The virus-induced cyclic dinucleotide 2'3'-c-di-GMP mediates STING-dependent antiviral immunity in Drosophila

Abstract

In mammals, the enzyme cGAS senses the presence of cytosolic DNA and synthesizes the cyclic dinucleotide (CDN) 2'3'-cGAMP, which triggers STING-dependent immunity. In Drosophila melanogaster, two cGAS-like receptors (cGLRs) produce 3'2'-cGAMP and 2'3'-cGAMP to activate STING. We explored CDN-mediated immunity in 14 Drosophila species covering 50 million years of evolution and found that 2'3'-cGAMP and 3'2'-cGAMP failed to control infection by Drosophila C virus in D. serrata and two other species. We discovered diverse CDNs produced in a cGLR-dependent manner in response to viral infection in D. melanogaster, including 2'3'-c-di-GMP. This CDN was a more potent STING agonist than cGAMP in D. melanogaster and it also activated a strong antiviral transcriptional response in D. serrata. Our results shed light on the evolution of cGLRs in flies and provide a basis for understanding the function and regulation of this emerging family of pattern recognition receptors in animal innate immunity.

Keywords: Drosophila; STING; c-di-GMP; cGAMP; cGAS; cGLR; cyclic dinucleotide; evolution; pattern recognition receptor; virus.

Figure 1:. 3′2′-cGAMP induces antiviral immunity in most, but not all, Drosophila species

A. Phylogeny of the 14 Drosophila species used. Drosophila and Sophophora subgenus are indicated. B. Summary of the antiviral effect of the indicated CDNs in the Drosophila species. Heatmap showing the log2 fold change of DCV RNA load at two- and three-days post-infection of male flies that had been pre-injected with the indicated CDNs or Tris. Significant changes (adjusted p value ≤0.05; pairwise permutation test with FDR method) are highlighted with dots. C-F. Relative DCV RNA loads at two or three days post DCV infection in flies from the indicated species. Data are from at least three independent experiments, each performed in biological triplicates and shown with boxplot with scatter plot. Analysis using pairwise permutation test with FDR method upon CDN- compared to Tris-injection is shown (ns= non significant, *<0.05, **<0.01 and ***<0.001).

Figure 2:. cGLR1 and cGLR2 produce several CDNs in D. melanogaster flies

A-B. Concentration of CDNs measured by LC-MS in the hemolymph (A) or in whole fly lysates (B) collected from male transgenic flies ectopically expressing cGLR1, cGLR2, their inactive cGLR^AFA versions or a control transgene (GFP). C. Concentration of CDNs measured by LC-MS in the lysate of HEK293T cells overexpressing cGLR2 or GFP. Data are representative of one (A), three (B) or four (C) independent experiments and shown with dot plots with median and quartile. nd, non-detected. Concentration of 2'3'-c-di-AMP in whole flies was analysed by pairwise permutation test with FDR method (***<0.001). D. Thin-layer chromatography (TLC) analysis of α-³²P labeled Dp-cGLR2 reaction products and treatment with P1 nuclease, which cleaves 3′–5′ phosphodiester linkages. Specific labeling reveals that only G nucleobases are incorporated in the cGLR2 product and P1 treatment supports that the cGLR2 product contains one 3′–5′ bond and one P1-insensitive bond. Data are representative of n= 3 independent experiments. E. HPLC chromatogram showing C18 elution profile of Dp-cGLR2 reaction; dashed lined indicates retention times of synthetic nucleotide standards. Data are representative of n= 3 independent experiments. F. High-resolution mass spectrometry confirms the major Dp-cGLR2 product as 2′3′-c-di-GMP. See also Figures S1, S2, S3, S4 and S5.

Figure 3:. DCV infection triggers cGLR-dependent production of CDNs in D. melanogaster

A. Concentration of CDNs measured by LC-MS in whole fly lysates from Tris- or DCV- injected flies. Data are from at least four independent experiments. B. Concentration of CDNs in the hemolymph of Tris- or DCV-injected dSTING knockout flies. Data are from three independent experiments. Each experiment involves 1000 flies. C. Concentration of CDNs in whole fly lysates from Nora-infected dSTING knockout flies. Data are from four independent experiments, each involving 500 flies. All data are shown with boxplot with scatter plot and were analyzed using permutation test with FDR method (ns= non significant, *<0.05, **<0.01, ***<0.001 and ****<0.0001), non-detected (nd). See also Figure S4.

Figure 4:. dsRNA binding of Drosophila cGLR2 facilitates its production of 2'3'-c-di-GMP

A. TLC analysis and quantification of GTP conversion to 2′3′-c-di-GMP by Dp-cGLR2 in the presence of different nucleic acid ligands. Data are mean +/- s.e.m. of n= 3 individual experiments. B. HPLC chromatogram showing 2′3′-c-di-GMP synthesis by Dp-cGLR2 in the presence of dsRNA mimic Poly I:C or a buffer control. Data are representative of n= 3 independent experiments. C. Surface electrostatic view of predicted Dp-cGLR2 structure modeled with a 19 bp dsRNA ligand, highlighting predicted interacting residues selected for mutagenesis. D. TLC analysis and quantification of wildtype (WT) and mutant Dp-cGLR2 activity in the absence and presence of dsRNA mimic Poly I:C. For quantification, % GTP conversion to 2′3′-c-di-GMP for each reaction was normalized to the wildtype control mean value for either buffer or Poly I:C reactions. Data are mean +/- s.e.m. of n= 3 individual experiments. See also Figures S5 and S6.

Figure 5:. 2'3'-c-di-GMP triggers a potent antiviral response in D. melanogaster

A. Survival of control (isogenic w¹¹¹⁸) and dSTING knockout flies pretreated with the indicated CDNs before systemic DCV infection. Data are from four independent experiments, each with three independent groups of around ten flies and analyzed by Log-rank test (ns= non significant and ****<0.0001). B. Control flies pre-injected with the indicated CDNs or Tris were challenged with VSV and viral RNA load was monitored after 4- or 5-days post injection. Data are from four independent experiments, each performed in biological triplicates (n = 12). Data are shown with boxplot with scatter plot and were analyzed using permutation test with FDR method (ns= non significant, *<0.05, **<0.01, ***<0.001 and ****<0.0001). C. Relative gene expression of the STING-regulated genes dSTING, srg1, srg2 and srg3 seven days after injection of the indicated CDNs in control and dSTING knockout flies. Samples were collected from three independent experiments, each involving three groups of six flies including three males and three females. Data are shown with boxplot with scatter plot and were analysed using permutation test with FDR method (ns= non significant, *<0.05, **<0.01 and ***<0.001). D. Expression profiles of control male D. melanogaster (isogenic w¹¹¹⁸) seven days after injection with Tris, 3′2′-cGAMP or 2′3′-c-di-GMP. All differentially expressed genes (DEGs) including upregulated and downregulated genes are shown in log2(FPKM+1). Data are from three independent experiments.

Figure 6:. 2'3'-c-di-GMP triggers antiviral immunity in D. serrata, D. sechellia and D. mojavensis

A-C. Relative DCV RNA loads at two or three days post-DCV infection in D. serrata (A), D. sechellia (B) and D. mojavensis (C) male flies pre-injected with Tris or the indicated CDNs. Data are from four independent experiments. Data are shown with boxplot with scatter plot and were analysed using pairwise permutation test with FDR method (ns= non significant, *<0.05, **<0.01 and ***<0.001). D. Expression profiles of D. serrata male flies seven days after injection with Tris, 3′2′-cGAMP or 2′3′-c-di-GMP. All differentially expressed genes (DEGs) including upregulated and downregulated genes are shown in log2(FPKM+1). Data are from three independent experiments. E-F. Analysis of the change in thermophoresis in dSTING::eGFP fusion proteins upon incubation with increasing doses of the indicated CDNs in lysates from S2 cells transfected with expression vectors for the D. melanogaster (E) or D. serrata (F) proteins. The three data points corresponding to the most concentrated doses of 2′3′-c-di-GMP were not taken into account for the calculation of the Kd. Data are from 2 to 4 independent experiments. See also Figure S7.

Figure 7:. Rapid evolution of cGLR-STING signaling in Drosophila

A. Phylogenetic tree of the cGLRs identified in Drosophila genomes (n= 207). Sequences of predicted cGLR were aligned using MAFFT and the aligned sequences were used to construct the tree. The names of the cGLRs from the 14 Drosophila species are indicated and those able to activate STING signaling are labelled in red while the inactive ones are in black. The red stars indicate cGLRs from D. melanogaster. The 8 cGLRs in grey were not tested. Isoelectric point (pI) annotated by color gradient where acidic pIs are shown in red and basic pIs are shown in blue. B. The predicted cGLRs encoded in the genome of the 14 Drosophila species. The cGLRs able to activate STING signaling are labelled in red. C. Heatmap showing log2 fold change intensities in dSTING luciferase reporter gene activity in S2 cells expressing the indicated 52 candidate cGLRs in comparison to control. Four conditions were tested with empty vector, STING, cGLR and cGLR+STING from the same species. 15 positive candidates highlighted with dots were selected by mixed-effect model (RELM) with p value ≤0.05 (Tukey correction for multiple hypothesis testing) and with fold change > 2 in cGLR or cGLR with STING conditions. Data are from 2 to 4 independent experiments. D. Concentration of 2′3′-c-di-GMP measured by LC-MS/MS in lysates from dSTING knockout S2 cells ectopically expressing the indicated cGLRs. Data are from 5 independent experiments.