LRRC8A:C/E Heteromeric Channels Are Ubiquitous Transporters of cGAMP

Abstract

Extracellular 2'3'-cyclic-GMP-AMP (cGAMP) is an immunotransmitter exported by diseased cells and imported into host cells to activate the innate immune STING pathway. We previously identified SLC19A1 as a cGAMP importer, but its use across human cell lines is limited. Here, we identify LRRC8A heteromeric channels, better known as volume-regulated anion channels (VRAC), as widely expressed cGAMP transporters. LRRC8A forms complexes with LRRC8C and/or LRRC8E, depending on their expression levels, to transport cGAMP and other 2'3'-cyclic dinucleotides. In contrast, LRRC8D inhibits cGAMP transport. We demonstrate that cGAMP is effluxed or influxed via LRRC8 channels, as dictated by the cGAMP electrochemical gradient. Activation of LRRC8A channels, which can occur under diverse stresses, strongly potentiates cGAMP transport. We identify activator sphingosine 1-phosphate and inhibitor DCPIB as chemical tools to manipulate channel-mediated cGAMP transport. Finally, LRRC8A channels are key cGAMP transporters in resting primary human vasculature cells and universal human cGAMP transporters when activated.

Keywords: 2’3’-cGAMP; LRRC8A; LRRC8C; LRRC8D; STING; VRAC; cGAMP; cyclic dinucleotide; transporter; vasculature.

Figure 1.. A Genome-Wide CRISPR Screen Identifies LRRC8A as a Positive Regulator of Extracellular cGAMP-Mediated STING Pathway Activation

(A) Scheme of extracellular cGAMP-STING signaling in U937 SLC19A1^−/− cells. Imported cGAMP binds to and activates STING, which results in phosphorylation of STING, TBK1, and the transcription factor IRF3, and ultimately leads to interferon induction and cell death. (B) CRISPR screen strategy. U937 Cas9-SLC19A1^−/− cells were transduced with a whole-genome sgRNA lentiviral library. In two replicates, library cells were treated with cGAMP or untreated for 12 days. Genomic DNA was harvested, deep sequenced, and analyzed for sgRNA depletion or enrichment. (C) Plot of casTLE score for each gene in replicate 1 versus replicate 2 with displayed correlation across all genes (R² = 0.86). Top scoring hits are annotated. (D) Volcano plot of casTLE effect size versus false discovery rate (FDR) for all genes across replicates. Top hits and all LRRC8 genes are annotated. A 5% FDR significance threshold is indicated. See also Figure S1.

Figure 2.. LRRC8A and LRRC8 Paralogs Differentially Facilitate cGAMP Import

(A) U937 SLC19A1^−/− subclones expressing scramble or LRRC8A–LRRC8E sgRNAs were treated with 100 μM cGAMP for 4 h and signaling was assessed by Western blot (see Figure S2 for display of all blots). Summarized results for unique subclones (confirmed heterozygous or homozygous knockouts) are plotted relative to the respective scramble control average (n = 4–9 subclones tested in 1 experimental replicate). Significance calculated by two-tailed t-test; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001. (B) U937 SLC19A1^−/−-scramble or -LRRC8A^−/− through -LRRC8E^−/− clones were electroporated with 100 nM cGAMP, cultured for 2 h, and signaling was assessed by Western blot (n = 2 biological replicates). For (A)–(B), plotted range is mean ± SD. See also Figure S2.

Figure 3.. LRRC8A-Containing VRAC Channels Directly Transport cGAMP

(A) HEK293 scramble or LRRC8A–LRRC8E knockout pools were treated with 100 μM cGAMP for 2 h and signaling was assessed by Western blot. Representative blots displayed with quantification (n = 4 biological replicates). (B) HEK293 scramble or LRRC8A–LRRC8E knockout pools were treated with 20 μM cGAMP for 1 h in isotonic or hypotonic solution and signaling assessed by Western blot. Representative blots displayed with quantification (n = 3 biological replicates). (C) Scramble or LRRC8A knockout cells were transfected for expression of mSTING. After 24 h, cells were treated with 250 μM DMXAA or 100 μM GAMP in isotonic or hypotonic buffer for 1 h. Activation of p-mSTING or p-hSTING signaling was assessed by Western blot. Representative blots displayed with quantification (n = 2 technical replicates). ND = not detected. (D) Extracellular cGAMP concentrations from HEK293 scramble or LRRC8A–LRRC8E knockout pools were measured following cGAS plasmid transfection from media (24 h) or after hypotonic buffer stimulation (20 min) using STING-CAP and the cGAMP-Luc assay (n = 3 biological replicates). (E) Representative whole-cell patch clamp current traces measured in HEK293 cells upon hypotonic stimulation in the absence or presence of cGAMP (100 μM). Overlays of the current-voltage relationship at the indicated points are also displayed. For additional traces, see Figure S3. (F) Summary data for inhibition of VRAC current by 100 μM extracellular cGAMP at +150 mV (n = 8 measured in 5 cells). For (A)–(D), and (F), plotted range is mean ± SD. Significance calculated using two-tailed t-test; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, not significant (NS). See also Figure S3.

Figure 4.. The LRRC8A:C Channel Is a Major cGAMP Importer in Microvasculature Cells

(A) TIME LRRC8A–E and SLC19A1 knockout pools were treated with 50 μM cGAMP for 2 h and signaling was assessed by Western blot (n = 3 biological replicates). (B) TIME scramble or LRRC8A–LRRC8E knockout pools were treated with 10 μM cGAMP for 1 h in isotonic or hypotonic solution and signaling assessed by Western blot (n = 3 biological replicates). (C) TIME LRRC8A, LRRC8C, and LRRC8A:C knockout pools were treated with 50 μM cGAMP for 2 h and signaling was assessed by Western blot (n = 3 biological replicates). (D) Cryo-EM structure of human homo-hexameric LRRC8A (PDB: 5ZSU) (Kasuya et al., 2018). Surface display of the complex, with ribbon representation of two opposing LRRC8A subunits and zoom to pore gating residue R103. (E) TIME knockout cells were induced to express Flag-tagged LRRC8A, LRRC8A-R103L, or LRRC8C, then treated with 50 μM cGAMP for 2 h and signaling assessed by Western blot (n = 3 biological replicates). For (A)–(C) and (E), representative blots are shown with quantification of all experiments and range plotted as mean ± SD. Significance calculated using two-tailed t-test; *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, not significant (NS). See also Figure S4.

Figure 5.. LRRC8A Channels Transport Other 2’3’-Cyclic Dinucleotides

(A) Chemical structures of 2’3’- and 3’3’-cyclic dinucleotide (CDN) STING agonists. (B) TIME scramble and LRRC8A knockout pools were treated with 50 μM cGAMP, 25 μM 2’3’-cG^SA^SMP, 100 μM 2’3’-CDA, 25 μM 2’3’-CDA^S, 100 μM 3’3’-cGAMP, 200 μM 3’3’-CDA, or 200 μM 3’3’-CDG for 2 h and signaling assessed by Western blot (n = 2 biological replicates). Representative blots are shown with quantification of all experiments and range plotted as mean ± SD. ND = not detected. (C) TIME scramble control or LRRC8A knockout pools were treated with cGAMP or 2’3’-cG^SA^SMP at the indicated concentrations for 3 h. Induction of IFNB1 relative to untreated cells was quantified by RT-qPCR (Figure S5) and fold change in IFNB1 induction between scramble versus LRRC8A-KO cells calculated. Plotted range is mean ± SD (n=2 biological replicates). See also Figure S5.

Figure 6.. Pharmacologic and Genetic Perturbations Reveal LRRC8A-Containing Channels are Major cGAMP Transporters in Primary Human Endothelial Cells

(A) Chemical structure of DCPIB. (B) TIME scramble and LRRC8A knockout pools were treated with cGAMP (50 μM), with or without S1P (10 or 100 nM) and DCPIB (20 μM), for 1 h and signaling assessed by Western blot (n = 3 biological replicates). Significance calculated using two-tailed t-test; *P ≤ 0.05, **P ≤ 0.01. (C) Following siRNA knockdown of LRRC8A expression in HUVEC cells pooled from six donors, cells were treated with cGAMP (100 μM) for 2 h. Western blots shown with quantification (n = 1). (D) HUVEC cells pooled from 6 donors were treated with cGAMP (50 μM), with or without S1P (10 or 100 nM) and DCPIB (20 μM) for 1 h and signaling assed by Western blot (n = 2 biological replicates). For (B) and (D), representative blots are shown with quantification of all experiments and range plotted as mean ± SD. See also Figure S6.