Lysyl-tRNA synthetase produces diadenosine tetraphosphate to curb STING-dependent inflammation

Abstract

Inflammation is an essential part of immunity against pathogens and tumors but can promote disease if not tightly regulated. Self and non-self-nucleic acids can trigger inflammation, through recognition by the cyclic GMP-AMP (cGAMP) synthetase (cGAS) and subsequent activation of the stimulator of interferon genes (STING) protein. Here, we show that RNA:DNA hybrids can be detected by cGAS and that the Lysyl-tRNA synthetase (LysRS) inhibits STING activation through two complementary mechanisms. First, LysRS interacts with RNA:DNA hybrids, delaying recognition by cGAS and impeding cGAMP production. Second, RNA:DNA hybrids stimulate LysRS-dependent production of diadenosine tetraphosphate (Ap₄A) that in turn attenuates STING-dependent signaling. We propose a model whereby these mechanisms cooperate to buffer STING activation. Consequently, modulation of the LysRS-Ap₄A axis in vitro or in vivo interferes with inflammatory responses. Thus, altogether, we establish LysRS and Ap₄A as pharmacological targets to control STING signaling and treat inflammatory diseases.

Fig. 1. MSC is recruited to RNA:DNA hybrids.

(A) Mean (±SEM), Ifnβ mRNA levels in WT-MEF, MEF^Sting−/−, and MEF^cGAS−/− transfected or not with RNA:DNA hybrids (R:D); n = 8. t test, ****P < 0.0001. (B) WB analysis of whole-cell extracts of cells treated as in (A). Membranes were probed with the indicated antibodies. (C) Left: Experimental scheme. MS, mass spectrometry; ss^BRNA, biotinylated ssRNA; ^BR:D, biotinylated RNA:DNA hybrids. Right: Silver staining of samples obtained following the experimental scheme. Numbers indicate molecular weight (MW) in kDa. (D) WB analysis of pull-down performed as in (C), except that biotinylated ssDNA (ss^BDNA) was included as a control. (E) HeLa cells were transfected or not with biotinylated ^BR:D, ss^BRNA, or biotinylated dsDNA (^BD:D) before whole-cell extract preparation and pull-down using streptavidin affinity beads. Input and eluates were analyzed by WB using the indicated antibodies. (F) As in (E), except that WT-MEF were transfected with ^BR:D before pull-down. Input and eluates were analyzed by WB using the indicated antibodies. All immunoblots show representative experiments.

Fig. 2. The Lyslyl tRNA synthetase interacts directly with RNA:DNA hybrids.

(A) In vitro interaction assay. Three, 10, and 30 pmol of recombinant LysRS, AspRS, or p43 were incubated with 20 pmol of ^BR:D before pull-down using streptavidin affinity beads. Input and eluates were analyzed by WB using anti-GST antibody. (B) As in (A), except that 3 pmol of recombinant LysRS was incubated with 20 pmol of ^BR:D before pull-down on streptavidin affinity beads. Where indicated, nucleic acids were treated with RNaseH before pull-down. Input and eluates were analyzed by WB using anti-GST antibody. (C) As in (A), except that 3, 10, and 30 pmol of recombinant LysRS, LysRSΔN, or LysRSΔC were incubated with 20 pmol of ^BR:D before pull-down using streptavidin affinity beads. Input and eluates were analyzed by WB using anti-GST antibody. (D) Three pmol of recombinant LysRS, LysRS∆C, or GST were incubated with 20 pmol of ^BR:D before pull-down using streptavidin affinity beads. Where indicated, nucleic acids were treated with RNaseH before pull-down. Input and eluates were analyzed by WB using anti-GST antibody. (E) Reciprocal immunoprecipitation. F/HA-LysRS was Flag-purified from the cytosolic fraction of HeLa cells and used as input material for pull-down using 20 pmol of ^BR:D and streptavidin affinity beads. Where indicated, nucleic acids were treated with RNaseH before pull-down. Input and eluates were analyzed by WB using the indicated antibodies. (F) Whole-cell extracts from A549 cells stably expressing shLuc or shLysRS were used in streptavidin pull-downs using ^BR:D. Input and eluates were analyzed by WB using the indicated antibodies. (G) Whole cell extracts (WCEs) from paraformaldehyde (PFA) cross-linked HeLa cells were immunoprecipitated with the S9.6 anti–RNA:DNA hybrids antibody or mock immunoglobulin G (IgG) (left). Input and eluates were analyzed by WB using an anti-LysRS antibody (right). All immunoblots show representative experiments. IP, immunoprecipitation.

Fig. 3. The Lysyl tRNA synthetase inhibits RNA:DNA hybrid–induced IFN production.

(A) Mean (±SEM) Ifnβ mRNA levels in shLysRS-expressing versus shLuc-expressing WT-MEF transfected or not with R:D for 3, 6, 9, and 12 hours. Representative graph (n = 4). (B) Whole-cell extracts from cells treated as in (A) were analyzed by WB using indicated antibodies. (C) Mean (±SEM) Ifnβ mRNA levels in shLysRS-expressing versus shLuc-expressing MEF^Trex1−/− transfected or not with R:D for 6 hours (n = 7). (D) Whole-cell extracts from cells treated as in (C) were analyzed by WB using indicated antibodies.(E) Mean (±SEM) Ifnβ mRNA levels in MEF^Trex1−/− overexpressing F/HA-LysRS following transfection or not with R:D for 6 hours (n = 4). (F) Whole-cell extracts from cells treated as in (E) were analyzed by WB using indicated antibodies. (G) Ifnβ mRNA levels in WT-MNF (mouse neonatal fibroblast) and MNF^{LysR +/−} (cells derived from two independent 1-day-old mice). Different colors indicate different mice. (H and I) WT-MNF and MNF^{LysR +/−} were infected with HSV-1 (H) or HSV-2 (I) at multiplicity of infection (MOI) =1, and viral titers were measured 24 hours later. Data represent biological triplicates of cells derived from two independent 1-day-old mice. Different colors indicate different mice. (J and K) Mean (±SEM) Ifnβ mRNA levels in MEF^Trex1−/− transduced with shLuc, shLysRS, shSting, or shLysRS and shSting prior (J) or after (K) transfection with R:D for 6 hours. Data are expressed relative to shLuc-expressing cells (n = 7). (L) WB analysis of whole-cell extracts from experiment performed as in (J) and (K). (M) Ifnβ mRNA levels in shLuc versus shLysRS-treated MEF^Trex1−/− after transfection with R:D for 6 hours. (N) Ifnβ mRNA levels in shLuc versus shLysRS-treated MEF^Trex1−/− after transfection with dsDNA (D:D) for 6 hours. (O) Ifnβ mRNA levels in shLuc versus shLysRS-treated MEF^Trex1−/− after transfection with dsRNA (R:R) for 6 hours. Results in (M) to (O) are presented as mean Ifnβ mRNA levels from technical duplicates (representative experiments; n = 3). (C, E, I, J, and K) Unpaired t test, *P < 0.05, **P < 0.01, and ****P < 0.0001. All immunoblots show representative experiments. PFU, plaque-forming units.

Fig. 4. LysRS delays the detection of RNA:DNA hybrids by cGAS.

(A) In vitro interaction. Left: Experimental scheme. T98G cells stably expressing F/HA-cGAS (T98G^F/HA-cGAS) were fractionated. The obtained cytosolic fraction was incubated with either 20 pmol of ^BR:D, ^BD:D, or ss^BDNA before streptavidin affinity pull-down. Right: Input and eluates were immunoblotted using anti-HA antibody to detect F/HA-cGAS and the indicated antibodies. (B) In cell interaction. Left: Experimental scheme. T98G^F/HA-cGAS cells were transfected or not with ^BR:D, and ^BD:D. Six hours after transfection, cells were harvested, and whole-cell extracts were prepared and used for streptavidin affinity pull-down. Right: Inputs and eluates were immunoblotted using anti-HA antibody. (C) In vitro streptavidin pull-downs were performed using 20 pmol of ^BR:D and increasing amounts of recombinant proteins (2.5, 5, 10, 20, 40, 80, and 160 nM). Biotin (20 pmol) was used as mock ligand at the lowest and highest doses of recombinant proteins. Input and eluates were immunoblotted with anti-GST antibody. (D) Signals from WB in (C) were quantified by ImageJ and plotted in Prism, and data were fitted by the nonlinear regression method for one-site specific binding. x axis: amounts (nM) of recombinant protein and y axis: absolute binding expressed as arbitrary units (AU). Representative graph (n = 4). (E and F) Competition experiments. Streptavidin-immobilized ^BR:D were incubated with 10 nM LysRS and increasing doses of cGAS (2.5, 5, 10, 20, 40, 80, and 160 nM) (E) or with 10 nM cGAS and increasing doses of LysRS (2.5, 5, 10, 20, 40, 80, and 160 nM) (F). Input and eluates were immunoblotted with either anti-LysRS or anti-cGAS antibodies as indicated. (G) LysRS was knocked down in MEF^Trex1−/− before transfection with R:D for 3, 6, and 9 hours. Cells were harvested, and cell pellets were processed for intracellular cGAMP measurements using the cGAMP ELISA Kit (CISBIO). Graph represents mean ± SD from technical duplicates of pg cGAMP per 10⁶ cells. Representative graph (n = 3). (H) Ifnβ mRNA levels, measured in cells treated as in (G), are expressed as mean (±SEM) from technical duplicates. Data are expressed relative to shLuc-expressing cells. Representative graph (n = 3). (I) Whole-cell extracts from experiment presented in (G) were analyzed by WB using the indicated antibodies. All immunoblots show representative experiments.

Fig. 5. Ap₄A inhibits STING-dependent Ifnβ expression.

(A) Mean (±SEM) Ap₄A levels in WT-MEFs transfected or not with R:D for 3, 6, 9, or 12 hours are expressed as pmol of Ap₄A per 10⁶ cells (n = 3). (B) Whole-cell extracts from cells treated as in (A) were analyzed by WB using the indicated antibodies. (C) Mean (±SEM) Ap₄A levels in HeLa cells stably expressing shLuc or shLysRS transfected or not with R:D hybrids for 12 hours are expressed as pmol of Ap₄A per 10⁶ cells (n = 5). (D) Whole-cell extracts from cells treated as (C) were analyzed by WB using the indicated antibodies. (E) Mean (±SEM) Ifnβ mRNA levels in untreated MEF^Trex1−/− versus MEF^Trex1−/− treated with 500 μM Ap₄A for 45, 90, or 180 min (n = 6). (F) Mean (±SEM) Ifnβ mRNA levels in WT-MEF stimulated with 200 μM DMXAA and treated, or not, with 500 μM Ap₄A for 2 hours (n = 13). Data are normalized to nontreated control. (G) Mean (±SEM) Ifnβ mRNA levels in MEF^Sting−/− stimulated with 200 μM DMXAA and treated, or not, with 500 μM Ap₄A for 2 hours (n = 3). Data are normalized to nontreated control. (H) Mean (±SEM) Ifnβ mRNA levels in MEF^Trex1−/− treated with 500 μM Ap₄A or 500 μM JB419 for 3 hours versus untreated MEF^Trex1−/− (n = 4). (I) The 3D modeling of docked cGAMP (left) or Ap₄A (right) in the cleft of STING obtained using Amber force field and the crystal structure of human STING complexed to cGAMP (PDB: 4KSY). Ap₄A has adopted a horseshoe-like conformation. (J) Predicted molecular interactions that stabilize Ap₄A in the active site of STING. Atoms are pink spheres (acidic, red outline; basic, blue outline). Interactions are depicted by arrows showing the direction of the electron exchange (donor or acceptor), with backbone interactions in blue and sidechain interactions in green. The dotted contour defines the proximity of the available conformational space for the depicted compound. (K) WB analysis of in vitro binding experiment using 100 μg of GST, GST-HINT or GST-STING_139–378, and 10 μM streptavidin-immobilized biotinylated Ap₄A. (L) WB analysis of in vitro binding experiment performed using 100 μg of GST or GST-STING_139–378 and 10 μM streptavidin-immobilized biotinylated Ap₄A. Competition was performed with 100 μM cGAMP, DMXAA, JB419, or Ap₄A as indicated. (M) Proposed model. At steady state (left), LysRS interacts with R:D, delaying the recruitment of cGAS to R:D. LysRS interaction with R:D also stimulates Ap₄A production, which buffers STING activation. In the absence of LysRS (right), cGAS detects R:D, producing cGAMP and resulting in STING activation. IFN is produced, inducing the establishment of a positive feedback loop, which increases the total amount of cGAS and additional ISGs. (A, C, E, F, G, and H) Unpaired t test: ns, nonsignificant; *P < 0.05, **P < 0.01, and ****P < 0.0001. All immunoblots show representative experiments.

Fig. 6. LysRS-mediated negative regulation of Ifnβ expression is conserved in zebrafish.

(A) WB analysis of pull-down experiments performed as in Fig. 2D, except that 1, 3, and 10 pmol of recombinant zebrafish LysRS (zlysrs) protein were used; (n = 3). (B) Schematic representation of the zebrafish lysrs locus. The region targeted by lysrs-targeting morpholino oligo (MO-lysrs) and primers used for amplification of lysrs are indicated by a red line or black arrows, respectively. Boxes, exons; straight lines, introns. (C) Larvae injected with MO-lysrs or with control MO (MO-Ctl). (D) lysrs, ifnφ1, and isg15 mRNA levels were quantified by RT-qPCR in larvae injected with MO-lysrs or with MO-Ctl. Median (±SEM) of 12 pools of three independent larvae per condition. Unpaired t test. ***P < 0.001, ****P < 0.01. All blots show representative experiments.