ENPP1's regulation of extracellular cGAMP is a ubiquitous mechanism of attenuating STING signaling

Abstract

The metazoan innate immune second messenger 2′3′-cGAMP is present both inside and outside cells. However, only extracellular cGAMP can be negatively regulated by the extracellular hydrolase ENPP1. Here, we determine whether ENPP1’s regulation of extracellular cGAMP is a ubiquitous mechanism of attenuating stimulator of interferon genes (STING) signaling. We identified ENPP1H362A, a point mutation that cannot degrade the 2′-5′ linkage in cGAMP while maintaining otherwise normal function. The selectivity of this histidine is conserved down to bacterial nucleotide pyrophosphatase/phosphodiesterase (NPP), allowing structural analysis and suggesting an unexplored ancient history of 2′-5′ cyclic dinucleotides. Enpp1H362A mice demonstrated that extracellular cGAMP is not responsible for the devastating phenotype in ENPP1-null humans and mice but is responsible for antiviral immunity and systemic inflammation. Our data define extracellular cGAMP as a pivotal STING activator, identify an evolutionarily critical role for ENPP1 in regulating inflammation, and suggest a therapeutic strategy for viral and inflammatory conditions by manipulating ENPP1 activity.

Keywords: ENPP1; STING; cGAMP; extracellular cGAMP; immunotransmitter.

Fig. 1.

Discovery and characterization of ENPP1^H362A, a mutation that degrades ATP but not cGAMP. (A and B) Schematic illustration of the ENPP1 active site bound to substrates cGAMP (A) and ATP (B). (C) Schematic illustration of proposed mutations (red Xs) of the residues in the guanosine-adjacent site (red). (D) Heat map showing the initial velocity relative to WT of guanosine-adjacent residue mutations for the substrates cGAMP and ATP at pH 9. The mean initial velocity was calculated from a linear fit of the degradation reactions during early time points; n = 4 independent reactions. (E) Schematic illustration of proposed mutations (red Xs) of the zinc-binding residues (zincs shown as dark gray spheres). (F) The catalytic center of mouse ENPP1 (PDB ID code: 4GTW). Zinc-binding residues shown as gray (D200, T238, D358, D405, H406, and H517) or orange (H362) sticks; zincs shown as dark gray spheres. (G) Heat map showing the initial velocity relative to WT of zinc-binding residues for the substrates cGAMP and ATP at pH 9. The mean initial velocity was calculated from a linear fit of the degradation reactions during early time points; n = 3 independent reactions. (H) Heat map showing the initial velocity relative to WT of H362X mutations (where X is each amino acid), for the substrates cGAMP and ATP at pH 9. The mean initial velocity was calculated from a linear fit of the degradation reactions during early time points; n = 3 independent reactions. (I) Kinetic analysis of cGAMP activity monitoring degradation products by thin-layer chromatography (TLC) using recombinant purified ENPP1; n = 3 independent reactions, mean ± SD with some error bars too small to visualize. (J) Michaelis–Menten plots of ATP activity monitoring AMP production using recombinant purified ENPP1; n = 3 independent reactions, mean ± SEM. WT: K_m = 12.1 µM, k_cat = 0.76 s⁻¹, k_cat/K_m = 6.3 × 10⁴ M⁻¹s⁻¹. H362A: K_m = 11.1 µM, k_cat = 0.79 s⁻¹, k_cat/K_m = 7.1 × 10⁴ M⁻¹s⁻¹.

Fig. 2.

Bacterial NPP selectively cleaves 2′-5′ linkages in cyclic dinucleotides using the conserved histidine. (A) Sequence alignment of NPP from a range of species showing the conserved histidine (highlighted in green box). (B and C) TLC of ATP (B) and cGAMP (C) degradation by Xac NPP^WT compared to Xac NPP^H214A (both at 3 µM enzyme concentration). In B, n = 3 independent reactions, mean ± SD with some error bars too small to visualize; C is representative of three independent reactions from two independent protein purifications. (D–F) Degradation of 3′2′-cGAMP (D), 2′3′-CDA and 3′3′-CDA (E), and 2′3′-CDG and 3′3′-CDG (F) by Xac NPP^WT and NPP^H214A. Product formation was measured by coupling the reaction to 1 U alkaline phosphatase and measuring resultant P_i; n = 3 independent reactions, mean ± SD with some error bars too small to visualize. Data were fit to one-phase decay.

Fig. 3.

Structure-guided elucidation of the substrate-selective degradation mechanism of ENPP1^H362A. (A) Cocrystal structure of Xac NPP^T90A (gray cartoon and sticks) with linear cGAMP intermediate pApG (gray spheres and sticks). Zinc atoms are depicted as dark gray spheres. (B) Comparison of pApG bound in Xac NPP^T90A (gray) versus mouse ENPP1^T238A (green). (C) Competitive inhibition of ATP hydrolysis with cGAMP at pH 7.5; n = 3 independent reactions, mean ± SD. WT: K_i = 120 µM. H362A: K_i = 320 µM. (D and E) Crystal structures of Xac NPP^H214A with one zinc bound (D) and Xac NPP^WT [PDB ID code: 2GSN (42)] with two zincs bound (E). (F and G) Zinc rescue experiments with ATP (F) or cGAMP (G) as the substrate; n = 3 independent reactions. Mean ± SD shown for ATP (F), representative reaction shown for cGAMP (G). (H and I) Model of mouse ENPP1^H362A (gray cartoon and sticks) with intact ATP (gray sticks and spheres) chelating Zn1 (H) compared to mouse ENPP1^WT (gold cartoon and sticks) with H362 chelating Zn1 (I). (J and K) Model of Xac NPP^H214A bound to 3′3′-cGAMP (J) and 2′3′-cGAMP (K).

Fig. 4.

Enpp1^H362A mice do not exhibit the severe systemic calcification seen in ENPP1-null humans and mice. (A) Ex vivo organ lysate (1 mg/mL) cGAMP degradation at pH 9 assessed by TLC after 4 h (liver, brain) or 2 h (lung, spleen); n = 1 mouse per genotype. (B) Ex vivo plasma cGAMP degradation assessed by TLC after 4 h; n = 3 mice per genotype. (C) In vivo cGAMP degradation: 5 mg/kg of cGAMP was injected subcutaneously into each mouse. Blood was collected from each mouse after 30 min and cGAMP was measured by LC-MS/MS; n = 3 mice per genotype. (D) Ex vivo liver lysate (1 mg/mL) ATP degradation at pH 7.5 assessed after 20 min; n = 4 mice per genotype. (E) Photos of Enpp1^WT, Enpp1^H362A, and Enpp1^asj mouse paws. The Enpp1^asj mouse paws are unable to relax due to the calcified joints. (F) Twenty-week-old mice were euthanized and the ears, airways, and vibrissae were fixed and stained with Alizarin red to identify calcium deposits. Representative images are shown from one to two mice per genotype. (G and H) Plasma calcium (G) and phosphate (H) concentrations in Enpp1^WT, Enpp1^asj, and Enpp1^H362A mice; n = 4 Enpp1^WT, 6 Enpp1^asj, and 3 Enpp1^H362A mice. (I) Plasma pyrophosphate concentrations in Enpp1^WT, Enpp1^asj, and Enpp1^H362A mice; n = 6 Enpp1^WT, 6 Enpp1^asj, and 3 Enpp1^H362A mice. (J) Ex vivo plasma ATP degradation at pH 7.5 assessed by luciferase assay after 45 min in Enpp1^WT, Enpp1^asj, and Enpp1^H362A mice; n = 4 Enpp1^WT, 4 Enpp1^asj, and 5 Enpp1^H362A mice. (K and L) In vitro ATP degradation comparing overexpressed ENPP1^WT and ENPP1^H362A as cell-surface proteins from cell lysate (K) and as secreted proteins from cell supernatant (L). For C–J, data are shown as the mean ± SD; P values were calculated by unpaired t test with Welch’s correction. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 5.

Enhanced extracellular cGAMP signaling confers resistance to HSV-1 infection. (A) Mice were infected with 2.5 × 10⁷ PFU per mouse HSV-1 through intravenous injection and body weight was monitored over 6 d. Percent body weight change is plotted for each day postinfection (Left) and as an average of days 1 to 6 postinfection (Right); n = 6 for Enpp1^WT and n = 5 for Enpp1^H362A. B–D Mice were infected with 2.5 × 10⁷ PFU per mouse of HSV-1 through intravenous injection and euthanized at 6 or 12 hpi. RNA was isolated from the spleen, liver, and lungs. qRT-PCR was performed to determine the expression levels of HSV-gB (B), Ifnb1 (C), and Il6 (D); n = 6 (6 hpi) or 8 (12 hpi) infected mice per genotype. (E and F) BMDMs isolated from Enpp1^WT, Enpp1^H362A, Sting1^−/−, and Sting1^−/−/Enpp1^H362A mice were infected at a multiplicity of infection (MOI) = 0.5 or 5. qRT-PCR was performed to determine the expression levels of HSV-gB (E) and Ifnb1 (F). For all qRT-PCR data, transcript levels were normalized to Actb. Cytokine transcript levels were also normalized to the average of two uninfected controls per genotype. Data are shown as the mean ± SD; P values were calculated using the nonparametric Mann–Whitney U test, appropriate for bimodal data. *P < 0.05, **P < 0.01, ***P < 0.001; ****P < 0.0001; P value is shown if between 0.05 and 0.1.

Fig. 6.

Enhanced extracellular cGAMP exacerbates radiation-induced inflammation. (A–D) Enpp1^WT and Enpp1^H362A mice were irradiated with 8 or 9 Gy of total body irradiation and then weighed daily. Mice were euthanized if they exhibited greater than 20% weight loss for 2 consecutive days; n = 11 mice per genotype (6 mice at 8 Gy and 5 mice at 9 Gy). (A) Kaplan–Meier plot showing the probability of survival for each genotype. P value was calculated using a log rank (Mantel–Cox) test. (B) Blood was drawn from each mouse 5 d postirradiation. Plasma IFN-β concentration was determined using a high-sensitivity IFN-β ELISA. P value was calculated using an unpaired t test. (C and D) Spleens were harvested from euthanized mice at experiment endpoint. RNA was extracted and qRT-PCR was used to determine expression levels of Ifnb1 (C) and Irf7 (D). One splenic H362A sample was excluded from C and D as an outlier based on the ROUT method (Q = 1%). (E–H) Sting1^−/− and Sting1^−/−/Enpp1^H362A mice were irradiated with 9 Gy of total body irradiation and then weighed daily; n = 6 mice per genotype. Procedures described in A–D were repeated to obtain plasma IFN-β concentrations (F) and splenic expression levels of Ifnb1 (G) and Irf7 (H). P values were calculated using an unpaired t test. *P < 0.05, **P < 0.01, ****P < 0.0001. (I) Model of cGAMP as a paracrine immunotransmitter regulated by ENPP1 during viral infection and radiation-induced inflammation. Infected or irradiated cells produce and export cGAMP in response to cytosolic double-stranded DNA. When ENPP1^WT is present, extracellular cGAMP is degraded, diminishing STING pathway activation. When cGAMP-specific ENPP1^H362A is present, extracellular cGAMP accumulates, leading to an enhanced response to viral infection or an exacerbated response to radiation-induced inflammation.