SRF617 Is a Potent Inhibitor of CD39 with Immunomodulatory and Antitumor Properties

The purine nucleoside adenosine plays a critical role in dampening innate and adaptive immune responses under inflammatory conditions (1). In contrast, high levels of extracellular ATP resulting from tissue damage or immunogenic cell death can initiate proinflammatory responses (2). Extracellular adenosine accumulates in cancerous tissues through ATP degradation and constitutes an important mechanism for tumor immune escape, angiogenesis induction, and metastasis (3). Early studies demonstrated that adenosine could inhibit CD3⁺ T cell and NK cell function and that endogenous adenosine levels in tumors would be sufficient to mediate immunosuppression (4). Other studies have shown that high levels of ATP can act as a potent chemoattractant for innate immune cells and mediate robust inflammation, leading to adaptive immune response activation (2). Thus, both extracellular adenosine and ATP play important roles in cancer maintenance and progression. Inhibition of extracellular adenosine accumulation while maintaining high levels of ATP in the tumor microenvironment (TME) may therefore have anticancer therapeutic effects.

Reduction of ATP and subsequent increase of extracellular adenosine production in the TME depend on the concerted enzymatic activity of two ectonucleotidases: CD39 (ENTPD1) and CD73 (NT5E). CD39 catalyzes the hydrolysis of ATP to AMP, and CD73 hydrolyzes AMP into adenosine. Inhibition or loss of CD39 has been shown to have antitumor effects in multiple preclinical studies. Notably, growth of MC38 colon cancer cells and B16 melanoma hepatic metastatic tumors was significantly inhibited in CD39 knockout mice compared with wild-type mice (5). Interestingly, hepatic metastatic growth was reduced in chimeric mice reconstituted with CD39^−/− bone marrow–derived cells, suggesting that CD39 on immune cells alone contributes to tumor growth. A complementary study reported CD39 transgenic mice overexpressing CD39 to have larger, more rapidly growing metastatic tumors after injection with CT26 colorectal cells (6). Emerging clinical study data demonstrate the importance of extracellular adenosine generation and signaling in human cancer progression and that targeting this pathway can benefit cancer treatment. Prolonged disease stabilization in humans has been observed in clinical studies of adenosine receptor inhibitors and CD73 inhibitors (targets downstream from CD39 in the adenosine pathway) as monotherapies and in combination with other therapies, including chemotherapy regimens and immune checkpoint inhibition (7–12). Taken together, these observations demonstrate the potential for targeting CD39 in cancer, further validating the importance of the extracellular adenosinergic pathway in promoting tumor growth.

SRF617 is a fully human anti-CD39 Ab that prevents CD39-mediated conversion of ATP to AMP and phosphate. We demonstrate that SRF617 is a potent inhibitor of CD39 enzymatic activity and can affect immune cell function in vitro and in vivo. We also demonstrate through mouse models that CD39 inhibition in vivo has antitumor effects while lowering adenosine levels and increasing inflammation in the TME. SRF617 is currently being evaluated in clinical studies across multiple tumor types as a monotherapy and in combinations (NCT04336098 and NCT05177770).

The CD39 binding portions of SRF617 were initially discovered using a selection campaign in yeast starting with de novo synthetic human Ab diversity (for review of diversity design and library assembly, see Xu et al. [13]). A library totaling 10¹⁰ in size was initiated to discover Abs with CD39 binding specificity (Adimab). Using a biotinylated form of recombinant CD39 together with MACS on the first two rounds followed by FACS, successive rounds of enrichment narrowed the Ab library diversity to those with the desired CD39 specificity (for a review of techniques, see Chao et al. [14]). This selection was followed by subsequent rounds of affinity maturation of selected Abs with anti-CD39 enzymatic properties, leading to the discovery of SRF617.

The MOLP-8 cell line was obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen and maintained in 10% FBS in RPMI plus 1% penicillin/streptomycin. Human buffy coat and whole blood were obtained from healthy donors at Research Blood Components, LLC. Human B cells were isolated from buffy coat using RosetteSep Human B Cell Enrichment Cocktail (STEMCELL Technologies) followed by Ficoll-Paque (GE Healthcare) density gradient purification centrifugation. Monocytes were isolated from healthy human donor blood using RosetteSep Human Monocyte Enrichment Cocktail (STEMCELL Technologies). HEK293, CT26, and H520 cell lines were purchased from American Type Culture Collection and maintained in 10% FBS in RPMI plus 1% penicillin/streptomycin.

SRF617 Ab was generated by ProBioGen. PE-anti-human CD39 Ab A1 clone (PE-CD39.A1) positive control, anti-huIgG4 isotype control Ab, PE-huIgG Fc secondary Ab, rat anti-murine CD39 (Duha59), and purified anti-human CD16, CD32, and CD64 Abs were purchased from BioLegend. DNP IgG4 single mutant isotype control Ab was generated at WuXi Biologics.

All kinetic analyses on the ForteBio Octet QK^e were performed using 1× kinetics buffer (10× kinetics buffer [ForteBio] diluted in PBS, pH 7.4) as the diluent for the SRF617 ligand and the CD39 analytes. Kinetic analysis by biolayer interferometry (BLI) for recombinant human and cynomolgus monkey CD39 was performed by capturing SRF617 ligand at 1 µg/ml for 5 min using anti-human Fc Octet biosensors followed by a single baseline step of 60 s in 1× kinetics buffer. SRF617-captured biosensors were dipped in wells containing 0 to 1.25 µg/ml of either recombinant human or cynomolgus monkey CD39 analytes for 10 min of association followed by 15 min of dissociation in 1× kinetics buffer. A no-analyte SRF617-captured reference biosensor was included to compensate for the natural dissociation of the captured SRF617 during the analysis. Sample plates were on an orbital shaker (1000 rpm) at 25°C during the analysis.

Kinetic analysis by BLI for recombinant mouse CD39 was performed by covalently capturing 20 µg/ml of either SRF617 or Duha59 ligand in 10 mM acetate, pH 6.0, using the amine-reactive second-generation kit and AR2G biosensors (ForteBio). Following a 120-s baseline step, a 10-min association of the SRF617- or Duha59-captured biosensors was done by dipping into wells containing 0 to 1.25 µg/ml recombinant mouse CD39 followed by a 20-min dissociation in 1× kinetics buffer. A no-analyte SRF617 or Duha59-captured reference biosensor was included to compensate for any dissociation of captured Abs during analysis. Sample plates were on an orbital shaker (1000 rpm) at 25°C during analysis.

All data from 0 to 1.25 µg/ml analyte concentrations were subtracted from the no-analyte (no recombinant CD39) reference biosensor and smoothed, aligned, and fitted using Data Analysis software version 10.0.3.7 (ForteBio) using a 1:1 global binding model.

Cells were plated at 2 × 10⁵ cells/well in a 96-well plate (Thermo Fisher). MOLP-8 cells were treated with a Human TruStain FcX block (BioLegend). B cells were treated with an Fc block mixture containing anti-human CD16, CD32, and CD64 Abs. Cells were incubated with 1.5 ng/ml to 30 μg/ml of either SRF617 or huIgG4 isotype control Ab. PE-huIgG4 Fc secondary Ab was used for detection. Cells were analyzed by flow cytometry, and geometric mean fluorescence intensity was calculated using FlowJo software.

MOLP-8 (5000 cells/well) or RBC-lysed whole blood was added to a 96-well plate (Corning) and allowed to settle for 30 min, after which 0.5 ng/ml to 30 μg/ml of either SRF617 or isotype control Ab was incubated with the cells in duplicate for 60 min, followed by 25 μM ATP (Thermo Fisher) for an additional 60 min. During the incubation, cellular CD39 catalyzed the hydrolysis of ATP to AMP, thereby releasing free phosphate. Malachite green dye (Enzo Life Sciences) was added to supernatants to react with free phosphate, resulting in increased absorbance at 620 nm. Control wells did not contain any Ab. One set of control wells contained cells and ATP added at the beginning of the enzymatic reaction (high absorbance, 0% inhibition); the other set contained cells and ATP added at the end of the reaction (low absorbance, 100% inhibition). Absorbance values were used to calculate the percentage inhibition of ATP hydrolysis and the IC₅₀. Assays were performed in phosphate-free buffer.

Cryopreserved CD4⁺ T cells isolated from PBMCs of three individual donors using a CD4⁺ isolation kit (STEMCELL Technologies) were pooled and prestained with proliferation detection dye using a CellTrace Violet Cell Proliferation Kit (Thermo Fisher). Prestained cells were added to a 96-well plate (Costar) and incubated with 10 μg/ml of SRF617 or isotype control Ab, 250 μM ATP (Thermo Fisher), and anti-CD3/anti-CD28 activating Dynabeads (Thermo Fisher) for 3 d. Human TruStain FcX was used to block unwanted Fc receptor staining, and anti-CD4 allophycocyanin conjugated (BD Pharma) and anti-CD39 PE conjugated (BD Pharma) were used for gating cells during flow cytometry. Gating cells for analysis showed that 14.6% of CD4⁺ cells were CD39⁺. Cells were analyzed for proliferation. For calculation of the proliferation index, each discrete generation of proliferation received a weighted factor; increased proliferation resulted in a larger factor. The percentage of CD4⁺ T cells in each discrete generational gate was multiplied by this weighted factor. The formula for the weighted factor was 2 ⁁ (Generation # − 1). Thus, for Generation 1, 2 ⁁ (1, 1) = 1, and for Generation 4, 2 ⁁ (4, 1) = 8. Proliferation index values were plotted in GraphPad Prism software against each Ab calculation to create a concentration (x-value) versus proliferation index (y-value) plot. A four-parameter nonlinear fit was achieved using GraphPad Prism software.

Monocyte-derived dendritic cells (DCs) were generated by differentiating monocytes from unpurified buffy coats of three individual human donors for 6 d with 40 ng/ml IL-4 (R&D Systems) and 200 ng/ml granulocyte M-CSF (R&D Systems). In the CD86 and HLA-DR isotype expression experiment, cells from all three donors were treated with 10 μg/ml SRF617 or isotype control Ab and incubated in the presence or absence of 300 μM ATP (Thermo Fisher) for 24 h in 96-well plates (Corning). Upon completion of the experiment, cells were incubated with Human TruStain FcX (BioLegend) to prevent nonspecific Ab binding and with LIVE/DEAD Fixable Near-Infrared Dead Cell Stain (Thermo Fisher) to exclude dead cells from downstream FACS analysis. To measure CD86 expression, cells were stained with a PE-conjugated CD86 Ab (BioLegend) and an allophycocyanin-conjugated HLA-DR Ab; the readout was the geometric mean of CD86 and HLA-DR. Cells were analyzed on a BD LSRFortessa X-20. Mean ± SD values for CD86 and HLA-DR geometric mean were calculated. Raw data values were exported from FlowJo. Data were plotted, and an unpaired t test was performed between isotype control Ab and SRF617 treatment for each donor and ATP condition using GraphPad Prism software.

In the IL-16 and IL-12/IL-23p40 secretion experiment, monocyte-derived DCs from donors were treated with 10 μg/ml SRF617 or isotype control Ab and incubated in the presence or absence of 300 μM ATP for 48 h in 96-well plates. Cell supernatants were collected and transferred to a Meso Scale Discovery (MSD) plate, and the MSD protocol was followed to measure IL-16 and IL-12/IL-23p40 secretion. Plates were read on a Meso Quick Plex SQ120.

Monocytes were plated in a 96-well flat-bottomed plate and incubated in RPMI medium containing 10% FBS, 1% penicillin/streptomycin, and 400 ng/ml granulocyte M-CSF (R&D Systems) for 6 d in a 37°C incubator. On day 6, 0.1 × 10⁶ cells/well were replated in a 96-well flat-bottomed plate and allowed to attach overnight in a 37°C incubator. Macrophages were then treated with 10 µg/ml isotype control Ab or SRF617 for 1 h, followed by 10 ng/ml LPS (Sigma-Aldrich) (Escherichia coli serotype 0111:B4) for 3 h and 0, 500, or 1000 µM ATP (Thermo Fisher) for 2 h at 37°C. Supernatant was removed and analyzed for IL-1β and IL-18 by MSD multiplex assay.

Cryopreserved human PBMCs were thawed in RPMI plus 10% FBS, counted, and plated in a 96-well round-bottomed plate at 200,000 cells per well. Cells were placed in a 37°C incubator and rested overnight. Cells were treated with SRF617 at a range of concentrations, returned to the incubator, and rested for 2 d. An additional set of cells was treated just prior to analysis to serve as a T = 0 control to rule out the possibility of Ab cross-blocking. Cells were washed once with flow buffer (PBS plus 2% FBS); 100 μl of FcX blocking solution was added to all wells; and 100 μl of flow staining mixture containing LIVE/DEAD Fixable Near-IR (780) (Thermo Fisher), anti-CD19, anti-CD39, and anti-CD45 was added to all wells. After incubating at 4°C for 30 min, 100 μl of flow buffer was added as a quench. Cells were washed twice with flow buffer, fixed with 1% paraformaldehyde, and analyzed on a BD LSRFortessa X-20.

All animal studies were carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health under protocols approved by the Charles River Accelerator and Development Lab (protocol number CR-008) or Mispro Biotech Services (protocol number 2020-SUR-2) institutional animal care and use committees, and all efforts were made to minimize suffering. Orthotopic pancreatic cancer experiments with human CD39 knock-in (hCD39 KI) mice were performed in accordance with the Canadian laws for animal protection and were approved by the Institutional Animal Protection Committee (Comité Institutionnel de Protection des Animaux, Centre hospitalier de l’Université de Montréal, Montreal, QC, Canada).

All animals completed the study as planned, and behavior and health observations were normal. No animals or data points were excluded from the calculations. Data were analyzed using GraphPad Prism software (version 6.07). Mean ± SEM values were calculated, and treated groups were compared with the isotype control Ab group using two-way ANOVA with Dunnett’s multiple comparison test (MOLP-8 tumor growth inhibition) or two-tailed unpaired t test (H520, hCD39 KI, MOLP-8 immunohistochemistry [IHC], and KPC). A p value <0.05 was considered statistically significant.

Forty-three 5- to 7-wk-old female SCID mice were injected s.c. with 1 × 10⁷ MOLP-8 cells and randomized via matched distribution method using Study Director software into four unblinded groups when tumors reached mean volumes of 100 mm³. Groups in individual cages were treated in the same order i.p. twice per week with either 30 (n = 11), 10 (n = 10), or 3 (n = 10) mg/kg SRF617 or 30 mg/kg (n = 11) huIgG4 isotype control Ab. Tumors were measured in two dimensions using a caliper twice per week, and tumor volumes were calculated as volume = (length × width × width)/2.

Thirty-two 5- to 7-wk-old female SCID mice were injected with MOLP-8 cells and divided into seven unblinded groups (six groups with and one control group without treatment, n = 3/group) 14 d after tumor inoculation. Mice were treated i.p. with a single dose of 10 mg/kg SRF617, euthanized, and tumor and blood samples were collected at 0, 2, 4, 24, 72, 168, and 336 h. Serum prepared from blood samples was tested for the presence of SRF617 by huIgG4 ELISA (Invitrogen). Tumor samples were split into two. One half was embedded in optimal cutting temperature (OCT) compound and flash frozen, and cryosections were prepared for a CD39 in situ ATPase assay. The other half was dissociated and analyzed by flow cytometry for the presence of SRF617 on MOLP-8 cells using PE-conjugated anti-huIgG4 Ab (clone HP6023; SouthernBiotech). A portion of dissociated cells was spiked with 30 mg/ml SRF617 just prior to staining. The percentage target occupancy on CD39 was determined by comparing the spiked samples, representing 100% occupancy, to the test samples.

In situ ATPase activity on OCT-embedded tissues was performed using a previously published protocol (15). Briefly, OCT-embedded, sectioned tissues were incubated in a medium containing ATP (as substrate) in the presence of other ATPase inhibitors and Pb(NO₃)₂ at room temperature for 30 min. Precipitated lead sulfite was imaged for ATPase activity.

Thirty-five Ncr nu/nu mice were injected s.c. with H520 cells and divided into four unblinded groups when tumors reached 100 mm³. Groups in individual cages were treated in the same order i.p. with 15 or 30 mg/kg SRF617 or isotype control (n = 10/group) twice per week. Naive mice (n = 5) were included to determine baseline levels of adenosine. On day 10, blood was collected, and plasma was isolated and analyzed for adenosine by liquid chromatography tandem mass spectrometry.

hCD39 KI mice were purchased from Genoway. These mice were developed from a C57BL/6 background by knocking in the mouse CD39 locus with a chimeric CD39 with human extracellular and murine transmembrane and intracellular domains. Female BALB/c hCD39 KI mice were administered 29 mg/kg SRF617 (n = 3) or sterile Dulbecco’s PBS (n = 3) i.v., and, after 48 h, blood and spleen were collected. Spleen was mechanically dissociated by pushing through a 40-μm nylon mesh filter (Fisher Scientific), and ammonia-chloride-potassium RBC lysis (Thermo Fisher) was performed on resulting splenocyte suspension and isolated blood. Splenocytes and PBMCs were stained with TruStain FcX (BioLegend) for 10 min at 4°C. Samples were stained with viability stain, FITC-conjugated anti-mouse CD19 (clone 65D; BioLegend), Brilliant Violet 421–conjugated anti-mouse CD4 (clone RM4-5; BioLegend), PE-conjugated anti-mouse CD8a (clone 53-6.7; BioLegend), Brilliant Violet 711–conjugated anti-mouse CD11b (clone M1/70; BioLegend), and allophycocyanin-conjugated anti-human CD39 (clone A1; BioLegend) for 30 min at 4°C and analyzed by flow cytometry.

Spleens from 12 female BALB/c hCD39 KI mice were isolated 48 h after the mice received a single unblinded i.v. dose of PBS (n = 3) or SRF617 (10, 20, or 29 mg/kg [n = 3/group]), then formalin fixed and paraffin embedded (FFPE). Dewaxing, Ag retrieval, and staining steps were performed on a Leica Bond Rx instrument. For CD39 staining, FFPE slides were exposed to ER2 solution for 20 min at 100°C for Ag retrieval followed by a 1-h incubation with 63 ng/ml CD39 Ab (clone EPR2046, ab223843; Abcam). For huIgG4 staining, the same Ag retrieval protocol was used, followed by a 1-h incubation with 150 ng/ml anti-huIgG4 Ab (Ab109493; Abcam). Tissues were dehydrated, mounted, and scanned with an Aperio Systems digital scanner. CD39 IHC was quantitated on digitally scanned slides using the Area Quantification module in HALO version 3 (Indica Labs). The result of CD39 IHC quantitation is expressed as the CD39 IHC H-index, which is equal to the percentage of tissue positive for CD39 multiplied by the average OD of tissue positive for CD39 (16).

SCID mice were injected s.c. with MOLP-8 cells. When tumors reached 100 mm³, mice (n = 11/group) were treated i.p. with 20 mg/kg SRF617 or isotype control Ab twice per week for 2 wk. Tumors were collected on day 14 and either formalin fixed (n = 3 per group), lysed for cytokine analysis (n = 3 per group), or dissociated to single-cell suspensions (n = 5 per group). Dissociated tumors were stained with BV421-muCD45, BV711muCD11b, and AF488-muF4/80 for 30 min and analyzed by flow cytometry. The percentage of CD45⁺ cells that was CD11b⁺, F4/80⁺ was gated as macrophages.

Unstained FFPE slides were dewaxed and pretreated with Bond Epitope Retrieval 2 solution (Leica Biosystems) for 20 min on a Leica Bond RX autostainer (Leica Biosystems) following the manufacturer’s recommended protocol. DAKO protein block (Agilent Technologies) was applied to slides and incubated for 15 min. Rabbit anti-human Ki67 primary mAb (Abcam) or rabbit anti-CC3 primary mAb (Cell Signaling Technology) was diluted at 0.4 µg/ml concentration (for Ki-67) or 1:400 (for CC3) in Bond Ab diluent (Leica Biosystems), and 150 µl diluted Ab solution was dispensed on each slide and incubated for 60 min at room temperature. Signal was detected by using a Bond Refine Detection kit (Leica Biosystems). Briefly, slides were incubated with anti-rabbit-HRP polymer for 8 min, with diaminobenzidine for 10 min, and with diaminobenzidine enhancer for 10 min. Slides were counterstained with hematoxylin for 20 min, followed by bluing reagent incubation for 3 min. Slides were manually dehydrated, coverslipped with SignalStain Mounting Medium (Cell Signaling Technology), and digitally scanned at 20× using an Aperio Versa 200 scanner (Leica Biosystems). Quantitative image analysis was performed using the Multiplex IHC algorithm version 3.2.3 (for Ki-67) and Area Quantification algorithm version 2.4.2 (for CC3) in HALO version 3.5 (Indica Labs). The Tissue Classifier algorithm was used to exclude areas of necrosis. The proliferation index indicates the percentage of Ki-67-positive cells. The CCR3 H-index indicates the percentage of tissue that is positive for CC3 multiplied by the average OD of the tissue that is positive for CC3.

F4/80 staining on fixed and sectioned MOLP-8 xenograft tumors was performed with rat anti-mouse F4/80 (Bio-Rad Laboratories). Slides were scanned on an Aperio AT2 whole-slide scanner. Image analysis on digital slide images with Visiopharm software was used to determine F4/80 expression as a percentage positive value within the tumor area region of interest.

For the cytokine analysis, a portion of each tumor was lysed and homogenized with radioimmunoprecipitation assay lysis and extraction buffer (Thermo Fisher) containing Halt protease inhibitor mixture (Thermo Fisher) in a Fast Prep tissue homogenizer (MPBio). After lysis and homogenization, protein concentration was determined by using a bicinchoninic acid assay kit (Thermo Fisher). Each lysate (1 μg/well) was analyzed for murine MCP-1, MIP-1α, and MIP-1β by MSD multiplex assay.

KPC 1245 pancreatic tumor cells (KPC cells) were a kind gift from Prof. David A. Tuveson (Cold Spring Harbor Laboratory). KPC cells were stably transfected by lentiviral vectors containing pCDH-EF1-cOVAv2-T2A-copGFP (donated by Prof. Jonathan Weissman, Stanford University) and CSCW-Gluc-CFP (donated by Dr. Bakhos A. Tannous, Massachusetts General Hospital) to create a tumor cell line expressing OVA and secreting Gaussia luciferase (KPC OVA Gluc). Tumor cells were grown in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Cells were maintained at 37°C in an incubator with a controlled humidified atmosphere composed of 95% air and 5% CO₂.

For orthotopic tumor injections, KPC cells were resuspended in PBS and Corning Matrigel Matrix High Concentration at a ratio of 1:1. A total of 50,000 cells in 10 μl were injected into the pancreas of twenty 8- to 10-wk-old female hCD39 KI mice. Mice received four unblinded doses of SRF617 (30 mg/kg, n = 10) or huIgG4 isotype control Ab (n = 10) by i.p. injections every 3 d starting on day 4. Tumor growth was monitored by measuring the concentration of secreted Gluc in the blood (17). Pancreatic tumors were harvested 15 d after orthotopic tumor implantation and cut in half. One half was processed for FACS analysis, and the other was fixed in 10% formaldehyde for IHC staining for huIgG4.

Tumors for flow analysis were cut with scissors and subsequently digested using collagenase type IV and DNase I. Cell suspensions were filtered using a 70-µm cell strainer and resuspended in PBS, 2% FBS, 5 mM EDTA (FACS buffer). Following Fc block (anti-CD16/32 mAb for 15 min at 4°C), Abs against cell surface Ags were added (CD45 and CD8) (30 min at 4°C) in Brilliant Stain Buffer (BD Biosciences), followed by viability staining in PBS. Fluorescence minus 1 controls were always included as negative controls. Flow cytometry was performed on an LSRFortessa X-20 device (BD Biosciences), and FlowJo software was used for analysis.

SRF617 was discovered using a yeast Ab display selection campaign. After multiple rounds of selection using biotinylated recombinant human CD39 on yeast cells expressing a diverse library of human Abs, SRF617 was selected from a panel of Abs with high-affinity binding to human CD39. Analysis using BLI for SRF617 demonstrated high-affinity binding to recombinant human and cynomolgus monkey CD39, but not murine CD39 (Table I). To assess binding of SRF617 to cellular CD39, flow cytometric analysis was performed on cells expressing high levels of surface CD39 incubated with various concentrations of SRF617. SRF617 binds specifically to HEK293 cells engineered to overexpress full-length human CD39 (HEK293 OE) as compared with lack of binding to the parental HEK293 cells and huIgG4 isotype control Ab (Fig. 1A). SRF617 binds in a specific dose-dependent manner to primary human B cells and the human multiple myeloma cell line MOLP-8 (Fig. 1B, 1C).

Ab-mediated inhibition of ATPase activity was assessed using a malachite green assay to detect phosphate released from ATP hydrolysis. HEK293 OE cells have increased enzymatic activity compared with their parental counterparts (Fig. 1D). SRF617 inhibited the enzymatic ATPase activity of these cells in a concentration-dependent manner (IC₅₀ of 0.3 μg/ml [1.9 nM]), whereas an isotype control Ab did not (Fig. 1E). Similar anti-ATPase activity for SRF617 was observed on MOLP-8 tumor cells (IC₅₀ of 0.1 μg/ml [0.7 nM]) and in PBMCs from whole blood (IC₅₀ of 0.2 μg/ml [1.2 nM]) (Fig. 1F, 1G). In additional experiments, CD39 inhibition by SRF617 was shown to be noncompetitive. Equivalent inhibition was observed with a range of ATP present (Supplemental Fig. 1).

Expression of both CD39 and CD73 on CD4⁺ T cells isolated from PBMCs has been described previously (18, 19). Exogenous ATP, when not degraded, increases proliferation of activated CD4⁺ T cells, but cellular CD39 hydrolyzes exogenous ATP to AMP. AMP is further reduced to adenosine, leading to potent suppression of the proliferation response (20). Therefore, inhibition of CD39 after exogenous addition of ATP should lead to maintenance of ATP levels and reduce the production of adenosine, leading to a stimulatory effect and a loss of suppression on CD4⁺ T cells. To examine the potential of SRF617 to inhibit CD39 and enhance T-cell proliferation in the presence of ATP, a flow cytometry assay of T-cell proliferation was used. Briefly, isolated CD4⁺ T cells were stimulated with anti-CD3/CD28 beads in the presence or absence of ATP and SRF617. Results from three individual donors showed that SRF617 enhanced the proliferation index of CD4⁺ T cells in the presence of ATP by roughly 2-fold (Fig. 2A) compared with an isotype control Ab, demonstrating that SRF617-driven CD39 inhibition prevented the expected decrease in T-cell proliferation caused by ATP reduction to adenosine in vitro.

Markers of DC maturation, such as CD86 expression, have been shown to increase with exogenous ATP stimulation (21). DCs are known to express high levels of CD39 that can rapidly convert ATP to AMP, reducing these ATP-induced effects (21–23). To assess the functional outcome of SRF617-induced CD39 inhibition on DCs in vitro, three markers of DC maturation (CD86 expression, IL-12/IL-23p40 secretion, and IL-16 secretion) were examined on monocyte-derived DCs isolated from buffy coats of individual human donors. SRF617 raised the levels of CD86 on the surface of DCs and induced higher levels of IL-16 and IL-12/IL-23p40 secretion in the presence of ATP, indicating that SRF617 enhanced DC maturation and activity in vitro (Fig. 2B–2D).

Extracellular ATP, in combination with LPS, is known to induce inflammatory responses in macrophages through activation of P2RX7 and subsequent induction of NLRP3 inflammasomes (24). Previous studies have also shown that inhibition of CD39 on macrophages can enhance inflammasome response by elevating or maintaining high extracellular ATP levels, which leads to release of IL-1β and IL-18 (25). To demonstrate the ability of SRF617 to increase inflammasome activity, levels of IL-1β and IL-18 were measured in supernatants from macrophages exposed to LPS and ATP (Fig. 2E, 2F).

To examine the antitumor activity of SRF617 via inhibition of CD39 enzymatic activity in the TME, a MOLP-8 xenograft s.c. tumor model was used. The MOLP-8 model was chosen due to high levels of human CD39 expression on MOLP-8 cells. Flow cytometric analysis to examine SRF617 bound to CD39 on isolated, dissociated tumor cells demonstrated increased target occupancy concomitant to Ab serum levels correlating well with the half-life of SRF617 of 256.6 h (Fig. 3A). SRF617 significantly reduced MOLP-8 tumor growth compared with an isotype control Ab (Fig. 3B). In addition, the use of an in situ enzymatic activity assay in tumor sections from mice treated with SRF617 demonstrated strong inhibition of ATPase activity in the TME within 24 h (Fig. 3C).

The ability of SFR617 to lower levels of circulating adenosine in tumor-bearing mice was assessed using the H520 xenograft model, in which high levels of CD39 are expressed on H520 cells. Significantly higher levels of circulating adenosine were detected in tumor-bearing versus naive mice. SRF617 significantly decreased plasma adenosine levels in tumor-bearing mice compared with isotype control Ab–treated animals (Fig. 3D). Quantitation of cellular proliferation and apoptosis in the MOLP-8 tumors was examined by quantitation of IHC staining for the markers Ki-67 and CC3, respectively. No significant differences in proliferation or apoptosis were observed in xenograft tumor sections from tumors treated with two different dose levels (Fig. 4). In addition, in vitro experiments to examine the potential for direct effect of SRF617 on MOLP-8 cell proliferation were performed. Treatment of MOLP-8 cells with SRF617 in vitro did not modulate their proliferation (Supplemental Fig. 2).

To better understand SRF617-driven tumor growth inhibition, we characterized immune cell infiltration in the TME. MOLP-8 xenograft tumors were examined by IHC and flow cytometry for the presence of CD45⁺ cells and F4/80 macrophages following treatment. Flow cytometric analysis of dissociated tumors from SRF617-treated animals revealed significantly higher levels of CD45⁺ immune cell infiltration compared with isotype control Ab–treated tumors (Fig. 5A). IHC imaging revealed increased staining of F4/80 macrophages in the tumors of SRF617-treated mice (Fig. 5B, 5C). To elucidate the mechanism of macrophage recruitment, tumor lysates were prepared and analyzed for murine cytokine levels by MSD. Murine macrophage-attracting cytokines MIP-1α, MIP-1β, and MCP-1 were all significantly elevated in the SRF617-treated tumors compared with isotype controls (Fig. 5D–5F).

To further characterize the mechanistic effects of SRF617 in syngeneic mouse tumor models, hCD39 KI mice were used. After a single dose of SRF617, flow cytometry revealed significant reduction of CD39 expression on circulating CD19⁺ and CD11b⁺ cells and the same cells from isolated spleens (Fig. 6A). Dose-dependent reduction of CD39 expression was similarly observed on CD19⁺ B cells from human PBMCs after treatment with SRF617 (Fig. 6B). The T = 0 cross-blocking control curve demonstrates that the flow Abs for all flow studies that were used do not significantly interfere with each other’s binding. No change in viability of B cells was observed. To further investigate the reduction of CD39 expression, spleens from hCD39 KI mice treated with a single dose of SRF617 were harvested 3, 7, and 14 d after administration and analyzed by IHC for CD39 expression. Significant decreases in CD39 expression were observed for all doses tested on days 3 and 7. At 14 d, CD39 expression had recovered, especially at the lower 10 mg/kg dose (Fig. 6C, 6D).

To observe the ability of SRF617 to achieve exposure in an orthotopic tumor setting, hCD39 KI mice were implanted orthotopically in the pancreas with KPC tumor cells and treated with multiple doses of SRF617 or isotype control Ab. The resulting tumors were examined 15 d after implantation by IHC to observe SRF617 staining (huIgG4) and analyzed by flow cytometry to measure CD8⁺ T-cell infiltration. The orthotopic tumors showed strong SRF617 staining in the viable tumor regions, as evidenced by IHC signal for huIgG4, whereas nonspecific isotype control Ab staining was observed only in the necrotic core (Fig. 7A). Flow cytometric analysis of dissociated tumor revealed that SRF617 induced a significantly higher level of CD8⁺ T-cell tumor infiltration than isotype control Ab when measured as CD8⁺ T cells as a percentage of total immune infiltrate or absolute count per milligram of tumor (Fig. 7B).

Targeting the adenosine pathway in combination with other therapies is emerging as an attractive approach to treating cancer. Investigational drugs in development include inhibitors of major components in the pathway (i.e., the ectoenzymes CD39 and CD73 and adenosine receptors A2a/A2b). Multiple clinical studies with CD73 inhibitors and adenosine receptor antagonists have shown positive clinical outcomes as monotherapy and in combination with chemotherapy and/or immune checkpoint inhibitors (7–12). In a randomized phase 2 study in non–small cell lung cancer, the anti-CD73 Ab oleclumab in combination with anti-PD-L1 agent durvalumab provided a significant improvement in progression-free survival over durvalumab alone (hazard ratio, 0.44) (26). These results demonstrate great potential for targeting adenosine production in the TME to relieve immune suppression. Although targeting CD73 or adenosine receptors can attenuate immune suppression in the TME, targeting of CD39 provides an additional immune-stimulatory effect by stabilizing ATP levels in tumors. This study describes the mechanism of action and antitumor potential of SRF617. SRF617 inhibits CD39 ATPase activity in the TME, which concomitantly stabilizes ATP to increase inflammatory response and lowers adenosine to remove immune suppression.

High levels of ATP (up to millimolar amounts) can be produced in the TME due to cellular stress, cytotoxicity, and cell damage (2, 27). Extracellular ATP, a potent “danger signal” acting mostly through the P2RX7 purinergic receptor on immune cells, stimulates immune responses on multiple cell types, primarily those of myeloid lineage (28–31). P2RX7 receptors have a relatively low affinity for ATP, so the inflammation response occurs only if high levels of ATP are maintained. CD39 is present at high levels in many types of tumors and can rapidly convert ATP to AMP, contributing to immune evasion by lowering ATP levels. We demonstrate that SRF617 can potentiate ATP effects in vitro in two ways: by stimulating maturation and cytokine secretion of DCs and by enhancing the release of IL-1β and IL-18 from macrophages. ATP-activated DCs have been shown to activate cytotoxic T cells more potently (31). Release of the cytokines IL-1β and IL-18 after exposure to extracellular danger signals initiates a powerful inflammatory response that leads to recruitment of immune cells, pyroptosis, and potentiation of CD8⁺ T cell effector function (28). In tumors, ATP signaling through P2RX7 and NALP3 brought about by inhibiting CD39 has been shown to stimulate these responses in the TME and is intrinsic to the antitumor activity by blocking CD39 enzymatic activity (32). In the MOLP-8 xenograft model, we show a large increase in tumor macrophage infiltration upon treatment with SRF617 associated with SRF617 binding to CD39 on MOLP-8 cells and inhibition of CD39 ATPase activity. ATP is a potent chemoattractant for macrophages, making it likely that these cells are responding to an increased steady-state level of ATP (29). In addition, in tumors treated with SRF617, we observed an increase of murine cytokines that attract macrophages (MCP-1, MIP-1α, and MIP-1β). Given that the cytokines detected are murine proteins, they are likely originating from the stromal compartment within the TME and are indicative of an enhanced state of inflammation. ATP can also modulate the function of other immune cells, including CD4⁺ and CD8⁺ T cells, by boosting proliferation and inducing chemotaxis (2, 27, 28, 31, 33). Finally, ATP has been shown to enhance metabolism in memory CD8⁺ T cells (33). We demonstrate that SRF617 enhances CD4⁺ T-cell proliferation in the presence of ATP and induces higher numbers of CD8⁺ T cells in the TME of an orthotopic pancreatic tumor in vivo. The increase in CD8⁺ T cells in the orthotopic tumor model could be attributed to either elevated ATP levels causing increased chemotaxis or a decrease in adenosine causing less immune suppression in the TME or both. At a minimum, it suggests that inhibition of CD39 in the TME is leading to a higher state of inflammation. Further studies into the exact mechanism for increased CD8⁺ T cell infiltration will be required to fully understand this phenomenon.

On the other end of the pathway, adenosine is a potent immune suppressor with wide-reaching effects on multiple immune cell types, including T, NK, and myeloid cells (20). Reduction of CD4⁺ T-cell proliferation by ATP is driven mostly by the rapid production of adenosine and subsequent signaling via the inhibitory A2aR. T-cell proliferation can also be suppressed by AMP or adenosine and can be reversed by CD73 enzymatic inhibition or adenosine receptor antagonism (20). We show that SRF617 can reduce the production of adenosine from ATP and rescue proliferation of CD4⁺ T cells in the presence of ATP. Furthermore, we demonstrate that SRF617 can lower systemic adenosine levels in the H520 tumor model, where adenosine levels are elevated by the presence of tumor.

We show that SRF617 is a potent inhibitor of the ATPase activity of CD39 and observe that SRF617 can lower expression of CD39 on immune cells, further lessening its effectiveness as an ATP degrader. Loss of cell surface CD39 was observed in circulating immune cells from SRF617-treated hCD39 KI mice and on ex vivo cultured primary human CD19⁺ B cells that cannot be attributed to detection Ab cross-blocking or cellular toxicity. It is also unlikely that SRF617 could cause significant Ab-mediated cell death because it is an Fc silent IgG4 Ab. Loss of CD39 expression on B cells and other CD39-expressing cells is a phenomenon also observed in an ongoing SRF617 phase 1 clinical study (NCT04336098). In humans treated with SRF617, a dose-dependent reduction of CD39 was observed on circulating B cells correlating with CD39 target occupancy. In addition, a reduction in CD39 expression was observed in matched tumor biopsies from before and after SRF617 treatment (34).

In murine tumor models (s.c. and orthotopic), SRF617 penetrates the TME and shows high levels of exposure that correlate with serum levels. We observed inhibition of the in situ enzymatic activity of CD39 correlating with maximal SRF617 occupancy in MOLP-8 tumors. In an orthotopic KPC tumor model study conducted in hCD39 KI mice, we observed stromal IHC staining for huIgG4 in the viable compartment of the TME of SRF617-treated tumors and no staining in the TME of isotype control–treated tumors. The presence of huIgG4 in these tumors can be attributed only to the presence of SRF617. Although staining is observed in the necrotic compartments of both treatment groups, this is likely due to nonspecific staining that is common in necrotic tissue. Taken together, these results provide strong evidence for SRF617 tumor exposure, enzymatic inhibition, and increased tumor immune cell infiltration in the TME.

The properties of SRF617 make it an excellent drug development candidate targeting the ectoenzyme CD39. Furthermore, by targeting the first enzyme in the ATP-degrading adenosine pathway, and thereby both maintaining high levels of ATP and reducing levels of adenosine in the TME, SRF617 has great potential for cancer treatment, especially in combination with immunogenic agents or other immuno-oncology–based treatments.

M.C.W., S.M., M.R., M.P., R.J.H., A.D., D.B., S.D.Y., S.K., R.M., Y.Y., R.P., J. Strand, T.Z., E.D., B.H.L., P.H., V.J.P., and A.C.L. are current or former employees and shareholders, and N.R. is an employee of Surface Oncology. J. Stagg is a Scientific Advisory Board member at Surface Oncology and received a research grant and consulting fees for this work. M.C.W., M.R., A.D., S.D.Y., T.Z., E.D., P.H., and A.C.L. are inventors on issued patents and pending patent applications relating to Antibodies that bind CD39 and uses thereof and anti-CD39 antibody compositions and methods. C.J.F. has no financial conflicts of interest.

We are grateful to Dr. Robert Ross and Dr. Jonathan Hill for intellectual input and to Karen D’Amour for editing. We are also grateful to Dr. David Tuveson for supplying KPC cells for the orthotopic studies.