ENPP1 inhibitor with ultralong drug-target residence time as an innate immune checkpoint blockade cancer therapy

Abstract

Only one in five patients respond to immune checkpoint inhibitors, which primarily target adaptive immunity. Ectonucleotide pyrophosphatase/phophodiesterase 1 (ENPP1), the dominant hydrolase of 2'3'-cyclic-GMP-AMP (cGAMP) that suppresses downstream stimulator of interferon genes (STING) signaling, has emerged as a promising innate immunotherapy target. However, existing ENPP1 inhibitors have been optimized for prolonged systemic residence time rather than effective target inhibition within tumors. Here, we report the characterization of STF-1623, a highly potent ENPP1 inhibitor with an exceptionally long tumor residence time despite rapid systemic clearance, enabled by its high ENPP1 binding affinity and slow dissociation rate. We show that membrane-bound ENPP1 on tumor cells, not the abundant soluble ENPP1 in serum, drives tumor progression. Consequently, STF-1623 unleashes anti-tumor immunity to produce robust anti-tumor and anti-metastatic effects across multiple tumor models. Conceptually, this work establishes a noncovalent small-molecule inhibitor of ENPP1 with ultralong drug-target engagement as a safe and precise strategy to activate STING within tumors.

Keywords: STING agonist; breast cancer; colorectal cancer; glioblastoma; immunotransmitter; pancreatic cancer.

Graphical abstract

Figure 1

Systemically administered STF-1623 concentrates in ENPP1-expressing tumors (A) Chemical structure of STF-1623. (B and C) Pharmacometrics of STF-1623 (50 mg/kg) administrated to mice bearing subcutaneous EMT6 tumors (B) or MC38 tumors (C). IC₉₅ = 14 ng/mL or g; below detection limit (bdl) = 1 ng/mL or g. Points bdl were not plotted. Mean ± SEM is plotted, n = 2–3 mice. (D) Enpp1 RNA expression in EMT6 and MC38 cell lines. Mean ± SD is plotted. p value is determined by unpaired two-sided t test. TPM, transcript per million. (E) Single-cycle kinetics analysis with surface plasmon resonance of the direct binding of 0.12, 0.36, 1.11, 3.33, and 10 nM to N-terminally Avi-tagged and biotinylated mouse (left) or human (right) ENPP1 immobilized on an streptavidin chip. Representative curves are plotted, n = 3 for moues ENPP1 and n = 4 for human ENPP1. Drug-target residence time (τ) is calculated as reciprocal of the dissociation rate constant (K_off). ∗∗∗∗p < 0.0001. See also Figure S1.

Figure 2

Co-crystal structure reveals the molecular determinants of STF-1623’s potency and specificity toward ENPP1 (A) The 2.7 Å crystal structure of STF-1623 (pink spheres) bound to human fxENPP1 (teal surface: nuclease-like domain; purple surface: phosphodiesterase domain; gray surface: somatomedin B domain). (B) Structural alignment between fxENPP1 (=ENPP3^Q244K/E275D, purple, shown without ligand) and ENPP1 (wheat, PDB: 6WFJ). (C) Expanded view of STF-1623 (pink sticks/spheres) bound in the active site of fxENPP1 (purple sticks/cartoon). Zincs are shown as dark purple spheres. Electron density maps of STF-1623 are shown as black mesh (2Fo-Fc, 0.7 root-mean-square deviation [RMSD]) and cyan mesh (Fo-Fc, 2.3 RMSD). (D) Schematic drawing of interactions formed between STF-1623 (pink) and the fxENPP1 active-site residues (black). Residues within 4 Å are shown (not including the zinc-coordinating residues). Metal coordination shown as gray dashed lines, hydrogen bonds shown as black dash lines, aromatic interactions shown as black wedged lines, and hydrophobic or polar interactions shown as spokes. (E) STF-1623 (pink sticks) interacting with fxENPP1 residues K244 and D275 (gray sticks) overlaid with apo ENPP3 (PDB: 6C01) residues Q244 and E275 (purple sticks). (F) In vitro dose-inhibition curves to determine IC₅₀ values of STF-1623 against purified human ENPP proteins. Enzyme concentrations of 15 pM (ENPP1) or 250 pM (all others) were incubated with 2 μM cGAMP and various concentrations of STF-1623 overnight. cGAMP degradation was measured with cGAMP-luc assay. Mean and standard deviation of 3–4 biological replicates are shown. Data from ENPP1, ENPP3^WT, and fxENPP1 are replotted from Figure S2C. IC₅₀ values: ENPP1 = 0.6 nM, ENPP3^WT = 800 nM, ENPP3^Q244K = 13 nM, ENPP3^E275D = 130 nM, fxENPP1 (=ENPP3^Q244K/E275D) = 1.4 nM. (G) Overlay of STF-1623 (pink) with STF-1084 (blue, PDB: 6XKD), bound to human fxENPP1 and mouse ENPP1, respectively. Ligands are shown as sticks. Protein residues K244 and D275 (fxENPP1), K277 and D308 (mENPP1), and zinc atoms are shown. (H) Structures of STF-1623 with fxENPP1 and STF-1084 with mENPP1 (top) and modeled ligands in their non-preferred poses. Modeling was performed in Pymol by building the appropriate methoxy groups in the desired ligand pose.

Figure 3

Intratumoral ENPP1 membrane retention determines tumor progression (A) Expression of mENPP1, hENPP1, and hENPP3 in supernatant and lysate of 293T ENPP1^−/− cells assessed by reducing western blotting. (B) Activity of mENPP1 and hENPP1 in the supernatant and lysate assessed by [³²P] cGAMP hydrolysis by thin-layer chromatography after 4 h of reaction at pH 9.0. (C) Kinetics of cGAMP hydrolysis by 10 nM purified memENPP1 and secENPP1 at pH 9.0. Mean ± SEM is plotted, n = 2 independent experiments for memENPP1; n = 3 technical replicates from 2 independent experiments for secENPP1. Data are fit with Michaelis-Menten model. (D and E) Expression of WT ENPP1, MPD residue mutants (D), and A84 residue mutants (E) in the supernatant and lysate of 293T cGAS ENPP1^−/− cells assessed by reducing western blotting. memENPP1, secENPP1, and cleaved ENPP1 are indicated as red, blue closed, and blue open triangles, respectively. Relative secENPP1 expression is calculated as secENPP1 over the sum of secENPP1 and memENPP1. Quantification is from two independent experiment, with blots from one representative experiment shown. (F) 4T1 ENPP1^WT-OE (n = 20 mice from two independent experiments), 4T1 ENPP1^A84S-OE (n = 25 mice from two independent experiments) cells were orthotopically injected in WT BALB/c mice. Mean ± SEM of tumor volume is plotted. The p value for tumor volume was determined by unpaired t test, while for Kaplan-Meier curve, it was determined by log rank Mantel-Cox test. (G and H) Relative cGAMP hydrolysis activity from tumor lysates (G) and sera (H) from randomly selected mice in (F) reaching experimental endpoints (n = 15, 22 for 4T1 ENPP1^WT-OE and 4T1^A84S-OE, respectively). Mean ± SD is plotted. p value is determined by unpaired two-sided t test. mENPP1, mouse ENPP1; hENPP1, human ENPP1; hENPP3, human ENPP1; S, supernatant; L, lysate; memENPP1, transmembrane ENPP1; secENPP1, secreted ENPP1; MPD, membrane proximal domain. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figures S2 and S3.

Figure 4

STF-1623 boosts intratumoral cGAMP levels to suppress murine breast cancer growth and metastasis (A–C) cGAMP levels (A), IFN-γ levels (B), and relative Ifn-γ mRNA expression (C) in subcutaneous EMT6 (top) or EMT6 shEnpp1 (bottom) tumors at time points indicated after STF-1623 (50 mg/kg) injection. For cGAMP measurement, background measured in vehicle controls was subtracted from the measurements. Mean ± SEM is plotted, n = 3 for EMT6 tumors; n = 4–5 for EMT6 shEnpp1 tumors. (D) BALB/c female mice with established subcutaneous EMT6 tumors received treatment as indicated. Mean ± SEM is plotted. (E) Percent lung metastatic burden determined by hematoxylin and eosin staining of mice in (D) on day 29 of the study. Mean ± SD is plotted, n = 9–10 mice. (F) Immunohistochemistry of CD8⁺ T cells in randomly selected tumors of mice in (D) on day 29 of the study. Mean ± SD is plotted, n = 5 mice. (G) BALB/c female mice with established orthotopic 4T1 tumors received treatment as indicated. Mean ± SEM of tumor volume is plotted. (H) BALB/c female mice with established orthotopic 4T1 tumors received treatment as indicated. Mice were sacrificed when exhibiting symptoms related to metastasis. The p value for the Kaplan-Meier analysis was determined by log rank test. p values were determined by two-sided unpaired t test unless otherwise mentioned. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; p value is shown if between 0.05 and 0.1. See also Figure S4.

Figure 5

STF-1623 delays murine colorectal tumor growth (A) C57BL/6 female mice with established subcutaneous MC38 tumors received treatment as indicated. Mean ± SEM of tumor volume is plotted. (B) Percent weight change of mice in (A) on day 14 compared to day 1 of the study. Mean ± SD is plotted, n = 10 mice per group. (C) ENPP1 activity of WT, ENPP1^WT-OE, and ENPP1^T238A-OE MC38 cell lysates assessed by [³²P] cGAMP hydrolysis by thin-layer chromatography. (D) MC38 ENPP1^WT-OE and MC38 ENPP1^T238A-OE cells were subcutaneously injected in WT female BALB/c mice. n = 9 for ENPP1^WT-OE and n = 10 for ENPP1^T238A-OE. Mean ± SEM of tumor volume is plotted. (E) C57BL/6 female mice with established subcutaneous MC38 ENPP1^WT-OE tumors received treatment as indicated. Mean ± SEM of tumor volume is plotted, n = 8–9 mice. (F) Percent weight change of mice in (E) on day 14 of study compared to day 1 of the study. Mean ± SD is plotted, n = 8 or 9 mice per group. (G) BALB/c mice with established CT26 tumors received treatment as indicated. Mean ± SEM is plotted, n = 10 mice. (H) Percent weight change of mice in (G) on day 13 compared to day 1 of the study. Mean ± SD is plotted, n = 10 mice. (I) BALB/c mice with established subcutaneous CT26 tumors received 5 or 50 mg/kg of STF-1632 on days 9–11, 16–18 post inoculation (pi), 10 mg/kg anti-PD-1 intraperitoneally on day 9, 12, 16, and 19 pi, or a combination of the two. On day 23 of the study, tumors were isolated and processed for flow cytometry. The number of CD3⁺ T cells, CD4⁺CD3⁺ T cells, and CD8⁺CD3⁺ T cells per gram of tumors is shown. Mean ± SEM is plotted, n = 10 mice. p values were calculated by two-sided unpaired t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 6

STF-1623 controls Panc02 pancreatic tumor growth by activating anti-cancer immunity (A and B) BALB/c female mice with established subcutaneous Panc02 tumors received treatment indicated. On day 12 of the study, tumors were processed for flow cytometry: the percentage of IA-IE⁺CD206^low M1 macrophage (F4/80⁺GR-1⁻) over IA-IE⁻CD206^high M2 macrophage, the number of CD335⁺CD3⁻ NK cells, CD335⁺CD3⁺ NKT cells, CD4⁺CD3⁺ T cells, and CD8⁺CD3⁺ T cells per gram of tumors, and the ratio of CD8⁺ T cells over Foxp3⁺CD4⁺ Treg cells (A), CD69⁺CD4⁺ T cells, PD-1⁺CD4⁺ T cells, Ki67⁺CD4⁺ T cells, CD69⁺CD8⁺ T cells, PD-1⁺CD8⁺ T cells, and Ki67⁺CD8⁺ T cells (B). Mean ± SD is plotted, n = 6 mice for vehicle and STF-1623 groups, and n = 3 for all other combination therapy groups. (C) BALB/c female mice with established Panc02 tumors received the treatment as indicated. Mean ± SEM of tumor volume is plotted, n = 9–12 mice. Spider plots of individual tumor growth are shown. (D) Percent weight change of mice in (C) on day 45 compared to day 1 of the study. Mean ± SD is plotted, n = 9–12 mice. NK, natural killer; NKT, natural killer T; Treg, regulatory T cells; IR. ionizing radiation. p values were calculated by two-sided unpaired t test unless otherwise noted. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; p value is shown if between 0.05 and 0.1.

Figure 7

STF-1623 crosses the BBB and exhibits efficacy in delaying GL-261 GBM growth (A) Concentration of STF-1623 in serum, cerebrospinal fluid (CSF), and brain of mice after one dose of STF-1623 (50 or 100 mg/kg) injection. IC₉₅ = 14 ng/mL or g; below detection limit (bdl) = 1 ng/mL or g. Measurements bdl were not plotted. Mean ± SEM is plotted, n = 2–3 mice. (B) C57BL/6 mice with established intracranial GL261 Red-Fluc tumors received the treatment as indicated. Kaplan-Meier of survival determined to death or humane endpoint is plotted, n = 8 mice. The p value was determined by log rank Mantel-Cox test. (C) Quantification and raw images of the total flux (p/s) of mice in (B) on day 28 of the study of groups with TMZ. Mean ± SD is plotted, n = 8 mice. (D) Percent weight change of mice in (B) on day 11 of the study. Mean ± SD is plotted, n = 9–10 mice. IR, ionizing radiation; TMZ, temozolomide. p values were calculated by two-sided unpaired t test unless otherwise noted. ∗p < 0.05. ∗∗p < 0.01.