cGLRs are a diverse family of pattern recognition receptors in innate immunity

Abstract

Cyclic GMP-AMP synthase (cGAS) is an enzyme in human cells that controls an immune response to cytosolic DNA. Upon binding DNA, cGAS synthesizes a nucleotide signal 2'3'-cGAMP that activates STING-dependent downstream immunity. Here, we discover that cGAS-like receptors (cGLRs) constitute a major family of pattern recognition receptors in innate immunity. Building on recent analysis in Drosophila, we identify >3,000 cGLRs present in nearly all metazoan phyla. A forward biochemical screening of 150 animal cGLRs reveals a conserved mechanism of signaling including response to dsDNA and dsRNA ligands and synthesis of isomers of the nucleotide signals cGAMP, c-UMP-AMP, and c-di-AMP. Combining structural biology and in vivo analysis in coral and oyster animals, we explain how synthesis of distinct nucleotide signals enables cells to control discrete cGLR-STING signaling pathways. Our results reveal cGLRs as a widespread family of pattern recognition receptors and establish molecular rules that govern nucleotide signaling in animal immunity.

Keywords: STING; cGAS; cGLR; cyclic dinucleotides; innate immunity; pattern recognition receptor.

Figure 1.. cGLRs are a widespread family of signaling enzymes in animal immunity

(A) Bioinformatic identification and phylogenetic tree of ~3,000 predicted cGLRs from nearly all major animal phyla. Characterized cGLRs including vertebrate cGAS proteins, N. vectensis cGLR, and Drosophila cGLR1 are denoted with an orange star. cGLRs with published crystal structures are denoted with a green triangle at the tree edge. Closely related Mab21L1/2/3 genes are included but do not contain a predicted functional active site. Yellow circles represent cGLRs cloned for the biochemical screen, and purple circles denote active cGLRs selected for in-depth analysis. For additional information, see Figure S3 and Table S1. (B) Number of cGLR genes encoded in individual species categorized by animal phylum. Data are mean ± SEM with individual data points shown as dots. (C) Domain organization of cGLR proteins. The prevalence of each domain architecture in sequenced animal genomes is listed as a percentage of all cGLR proteins (ANK = ankyrin repeat motif, TPR = tetratricopeptide repeat motif). Domain architectures that account <0.5% of all cGLR proteins are represented as “Other”. For additional details, see Table S1. (D) Crystal structures and surface electrostatic views of the metazoan cGLRs Cv-cGLR1 from the oyster C. virginica and Sp-cGLR1 from the coral S. pistillata. Structural comparison with the human cGAS–dsDNA complex (PDB 6CTA) confirms that metazoan cGLRs adopt a conserved overall architecture and shared primary ligand-binding surface (indicated by dashed lines).

Figure 2.. Divergent metazoan cGLRs respond to the common PAMPs dsDNA and dsRNA

(A) Thin layer chromatography of cGLR nucleotide second messenger products. Major and minor products were identified with LC-MS/MS analysis in comparison to synthetic standards (Figure S3 and S4). cGLRs that synthesize a major product that could not be matched to a known nucleotide second messenger are denoted as “unknown”. Human cGAS was used as a control and is denoted as “C”; active cGLR number designations correspond to labels in Figure 1A. Data are representative of n = 3 independent experiments. (B) Thin layer chromatography analysis of animal cGLR activation in the presence of a 45 bp dsDNA or polyI:C dsRNA. Human cGAS was used as a control and is denoted as “C”; active cGLR number designations correspond to labels in panel A. Data are representative of n = 3 independent experiments. See Figure S5B for biochemical deconvolution of the activating ligand specificity of all other active cGLR enzymes. (C) Summary of specific activating ligand and nucleotide second messenger product details for known cGLRs and each active cGLR enzyme identified here. See Figures 3, S3 and S4 for analysis of nucleotide second messenger product identification for Cg-cGLR1 (07), Cv-cGLR1 (08), and Sp-cGLR1 (09). (D) Cv-cGLR1 and Sp-cGLR1 in vitro activity was analyzed in the presence of a panel of dsDNA or dsRNA activating ligands with various length. Data are the mean ± SD of n = 3 independent experiments.

Figure 3.. Metazoan cGLRs produce diverse cyclic di-purine and purine-pyrimidine signals

(A) Thin layer chromatography analysis of Cv-cGLR1, Sp-cGLR1 and Sp-cGLR3 reactions labeled with individual α³²P-NTPs and treated as indicated with nuclease P1 (a 3′–5′ phosphodiester bond-specific nuclease) and CIP (a phosphatase that removes terminal phosphate groups from nucleotides). Nucleotide labeling and P1 digestion suggest that Cv-cGLR1 and Sp-cGLR1 synthesize 2′3′-cUA and 3′3′-cUA while Sp-cGLR3 synthesizes 3′3′-cAA. Data are representative of n = 3 independent experiments. (B) HPLC analysis Cv-cGLR1, Sp-cGLR1 and Sp-cGLR3 reactions compared to 2′3′-cUA and 3′3′-cUA synthetic standards. Data are representative of n = 3 independent experiments. Multiple products can be identified from the Sp-cGLR1 reaction, the ratio of HPLC-integrated concentration of 3′3′-cUA, 3′3′-cGAMP and 3′3′-cAA is 3.3 : 1 : 2.1 (corrected with extinction coefficient 22500 M⁻¹cm⁻¹, 25050 M⁻¹cm⁻¹ , 27000 M⁻¹cm⁻¹ respectively), confirming 3′3′-cUA is the major product in vitro. See also Figure S3B for HPLC trace of Sp-cGLR1 reaction supplemented with only ATP and UTP, where the ratio of HPLC-integrated concentration of 3′3′-cUA and 3′3′-cAA is 2.5 : 1. The peak eluted around 3 min in Sp-cGLR3 reaction is likely an intermediate derived from ATP, as the same peak was also observed when the reaction was supplemented with only ATP (see Figure S3B). (C) Diversity of animal cGLR nucleotide products. cGLR analysis supports that although 2′3′-cGAMP is the most common nucleotide second messenger product, cGLRs are capable of synthesizing diverse immune signals including 2′3′-cUA, 3′3′-cUA and 3′3′-cAA. Detailed LC-MS/MS and NMR verification of the Cv-cGLR1, Sp-cGLR1 and Sp-cGLR3 can be found in Figure S4.

Figure 4.. Animals encode STING receptors with distinct nucleotide second messenger preferences

(A) cGLR and STING protein diversity in select representative organisms from different animal phyla. Mab21L1-like proteins are excluded. (B) Analysis of cGLR and STING gene copy number in 381 animal species. Bubble size corresponds to species frequency. Animal genomes demonstrating significant cGLR and STING gene radiation are highlighted in a purple box and the derived taxa are plotted in the corresponding pie chart. Full data of the number of cGLR and STING genes in each animal species are included in Table S3. (C) Diversity of cGLR proteins in the stony coral S. pistillata including three active enzymes Sp-cGLR1, Sp-cGLR2 and Sp-cGLR3 identified in the biochemical screen. Sp-cGLR1 and Sp-cGLR3 produce 3′3′-cUA and 3′3′-cAA respectively in response to dsRNA and Sp-cGLR2 produces 2′3′-cGAMP in response to an unknown ligand. S. pistillata cGLRs fused to ankyrin repeats (ANK) and death domains (DD) potentially involved in protein–protein and protein–ligand interactions are annotated. Isoelectric point (pI) of each cGLR is annotated by heatmap with red representing low pI value (5.0) and blue representing high pI (9.7). (D) S. pistillata encodes seven STING proteins, each predicted to contain an N-terminal transmembrane domains (TM) and a C-terminal cyclic dinucleotide binding domain (CBD). Sp-STING proteins share ~54–75% CBD sequence identity. (E) Dissociation constants of five Sp-STING receptors with five different cyclic dinucleotide messengers calculated from electrophoretic mobility shift assay (EMSA) analysis. Data are the mean ± std of n = 2 or 3 independent experiments. See Figures S6C for EMSA images and binding curves. (F,G) Quantification of EMSA analysis of affinity of Sp-STING1 and Sp-STING3 with the nucleotide second messengers 2′3′-cGAMP, 2′3′-cUA, 3′3′-cGAMP, 3′3′-cGG and 3′3′-cAA demonstrates that Sp-STING1 preferentially binds 2′3′-linked cyclic dinucleotides and Sp-STING3 preferentially binds 3′3′-linked cyclic dinucleotides. Results are plotted as fraction bound (shifted/total signal) as a function of increasing protein concentration and fit to a single binding isotherm. Data are the mean ± std of n = 2 or 3 independent experiments.

Figure 5.. Molecular mechanism of STING ligand recognition in S. pistillata

(A) Crystal structures of Sp-STING1 in complex with 2′3′-cGAMP and Sp-STING3 in complex with 3′3′-cGAMP in comparison to human STING-2′3′-cGAMP (PDB: 4KSY) and human STING-3′3′-cGG (PDB: 4F9G) complexes. The Sp-STING1–2′3′-cGAMP and Sp-STING3–3′3′-cGAMP structures adopt a tightly closed conformation with an ordered β-strand lid most similar to the human STING–2′3′-cGAMP complex, supporting high-affinity recognition of an endogenous cGLR nucleotide second messenger signal. (B) Sequence alignment of Sp-STING receptors with STING proteins from representative animal species and the bacterium S. faecium reveals that Sp-STING3 has unique substitutions at key residues involved in cyclic dinucleotide binding. (C,D) Structural comparison of the bacterial Sf-STING–3′3′-cGG complex, coral Sp-STING3–3′3′-cGAMP complex, Sp-STING1–2′3′-cGAMP complex and human STING–2′3′-cGAMP complex reveals critical difference in phosphodiester linker recognition that explain specificity for 2′3′-linked and 3′3′-linked cGLR nucleotide second messenger signals. (E) Crystal structures of Sp-STING1 and human STING in complex with 2′3′-cUA. Both complexes adopt a tightly closed conformation with an ordered β-strand lid most similar to the human STING–2′3′-cGAMP complex. (F) Structural comparison of Sp-STING1–2′3′-cUA complex and human STING–2′3′-cUA complex reveals conservation of specific cyclic dinucleotide contacts.

Figure 6.. cGLR nucleotide signals induce activation of a conserved immune gene program

(A) Volcano plots showing differentially expressed genes (DEGs) in S. pistillata treated with 3′3′-cUA or 2′3′-cGAMP for 18 h versus seawater control. Significantly up- and down-regulated genes (adjusted p value < 0.05 and absolute log2 fold change > 1) are highlighted in purple and orange, respectively. Among those, genes of the STING and cGLR pathways are highlighted as squares and triangles, respectively. DEGs with homologs known to be involved in immunity in human and other organisms are labeled. (B) DEGs were categorized based on known KEGG signaling pathways. The most significantly perturbed KEGG pathways (adjusted p value < 0.01) in S. pistillata samples treated with 3′3′-cUA or 2′3′-cGAMP were shown in a bubble plot. (C) Euler diagrams showing intersection of up- (purple circles) and down-regulated (orange circles) DEGs in S. pistillata samples treated with 3′3′-cUA or 2′3′-cGAMP. (D) Volcano plots showing DEGs in C. virginica treated with 2′3′-cUA or 2′3′-cGAMP versus seawater control. (E) Bubble plot showing the most significantly perturbed KEGG pathways (adjusted p value < 0.01) in C. virginica samples treated with 2′3′-cUA. This analysis was not done for samples treated with 2′3′-cGAMP as only 2 DEGs were induced. (F) Number of unique and shared up-regulated orthologs between S. pistillata and C. virginica. Selected examples of up-regulated orthologs’ expression are shown in the bubble plot. (G) Model of cGLR signaling pathway in animal immunity. Animals encode multiple cGLRs that recognize diverse PAMPs and produce distinct nucleotide second messengers. STING receptor duplication and cyclic dinucleotide-specific adaptations enables creation of complex cGLR-STING signaling networks.