TAM Family Receptor Kinase Inhibition Reverses MDSC-Mediated Suppression and Augments Anti–PD-1 Therapy in Melanoma

A high number of tumor-infiltrating lymphocytes are associated with a better prognosis for advanced melanoma, breast and lung cancer (1–3). This and emerging data suggest targeting the host immune system can enhance cancer therapies (4). Although some patients have a positive response to immunotherapy, the majority remain refractory (5). A better understanding of the entire immunosuppressive landscape—both the adaptive and the innate systems—is necessary to improve this promising treatment modality.

The overall melanoma response rate to the CTLA-4 antagonist, ipilimumab, is ∼10% and the PD-1 antagonist pembrolizumab exceeds 35% (6, 7). A lack of functional or tumor-infiltrating T cells is often cited as a reason for therapeutic failure (8, 9). Several additional T-cell–centric therapies are currently being investigated in clinical trials (10–14). Although many acknowledge the importance of the myeloid compartment in the tumor microenvironment (TME), including tumor-associated macrophages (15, 16), myeloid-derived suppressor cells (MDSC; refs. 17, 18), and tolerizing or inactive antigen-presenting cells (APC; refs. 19, 20), cancer therapeutics based on the innate immune system are less common (e.g., IDO and PI3K-γ inhibitors; refs. 21–23). To increase the efficacy and durability of immunotherapy, additional methods reversing the immunosuppressive TME and a better understanding of immunotherapy resistance mechanisms are needed.

TYRO3, AXL, MERTK transmembrane receptor tyrosine kinases (TAM RTK) regulate the innate immune system, dampening inflammatory responses to prevent chronic inflammation and autoimmunity (24, 25). The activity of these receptors is stimulated by protein ligands, e.g., GAS6 and PROTEIN S (26, 27). The protein ligands have a higher affinity and are most effective when they bridge the cell-surface TAM receptors to a source of lipid phosphatidylserine (PtdSer) exposed on apoptotic cells, aggregating platelets, exosomes, and certain viruses that expose PtdSer on their surface (apoptotic mimicry; ref. 24). GAS6 and PROTEIN S are expressed in multiple cell types, including epithelial and immune cells (28), and can be secreted by tumor cells dampening the immune response (29). Genetic deletion of MERTK in mice leads to heightened inflammatory responses and mild autoimmunity (30). When tumors are implanted in Mertk^−/− mice, tumor growth is slower and metastasis is decreased (31).

MDSCs are a heterogeneous population of bone marrow–derived, immature myeloid cells (17, 32–34). MDSCs are potently immunosuppressive and have been implicated in negative regulation of immune responses during infection, cancer and the maternal immune response to fetuses (35–37). MDSCs promote tumor growth and metastasis by suppressing the antitumor immune response and through paracrine secretions that stimulate tumor cell proliferation, motility, and angiogenesis (32). MDSCs correlate with poor prognosis in human melanoma and with lymph node metastasis in breast cancer patients (38, 39). In mice, there are two MDSC subsets: monocytic (M-MDSC) and granulocytic or polymorphonuclear (PMN-MDSC). M-MDSCs are characterized by expression of CD11b⁺Ly6G⁻Ly6C^high and PMN-MDSCs by CD11b⁺Ly6G^highLy6C^low (32). M- and PMN-MDSCs utilize diverse mechanisms to thwart T-cell proliferation including increased activity of arginase 1 (Arg), indoleamine-2,3-dioxygenase (IDO), inducible nitric oxide synthase (iNOS), production of reactive oxygen species (ROS) as well as antigen-specific suppression (32). One factor controlling MDSC biology is the transcription factor STAT3 (signal transducer and activator of transcription 3), which regulates MDSC expansion and promotes MDSC survival (32). STAT3 also prevents differentiation of MDSCs to macrophages and dendritic cells (DC; ref. 40), in effect maintaining a more immunosuppressive microenvironment.

We investigated the contribution of TAM RTKs and ligands in MDSC function using Mertk^−/−, Axl^−/−, and Tyro3^−/− mice and TYRO3, AXL, and MERTK inhibitors. Genetic deletion or pharmacologic inhibition of TYRO3, AXL, and MERTK diminished MDSC suppression in a STAT3-dependent manner. Additionally, inhibition of TYRO3, AXL, and MERTK delayed the growth of a syngeneic melanoma model, enhanced T-cell infiltration into the TME, and augmented the efficacy of anti–PD-1 therapy. We showed that metastatic melanoma patients exhibit increased numbers of TYRO3⁺, AXL⁺, and MERTK⁺ MDSCs, with MERTK⁺ MDSCs predominating. These results suggest that TYRO3, AXL, and MERTK cooperate as immunosuppressive, innate immune-checkpoint genes across the myeloid spectrum with potential therapeutic import.

C57BL/6J and OT-I (C57BL/6-Tg(TcraTcrb)1100Mjb/J) mice were purchased from The Jackson Labs. Mertk^−/−, Axl^−/−, and Tyro3^−/− mice are on a C57BL/6 background. Mertk^−/− mice were derived by our lab and Axl^−/− and Tyro3^−/− mice were provided by Douglas K. Graham (Emory University, USA). Mice were genotyped by PCR to confirm gene deletion.

BRAF^V600EPTEN^−/− cells (500,000) were subcutaneously injected into syngeneic mice and allowed to form tumors. Tumors were measured by caliper every other day. Tumor volume was calculated according to the formula: cm³ = [(length (cm) × (width (cm))²]/2. UNC4241 (25 mg/kg) dissolved in saline or a saline vehicle was administered daily by oral gavage in a 100 μL volume.

Tumor-bearing mice were injected with 200 μg rat IgG2b isotype control or anti-mouse Ly6G/Ly6C (Gr-1) antibody (Clone RB6-8C5, Bio X Cell) every 3 days (36).

Tumor-bearing mice were injected with 250 μg rat IgG2a isotype control or anti–PD-1 antibody (Clone RMP1-14, Bio X Cell) every 3 days (41).

FACS-sorted MDSCs and BRAF^V600EPTEN^−/− cells were mixed at a 1:10 MDSC:tumor cell ratio to mimic physiologic conditions (42) and injected into syngeneic mice. Tumor volume was assessed as above.

All experimental groups included randomly chosen littermates, 8 to 10 weeks old of both sexes. Animal protocols (IACUC Protocol # 16-175) were approved by the Institutional Animal Care and Use Committee at UNC.

The BRAF^V600EPTEN^−/− melanoma cell line was previously generated (43). Cell lines were cultured no longer than 2 weeks and were received in 2013 from the Hanks Lab (Duke University). Cells were not authenticated in the past year. Cells are cultured with a mycoplasma removal agent (MP Bio) to prevent contamination and were tested monthly with a PCR-based assay (Universal Mycoplasma Detection Kit, ATCC 30-1012K).

One million tumor, spleen, or lymph node cells (obtained as described below) were stained with Live Dead stain (Thermo Fisher L34957) in PBS for 20 minutes. The reaction was quenched by adding RPMI with 5% fetal bovine serum (FBS) and centrifuging for 5 minutes at 1,700 rpm at room temperature. Cell pellets were resuspended in flow buffer [PBS with 2% bovine serum albumin (BSA)] 1 μg of each fluorochrome-conjugated antibody was added and incubated for 1 hour at room temperature in the dark. After staining, 2 mL of flow buffer was added and centrifuged for 5 minutes at 1,700 rpm at room temperature. Samples were fixed with 2% paraformaldehyde in flow buffer for 15 minutes then washed with flow buffer. Samples were resuspended in flow buffer and analyzed using a LSRFortessa (Becton Dickinson). Data were analyzed using FlowJo software (Becton Dickinson) or FCS Express (De Novo Software).

Mice were subcutaneously injected with 500,000 BRAF^V600EPTEN^−/− cells and allowed to form tumors. When tumors reached 1,000 mm³ the mice were euthanized, and their spleens were harvested. We enriched CD11b⁺ cells using the EasySep Mouse CD11b-positive Selection Kit (STEMCELL Technologies, 18970). Briefly, the selection cocktail was added to cells and incubated for 5 minutes before RapidSpheres were added, and the mixture was incubated for 3 minutes. The sample tubes were inserted into the magnet (“The Big Easy” EasySep Magnet, STEMCELL Technologies, 18001) for 5 minutes and the supernatant discarded. Remaining cells were washed with column buffer and subjected to magnetic isolation again. Isolated cells were then stained for Ly6C and Ly6G as described above and sorted on a FACSAria III (Becton Dickinson) instrument. See Supplementary Fig. S1 for gating strategy.

Tumors were harvested, minced with a razor blade and digested with RPMI containing collagenase IV (1 mg/mL), hyaluronidase (0.1 mg/mL), and deoxyribonuclease (20 U/mL, Sigma) at 37°C for 1 hour then filtered through a 40-μm filter for downstream applications.

Total RNA was isolated using the RNeasy Plus Mini Kit (Qiagen, 74134) and quantified with a NanoDrop. RNA (100 ng) was reverse transcribed to cDNA using SuperScript IV VILO Master Mix (Thermo Fisher, 11756050). Real-time PCR was performed in triplicate using 5 ng RNA on a ViiA 7 Real-Time PCR System (Thermo Fisher). TaqMan probes (Tyro3 Mm00444547_m1), Axl (Mm00437221_m1), Mertk (Mm00434920_m1), Gas6 (Mm00490378_m1), Pros1 (Mm01343426_m1; Thermo Fisher) or validated primers (below) and PowerUp SYBR Green Master Mix (Thermo Fisher, A25742) were used for conventional qPCR. GAPDH was used as a reference gene for normalization, and gene expression was determined using the 2^−ΔΔCT method.

Arg Forward: 5′ CTCCAAGCCAAAGTCCTTAGAG 3′ Reverse: 5′ AGGAGCTGTCATTAGGGACATC 3′

Nos2 Forward: 5′ AGGAAGTGGGCCGAAGGAT 3′ Reverse: 5′ GAAACTATGGAGCACAGCCACAT 3′

Gapdh Forward: 5′ GTCTACATGTTCCAGTATGACTCC 3′ Reverse: 5′ AGTGAGTTGTCATATTTCTCGTCGT 3′

Ido1 Forward: 5′ CAGGCCAGAGCAGCATCTTC 3′ Reverse: 5′ GCCAGCCTCGTGTTTTATTCC 3′

Tgfb1 Forward: 5′ CTCCCGTGGCTTCTAGTGC 3′ Reverse: 5′ GCCTTAGTTTGGACAGGATCTG 3′

Ten thousand MDSCs were plated in a black 96-well plate (Costar, 3916) and incubated overnight at 37°C with 5% CO₂. ROS production was measured by the DCFDA cellular ROS detection assay kit (Abcam, ab113851) according to the manufacturer's directions. Fluorescence was measured at 535 nm with an EnSpire plate reader (PerkinElmer).

Murine blood was obtained by terminal cardiac puncture and contained in heparin-coated tubes. Cells were isolated by density gradient centrifugation using Lympholyte-M (Cederlane Labs).

Twenty milliliters blood was collected from 25 healthy donors and 18 metastatic melanoma patients under protocol LCCC 1715 after written consent was obtained. Inclusion criteria for healthy cohort were as follows: (i) Healthy donors were older than 18 years of age. (ii) Male or female. (iii) No documented diagnosis of cancer. (iv) Donors consented to providing a one-time blood sample of up to 20 mL. (v) Consented to abstraction of their medical records as evidence by signed HIPAA form. (vi) Signed institutional review board (IRB)–approved informed consent document. Inclusion criteria for melanoma cohort were as follows: (i) 18 years of age or older. (ii) Male or Female. (iii) English speaking. (iv) Appointment at the NC Cancer Hospital (NCCH) for treatment of a pathologically or diagnostically confirmed diagnosis of metastatic melanoma. (v) Patient consented to providing a baseline blood sample of up to 20 mL. (vi) Consents to abstraction of their medical records as evidence by signed HIPAA form for this protocol. (vii) Signed an IRB-approved informed consent document for this protocol. Exclusion criteria for both cohorts were as follows: (i) Younger than 18 years of age. (ii) Dementia, altered mental status, or any psychiatric condition that would prohibit the understanding or rendering of informed consent. (iii) Diagnosis of autoimmune diseases, or chronic or acute infection. (iv) Non-English speaking. Cells were immediately isolated by density gradient centrifugation using Lympholyte-poly (Cederlane Labs, CL5071) and then stained for flow cytometry analysis.

Assay was adapted from (44). Spleens and lymph nodes were harvested from OT-I mice [C57BL/6-Tg(TcraTcrb)1100Mjb/J, Jackson Labs] and homogenized by pushing cells through a 70-μm mesh strainer with a syringe plunger. Red blood cells were lysed using Red Blood Cell Lysis Buffer (Sigma-Aldrich, R7757). Splenocytes were stained with Cell Proliferation Dye eFluor 450 (eBioscience, 65-0842-85) according to the manufacturer's protocol. Briefly, splenocytes were stained with 10 μmol/L dye in the dark at room temperature for 20 minutes. Serum-containing RPMI was added to quench the reaction on ice for 5 minutes. Stained splenocytes were incubated with FACS-sorted MDSCs at a 1:1 ratio in a 96-well v-bottom plate (Costar, 3894). Proliferation was stimulated with OVA-257 (SIINFEKL) peptide (Sigma-Aldrich) at a final concentration of 1 μg/mL. Influenza peptide was used as an irrelevant peptide negative control. Cells were incubated in a tissue culture incubator at 37°C for 5 days. Dye dilution was measured by flow cytometry on a BD LSRFortessa using FACSDiva.

FACS-sorted MDSCs and OT-I splenocytes were incubated at a 1:1 ratio on ELISPOT plates, stimulated with OVA-257 or an influenza peptide control and incubated for 5 days in a tissue culture incubator at 37°C. Mouse IFNγ ELISpot (MABTECH) was performed according to the manufacturer's guidelines. Briefly, cells were incubated in an antibody-coated plate for 48 hours in a tissue culture incubator at 37°C. Cells were then removed, and detection antibody was added and the plate incubated for 2 hours at room temperature. Streptavidin-HRP was added after the plate was washed and incubated for 1 hour at room temperature. The plate was washed, and substrate solution was added for 10 minutes until spots developed at room temperature. The plate was then washed in deionized water. Spots were imaged and quantified using an ELISPOT Reader System (AID, Autoimmun Diagnostika GMBH).

In vitro: FACS-sorted MDSCs were incubated in a tissue culture incubator at 37°C with 5% CO₂ with 10 ng/mL GM-CSF (R&D Systems) for 3 days and then stained for 1 hour at room temperature in the dark and analyzed by flow cytometry on a BD LSRFortessa using FACSDiva for markers of DCs (CD11c⁺MHCII⁺) and macrophages (CD11b⁺F4/80⁺).

In vivo: FACS-sorted MDSCs were stained with eFluor 450 proliferation dye and then implanted into mice with tumor cells at a 1:10 MDSC:tumor cell ratio. Mice were euthanized and tumors were resected 3 days later and eFluor 450 proliferation dye positive cells were analyzed by flow cytometry for markers of DCs (CD11c⁺MHCII⁺) and macrophages (CD11b⁺F4/80⁺).

FACS-sorted MDSCs were stained with eFluor 450 proliferation dye then implanted into the right flank of mice with tumor cells at a 1:10 MDSC:tumor cell ratio. Mice were euthanized, and tumors and tumor-draining lymph nodes (TDLN) were resected 72 hours later and analyzed by flow cytometry for eFluor 450 proliferation dye–positive cells.

BM-MDSCs were derived as previously described (45). Briefly, bone marrow was harvested and incubated in a tissue culture incubator at 37°C with 5% CO₂ for 5 days with 40 ng/mL GM-CSF, 40ng/mL G-CSF and 40 ng/mL IL6 in RPMI with FBS. Media and cytokines were replaced every 3 days.

Flow cytometry—mouse: CD11b (BD Biosciences 562605), Ly6C (BD Biosciences 561237), Ly6G (BD Biosciences 563978), Mer (Biolegend 151506), Axl (R&D Systems FAB8541A), Tyro3 (R&D Systems FAB759G), F4/80 (BD Biosciences 123141), CD11c (BD Biosciences 564080), MHC II (BD Biosciences 743872), CD8 (Miltenyi Biotec 130-109-250), CD45 (BD Biosciences564279). Flow cytometry—human: CD11b (BD Biosciences 562723), CD14 (Miltenyi Biotec 130-110-521), CD15 (Miltenyi Biotec 130-104-939), CD66b (BD Biosciences 561927), HLA-DR (BD Biosciences 565972), CD33 (BD Biosciences 563171), Lineage Cocktail (BioLegend 348807), Mer (BioLegend 367603), Axl (R&D Systems FAB154A), Tyro3 (R&D Systems FAB859P), CD16 (BD Biosciences 564434). Western blot: p-Stat3 (Cell Signaling Technology 49081), Stat3 (Cell Signaling Technology 9139), Actin (Cell Signaling Technology 3700), N14 (Rockland scientific). Immunoprecipitation: MERTK (Invitrogen 14-5751-82). Immunofluorescence: Stat3 (Abnova PAB6030).

Cells were lysed in coimmunoprecipitation (co-IP) lysis buffer (20 mmol/L Hepes, 2 mmol/L EDTA, 10 mmol/L NaF, 150 mmol/L NaCl, 10% glycerol, 0.5% Nonidet P-40) containing 1 tablet each of protease and phosphatase inhibitors (Roche, Complete protease inhibitor cocktail 11697498001, PhosSTOP 4906845001) and lysates were precleared by centrifuging at 10,000 rpm for 30 minutes at 4°C. Lysates were incubated with PA/GS beads (Santa Cruz Biotechnology, sc-2003) and the appropriate primary antibody on a rotator overnight at 4°C. Immunoprecipitates were washed 3 times in co-IP lysis buffer and subjected to Western blot analysis. Briefly, proteins were resolved by SDS-PAGE and transferred to a PVDF membrane and incubated overnight at 4°C with the indicated antibody. Blots were developed using ECL (Pierce, 34095) and imaged on a Bio-Rad Imaging System (Bio-Rad ChemiDoc XRS+).

MDSCs were plated on slides and then fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton-X 100. These slides were blocked in 5% BSA dissolved in PBS, incubated with primary antibodies at a 1:500 dilution in PBS with BSA, washed with PBS with BSA, and incubated with secondary antibodies at a 1:500 dilution. After another wash with PBS with BSA, the slides were mounted in Prolong gold (Invitrogen-Life Technologies, P36930). Immunofluorescently labeled cells were imaged using an LSM 710 Spectral Confocal Laser Scanning Microscope (Zeiss). Fifteen independent images for each treatment were obtained, and colocalization analysis of the confocal images was performed using ImageJ software (NIH).

STAT3 inhibitor NSC 74859 (Selleckchem S1155) was dissolved in DMSO, STAT3 activator Colivelin (Tocris 3945) was dissolved in 20% ethanol in water, p38 MAPK Inhibitor XIX Skepinone-L (Calbiochem 506174) was dissolved in DMSO, ERK Inhibitor II (Millipore Sigma 328007) was dissolved in DMSO, JNK Inhibitor VIII (Millipore Sigma 420135) was dissolved in DMSO. 1 μmol/L of each inhibitor was added to cells in RPMI with FBS and incubated for 2 hours in a tissue culture incubator with 5% CO₂ at 37°C.

BRAF^V600EPTEN^−/− cells were incubated with the indicated concentration of the TYRO3, AXL, and MERTK inhibitor UNC4241 for 24, 48, or 72 hours. The MTS assay was performed by adding 20 μL CellTiter reagent and incubating the plate for 4 hours in a tissue culture incubator with 5% CO₂ at 37°C (Cell Titer, Promega, G3582). Absorbance was then read at 490 nm.

IHC for CD8 was performed by the Animal Histopathology and Laboratory Medicine Core at the University of North Carolina, which is supported in part by an NCI Center Core Support Grant (5P30CA016086-41) to the UNC Lineberger Comprehensive Cancer Center. Briefly, IHC analysis was performed on 4-μm thick paraffin sections on the Roche DISCOVERY Ultra automated platform. Automated antigen retrieval was performed using CC1, pH 8.5 (Roche, 950-124) for 64 minutes at 100° Celsius. The CD8a primary antibody [Cell Signaling Technology, 98941, CD8α (D4W2Z) XP Rabbit mAb (Mouse Specific)] was incubated for 2 hours at room temperature at a final dilution of approximately 1:400 titrated in DISCOVERY PSS Diluent (Roche, 760-212). This was followed by a post-primary hydrogen peroxidase incubation for 32 minutes, secondary antibody (DISCOVERY OmniMap anti-Rabbit HRP, ready-to-use dilution, Roche 760-4311) for 32 minutes, DISCOVERY Purple chromagen (Roche, 760-229) for 32 minutes, hematoxylin II (Roche, 790-2208) for 12 minutes, and bluing reagent (Roche,760-2037) for 4 minutes, all at room temperature.

The Translation Pathology Lab scanned tissue sections stained for single target detection of CD8 on the ScanScope XT (Leica Biosystems) with a 20 × power Olympus UPlanSApo objective. Exposure times were set according to the calibration data from the line camera. The output was in the format of ScanScope Virtual Slide (.svs) files saved with 8-bit image depth and JPEG2000 compression (compression quality set to 70). The ScanScope XT uses a halogen light source (TechniQuip).

Specific statistical tests are reported in the figure legends. GraphPad Prism 6 was used for all statistical analyses. Unpaired two-tailed t tests were used to compare mean differences between control and treatment groups. Mann–Whitney tests were used for human data analysis. A P value of less than 0.05 was considered significant.

See Supplementary Fig. S2A and S2B for schematic.

To a suspension of 1 (30.6 g, 100 mmol/L) and 4-fluoroaniline (14.1 mL, 150 mmol/L) in acetonitrile (350 mL) was added a 4.0 mol/L HCl solution in 1,4-dioxane (90 mL, 360 mmol/L). The resulting mixture was heated at 82°C (under reflux) overnight. After cooled to room temperature, the suspension was filtered, and the precipitate was washed with CH₂Cl₂ (2×). The filtrate was concentrated to provide the HCl salt of the desired compound 2 as a pale white solid (42 g, 100%). MS (ESI) for [M+H]⁺ (C₁₆H₁₉BrFN₄O⁺): calcd. m/z 381.06; found m/z 381.10.

A solution of 2 (6.6 g, 15.8 mmol/L) in dimethylformamide (DMF; 72 mL) was added an aqueous solution of K₂CO₃ (8.72 g in 24 mL water) at room temperature. The mixture was stirred vigorously in a three-neck flask equipped with a condenser opening to the air and was added 3 (10.38 g, 47.4 mmol/L). After stirring at room temperature for 5 minutes, the reaction mixture was added PdCl₂(dppf)·CH₂Cl₂ (1.29 g, 1.58 mmol/L) and CuBr (452 mg, 3.16 mmol/L), stirred for another 2 to 3 minutes, then heated at 120°C for 30 minutes in a preheated oil bath. The mixture was cooled to room temperature, then filtered through a pad of celite, and washed with DMF. The filtrate was concentrated under vacuum pressure. After drying under high vacuum for 30 minutes, the mixture was added CH₂Cl₂ and was sonicated for 1 hour. The precipitate was filtered with a Buchner glass funnel and provided the desired compound 4 (5.0 g, 70%) as a pale yellow solid.

To compound 4 (12.0 g, 26.6 mmol) in a round flask was added a 4 N HCl aqueous solution (300 mL). The mixture was heated at 100°C (under reflux) for 3 hours, cooled to room temperature, then was poured into an ice-water bath. Sodium hydroxide (45.5 g, 1.14 mol/L) was added portion wise followed by potassium carbonate until the pH of the mixture was 8–9. The precipitated yellow solid was filtered to yield the desired compound 5 (6.7 g, 62%).

A suspension of 5 (6.6 g, 16.2 mmol/L) in CH₂Cl₂ (150 mL) was added pyrrolidine (2.6 mL, 32.4 mmol/L), sodium triacetoxyborohydride (6.87 g, 32.4 mmol/L), and acetic acid (1 mL, 17.5 mmol/L). The reaction mixture was stirred at room temperature overnight and then was added to pyrrolidine (1.3 mL, 16.2 mmol/L), sodium triacetoxyborohydride (3.44 g, 16.2 mmol), and acetic acid (0.5 mL, 8.7 mmol/L). After 2 hours, the mixture was filtered. The filtrate was basified with a 1 mol/L NaOH aqueous solution and then was extracted with a mixture of CH₂Cl₂ and MeOH (10:1; 3×). The combined organic layers were dried (using MgSO₄), concentrated, and purified by column chromatography with ISCO system (0%–10% MeOH in DCM+0.5% NH₃) to yield crude material, which was further purified by pre-HPLC on a reverse phase column (10%–100% MeOH/water) to provide the TFA salt of the desired product. The TFA salt was dissolved in a 2 mol/L HCl aqueous solution, sonicated for 30 minutes, and concentrated to yield the HCl salt of UNC4241 (4.7 g, 51%) as an off-white solid. ¹H NMR (400 MHz, CD₃OD) δ 8.88 (s, 1H), 8.54 (s, 1H), 8.27 (d, J = 8.0 Hz, 1H), 8.10 (d, J = 8.2 Hz, 1H), 7.67 – 7.54 (m, 2H), 7.25–7.15 (m, 2H), 4.57 (s, 2H), 4.12–3.94 (m, 1H), 3.72–3.51 (m, 3H), 3.30–3.20 (m, 2H), 2.32–2.17 (m, 2H), 2.17–1.96 (m, 6H), 1.62–1.45 (m, 2H), 1.45–1.30 (m, 2H); ¹³C NMR (101 MHz, CD₃OD) δ 163.16, 160.97, 160.72, 154.23, 152.20, 150.27, 142.77, 141.94, 133.74, 127.71, 126.36, 122.24, 116.95, 116.73, 107.41, 69.87, 55.74, 55.04, 51.83, 34.42, 30.53, 23.96; MS (ESI) for [M+H]⁺ (C₂₆H₃₂FN₆O ⁺): calculated. m/z 463.26; found m/z 463.30.

Knowing that the TAM RTKs regulate some innate immune cells, we investigated their role in MDSCs. First, we measured the effect of tumor growth on MDSC receptor and ligand expression. MDSCs were purified by FACS from the spleen and Tyro3, Axl, and Mertk, and ligand expression were compared in BRAF^V600EPTEN^−/− tumor-bearing and non–tumor-bearing mice. We observed dramatic increases in expression of all three receptors and the ligands in MDSCs from tumor-bearing mice (Fig. 1A). Next, we quantified the surface expression of TYRO3, AXL, and MERTK MDSCs in non–tumor-bearing mouse blood, and tumor-bearing mouse blood and tumors. There was no change in AXL expression in the blood of control and tumor-bearing mice but both MERTK and TYRO3 increased in M- and PMN-MDSCs in the blood of tumor bearing mice (Fig. 1B and C). All three receptors showed increased MDSC surface expression in the tumors compared with their level in the blood of the same animals (Fig. 1B and C).

To study the effect of each individual receptor, splenic (Supplementary Fig. S1A and S1B) and intratumoral (Fig. 1D; Supplementary Fig. S1C; gating strategy in Supplementary Fig. S1D) MDSCs were isolated from wild-type (WT), Axl^−/−, Mertk^−/−, and Tyro3^−/− mice bearing syngeneic BRAF^V600EPTEN^−/− tumors. The number of splenic MDSCs increased in the different knockout mice bearing tumors by a small increment, although the Axl^−/− mouse showed a slight decrease in PMN-MDSCs (Supplementary Fig. S1B). In contrast, although the percentage of intratumoral M-MDSCs was similar in WT and Axl^−/ − mice, there was a decrease in Mertk^−/− and Tyro3^−/− mice suggesting a migration, homing or survival defect in moving to or surviving in the tumor in the absence of either MERTK or TYRO3 (Fig. 1D).

To test if TYRO3, AXL, and MERTK alter the suppressive abilities of MDSCs, MDSCs from the spleens of WT, Axl^−/−, Mertk^−/−, and Tyro3^−/− mice bearing BRAF^V600EPTEN^−/− tumors were FACS purified and RNA was extracted for quantitative real-time PCR (qRT-PCR) analysis. Arginase 1, iNOS, TGFβ, and IDO expression were substantially reduced in Axl^−/−, Mertk^−/−, and Tyro3^−/− M-MDSCs (Fig. 2A–D). Arginase expression was significantly decreased in Axl^−/−, Mertk^−/−, and Tyro3^−/− PMN-MDSCs, whereas TGFβ and IDO were more variable (Fig. 2C and D). ROS production, a major mechanism of PMN-MDSC activity, was measured enzymatically over 24 hours after isolation and found to be decreased in Axl^−/−, Mertk^−/−, and Tyro3^−/− PMN-MDSCs compared with WT PMN-MDSCs (Fig. 2E). These findings indicate that the absence of TYRO3, AXL, or MERTK decreases the suppressive capabilities of MDSCs with a greater influence on M-MDSCs than PMN-MDSCs.

To determine if Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs altered their ability to suppress CD8⁺ T-cell proliferation, we performed in vitro antigen-specific T-cell proliferation assays using the model antigen OVA and OVA-specific T cells. Axl^−/−, Mertk^−/−, and Tyro3^−/− M-MDSCs suppressed T-cell proliferation less compared with WT MDSCs (Fig. 2F). Consistent with this finding, Axl^−/−, Mertk^−/−, and Tyro3^−/− M-MDSCs allowed an enhancement in the generation of OVA antigen-specific T-cell responses based on IFNγ ELISPOT assays (Supplementary Fig. S3A).

To determine if MDSC differentiation into more mature, less suppressive macrophages and DCs (42) was affected by TAM RTKs, we FACS sorted M-MDSCs from the spleens of tumor-bearing mice, incubated them with GM-CSF, and analyzed cell-surface markers by flow cytometry 5 days later. Axl^−/−, Mertk^−/− and Tyro3^−/− MDSCs differentiated to DCs (CD11c⁺MHCII⁺; Fig. 2G; Supplementary Fig. S3B) and macrophages (CD11b⁺F4/80⁺; Fig. 2G; Supplementary Fig. S3C) at a higher frequency than their WT counterparts. To investigate if this was only an ex vivo phenomenon, we subcutaneously implanted FACS-sorted MDSCs stained with e450 proliferation dye, serving as a marker to distinguish implanted MDSCs from naïve MDSCs, into mice along with BRAF^V600EPTEN^−/− melanoma cells. Forty-eight hours later, the implanted e450 proliferation dye–positive cells were analyzed for cell-surface markers by flow cytometry. Axl^−/−, Mertk^−/−, and Tyro3^−/− e450⁺ cells each exhibited markers for DCs (CD11c⁺MHCII⁺; Fig. 2H; Supplementary Fig. S3D) and macrophages (CD11b⁺F4/80⁺; Fig. 2H; Supplementary Fig. S3E) at higher levels than WT. Taken together, these findings suggest that TYRO3, AXL, or MERTK maintains MDSC immaturity.

MDSCs are actively recruited to tumor sites by chemokines produced by the tumor. To test whether Mertk^−/− and Tyro3^−/− MDSCs have a general migratory or homing defect, we FACS sorted MDSCs from WT and Axl^−/−, Mertk^−/−, and Tyro3^−/− knockout mice and stained them with e450 proliferation dye. We mixed these MDSCs with BRAF^V600EPTEN^−/− melanoma cells and implanted the mixture into WT mice and 72 hours later collected the tumors and TDLNs and analyzed e450, CD11b, Ly6C, and Ly6G to determine the proportion of MDSCs that remained in the tumor versus those that migrated to the TDLN using flow cytometry. Mertk^−/− and Tyro3^−/− MDSCs remained in the tumor, while WT and Axl^−/− MDSCs were more likely to migrate to the TDLN (Fig. 2I). These results suggest that MERTK and TYRO3 influence MDSC migration from spleen to tumor and from tumor to TDLN—a major site of T-cell activation or suppression.

To determine if Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs affect tumor growth, we FACS sorted WT, Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs and coimplanted them into the right flank with BRAF^V600EPTEN^−/− melanoma cells in a 1:10 ratio (MDSC:tumor cells). Tumors were significantly smaller in mice receiving Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs compared with mice receiving WT MDSCs (Fig. 2J; Supplementary S3F). Thus, TYRO3, AXL, and MERTK modulate and maintain MDSC-suppressive function and Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs inhibit tumor growth, perhaps by creating a less-suppressive TME.

The above results suggest that AXL, MERTK, and TYRO3 regulate MDSC activity to promote an immunotolerant microenvironment. To determine if inhibiting TYRO3, AXL, and MERTK is therapeutically tractable, we used UNC4241, a pan-TAM small-molecule inhibitor, with pharmacokinetic properties in mice allowing once daily dosing by oral gavage (IC₅₀: MERTK 1.4 nmol/L; AXL 5.4 nmol/L and TYRO3 2.3 nmol/L; Supplementary Fig. S2A and S2B). MDSCs from tumor-bearing mice treated with vehicle or 25 mg/kg UNC4241 for 3 weeks were FACS sorted and analyzed for their suppressive capabilities. UNC4241 reduced arginase 1 (Fig. 3A), iNOS (Fig. 3B), and IDO (Fig. 3D), while modestly lowering TGFβ expression (Fig. 3C) in M-MDSCs and ROS production (Fig. 3E) in PMN-MDSCs compared with vehicle treatment.

To test if UNC4241 alters the characteristics of MDSCs in a short-term in vitro assay, we cultured FACS-sorted MDSCs with 300 nmol/L UNC4241 and assayed survival, differentiation and maintenance of Ly6C and Ly6G expression. UNC4241 had no effect on survival as measured by the MTS assay (Supplementary Fig. S2C). In addition, we observed a slight increase in DC but not in macrophage differentiation (Supplementary Fig. S2D), and Ly6C and Ly6G levels were stable (Supplementary Fig. S2E).

To test if UNC4241 treatment mirrors the effect of Axl, Mertk, and Tyro3 genetic deletion on T-cell proliferation, we studied MDSC suppression of T-cell proliferation in OVA-specific CD8 T-cell proliferation assays. Purified MDSCs from tumor-bearing mice treated with UNC4241 for 3 weeks were isolated and incubated with splenocytes from OT-I mice. Compared with saline vehicle, UNC4241 treatment in vivo yielded MDSCs that were significantly less suppressive allowing increased T-cell proliferation (Fig. 3F).

To test if UNC4241′s reduction of MDSC suppression (Fig. 3) altered the growth of BRAF^V600EPTEN^−/− syngeneic melanoma in vivo, we subcutaneously injected cells and treated mice daily with UNC4241 for 5 weeks. Although the cell line expresses all three RTKs (Supplementary Fig. S4A), UNC4241 did not inhibit proliferation in vitro (Supplementary Fig. S4B). In contrast, UNC4241 inhibited tumor growth compared with vehicle in vivo (Fig. 4A and B). To assess the UNC4241 immune consequences, IHC staining for CD8 was performed on fixed tumor sections and a greater than 4-fold increase in CD8 T-cell infiltration in UNC4241-treated tumors compared with vehicle-treated (Fig. 4C) was observed. To determine if UNC4241 treatment increased DCs and macrophages in tumors, we performed flow cytometry and found no change in DC frequency and a slight decrease in macrophage numbers in UNC4241-treated tumors (Fig. 4D; Supplementary Fig. S4D). To rule out a direct UNC4241 effect on CD8⁺ T cells, we performed in vitro T-cell proliferation assays with CD8⁺ T cells pretreated with either vehicle or UNC4241 and found that CD8 T-cell proliferation was unaffected (Supplementary Fig. S4C). To test if lessened MDSC suppression leads to the decrease in tumor growth, we repeated the tumor study with anti-Gr-1 antibody to deplete MDSCs. The antibody alone caused a nonsignificant reduction in tumor size, whereas UNC4241-treated tumors were significantly smaller than control tumors. However, in the context of MDSC depletion with anti-Gr-1, UNC4241 no longer inhibited tumor growth (Fig. 4E).

To investigate if UNC4241 treatment augmented checkpoint inhibitor therapy, mice were injected with a BRAF^V600EPTEN^−/− melanoma cell line and once tumors became palpable therapy was initiated with anti–PD-1 alone, UNC4241 alone, the combination of anti–PD-1 with UNC4241 or the vehicle treatment. Although UNC4241 alone again decreased tumor growth, the combination of anti–PD-1 with UNC4241 significantly decreased tumor growth and prolonged survival with the median survival of anti–PD-1 alone being 75 days and combination anti–PD-1 and UNC4241 survival equaling 96 days (Fig. 4F–H).

To define mechanisms downstream of MERTK that maintain or enhance MDSC activity, we investigated pathways involved in MDSC biology that intersect with MERTK signaling, specifically STAT3 signaling (24). We analyzed STAT3 phosphorylation in MDSCs derived from bone marrow precursors (BM-MDSC) of WT and Axl^−/−, Mertk^−/−, and Tyro3^−/− mice. STAT3 is activated by Tyr705 phosphorylation whereas transcriptional activation is regulated by Ser727 phosphorylation (46). The process of differentiating BM-MDSCs with cytokines (GM-CSF, G-CSF, and IL6; ref. 45) resulted STAT3 Tyr705 phosphorylation in all four (WT, Axl^−/−, Mertk^−/− and Tyro3^−/−) genotypes, presumably due to cytokine activation of JAK kinases (Fig. 5A). In contrast, STAT3 serine phosphorylation was markedly diminished in the Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs compared with WT (Fig. 5A). To investigate if the ligands GAS6 and PROTEIN S induces STAT3 serine phosphorylation, we treated BM-MDSCs with GAS6 or PROTEIN S and evaluated STAT3 phosphorylation. Although GAS6 induced serine phosphorylation in WT MDSCs, PROTEIN S did not (Fig. 5B). Additionally, GAS6-dependent STAT3 serine phosphorylation was lost in Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs. To see if GAS6 or PROTEIN S treatment induced complex formation between MERTK and STAT3, we treated BM-MDSCs and immunoprecipitated MERTK. Following GAS6 stimulation, STAT3 precipitated in a complex with MERTK and STAT3 exhibited increased serine phosphorylation; however, we saw no effect with PROTEIN S treatment (Fig. 5C). The increase in p-Ser STAT3 was inhibited by UNC4241 (Fig. 5C). Given the observed change in serine phosphorylation, we investigated serine kinases known to phosphorylate STAT3 that are downstream of MERTK. We determined that GAS6-dependent STAT3 serine phosphorylation was most inhibited by a p38 inhibitor, very modestly inhibited by a JNK inhibitor and not altered by an ERK inhibitor (Supplementary Fig. S5A–S5C). We also demonstrated that GAS6-stimulated MERTK immunoprecipitated with both STAT3 and p38 (Supplementary Fig. S5A). In addition, we examined STAT3 cellular localization by IHC and determined that GAS6-induced nuclear translocation of STAT3 is detectable at 40 minutes following GAS6 treatment (Fig. 5D). We then showed that STAT3 inhibition reversed the T-cell–suppressive capability of WT MDSCs from tumor-bearing mice (Fig. 5E). To show that STAT3 functions downstream of the TAM RTKs, we showed that pretreating MDSCs with a STAT pathway activator, reversed the loss of suppression seen in Axl^−/−, Mertk^−/−, and Tyro3^−/− MDSCs (Fig. 5F). The latter result indicates that TAM RTKs' control of STAT3 in part regulates MDSC immunosuppression. We also demonstrated that iNOS (Fig. 5G) and arginase (Fig. 5H) expression was induced by GAS6 and inhibited by UNC4241.

Circulating MDSCs can be detected in cancer patients, and the number of MDSCs negatively correlates with response to checkpoint inhibitor therapy (38, 47–49). Given the potential for targeting TYRO3, AXL, and MERTK in human disease, we studied their surface expression on the three major subclasses of human circulating MDSCs, M-MDSCs (CD33⁺CD14⁺HLA-DR^lo/−CD15⁻), PMN-MDSCs (CD14⁻CD11b⁺CD66b⁺), and e-MDSCs [Lin⁻(CD3/14/15/19/56) HLA-DR⁻CD33⁺; ref. 50]. We obtained blood from healthy donors and metastatic melanoma patients (IRB protocol LCCC 1715), isolated PBMCs, and analyzed by an eleven-color flow assay to assess TAM RTK⁺ MDSCs (gating strategy; Supplementary Fig. S6A). We observed an increase in circulating MDSCs in melanoma patients versus healthy donors (Fig. 6A). There were more MERTK⁺ MDSCs (20%–50%) than AXL⁺ or TYRO3⁺ MDSCs (<2%; Fig. 6B–D; Supplementary Fig. S6B–S6D). The relative MERTK⁺ MDSC numbers increased in all three MDSC populations in melanoma patients. The relative increase of TYRO3⁺ and AXL⁺ M- and PMN-MDSCs is somewhat greater than that of MERTK⁺ MDSCs, but the absolute levels are considerably lower (Fig. 6B–D; Supplementary Fig. S6B–S6D). Intriguingly, the percentage increase in MERTK⁺ e-MDSCs is the greatest of the three subtypes (>10-fold) reaching 50% of the e-MDSCs (Fig. 6D; Supplementary Fig. S6D). The percentage increase in TYRO3⁺ MDSCs of all three subtypes is also noteworthy (>10-fold), but there are fewer than in the MERTK⁺ subset. Taken together, these data suggest that MERTK⁺ MDSCs may serve as a marker of malignancy and potentially be a biomarker for response to therapy with MERTK inhibitors or other agents targeting the immune system.

This article demonstrated that the innate immune RTKs, TYRO3, AXL, and MERTK regulated MDSC-suppressive activity in a STAT3-dependent manner and inhibition of TAM RTKs induced more robust antitumor responses. Receptor and ligand levels were increased on MDSCs in tumor-bearing mice compared with non–tumor-bearing mice and this RTK family regulated arginase 1, iNOS, IDO, TGFβ, and ROS-suppressive mechanisms. The absence of TAM RTKs inhibited MDSC migration and allowed differentiation to less suppressive mature cells. By using a pan-TAM inhibitor, UNC4241, we decreased melanoma tumor growth alleviating MDSC suppression, promoting greater T-cell infiltration and augmenting anti–PD-1 activity. Lastly, we found that MERTK, AXL, and TYRO3 were upregulated on MDSCs in human melanoma patients.

Malignancies promote signaling networks leading to accumulation of MDSCs and their pathological action. (32, 42). Tumor-bearing mice had more AXL⁺, MERTK⁺, and TYRO3⁺ MDSCs compared with WT. Similarly, melanoma patients had an increase in TYRO3⁺, AXL⁺, and particularly MERTK⁺ MDSCs compared with healthy donors. These data suggest that malignancy (and perhaps other inflammatory and cytokine producing conditions) induces TYRO3, AXL, and MERTK expression on MDSCs helping to establish a suppressive TME and resistance to anticancer therapeutics.

MDSCs suppress T-cell activity through ROS, iNOS, arginase, IDO, and TGFβ activity (51). Our data demonstrated that loss of either MERTK, TYRO3, or AXL downregulated arginase 1 expression in both M-MDSCs and PMN-MDSCs while decreasing iNOS and ROS production. These results were recapitulated by receptor kinase inhibition with UNC4241. Downregulation of IDO and TGFβ by genetic deletion and drug inhibition was more striking in M-MDSCs and was variable in PMN-MDSCs. We demonstrated that MDSC-mediated antigen-specific suppression of T-cell proliferation and activation is reduced in Axl^−/−, Mertk^−/−, or Tyro3^−/− MDSCs and is again recapitulated by UNC4241 inhibition of TAM RTKs. We observed an increase of CD8⁺ T-cell infiltration into tumors in UNC4241-treated mice compared with vehicle-treated mice. To confirm that UNC4241 acted on MDSC suppression, we depleted MDSCs during UNC4241 treatment and showed a decreased response to UNC4241, indicating that in this in vivo model UNC4241's method of action is primarily on MDSCs. To complement our depletion study, we coimplanted Axl^−/−, Mertk^−/−, or Tyro3^−/− MDSCs with melanoma cells. Compared with tumors receiving WT MDSCs, tumors receiving null MDSCs were significantly smaller, suggesting that the lack of TYRO3, AXL, or MERTK in MDSCs provided a less favorable environment for tumor growth. Taken together, these data demonstrate that the lack of TAM RTKs enhances the antitumor T-cell response by reversing MDSC-mediated suppression and that the TAM RTKs are regulators of multiple suppression mechanisms.

Although the precise differences between the TAM RTKs' roles and intracellular pathways need further investigation, it is clear that they are not redundant. It remains somewhat surprising that deletion of any one of the three family members results in similar deficits. Previous reports indicate that the loss of any of the TAM RTKs leads to defects in platelet aggregation (52). We observed some differences between these receptors in MDSCs, for example, MERTK and TYRO3 deletion produces a deficit in MDSC egress from the tumor to the TDLN, whereas AXL deletion does not. There are several possible explanations for this phenomenon, including that each receptor has a nonoverlapping function. Another possibility is that not only may heterodimerization be important, but large oligomers of TAM RTKs may be required to optimally amplify the ligand-phosphatidyl serine-suppressive signal. Because this phenomenon has now been observed in several signaling models (platelets, ref. 52; MDSCs, this work; macrophages, ref. 29), the mechanism will require further elucidation.

Emerging research indicates that intratumoral MDSCs are much more suppressive than MDSCs in lymphoid organs (53). Thus, preventing MDSCs from accumulating in the TME would potentially lessen the suppressive TME, and potentially enhance antitumor therapies. Mertk^−/− and Tyro3^−/− MDSCs do not migrate to or survive as well in the TME, thereby lessening the MDSC contribution to the suppressive TME. This suggests that TAM RTK inhibition could augment immunotherapy by decreasing MDSC accumulation in the TME.

Another area of research focuses on promoting MDSC differentiation into less suppressive, mature myeloid cells. MDSCs differentiate into DCs and macrophages and a number of studies focus on drug treatment to promote maturation (17). Among others, all-trans retinoic acid (ATRA) and vitamin D promote myeloid differentiation and have been studied in the context of MDSC maturation; however, clinical trials have shown limited success (54, 55). Our data also show that WT MDSCs are less able to differentiate into macrophages or DCs, whereas loss of any single TAM RTK allows for more efficient differentiation, adding another mechanism through which TAM RTK inhibition can lessen immunosuppression and potentially augment responses to current standard-of-care therapies.

Consistent with TAM RTK knockout MDSCs' ability to differentiate more easily is the loss of STAT3 serine phosphorylation in these cells. STAT3 regulates differentiation status in MDSCs, and the loss of STAT3 gene expression allows for MDSC differentiation into DCs and macrophages (40). Therapeutic response may be linked to MDSCs differentiating to less suppressive myeloid cells. This suggests aiding differentiation via TAM RTK inhibition may be one mechanism to enhance T-cell–centric therapies.

STAT3 dimerization, nuclear translocation, and DNA binding properties are initiated by phosphorylation at Tyr705, transcriptional activation is dependent on Ser727 phosphorylation through MAP kinase family members (46). The ability to induce STAT3 serine phosphorylation with GAS6 treatment correlated with the induction of iNOS and arginase RNA expression, events that were inhibited with UNC4241. A STAT3 inhibitor reduced MDSC suppression of T-cell proliferation in WT MDSCs, whereas a STAT3 activator in TAM RTK-null MDSCs bypassed the TAM RTK dependency and restored MDSC suppression back to WT levels. These findings strongly indicate that TAM RTKs regulate various MDSC-suppressive and antidifferentiation mechanisms in part through TAM RTK-dependent STAT3 signaling. The TAM RTKs' ability to regulate multiple aspects of MDSC suppression is significant as it suggests the TAM RTKs function as a myeloid checkpoint and that inhibition of this checkpoint could function as a switch, effectively “turning off” MDSC suppression. Interestingly, PROTEIN S had no effect on STAT3 phosphorylation, suggesting GAS6 and PROTEIN S have nonredundant context-specific signaling roles and to fully appreciate the complexity of innate immune activity the differential signaling aspects of these ligands need to be explored further.

Although checkpoint inhibitor therapy is somewhat successful in melanoma patients, many do not exhibit long-term survival due to initial or acquired resistance (56, 57). A large effort is focused on determining mechanisms of resistance and combination therapies subverting resistance. Circulating MDSC levels are increased in cancer patients (39, 47, 48), negatively correlate with response to anti–PD-1 and anti–CTLA-4 therapy (38, 49) and correlate negatively with patients' overall survival (58). Targeting MDSCs in human cancer is an intriguing thought and being actively investigated (51, 55). Consistent with our murine data, we demonstrate that TAM RTKs are expressed at higher levels in human cancer patients compared with healthy donors. Data presented here suggest that elevated TAM RTK levels on MDSCs can be a potential biomarker and therapeutic target in patients. We demonstrated that inhibiting TAM RTKs with UNC4241 works on the myeloid compartment and augments T-cell directed therapy. These and other data advocate for manipulating both the innate and adaptive immune systems to provide a more efficacious and durable antitumor immune response.

X. Wang has ownership interest (including patents) in Meryx Inc. D.K. Graham is Treasurer, Board of Directors, of and has ownership interest (including patents) in Meryx Inc. S.V. Frye is President of and has ownership interest (including patents) in Meryx Inc. H.S. Earp is Founder and a board member of and has ownership interest (including patents) in Meryx Inc. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Holtzhausen, D.K. Graham, H.S. Earp

Development of methodology: A. Holtzhausen, Y. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Holtzhausen, E. Ubil, D.M. Hunter, D. Zhang, H.S. Earp

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Holtzhausen, D.M. Hunter, D. Zhang, H.S. Earp

Writing, review, and/or revision of the manuscript: A. Holtzhausen, X. Wang, D.K. Graham, H.S. Earp

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Holtzhausen, W. Harris, E. Ubil, J. Zhao, Q. Liu, X. Wang, H.S. Earp

Study supervision: A. Holtzhausen, H.S. Earp

Other (provided compounds tested): S.V. Frye

The authors acknowledge substantial use of the UNC Flow Cytometry (Nancy Fisher, Director), Animal Studies (Charlene Santos, Director), and Animal Histopathology (Stephanie Montgomery, Director) Core Facilities supported by P30 CA016086 Cancer Center Core Support Grant to the UNC Lineberger Comprehensive Cancer Center. Research resources were also supported by the Center for AIDS Research award number 5P30AI050410. The authors thank the Office of Clinical Translation Research (OCTR) for their help with patient samples (particularly Luz Cuaboy and Amy Garrett). Bing Li acquired the UNC4241 NMR spectra. A. Holtzhausen is supported by Postdoctoral Fellowship PF-15-044-01-LIB from the American Cancer Society – South Atlantic Division. H.S. Earp is supported by NIH R01-CA205398, the Saban Family Foundation Team Science Award from the Melanoma Research Alliance, and the Breast Cancer Research Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.