measured on day 28, the dose level of 5 mg/kg/day corresponded to a C_max of 813 and 570 ng/mL and an AUC_0–24 of 10,800 and 6,470 ng/mL*hour in male and female rats, respectively (Fig. 3B, for female rats).

The human pharmacokinetic parameters of OATD-02 were predicted by applying three-species allometric scaling approaches using nonclinical data obtained from animals supported by in vitro data (Supplementary Table S7; ref. 35). Estimates of allometric coefficient (i) and exponent (ii) were obtained from intercept and slope on a log-log plot, respectively (Fig. 4A; Supplementary Figs. S20–S22). Human plasma CL (CL_P) of OATD-02 was estimated to be 0.051 L/hr/kg, representing the average value of the two prediction approaches. Human volume of distribution (V_ss) of OATD-02 of 2.4 L/kg was predicted on the basis of f_u-normalized, that is, plasma binding corrected, allometric scaling of V_ss. The intravenous plasma half-life (t_1/2) of OATD-02 of approximately 33 hours was calculated on the basis of the predicted CL and V_ss. For simplicity, it was assumed that OATD-02 exhibits linear (first-order) pharmacokinetic behavior that can be described by a one-compartment model and, thus, shows a monophasic decline of plasma concentrations on a semilogarithmic scale. Calculated human oral bioavailability (F) of OATD-02 of 35% represents average of (maximum) F observed in preclinical species after administration of 10 mg/kg, which ranged from approximately 13% in mice, approximately 30% in rats, to approximately 61% in dogs. Predicted human pharmacokinetic parameters of OATD-02 are presented in Table 2.

The maximum recommended starting dose (MRSD) for the first-in-human (FIH) study was estimated to be 2.5 mg daily based on the human equivalent dose (HED) calculated using the respective NOAEL (5 mg/kg in rats) and implementing a safety factor of approximately 20. Because of the variability in pharmacokinetics observed across the preclinical species, a safety factor higher than 10 is recommended by the FDA (36). After that, we focused on the simulations of potential human exposures after repeated oral administration and estimation of a therapeutic window. Human plasma C, t profiles following repeated oral administration (N = 28, dosing interval τ = 24 hours) of OATD-02 at proposed clinical doses between 2.5 and 30 mg every day were simulated on the basis of the predicted pharmacokinetic parameters CL/F (0.15 L/hour/kg), V_ss/F (6.9 L/kg), and t_1/2a (0.5 hour) using a simplified, that is, one-compartment model. As far as applicable, absorption half-lives (t_1/2a) of OATD-02 were obtained by curve fitting (compartmental modeling) of the individual and mean oral concentration versus time (C, t) profiles in mice, rats, and dogs (Fig. 4B; Supplementary Fig. S23).

On the basis of these simulations, the intended starting dose of 2.5 mg every day in humans is expected to generate plasma OATD-02 concentrations at steady state above the IC₅₀ for hARG1, which is thought to translate into immunomodulatory and anticancer effects (Fig. 4B; Supplementary Fig. S23). At the simulated dose of 30 mg, the steady-state systemic concentrations of the drug should be much higher than IC₉₀ (123 nmol/L) for inhibition of hARG1 (over the entire dosing interval) and hARG2 (372 nmol/L) (for at least 20 hours postdose) as well as IC₅₀ toward intracellular hARG2 (Fig. 4B and C). On the basis of the murine models, the OATD-02 concentration in TME is higher than in the systemic circulation. Thanks to a long residence time on arginase, OATD-02 escapes a simple equilibrium and therefore achieves strong intracellular inhibition of ARG2. The projection of predicted human C_max,ss for a 30-mg daily dose was 534 nmol/L which is fourfold lower than C_max,ss observed at NOAEL in the most sensitive species (rats) at the toxicology studies and more than 20-fold lower than the concentration needed for ARG1 inhibition in primary human hepatocytes in vitro (based on IC₅₀; Fig. 4C). The correlation between predicted exposures and effectiveness/toxicity was calculated from dose-dependent efficacy studies in the CT26 mouse model of colorectal cancer, pharmacokinetics in mice, and toxicologic studies in rats (Fig. 4D; Supplementary Table S11; ref. 32). The presumed therapeutic window is from 2.5 mg every day to about 50 mg every day, with doses at 20 to 30 mg predicted to optimally balance safety and activity/efficacy.

The primary objective of our drug discovery program was to identify novel, intracellularly active, and pharmacologically active arginase inhibitors that could be developed as oral enhancers for cancer immunotherapy. To the best of our knowledge, the approach presented here to focus on compounds that are capable of inhibiting intracellular arginases in TME with limited undesirable intrahepatocytic ARG inhibition is unique and is particularly important in blocking immunosuppression and tumor progression (17–22).

As a result, we discovered OATD-02—a cyclic diamino acid boronic acid dual ARG1/2 inhibitor. Opposed to other ARG inhibitors under development, OATD-02 is not a dipeptidyl derivative, and thus it is not transported by the peptidyl transporting system (e.g., PEPT1) as confirmed in vitro. Recognition by PEPT1 may increase intestinal absorption of compound to some extent; however, this transporting system is also expressed in hepatocytes (but was not found in targeted immune cells) making this approach hard to be drugable (37, 38). OATD-02 demonstrated high inhibition of intracellular arginases in several cell lines. Compared with numidargistat, OATD-02 is approximately fourfold more potent toward inhibition of recombinant hARG1 and 10-fold toward hARG2; it also inhibits intracellular arginases at a much lower concentration (>60×) in a BMDM model (32). Importantly, OATD-02 inhibits arginase in human primary hepatocytes at a much higher micromolar concentration (IC₅₀ = ∼13 μmol/L) compared with the model cell lines (BMDM, CHO-K1), suggesting that the anti-immunosuppressive effect may be achievable at concentrations that do not affect the urea cycle, thus minimizing potential on-target liver toxicity (Fig. 2E; Table 2). OATD-02 is likely not to be actively taken up by hepatocytes presumable due to the lower expression of the transporting system responsible for the active uptake of OATD-02 from hepatocytes compared with other cell types.

OATD-02 was shown to be well absorbed in rat and dog, demonstrates low plasma clearance and moderate to high volumes of distribution resulting in long plasma half-lives. It generates a strong and a long pharmacodynamic effect (measured as increased of plasma l-arginine concentration) in animal models despite a decrease of plasma compound level. Such disconnections between drug plasma concentration and pharmacologic effect are often characteristic of so-called slow-offset compounds with slow drug-target binding kinetics (39). OATD-02 demonstrates a long dissociative half-life of a drug–target complex. Although other boronic acid arginase inhibitors also have long residence times, OATD-02 has the longest one; it is twice as long as numidargistat at physiologic pH (40, 41). Long residence time on the target is expectedly vitally important to achieve the in vivo activity and has recently attracted increasing attention in the drug discovery process (42). For an intracellular target (like ARG1 and ARG2), an unbound inhibitor in cells is compartmentalized at a high local concentration and is not removed from plasma before the next binding event (43). Slow offset and favorable pharmacokinetic (especially high V_ss) of OATD-02 may result in a long half-life in which arginases will remain inhibited throughout their lifetime, until new resynthesis by ribosomes.

In fact, the prognostic marker of anti-immunosuppressive efficiency is not systemic l-arginine concentration but both intracellular and extracellular arginine levels in the local TME. Systemically acting extracellular ARG inhibitors with fast clearance and low volume of distribution may not be able to maintain appropriately high l-arginine levels in TME, where the ARG activity is enhanced. In animal models, OATD-02 shows a strong dose-dependent increase in plasma l-arginine levels as well as TGI in the CT26 murine model of colon carcinoma; notably, the l-arginine concentration elevation in this model is even stronger in tumor homogenates compared with plasma (32). The anticancer effect of OATD-02 was previously demonstrated in several syngeneic mouse cancer models including Lewis lung carcinoma (LLC), ovarian cancer, and gliomas (8, 33–34). In addition, it has also been shown that OATD-02 potentiates the antitumor effects of anti-PD-1, and anti-PD-L1 antibodies, epacadostat (IDO inhibitor), and DMXAA (STING agonist; refs. 32–34). Moreover, OATD-02 restores activity and proliferation of cytotoxic T cells suppressed by ARG1 in vitro and in vivo (8, 33). Recently, the anticancer activity of OATD-02 was reported in the ARG2-dependent syngeneic Renca model of renal carcinoma and immune-independent K562 xenograft model of leukemia; it can be speculated that both observations prove the importance of intracellular ARG2 inhibition (32). In several in vivo mouse models of cancers, OATD-02 demonstrated superior activity compared to numidargistat. These include, for example, a syngeneic mouse model of colorectal cancer CT26 (TGI 48%, P = 0.0033 for OATD-02 vs. TGI 28%, P = 0.4390 for numidargistat) when compounds were used at the same doses despite much higher oral bioavailability of numidargistat (32). In xenograft mouse model of leukemia K562 that depends on ARG2 OATD-02 demonstrated an even stronger difference in efficacy (TGI 49%, P = 0.0172 for OATD-02 vs. TGI 14%, P = 0.578 for numidargistat) with half the dose used compared with numidargistat (32). Another nondirected literature example is syngeneic mouse model of lung carcinoma LLC when OATD-02 (referred as its previous code OAT-1746) administered twice daily by an intraperitoneal route at a dose of 20 mg/kg showed similar TGI to numidargistat dosed orally at a dose 100 mg/kg twice daily (33).

On the basis of the human pharmacokinetic predictions, OATD-02 is expected to show an overall favorable human pharmacokinetic profile with low plasma CL, medium V_ss (∼fourfold higher than the total body water in man), long plasma t_1/2, and moderate oral bioavailability. Oral exposure to OATD-02 is assumed to be restricted by the drug's permeability rather than by its solubility or first-pass metabolism/elimination in the gut and liver. The predicted long t_1/2 (33 hours) should allow once-daily dosing of OATD-02 in humans and should provide effective inhibition of systemically circulating arginases throughout treatment at the expected doses. On the other hand, the expected moderate volume of distribution (2.4 L/kg) should allow for adequate drug concentrations at the tumor site at simulated relatively low doses, despite the predicted moderate oral bioavailability (35%). Notably, numidargistat demonstrated a much shorter half-life and lower volume of distribution in humans (18). At the simulated dose of 30 mg every day, the concentration of OATD-02 in TME should be much higher than hARG1/2 IC₉₀ and intracellular hARG2 IC₅₀. The projection of C_max,ss for this dose was approximately fourfold lower than C_max,ss observed at NOAEL in rats indicating a reasonable therapeutic window (2.5–50 mg every day) achievable by OATD-02 in humans. The selected doses for FIH clinical trials should inhibit both, ARG 1 and 2 in TME (but not in hepatocytes), generating optimal anticancer immunity without developing clinically relevant hyperammonemia and/or argininemia.

In conclusion, the combination of intracellular inhibition of ARG1/2 together with a demonstrated moderate to high volume of distribution and slow binding kinetic makes OATD-02 the first-in-class dual inhibitor that is expected to demonstrate efficacy due to inhibition of both arginases. Compared with the clinically tested arginase inhibitor numidargistat, OATD-02 is a much more potent inhibitor capable of intracellularly blocking ARG2 and has a longer residence time at the target, higher efficacy in tumor models, and a much better predicted human pharmacokinetic profile (much longer half-life and much higher volume of distribution). A FIH clinical trial using OATD-02 as a novel cancer immunotherapy has been just initiated (ClinicalTrials.gov Identifier: NCT05759923).

B. Borek reports grants from National Centre for Research and Development during the conduct of the study and has a patent for US10391077B2 issued to Molecure SA. K. Jedrzejczak reports grants from National Centre for Research and Development during the conduct of the study and, K. Jedrzejczak has a patent for US10391077B2 issued to Molecure S.A. M. Grzybowski reports grants from National Centre for Research and Development during the conduct of the study. P. Pomper reports grants from National Centre for Research and Development during the conduct of the study. T. Rejczak reports grants from National Centre for Research and Development during the conduct of the study. K. Matyszewski reports grants from National Centre for Research and Development during the conduct of the study and has a patent for US10391077B2 issued to Molecure S.A. A. Golebiowski reports grants from National Center for Research and Development during the conduct of the study and has a patent for US10391077B2 issued to Molecure SA. J. Olczak reports grants from National Centre for Research and Development during the conduct of the study and has a patent for US10391077B2 issued to Molecure SA. K. Lisiecki reports grants from National Centre for Research and Development (Poland) during the conduct of the study. M. Tyszkiewicz reports grants from National Centre for Research and Development during the conduct of the study. M. Kania reports grants from The Nationale Centre for Research and Development during the conduct of the study. A. Cabaj reports grants from The National Centre for Research and Development during the conduct of the study. P. Dera reports grants from National Centre for Research and Development during the conduct of the study. J. Chrzanowski reports grants from National Centre for Research and Development during the conduct of the study. D. Kusmirek reports grants from National Centre for Research and Development during the conduct of the study. E. Sobolewska reports grants from National Centre for Research and Development during the conduct of the study. M. Magdycz reports grants from The National Centre for Research and Development during the conduct of the study. L. Mucha reports grants from The National Centre for Research and Development during the conduct of the study. M. Masnyk reports grants from National Centre for Research and Development during the conduct of the study. A. Napiorkowska-Gromadzka reports grants from National Centre for Research and Development during the conduct of the study. S. Pikul reports grants from National Centre for Research and Development during the conduct of the study. R. Jazwiec reports grants from NCBIR during the conduct of the study. P. Dobrzanski reports stock ownership in OncoArendi Therapeutics. M. Meyring reports personal fees from Molecure S.A. during the conduct of the study and personal fees from Molecure S.A. outside the submitted work. P. Iwanowski reports personal fees from Molecure during the conduct of the study and personal fees from Biomapas outside the submitted work. Z. Zaslona reports grants from The National Centre for Research and Development and European Funds Smart Growth and European Regional Development Fund during the conduct of the study and personal fees from board members outside the submitted work. R. Blaszczyk reports grants from The National Centre for Research and Development during the conduct of the study and has a patent for US10391077B2 issued to Molecure SA. No disclosures were reported by the other authors.

B. Borek: Investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Nowicka: Investigation, methodology, writing–review and editing. A. Gzik: Investigation, methodology, writing–review and editing. M. Dziegielewski: Investigation, methodology. K. Jedrzejczak: Investigation, methodology. J. Brzezinska: Investigation, methodology. M. Grzybowski: Conceptualization, investigation, visualization, methodology, writing–review and editing. P. Stanczak: Conceptualization, supervision, investigation, methodology, writing–review and editing. P. Pomper: Methodology. A. Zagozdzon: Methodology. T. Rejczak: Investigation, visualization, methodology, writing–review and editing. K. Matyszewski: Methodology. A. Golebiowski: Conceptualization, funding acquisition, investigation, methodology, writing–review and editing. J. Olczak: Conceptualization, funding acquisition, writing–review and editing. K. Lisiecki: Investigation, methodology. M. Tyszkiewicz: Supervision, investigation, methodology. M. Kania: Methodology. S. Piasecka: Methodology. A. Cabaj: Investigation. P. Dera: Investigation. K. Mulewski: Methodology. J. Chrzanowski: Visualization, methodology, writing–review and editing. D. Kusmirek: Methodology. E. Sobolewska: Methodology. M. Magdycz: Methodology. L. Mucha: Methodology. M. Masnyk: Methodology. J. Golab: Conceptualization. M. Nowotny: Investigation, visualization, methodology. E. Nowak: Visualization, methodology. A. Napiorkowska-Gromadzka: Methodology. S. Pikul: Funding acquisition, investigation, methodology, project administration. R. Jazwiec: Investigation, methodology. K. Dzwonek: Conceptualization, supervision, funding acquisition, project administration. P. Dobrzanski: Conceptualization, investigation. M. Meyring: Methodology. K. Skowronek: Visualization, methodology. P. Iwanowski: Writing–review and editing. Z. Zaslona: Conceptualization, writing–review and editing. R. Blaszczyk: Conceptualization, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

Studies supported by the project DIMUNO: “Development of new cancer therapies based on selective antitumor immunomodulators” (STRATEGMED2/265503/3/NCBR/15)—cofinanced by National Centre for Research and Development in the framework of STRATEGMED Program.

The study was supported by the project: “Preclinical and clinical development of arginase inhibitor for application in anticancer immunotherapy” (POIR.01.01.01-00-0415/17), acronym ARG cofinanced by the European Union in the framework of European Funds Smart Growth and European Regional Development Fund.

We would like to thank Dr. Anna Krause, Mrs. Malgorzata Borkowska, Dr. Elzbieta Pluta, Dr. Jolanta Peczkowicz-Szyszka, Dr. Nazan Cemre Güner-Chalimoniuk, Dr. Anna Siwinska, Dr. Katarzyna Drzewicka, Mrs. Anita Troc, Prof. Dominika Nowis, Dr. Rafal Kaminski, Mr. Mikolaj Welzer, Dr. Marcin Mazurkiewicz, Dr. Janusz Zukowski, Dr. Timi Oshodi, Mrs. Karolina Kosinska, Mr. Mariusz Kaminski, and Dr. Michal Mlacki for their exceptional support in various aspects of OATD-02 discovery and development stage as well as Mrs. Anna Bajera, Dr. Marta Dudek, Mrs. Olga Lozano-Solarska, Mrs. Kinga Jablonska, and Dr. Samson Fung, for their outstanding work on EMA Scientific Advice, IB, IMPD, clinical trial protocol and CTA preparation and submission.