Peripheral blood mononuclear cells (PBMCs) (n = 33) and FLS (n = 4) were obtained from blood and inflamed joints of RA patients, respectively. All subjects provided written informed consent, and the protocols were approved by the Ethics Committee of Galicia prior to patient inclusion in this study (studies numbers 2020/159 and 2021/03). RA patients fulfilled the American College of Rheumatology/European Alliance of Associations for Rheumatology 2010 classification criteria for RA (10). Clinical characteristics of patients are detailed in Supplementary Table S1.

PBMCs were obtained by Ficoll gradient (STEMCELL Technologies) and CD14⁺ monocytes were isolated by using the MagniSort Human pan-Monocyte isolation kit (Thermo Fisher Scientific). Monocytes were differentiated into RA macrophages (RA MØ) by culturing in Iscove’s Modified Dulbecco’s Medium (IMDM) supplemented with 10% of heat-inactivated fetal bovine serum (FBS, Corning™) and 10000 I.E penicillin-streptomycin (Lonza™ BioWhittaker™), in the presence of IFN-γ (10 ng/mL; R&D Systems; RA MØ_IFN-γ) or M-CSF (25 ng/mL; PeproTech; RA MØ_M-CSF) for 7 days. On one side, macrophages were differentiated in the presence or the absence of recombinant human (rh)Sema3B (200 ng/mL; R&D Systems) for 7 days. Conversely, RA MØ_IFN-γ and RA MØ_M-CSF were cultured for 7 days and stimulated for 24 h with LPS (10 ng/mL; InvivoGen) in the presence or absence of rhSema3B (200 ng/mL). RA MØ_IFN-γ and RA MØ_M-CSF were pre-incubated during 1 hour with neutralizing anti-Neuropilin-1 (α-NRP1) antibody (5µg/mL; R&D Systems), neutralizing anti-Neuropilin-2 (α-NRP2) antibody (5µg/mL; R&D Systems) or their respective isotype controls (sheep and goat IgG; 5µg/mL; R&D Systems), and stimulated with LPS (10 ng/mL) in the presence or absence of rhSema3B (200 ng/mL) for 24 h.

Alternatively, RA MØ_IFN-γ and RA MØ_M-CSF were also pre-incubated during 1 hour with DMSO or a specific MerTK inhibitor (100nM; UNC2881, Cayman Chemical) (11, 12) and stimulated with LPS in the presence or absence of rhSema3B for 24 h.

Cells were lysed for mRNA expression analysis or typsinized for Flow Cytometry. Cell-free supernatants were harvested for cytokine analysis.

RA FLS were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Lonza™ BioWhittaker™) containing 10% FBS, 200 mM Glutamine (Lonza™) and 10,000 U/mL penicillin-streptomycin (Thermo Fisher Scientific) and used between passages 6 to 10. Prior to stimulation, RA FLS were cultured overnight in DMEM containing 1% FBS. Afterwards, the cells were stimulated for 4 h with the supernatants (20%, v/v) from RA MØ_IFN-γ and RA MØ_M-CSF.

RNA from RA MØ and RA FLS was isolated employing the NucleoSpin RNA/Protein Mini kit (Macherey-Nagel). Total RNA was reverse-transcribed using iScript (Biorad). cDNA was amplified by qPCRs in duplicates using SYBR green (Biorad) and specific primers (Integrated DNA Technologies (IDT); Supplementary Table S2) with a CFX96 Touch Real-Time PCR Detection System (Biorad). Relative levels of gene expression were normalized to the expression of 2 housekeeping genes (GAPDH and B2M). The relative quantity (RQ) of mRNA was calculated by using the formula 2^–ΔΔCt.

IL-12p70, IL-6, TNF (R&D Systems) and Resolvin D1 (Cayman Chemical) protein levels were measured by ELISA in cell-free supernatants, according to the manufacturing instructions.

Data were acquired on a CytoFLEX S analyser (Beckman Coulter). 10% of Anti-Hu Fc Receptor (Thermo Fisher Scientific) was used for avoiding non-specific binding. Macrophages were stained with Fixable Viability Dye eFluor for dead cell exclusion (e450; Thermo Fisher Scientific) and antibodies for CD14-PerCPCy5.5, CD64-FITC, CD80-BV510, CD86-PECy7, CD163-APC-Cy7, CD206-APC, HLA-DR-BV605 and MerTK-PE (all Biolegend). In the case of MerTK, Fluorescence Minus One (FMO) enabled the tagging of the positive population. After excluding debris, doublets and dead cells, cell populations were analyzed using CytExpert software and Cytobank platform (Beckman Coulter). Results were expressed as the Median Fluorescence Intensity (MFI).

Macrophage phagocytic activity was determined by the uptake of the fluorogenic substrate DQ Red BSA (FITC; Thermo Fisher Scientific). Stimulated RA MØ_IFN-γ and RA MØ_M-CSF were typsinized and plated in conical well plates. DQ Red BSA (20 ng/mL) was added for 0, 30 and 60 minutes. Uptake data were acquired on a CytoFLEX S analyser (Beckman Coulter) and results were expressed as percentage of DQ-BSA uptake versus 30 minutes basal uptake.

Protein from RA MØ_M-CSF was isolated by Laemmli buffer. Equal quantities of total protein were submitted to electrophoresis on Polyacrylamide gels and transferred to PVDF Transfer membranes (Thermo Fisher Scientific). Membranes were incubated (4°C, overnight) with primary antibodies for (p)hosphoMerTK (FabGennix) and β-actin (R&D Systems) in 4% Milk-TBS/T, washed and incubated in 2% Milk-TBS/T containing HRP-conjugated secondary antibody (anti-mouse IgG, Thermo Fisher Scientific). Protein was developed with ECL Western Blotting Substrate (Thermo Fisher Scientific) employing a ChemiDocTM MP System (Biorad). Densitometry analysis was performed by ImageJ software and relative protein expression was normalized to the values for β-actin.

The gene expression of Sema3B receptors in bone marrow-derived macrophages (BMDM) from wild type (WT) and MerTK deficient (Mertk^-/-) mice was retrieved from array profiling data available at the Gene Expression Omnibus (GEO–NCBI; GSE205070) (13).

Statistical analysis was performed by using Windows GraphPad Prism 8 (GraphPad Software, Inc.). Normality was analysed by Shapiro-Wilk and Kolmogorov-Smirnov tests. The potential differences between experimental groups following normal distribution were analysed by One-way ANOVA and Paired t tests, as applicable. Data non-following normal distribution were analysed by Friedman’s test. P values < 0.05 were considered statistically significant.

We firstly analysed the effect of Sema3B in inflammatory macrophages (RA MØ_IFN-γ). Sema3B alone did not affect the expression of pro-inflammatory mediators, but significantly reduced the LPS-induced expression of IL12B, IL23, CD86, TNF and CCL2 , while moderately IL1B , IL6 and CXCL10 (Figure 1A). In the case of anti-inflammatory mediators, Sema3B induced the expression of STAB1, but did not modulate the expression of IL10, IL13 and TGFB (Supplementary Figure S1A).

The influence of Sema3B was also determined on RA MØ differentiated with M-CSF, which promotes an anti-inflammatory phenotype (14). Similarly to RA MØ_IFN-γ, Sema3B alone did not affect the expression of pro-inflammatory mediators in RA MØ_M-CSF, but significantly down-regulated the LPS-induced expression of IL12B , besides reducing CD86, TNF and CCL2 (Figure 1B). In contrast, the anti-inflammatory mediators STAB1 and TGFB were up-regulated by Sema3B, alone and in combination with LPS (Supplementary Figure S1B).

We validated these findings at the protein level in both RA MØ_IFN and RA MØ_M-CSF, in which Sema3B reduced the LPS-induced secretion of IL12p70, TNF-α and, in RA MØ_IFN-γ specifically, IL-6 (Figures 1C, D).

With the aim of determining whether the anti-inflammatory effect of Sema3B could have functional consequences on effector cells, RA FLS were stimulated with cell-free supernatants from RA MØ_IFN-γ and RA MØ_M-CSF. Interestingly, supernatants of those macrophages stimulated with LPS in combination with Sema3B reduced the expression of IL1B, IL6, IL8, TNF, CCL2, CXCL10, MMP1 and MMP3 compared to supernatants of LPS-stimulated macrophages, although differences were not significant for all mediators (Figures 1E, F).

The effect of Sema3B on the modulation of M1 (CD80, CD86 and HLA-DR) and M2 (CD163 and CD206) surface markers expressed by RA synovial macrophages was also evaluated (6). In RA MØ_IFN-γ, Sema3B alone promoted CD64 and CD206 expression and, in LPS-stimulated macrophages, reduced the expression of CD64, HLA-DR and, significatively, CD86 (Figure 2A). In the case of RA MØ_M-CSF, we found that, in combination with LPS, Sema3B decreased the expression of CD64 and a trend for CD80 (Figure 2B). We also determined whether Sema3B was able to modulate the expression of specific markers during the macrophage differentiation process. Sema3B significantly down-regulated the expression of HLA-DR in IFN-γ-differentiated macrophages, but it did not modulate M2 markers. During the differentiation process with M-CSF, Sema3B reduced the expression of HLA-DR and CD64, while it significantly induced the expression of CD163 (Supplementary Figure S2).

Altogether, these data demonstrate an anti-inflammatory effect of Sema3B in RA MØ, by reducing the expression of the inflammatory phenotype and modulating the phenotypic characteristics of RA MØ.

We next sought the mechanisms involved in the anti-inflammatory effect of Sema3B. Sema3B did not modulate the expression of its known receptors, PlexinA1 and PlexinA2, and co-receptors, Neuropilin-1 (NRP-1) and Neurolipin-2 (NRP-2) in either RA MØ_IFN-γ or RA MØ_M-CSF (15, 16) (Figures 3A, B). However, the neutralization of both NRP-1 and NRP-2 abolished the Sema3B-mediated decrease of IL12p70 in RA MØ_IFN-γ, demonstrating that both Sema3B co-receptors are involved in the anti-inflammatory role of Sema3B (Figure 3C).

Due to the crucial role of MerTK in resolving inflammation in RA (9, 17, 18), the involvement of Sema3B on the expression of this tyrosine kinase receptor was also evaluated. Sema3B alone, but not in combination with LPS, up-regulated the mRNA and protein expression of MerTK, and the mRNA expression of the MerTK ligand GAS6 (19–21) in both RA MØ_IFN-γ (Figure 4A) and RA MØ_M-CSF (Figure 4B), although differences were more pronounced in the latter. Sema3B also increased MerTK expression during the differentiation process into RA MØ_M-CSF (Supplementary Figure S2B). Importantly, Sema3B induced the activation of MerTK in RA MØ_M-CSF at different time points, being this effect significant at 4 hours (Figure 4C).

Since MerTK⁺ RA MØ are involved in the resolution of inflammation, we next analysed the effect of Sema3B in the phagocytic activity of macrophages and the synthesis of Resolvin D1, a lipid induced by macrophage MerTK signalling involved in resolving joint inflammation (22). In RA MØ_IFN-γ, Sema3B moderately increased Resolvin D1 secretion, although it did not modulate the macrophage phagocytic activity (Figure 4D). On their part, Sema3B-stimulated RA MØ_M-CSF significantly raised the secretion of Resolvin D1, as well as their phagocytosis capacity, although it resulted insufficient counteracting the effect of LPS (Figure 4E).

Ultimately, we tested if these responses were mediated by MerTK, using a specific MerTK (i)nhibitor. Sema3B reduced the LPS-mediated secretion of IL12p70 in both RA MØ_IFN-γ and RA MØ_M-CSF, and this effect was abrogated in the presence of the MerTKi (Figures 4F, G). Also, MerTK inhibition reversed the Sema3B-induced phagocytic activity of RA MØ_M-CSF (Figure 4G).

Therefore, these results suggest that the anti-inflammatory and pro-resolving responses induced by Sema3B are, at least in part, mediated by MerTK.