The study was carried out in compliance with the ARRIVE guidelines (http://nc3rs.org.uk/arrive-guidelines). The R96C point mutation was introduced into the C57Bl/6 mouse genome by homologous recombination using CRISPR/Cas9 technology. One single nucleotide was replaced by homologous recombination in pronuclear-stage zygotes, as previously described (25). The C57Bl/6 background was chosen to facilitate the production of further recombinant mice (gene reporter or knockout strains) of interest to elucidate the mechanisms altered by the point mutation. Eight- to 10-week-old female or male SOCS2^KI mice or their littermates WT mice were bred and housed in a specific pathogen-free facility (INSERM US 006 – CREFRE). Body growth was determined for female and male WT and SOCS2^KI mice. Experiments were performed in an accredited research animal facility of the UMR IHAP, ENVT, Toulouse, France. Mice were handled and cared for according to the ethical guidelines of our institution (APAFIS#22936-2019112515186332) following the Guide for the Care and Use of Laboratory Animals (National Research Council, 1996) and in compliance with European directive 2010/63/UE under the supervision of authorized investigators. Mice were euthanized by cervical dislocation and all efforts were made to minimize the pain and distress of the animals.

Anesthetized (Isoflurane, Virbac) 10-week-old male mice (n=6) were scanned using a small animal computed tomography (CT) system (NanoScan PET/CT Mediso). The nanoScan CT has a rotating gantry and the X-ray source and detector rotate around the object. The scanning protocol was 35 kVp, 800 µA, 450 ms integration time, 720 projections per 360°, scan duration of 5’33”. The reconstructed voxel size was 125 x 125 x 125 µm. From the acquired data, the length of the humerus, radius, ulna, femur, and tibia and the length and width of the skull were bilaterally determined. To ensure the reproducibility of the measurements, precise anatomical landmarks were defined beforehand.

Staphylococcus aureus subsp. aureus (ATCC^® 49525™) XEN36 (bioluminescent bacterial strain) or the HG001 strain, a GFP-expressing mutant (26), were grown overnight in TSB at 37°C with orbital shaking (200 rpm). The culture was further diluted 1:100 in TSB and grown to the mid-log phase (O.D. 600 nm _1). Bacterial cells were pelleted (2800 x g, 10 min, 4°C), washed twice in PBS, and resuspended in physiological serum. The concentration of the bacteria was estimated by measuring the absorbance at 600 nm (OD600 = 0.8 for 2×10⁸ CFU/mL) and confirmed by serial dilution in PBS Tween 20 (0.05%) and plating on LB-Agar for determination of the number of colony-forming units (CFU). CFUs were determined after 24 h of incubation at 37°C. Bacteria were freshly prepared before each experiment and adjusted to the desired concentration. The XEN36 strain was killed by heating for 1 h at 65°C. S. aureus inactivation was confirmed by LB culture. Male mice were infected intraperitoneally (i.p.). with S. aureus. at 10⁸ CFU/mouse (100 ul/mouse).

The peritoneal cavity was washed by injecting 3 mL PBS. After recovery, peritoneal exudates were centrifuged (300 x g, 5 min) and the supernatants were stored at -80°C for further analysis. PerCs were harvested for cell phenotyping by flow cytometry. The spleens and lymph nodes (LNs) were removed and the cells were isolated through 70 and 40 μm cell strainers, respectively, to obtain single-cell suspensions in PBS.

The number of cells obtained from the peritoneal exudates, spleens, and LNs was determined using a flow cytometry absolute counting system (MACSQuant Analyzer, Miltenyi Biotec, Germany). Cells (1-2 x 10⁶) were incubated in HBSS, 0.5% BSA, 10 mM Hepes containing mouse FcR Blocking Reagent (Miltenyi Biotec, Germany) following the manufacturer’s instructions. Cell viability was assessed using Viobility 488/520 Fixable Dye (Miltenyi Biotec, Germany). Antibodies were incubated at 4°C for 30 min in the dark. The antibodies used for flow cytometry are listed in Supplementary Table 1 . Flow cytometry was used to measure efferocytosis by quantifying the number of F4/80^int Ly6G⁺ macrophages in the peritoneal cavity from WT or SOCS2^KI mice after S. aureus infection. The acquisition was performed using a MACSQuant (Miltenyi Biotec, Germany) flow cytometer with MACS Quantify software. Flow cytometry data were analyzed using FlowJo (Tree Star, USA) software.

Femurs and tibias from male WT or SOCS2^KI mice were cut at both ends and the bone marrow was flushed out with PBS with a syringe mounted with a 26-gauge needle. The cells were collected and cultured in X-Vivo (BE02-060F, Ozyme, France) supplemented with 50 ng/mL M-CSF (Peprotech, France) at a density of 2 x10⁵ cells/mL. Cells were incubated at 37°C in 5% CO₂ and fresh medium was added on days 3 and 5 of culture. On day 7, the BMMs were harvested, counted, and primed for 24 h with either GM-CSF (Biolegend, France), IFN-γ (Peprotech, France), or IL-10 (Peprotech, France) at 3, 10, or 30 ng/ml. The TLR ligands FLS1 (Invivogen, France) and CRX (Invivogen, France) at various concentrations (3, 10, 30, or 100 nM) or heat-killed S. aureus (HKSA) (Invivogen, France) or heat-killed E. coli (HKEB) (Invivogen, France) at various MOIs (3, 10, 30, or 100) were added to the cell cultures, and the cultures were incubated for an additional 24 h. We used the mycoplasma lipoprotein FSL-1, recognized by the TLR2/TLR6 heterodimer, HKSA, recognized mainly by TLR2, CRX-527, a highly specific TLR4 agonist and an LPS-like molecule, and HKEB recognized by TLR2 and TLR4. Cell supernatants were then harvested for cytokine detection. BMMs were pulsed 24 h with GM-CSF at 10 ng/ml to induce SOCS2 expression followed by a 3 h chase. Then, GM-CSF (10 ng/ml) was added de novo to cells followed by incubation for 10, 30, 60, or 120 min. Cells were harvested to analyze SOCS2 and pSTAT5/total STAT5 expression by western blotting.

Cytokines in peritoneal exudates were quantified using a customized multiplex assay kit containing IL-1β, IL1-α, IL-10, IL-6, IFN-γ, IL-12, IL-17, CXCL1, CXCL10, CCL2, CCL3, CCL4, and TNF-α (Tumor necrosis factor) (Milliplex-MAP, Merck Millipore, France) and a xMAP instrument (MAGPIX, Luminex). Individual cytokine detection kits were also used to quantify mouse IL-6 and TNF-α (Bio-techne, France) and GM-CSF (Biolegend, San Diego, USA).

After stimulation, BMMs were washed once with cold PBS. Total protein was extracted with radio immune precipitation assay (RIPA) buffer containing phenylmethylsulphonyl fluoride (PMSF) (Roche, Switzerland) and a protease inhibitor cocktail (Roche) before storage at -80°C. Protein concentrations were determined using the BCA method. Cell lysates (30 µg per lane) were subjected to SDS polyacrylamide gel electrophoresis and then transferred to polyvinylidene difluoride membranes for western-blot analysis. After blocking with 5% fat-free milk dissolved in TBS-T (Tris-buffered solution with Tween 20) for 1 h at room temperature, membranes were incubated overnight with antibodies raised against pSTAT5, total STAT5, and SOCS2 (Cell Signaling Technology, Beverly, MA) according to the manufacturer’s instructions ( Supplementary Table 2 ). The binding of these primary antibodies was visualized using goat anti-rabbit/anti-mouse immunoglobulin coupled with horseradish peroxidase (Jackson ImmunoResearch). The measurement of β-actin served as a loading control.

To assess phagocytosis, BMMs were first primed with either GM-CSF (Biolegend, France), IFN-γ (Peprotech, France), or IL-10 (Peprotech, France) for 24 h and then infected at a MOI of bacteria to macrophages of 10:1 with S. aureus HG001-GFP for 1 h at 37°C or 4°C as a control. Each well was washed three times and extra-cellular fluorescence quenched by adding 0.2 uM syringe-filtrated trypan blue (Sigma). Cell viability was determined using 7-AAD dye (BioLegend). The amount of live KI and WT BMMs with internal (BT-) engulfed bacteria (GPF+) was measured by flow cytometry (MACSQuant, Miltenyi Biotech, Germany).

Male mice were infected i.p. with bioluminescent S. aureus XEN36 at 10⁸ CFU/mouse (100 ul/mouse). A neutrophil elastase solution (680 FAST, Perkin Elmer) was injected intravenously (4 nmol/100 ul/mouse) 3 h later. The animals were imaged by bioluminescence and fluorescence 24 h after S. aureus infection.

Prism v. 8.0.1 (GraphPad Software, San Diego, CA, USA) was used for all analyses. Significant differences were analyzed using the t-test or multiple t-tests of ANOVA. P<0.05 was considered statistically significant. Statistical computing was carried out and some of the graphics were generated using R software. Graphics were designed using the ggplot2 R package. Heatmaps were computed after the transformation of the cytokine concentrations into Z-scores and generated using the Pheatmap R package (Kolde Raivo). PLSDA was carried out using the MixOmics package (27).

For further investigating the R96C SOCS2 mutation in vivo, we used CRISPR/Cas9 gene editing to generate a new C57BL/6 mouse strain bearing the Arg96 to Cys mutation (SOCS2^KI mice). SOCS2^KI mice were indistinguishable from wildtype (WT) mice until weaning at three weeks of age, but they subsequently grew more rapidly ( Figure 1A ), and achieved higher body weight (average of 40%) by six weeks of age, in both male and female mice ( Figure 1B ). CT imaging of skeletons from 10-week old males revealed that the length of most long bones and skull was significantly increased ( Figure 1C ), showing that the gigantism phenotype of SOCS2^-/- mice is resumed in the R96C knock-in model. In line with our previous observation, the R96C mutation is associated with a loss of SOCS2 functions. We next characterized the immune-cell composition of various immune organs and compartments, such as the spleen, LNs, and peritoneal cavity of SOCS2^KI mice relative to those of WT mice ( Supplementary Figure 1 ). No significant difference was observed in the immune-cell composition of the spleen ( Figure 1D ), the LNs ( Figure 1E ), or the peritoneal cavity ( Figure 1F ). However, the cell number per unit of body weight was significantly lower in SOCS2^KI mice, as shown for the spleen ( Supplementary Figure 2 ).

Due to the previously reported important role of SOCS2 in these cell types, we next analyzed macrophage and DC populations in the peritoneal cavity and spleen of SOCS2^KI mice in comparison with WT mice ( Figure 1G ; Supplementary Figure 3 ). At the steady state, there were no significant differences in the cell composition, and the cell numbers of various macrophages or DCs types were similar in the two genotypes.

In mice, bone-marrow macrophages differentiated using M-CSF as a growth factor are a frequently used and convenient model to study signaling and functions in macrophages (28). After M-CSF culture of bone marrow stem cells, no difference in the phenotype is noticeable between WT and SOCS2^KI mice. Both BMM cultures showed a homogeneous population of F4/80⁺ cells, with similar levels of CD11b and MHCII molecule expression ( Supplementary Figure 4 ).

Evidence has now accumulated indicating that SOCS1, SOCS3, and CIS are induced after TLR engagement in macrophages and DCs, and this contributes to avoid overshooting TLR stimulation (29). Similarly, it was previously shown (30) that various TLR ligands can induce Socs2 gene expression in human DCs or thioglycolate-induced mouse peritoneal macrophages (31). We thus investigated SOCS2 protein expression after TLR stimulation by western blot in WT and SOCS2^KI M-CSF-derived macrophages. To our surprise, neither TLR-2 (FSL1, HKSA) nor TLR-4 (CRX) ligands induced SOCS2 expression in BMMs from both WT and SOCS2^KI mice after 24 h of culture ( Figure 2A ) in contrast to previous reports using GM-CSF as a differentiating factor. We also investigated the direct effect of TLR ligands on pro-inflammatory cytokine production in these conditions. There was no or few difference in the secretion of IL-6 in the cell culture supernatants of BMMs from WT and SOCS2^KI mice stimulated with TLR ligands like FSL1 or CRX, or heat-killed bacteria like HKSA or HKEB, regardless of the concentration we used ( Figure 2B ).

However, it is worth noting that FSL1 stimulation induced higher production of GM-CSF by the BMMs from SOCS2^KI than WT mice ( Supplementary Figure 5 ).

SOCS2 is a well-known feedback inhibitor of the JAK-STAT signaling, and STAT5-inducing cytokines, such as GM-CSF promote its expression (32, 33), whereas IFN-γ via STAT-1 and IL-10 via STAT-3 induces SOCS1 (34) and SOCS3, respectively (as reviewed in 35). Indeed, SOCS2 was found expressed in BMMs from WT mice after 24h culture with GM-CSF, but not with IFN-γ or IL-10 ( Figure 2C ). After GM-CSF culture, SOCS2 was also expressed by BMMs from SOCS2^KI mice at comparable levels as indicated by densitometry results on western blot data ( Figure 2D ). We then detected STAT5 in WT BMMs whatever the conditions, but phospho-STAT5 was only seen when GM-CSF was present in the culture medium ( Figure 2C ). Further stimulation with FSL-1 did not substantially change STAT5 expression, nor its phosphorylation. No difference in pSTAT5/STAT5 levels were noticed between the two SOCS2 variants, and SOCS2 was equally well detected ( Figure 2E ).

To investigate the possibility of a difference in the JAK-STAT signaling turnover rate, we next studied the kinetics of STAT5 phosphorylation/dephosphorylation following the stimulation with GM-CSF. For that purpose, both cell types were cultured for 24h with GM-CSF to induce SOCS2 expression, which was followed by a 3-h chase to return to the steady state and absence of phospho-STAT5, as shown in the control condition in Figure 3A . Phospho-STAT5 levels were monitored from 10 to 120 min after re-introduction of GM-CSF in the BMM medium ( Figure 3A ). Phosphorylation of STAT5 was already present at high levels in both KI and WT cells after 10 min of stimulation. After that, STAT5 phosphorylation decreased rapidly in WT BMMs while it was sustained at high levels in SOCS2^KI cells for up to 30 min post-stimulation; a significant difference was still present after one hour of stimulation ( Figure 3B ). Relative to WT, STAT5 dephosphorylation was significantly delayed in SOCS2^KI cells ( Figure 3C ), consistent with our working hypothesis of a difference in the regulation of the JAK/STAT pathway when SOCS2 is inactive.

Macrophages secrete both pro- and anti-inflammatory cytokines following initial exposure to microbial-associated molecular patterns (MAMPS). As such, deletion of STAT5 results in increased pro-inflammatory cytokine expression in mouse myeloid cells with PRR (pattern recognition receptor) expression (36), but results are less known if STAT5 signals are increased. We explored the consequences of the SOCS2 LOF mutation on the pro-inflammatory cytokine secretion after activation with TLR2-related ligands, FSL-1 and HKSA. For that, SOCS2^KI and WT BMMs were cultured for 24 h with different concentrations of GM-CSF to induce SOCS2, followed by a MAMPS stimulation. Cytokine concentrations ( Supplementary Figure 8 ) are depicted as a heatmap according to the FSL-1 (from 3 to 100 nM) and GM-CSF concentrations (from 3 to 30 ng/ml) in main Figure 4 . Lower concentrations of FSL-1, even in the presence of the lowest concentration of GM-CSF (3 ng/ml) tested here, are sufficient to induce a greater production of IL-6 in SOCS2^KI than WT BMMs ( Figure 4A ). Indeed, stimulation with FSL1 (at 30 and 100 nM) resulted in significantly higher production of IL-6 by BMMs from SOCS2^KI mice with a 4-fold increase. Interestingly, TNF-α secretion was not affected by the loss of SOCS2 function ( Figure 4B ).

Previous studies indicated that SOCS2 may regulate other SOCS proteins, in particular SOCS1 and SOCS3 (22), in a dose-dependent manner (37). Through this mechanism, SOCS2 can indirectly modulate cytokine-induced STAT activation by removing the negative regulation mediated by the two other SOCS proteins that are targeted.

We next investigated the possible interaction of SOCS2 on the respective role of IFNγ and IL-10 cytokines known to regulate SOCS1 and SOCS3, respectively. As done with GM-CSF, BMMs were first primed with IFN-γ and IL-10 before stimulation with the FSL-1 ligand. Recognition of FSL-1, a diacylated lipoprotein, is mediated by TLR2 which cooperates with TLR6 through their cytoplasmic domain to induce a signaling cascade leading to AP-1 and NF-κB activation and cytokine production. The cytokine response and phagocytic activity of cytokine-driven macrophages were quantified. Unexpectedly, IL-6 concentrations were significantly higher in culture supernatants from SOCS2^KI BMMs than those from WT BMMs after 30 nM of FSL-1 stimulation in the presence of increasing concentrations of IFN-γ and IL-10 (from 3 to 30 ng/ml) ( Figure 5A ). SOCS2^KI BMMs also produced higher levels of TNF-α in the IL-10 conditioning medium, even at the lowest FSL-1 concentration (3 nM) ( Figure 5B ). By contrast, phagocytosis was not significantly different between the two BMMs types, whatever the cytokines GM-CSF, IL-10, or IFN-γ used to pre-sensitize the BMMs ( Supplementary Figure 6 ). In a context with the secretion of various cytokines, SOCS2 may have several effects on the cell responsiveness to MAMPS that are difficult to predict in culture.

Peritoneal macrophages are central in the response to Staphylococcus aureus in a peritonitis model. To examine the impact of the LOF SOCS2 mutation, we inoculated SOCS2 KI and WT mice with a sublethal dose of S. aureus, as previously described (38) ( Figure 6A ). After 16 h of infection, peritoneal exudates were collected and the different cell types and cytokine concentrations were determined. The total cell number was higher in SOCS2^KI mice, due to a higher number of Ly6G⁺ neutrophils, and several macrophage subsets (defined as F4/80^int/+ cells). A partial lest-square differential analysis (PLSDA) showed a clear dichotomy between the two genetic backgrounds using cell composition data ( Figure 6B ) indicating that the inflammatory response differed according to the SOCS2 variant. In vivo imaging showed higher neutrophil recruitment in SOCS2^KI than in WT mice, confirming the difference in inflammatory cell recruitment upon infection ( Figure 6C ). Interestingly, injection of heat-killed S. aureus (HKSA) failed to recapitulate the observed inflammatory response pattern ( Supplementary Figure 7 ), suggesting that a dynamic response was established after inoculation of living bacteria, on the contrary to MAMPS injection. After 48 h of S. aureus infection, inflammatory cell recruitment was highly similar between the two types of mice ( Figure 6D ). As the cell numbers for neutrophils and macrophages were different at early stages of infection, we studied the bacterial fitness at 16, and 48 hours post-infection. In accordance with the increase in the number of neutrophils, there was a significantly lower number of live bacteria 16h post-infection ( Figure 7A ), associated with an increase in efferocytosis (as indicated by the increased number of F4/80⁺Ly6G⁺ cells) ( Figure 7B ). Interestingly, the peritoneal exudates of SOCS2^KI mice contained much more IFNγ and CXCL10 than those of WT mice ( Figure 7C ). By contrast, at 48h post-infection, the number of live bacteria was significantly higher in SOCS2^KI mice, indicating that, despite early recruitment of inflammatory cells, the infection was not contained in the SOCS2^KI mice and continued to progress.

Our results bring evidence that SOCS2 R96C mutation leads to a dysregulation of the inflammatory response. Along with infection with S. aureus, the initiation of the inflammatory reaction was similar between the two SOCS2 variants. However, later steps were characterized by increased production of cytokines, sustained recruitment of inflammatory cells, and a limited ability to control bacterial burden in R96C SOCS2 animals. Nevertheless, while we did not precisely identify the molecular and cellular mechanisms causing the impaired resolution phase in mutants, we showed in vitro that the R96C mutation led to a long-lasting phosphorylation of STAT5, which very likely contributes to inflammatory cytokines secretion in vivo. Further investigations will be necessary to fully decipher this process.