Hydrolysis of 2'3'-cGAMP by ENPP1 and design of nonhydrolyzable analogs

Abstract

Agonists of mouse STING (TMEM173) shrink and even cure solid tumors by activating innate immunity; human STING (hSTING) agonists are needed to test this therapeutic hypothesis in humans. The endogenous STING agonist is 2'3'-cGAMP, a second messenger that signals the presence of cytosolic double-stranded DNA. We report activity-guided partial purification and identification of ecto-nucleotide pyrophosphatase/phosphodiesterase (ENPP1) to be the dominant 2'3'-cGAMP hydrolyzing activity in cultured cells. The hydrolysis activity of ENPP1 was confirmed using recombinant protein and was depleted in tissue extracts and plasma from Enpp1(-/-) mice. We synthesized a hydrolysis-resistant bisphosphothioate analog of 2'3'-cGAMP (2'3'-cG(s)A(s)MP) that has similar affinity for hSTING in vitro and is ten times more potent at inducing IFN-β secretion from human THP1 monocytes. Studies in mouse Enpp1(-/-) lung fibroblasts indicate that resistance to hydrolysis contributes substantially to its higher potency. 2'3'-cG(s)A(s)MP is therefore improved over natural 2'3'-cGAMP as a model agonist and has potential as a vaccine adjuvant and cancer therapeutic.

Figure 1. There is one dominant hydrolase activity in the plasma membrane/heavy organelles and it is not due to PDE12

(a) 2′3′-cGAMP and 3′3′-cGAMP hydrolysis reactions. THP-1 cells were lysed using 1% NP-40. The ³²P-labeled CDNs were incubated with the cell lysate for 20 hours and the reaction was monitored using a TLC assay and visualized by autoradiography. (b) 2′3′-cGAMP hydrolase activity in differential centrifugation fractions of MDA-MB231 cells. Cells were suspended in isotonic buffer and lysed with a Dounce homogenizer. Supernatant and pellets from spins at the indicated speeds were solubilized and assayed in 1% NP-40, 20 mM Tris-HCl, pH 7.5, 1 mM Ca²⁺. (c) Knockdown of PDE12 does not decrease hydrolase activity. MDA-MB231 cells were transfected with four siRNA oligos against PDE12 and a control oligo. After 4 days, cells were lysed in the same buffer as (b) and assayed for activity. Upper panel: hydrolase activity in cells treated with different siRNA oligos. Lower panel: PDE12 levels in these cells. (d) Activity of purified PDE12 compared with that of MDA-MB231 whole cell lysate.

Figure 2. ENPP1 is an efficient hydrolase for 2′3′-cGAMP

(a) Ion dependency and (b) pH preference of the dominant hydrolase activity in MDA-MB231 cells. (c) Activity of recombinant ENPP1 alone or (d) coupled to alkaline phosphatase in buffer condition: 0.2% NP-40, 20 mM Tris-HCl, pH 9.0, 2 mM Ca²⁺, 200 μM Zn²⁺. (e) Kinetics of 2′3′-cGAMP and ATP hydrolysis by recombinant ENPP1. 1 nM ENPP1 was tested in the same buffer condition as (f) and (g). Data are presented as mean and standard error.

Figure 3. ENPP1 is the dominant hydrolase activity for 2′3′-cGAMP

(a) Knockdown of ENPP1 in MDA-MB231 cells diminished hydrolase activity. Four siRNA oligos against ENPP1 with a control siRNA sequence against PDE12 were used. 4 days after siRNA transfection, cells were lysed and assayed for activity (upper panel) and blotted for ENPP1 level (lower level). (b) Hydrolase activity in plasma from Enpp1^-/- mice and their littermates. (c) Western blot characterization of Enpp1^-/- mice. (d) Hydrolase activity in livers and spleens from Enpp1^-/- mice and their littermates. Livers and spleens were minced and Dounce homogenized in lysis buffer: 1% NP-40, 20 mM Tris-HCl, pH 7.5, protease inhibitor cocktail. The assay was conducted in 0.2% NP-40, 20 mM Tris-HCl, 150 mM KCl, 2 mM Ca²⁺, 2 mM Mg²⁺, 200 μM Zn²⁺ at the indicated pH. NaOAc buffer was used for pH 5.0-6.0; PIPES buffer was used for pH 6.5 and 7.0 and 2′3′-cGAMP runs at a lower R_f in this buffer; Tris-HCl buffer was used for pH 7.5-9.0, and Borate buffer was used pH 9.5. These buffer conditions were also used in liver and spleen extract studies.

Figure 4. Development of hydrolysis resistant hSTING agonists

(a) Scheme of enzymatic synthesis of 3′3′-cGAMP and 2′3′-cGAMP analogs. For 3′3′-cGAMP, 50 nM of DncV was incubated with 1 mM ATP and 1 mM GTP in 1 mL buffer containing 20 mM Tris-HCl, pH 8.0, 20 mM MgCl₂ for 3 h at room temperature. For 2′3′-cGAMP analogs, 1-10 μM mouse cGAS (amino acid residues 147-507) was incubated with 1 mM ATP, 1 mM GTP, and 0.1 mg/mL HT-DNA in 1 mL of the same buffer for 12 h at room temperature. (b) Hydrolysis reactions of ENPP1 (1 nM) with the analogs (10 μM). (c) Scintillation proximity assay to measure the binding affinity of the analogs towards hSTING (amino acid residues 139-379). Biotinylated hSTING (100 nM) was immobilized onto 96-well streptavidin coated SPA plates. Neat ³⁵S-labeled 2′3′-cGA^sMP (500 pM) was used as the probe. N=3 samples. Data are presented as mean and standard error. (d) IFN-β production in THP-1 cells stimulated with the analogs. THP-1 cells were incubated with the analogs at the indicated concentrations for 24 h. IFN-β in the media was measured using a HEK-SEAP cell line. Representative data from a biological triplicate is depicted.

Figure 5. ENPP1 dampens 2′3′-cGAMP signaling

Lung fibroblast cells from wild type and Enpp1^-/- female mice were incubated with 2′3′-cGAMP analogs and HSV-60 at the indicated concentrations for 24 h. IFN-β in the media was measured using a B16-SEAP cell line. N=3 samples. Data are presented as mean and standard error.