Previous studies reported that CD83 is transiently expressed on human Mφ after LPS stimulation and that it is associated with IL-4 induced gene expression in murine Mφ (7, 22). However, temporal regulation of CD83 expression after both pro- and anti-inflammatory stimulation of murine Mφ has not yet been investigated. Thus, we incubated murine BMDM either with pro-inflammatory or with anti-inflammatory mediators and assessed CD83 expression kinetics. Stimulation with IL-4 induced a strong and long-lasting CD83 expression on the cell surface of murine Mφ after 16 h, while neither LPS, IFN-γ nor TNF-α induced elevated surface expression levels ( Figure 1A , left bar graph). Representative FACS histograms are shown in Figure 1A , right graph. In agreement with these results, we detected no Cd83 regulation on mRNA level upon stimulation with IFN-γ or TNF-α after 16h ( Figure 1B ). In addition, LPS stimulation of murine Mφ results in significantly reduced Cd83 mRNA levels in comparison to unstimulated Mφ ( Figure 1B ). Again, only IL-4 induced significant Cd83 mRNA transcript levels after 16 h post stimulation ( Figure 1B ).

Subsequently, we investigated the temporal regulation of CD83 expression. Thus, we stimulated Mφ with LPS and other pro-inflammatory compounds and analyzed surface expression of CD83 over a 20 h time course ( Figure 1C ). LPS treatment caused an almost threefold increase of CD83 surface expression at the 2 h time point, followed by a rapid decline to baseline levels after six to eight hours. Similarly, the yeast cell wall component zymosan, which in contrast to LPS acts via TLR2, induced fourfold higher CD83 surface levels during the first two hours of stimulation. A slightly slower and less pronounced response was observed upon treatment with TNF-α, which resulted in a peak of CD83 expression after 4 h ( Figure 1C ). IFN-γ induced a rather delayed type of response, with a steady increase up to two-fold within the first six hours and a subsequent decrease until 16 h after stimulation ( Figure 1C ). Interestingly, IL-4 and IL-13 treatment caused a 2- to 3-fold CD83 induction as early as 4 h after stimulation, but in contrast to the pro-inflammatory mediators, this elevated expression did not revert to baseline levels even 20 h after stimulation ( Figure 1D ). Treatment with IL-10, another anti-inflammatory cytokine, had no influence on CD83 expression ( Figure 1D ). Thus, we observed a striking discrepancy in CD83 regulation after stimulation: while pro-inflammatory mediators induced a very rapid but transient increase in CD83 surface expression, stimulation of the IL-4R with either IL-4 or IL-13 resulted in a stable CD83 display on the cell surface. These data indicate an interesting functional role of CD83 in Mφ biology, especially for the resolving phenotype associated with IL-4 signaling.

Next, we aimed to characterize the biological function of CD83 expressed by murine Mφ by using a conditional knock-out (cKO) strategy. By crossing mice carrying floxed Cd83 alleles with a CX3CR1-Cre line, we generated a conditional line with an abrogated CD83 expression specifically in Mφ (herein termed CD83^ΔMφ). To test the efficacy of our KO strategy, we treated Mφ from CD83 cKO mice and wt mice with IFN-γ or IL-4 for 16 h and assessed the expression of Cd83 by qPCR ( Figure 2A ). As depicted in Figure 2A , murine CD83wt Mφ stimulated with IL-4 show a highly significant induction of Cd83 mRNA levels compared to unstimulated CD83wt Mφ, whilst no expression was observed in Mφ from cKO mice ( Figure 2A ). These results were confirmed by surface expression analyses of cKO derived Mφ after IL-4 treatment, whereas wt derived Mφ showed a significant and stable expression of CD83 ( Figure 2B ). Similarly, CD83 protein was also absent in cell lysates derived from cKO BMDMs as shown by Western blot analyses ( Figure 2C ). Interestingly, while stimulation with IFN-γ had no apparent effect on CD83 surface expression after 16 h, total protein levels were also elevated ( Figure 2C ), although to a lesser extent than IL-4 treatment. Next, we examined whether CD83-deletion affects cell viability ( Figure 2D ) or differentiation efficacy ( Figure 2E ), by flow cytometry and found that neither were affected. Furthermore, CD83 deletion has no effect on expression levels of F4/80 as well as CD11b on CD83^ΔMφ ( Figure 2F ). Thus, we conclude that CD83 deletion does not alter cell viability nor differentiation of murine mock-, IFN-γ- or IL-4-stimulated BMDM, generated from CD83wt or cKO mice.

In APCs, such as DCs and B cells, CD83 has been reported to stabilize surface MHC-II and CD86 expression by preventing their ubiquitin-dependent degradation (14, 17). Both molecules are hallmarks of a classic activation via IFN-γ, and consequently, we addressed the question how deletion of CD83 might affect their surface display. We observed an up-regulation of MHC-II and CD86 molecules on BMDM after stimulation with IFN-γ and to a lesser extent after IL-4 treatment ( Figure 3A ). In line with previous reports regarding the CD83-MARCH-MHC-II axis, Mφ from CD83^ΔMφ mice exhibited significantly lower surface expression levels of MHC-II and CD86 ( Figure 3A ). Since CD83 inhibits MHC-II ubiquitination via the interference with the ubiquitin-ligase March1 or March8 in DCs or thymic epithelial cells, respectively (14, 15), we tested which ubiquitin-ligase predominates in BMDM. Thus, BMDM were either left unstimulated or treated with IFN-γ, and we detected comparable levels of Marchf1 transcripts. Since Marchf8 expression is reduced by almost two orders of magnitude in comparison to Marchf1 (see. Figure 3B ), we conclude that CD86 as well as MHC-II stabilization is achieved by CD83-mediated inhibition of March1.

Next, we analyzed the overall phenotype of CD83-deficient Mφ. Since CD83 expression is tightly associated with an IL-4 mediated alternative activation ( Figure 1C ), we hypothesized that deletion of CD83 would mostly affect polarization of AAM. Indeed, we detected an altered phenotype in IL-4 stimulated CD83-deficient Mφ. This phenotype was characterized by significantly increased Dectin-1 expression on protein ( Figure 3C , left bar graph) and mRNA level ( Figure 3C , right bar graph), which is associated with a pro-inflammatory CAM polarization (23–26). Concomitantly, in IL-4-stimulated CD83-deficient Mφ we observed significantly reduced expression levels of the inhibitory receptor CD200R, which is known to limit pro-inflammatory cytokine secretion (27) ( Figure 3D , left bar graph). Moreover, pro-resolving MSR-1 was also significantly decreased in IL-4 stimulated CD83-deficient Mφ, compared to CD83wt control Mφ. Furthermore, qPCR analyses of Clec7a ( Figure 3C , right bar graph), Cd200R ( Figure 3D , right bar graph) and Msr-1 ( Figure 3E , right bar graph) are in line with the protein data. Msr-1 is known to be upregulated on AAM being important to phagocytose E.coli bacteria. Since we observed a reduction in Msr-1 expression, we next checked whether CD83-deficient Mφ were impaired in their phagocytic activity to engulf E.coli. In fact, using a gentamicin protection assay revealed a significantly impaired phagocytic activity of IL-4 stimulated, CD83-deficient Mφ ( Figure 3F ). Collectively, these data suggest a profound functional change of IL-4 stimulated, CD83-deficient Mφ.

Since we detected a phenotypic change on IL-4-stimulated Mφ derived from CD83 cKO mice, we next examined whether members of the IL-4 signaling cascade are also modulated. IL-4 binds to IL-4Rα that recruits the IL-2Rγ chain, which leads to the activation of the tyrosine kinases Jak1/Jak3 and phosphorylation of STAT6, which form pSTAT6-dimers and translocate to the nucleus and initiate transcription of target genes (28, 29). To analyze possible differences between CD83wt and CD83 KO Mφ, we generated Mφ from CD83wt as well as CD83 cKO mice and stimulated them with IL-4 for 15 or 30 min. Subsequently, whole-cell lysates were prepared and analyzed by Western blot, in respect to pSTAT6 and STAT6 levels. In fact, we detected a decreased phosphorylation status of STAT6 upon IL-4 stimulation in CD83-deficient Mφ, compared to CD83wt Mφ ( Figure 4A ). Quantified ratios of pSTAT6 to STAT6 are shown in Figure 4B . Next, we analyzed the expression of STAT6 target gene Gata3 (30), and its expression was downregulated in IL-4-stimulated CD83-deficient Mφ, when compared to wt derived Mφ ( Figure 4C ).

Next, we analyzed the impact of CD83-deficiency regarding the cytokine and chemokine production of IL-4- and IFN-γ-stimulated Mφ. As depicted in Figure 5A , IL-4-stimulated CD83-deficient Mφ showed a significantly increased secretion of pro-inflammatory mediators, including IL-6, TNF-α, CXCL1, and G-CSF ( Figure 5A upper bar graphs). Additionally, we verified these data by qPCR and observed significantly increased expression levels of the corresponding transcripts Il6, Tnfa, Cxcl1, and Csf3 ( Figure 5A lower bar graphs). In addition, we detected significantly increased levels of RANTES/CCL5 ( Figure 5 , left bar graph) and of MCP-1/CCL2 in IFN-γ-stimulated CD83-deficient Mφ ( Figure 5 , right bar graph), suggesting that CD83 also influences the function of IFN-γ stimulated Mφ. Collectively, these data support our hypothesis that CD83 deficiency modulates activation and function of Mφ. Recently we reported that CD83-deficient DCs enhance antigen-specific T cell proliferation and increase secretion of pro-inflammatory cytokines compared to co-cultures with CD83wt DCs (17). Since the data described above indicate a modulated pro-resolving phenotype of CD83-deficient Mφ, we next assessed the T cell stimulatory capacity of IFN-γ- and IL-4-stimulated, in comparison to unstimulated Mφ, generated from CD83wt and CD83 cKO mice. Thus, we co-cultured the differently stimulated Mφ with allogeneic splenocytes, derived from BALB/c mice, and T cell proliferation was assessed via tritium incorporation. As depicted in Figure 6 , T cell proliferation was enhanced in all co-cultures with CD83-deficient Mφ, regardless of their stimulus ( Figures 6A–C ). This observation is also reflected by enhanced clustering of T cells upon co-culture with CD83-deficient BMDMs (see representative microscopic images¸ Figures 6A–C ). In MLRs with CD83-deficient Mφ we observed significantly increased proliferative response of alloreactive T cells ( Figures 6A–C ). Next, we investigated if the composition of T cell subsets, present in co-cultures with mock-, IFN-γ or IL-4-treated Mφ generated from CD83wt or CD83 cKO mice, would be altered. As depicted in Figure 6D , flow cytometric analyses revealed significantly reduced frequencies of CD4⁺Foxp3⁺ Tregs in co-cultures of allogeneic splenocytes with mock, IFN-γ or IL-4-stimulated CD83-deficient Mφ. This indicates that the reduced numbers of Tregs present in the co-cultures may account for the observed increased T cell proliferation ( Figures 6A–C ).

In order to investigate the in vivo relevance of CD83-deficiency in CX3CR1⁺ Mφ, we performed full-thickness excisional wound healing experiments using cKO (CD83^ΔMφ) in comparison to wildtype control mice (CD83wt). As depicted in Figure 7A , we induced 6 mm biopsy punches in the dorsal skin and monitored wound closure until day 6 and collected skin biopsies on day 3 as well as day 6 after wound infliction. As shown in Figure 7B , on day 3 the wound closure was significantly enhanced in CD83^ΔMφ mice when compared to CD83wt mice. This indicates a boost of the initial inflammatory phase, which is crucial for wound closure. This upregulated inflammatory phase was also confirmed by qPCR analyses, since transcripts such as Il6 and Cxcl1, associated with a pro-inflammatory macrophage phenotype, were upregulated in CD83^ΔMφ mice ( Figure 7C ). Concomitantly, markers associated with pro-resolving Mφ, including CD200r, Msr-1 as well as Ym-1 were significantly reduced ( Figure 7D ). Surprisingly, regarding surface wound closure on day 6, no differences were detected between CD83^ΔMφ and wt animals. However, in samples derived from cKO animals we observed significantly increased expression levels of Tgfb, Acta-2 as well as Col1a1. This indicates a higher prevalence of myofibroblasts and fibrosis associated transcript in CD83^ΔMφ mice, which are associated with disturbed wound healing processes ( Figure 7E ). Indeed, histological analyses of day 6 skin biopsies showed an altered reconstitution process, as indicated by an expanded epidermis, absent dermis and a prominent inflamed tissue area in CD83^ΔMφ mice ( Figure 7F , upper panel). In contrast, histological analyses of skin biopsies from CD83wt mice showed a distinct epidermis alongside the dermis and a resolving inflammatory area ( Figure 7F , lower panel). Of note, in CD83wt mice, hair follicles have already migrated into the former sites of excisional wounds, indicating a progressed stage of wound regeneration, which was not observed in CD83^ΔMφ mice.

To generate mice with CD83-deficient Mφ, we used a conditional knock-out strategy (CD83 cKO) by mating mice with floxed Cd83 alleles (47), with the Cx3cr1-Cre line, which was kindly provided by Prof. Dr. Gerhard Krönke (Department of Medicine 3, University Hospital Erlangen, Erlangen, Germany). Cre-negative littermates as well as age-matched Cre-positive CD83wt mice served as controls (hereafter referred to as wt mice). Animal care and all experimental procedures of the present study were performed in accordance with the European Community Standards on the Care and Use of Laboratory Animals and were approved by the local ethics committee.

Bone-marrow derived Mφ were generated from murine bone-marrow precursor cells from CD83 cKO mice and wild type littermates, in D10 medium consisting of DMEM (Lonza), 10% FCS (Merck), Penicillin-Streptomycin-Glutamine-solution (Sigma Aldrich) and 50 µM β-mercaptoethanol. Bone-marrow cells were flushed from femur and tibia of mice and seeded for 1d in D10 medium containing M-CSF supernatant (10 - 30%). After overnight incubation, cells were harvested and seeded at a starting density of 3-4 x 10⁶ cells per 10 cm² dish (Falcon) in D10 medium + M-CSF supernatant. Fresh D10 medium + M-CSF was added on day 3. On day 6, Mφ were harvested and stimulated as described below.

On day 6, Mφ were harvested with 10 mM EDTA-PBS, washed with fresh medium and seeded in uncoated 24-well plates at a cell density of 2 x 10⁶ cells per ml. For phenotypic and functional characterization of CD83, Mφ were generated from wt or cKO mice and seeded for differentiation into classically activated Mφ (CAM) or alternatively activated Mφ (AAM) using IFN-γ (300 U/ml) or IL-4 (40 ng/ml, PeproTech), respectively. For time kinetic experiments, Mφ were stimulated with inflammatory activators such as IFNγ (300 U/ml, PeproTech), LPS (100 ng/ml, In vivogen) or TNF-α (1000U/ml) or alternatively with mediators, such as IL-4 (40 ng/ml), IL-13 (40 ng/ml) or IL-10 (10 ng/ml) for the indicated time period. Subsequently, cells were analyzed by flow cytometry.

Live/dead discrimination was performed using either 7-AAD or LIVE/DEAD™ Fixable Aqua Dead Cell Stain (ThermoFisher Scientific). Surface staining of BMDMs and cells used in MLR assays was performed in PBS-diluted appropriate antibodies for 30 minutes. In the case of live/dead discrimination with 7-AAD, the dye was added just before the flow cytometry measurement. For intracellular staining, cells were permeabilized and fixed in Permeabilization Reagent (Thermo Fisher Scientific, 00-5523-00). The following antibodies were used from BioLegend, others are stated: F4/80 (BM8), CD11b (M1/70), CD200R (OX-110), Msr-1 (M204PA; Invitrogen), MHCII (M5/114.15.2), CD86 (GL-1), CD206 (C068C2), MERTK (2B10C42), CD83 (Michel-19) RORɣT (Q31-378), GATA3 (L50-823;BDBiosciences), T-BET (O4-46,BD Pharmingen), FOXP3 (FJK-16s; Thermo Fisher Scientific). Afterwards, the cells were washed with PBS and subsequently analyzed by flow cytometry.

Supernatants of BMDM were analyzed using the LEGENDplex™ Mouse Macrophage/Microglia or LEGENDplex™ Mu Pro-inflammatory Chemokine Panel (BioLegend), respectively, according to the manufacturer’s instructions.

On day 6, BMDMs from wt controls and cKO mice were harvested, seeded in 96-well plates and stimulated with IFN-γ or IL-4 or left untreated. Allogeneic splenocytes derived from BALB/c mice (4 x 10⁵ cells/well) were co-cultured with BMDMs in 96-well plates for 72 hours in D10 medium (37°C, 5.5% CO₂), at different Mφ:splenocyte ratios (1:2, 1:4, 1:8). To analyze the allogeneic T cell proliferation capacity, cell cultures were subsequently pulsed with [³H]-thymidine (1 μC/well; PerkinElmer, Germany) for additional 8-16 h. Culture supernatants were harvested onto Glass Fiber Filter Mates using an ICH-110 harvester (Inotech, Switzerland), and filters were counted in a 1450-microplate (Wallac, Finland). Cells of cocultures were also harvested after 72 h and used for flow cytometric analyses to determine frequencies of different T cell subsets.

To analyze the ability of BMDMs to phagocytose and uptake E.coli bacteria, a gentamicin protection assay was performed. Bone marrow derived Mφ were generated from wt or cKO mice and seeded in 6 well plates in technical replicates in D10 medium without antibiotics. Cells were differentiated into CAM or AAM for 16h. Afterwards, Mφ were exposed to E. coli (DSM 1103) at an MOI = 10 for 1 hour. After rinsing the cultures three times in PBS to remove non-engulfed bacteria, cells were incubated for 1 h in fresh RPMI1640 medium containing 100 µg/ml of gentamicin to kill extracellular bacteria. Gentamicin was removed, and the cells were gently rinsed three times in PBS. BMDMs were lyzed by incubating them for three minutes in PBS containing 2 mM EDTA and 0.5% saponin, followed by transfer to Eppendorf tubes and high-speed vortexing for 30 s. Subsequently, cells were plated onto blood agar plates and the next day, E.coli colonies were counted.

To assess CD83 protein content in whole cell lysates of murine BMDM, Western blot analyses were performed. In addition, protein levels of pSTAT6 and STAT6 in lysates of CD83wt or CD83-deficient Mφ were also analyzed by Western blotting. Thus, protein-lysates (20-30 µg per lane) were separated via SDS – polyacrylamide gel electrophoresis and blotted onto a nitro-cellulose membrane (GE Healthcare). After blocking in blocking reagent (5% BSA-TBST) membranes were incubated with the following primary antibodies overnight (4°C): goat CD83 (Clone: AF1437, R&D systems), mouse β-actin (Clone: AC-74, Sigma Aldrich), rabbit p-STAT6 (Clone: D8S9Y, Cell Signaling), rabbit STAT6 (Clone: D3H4, Cell signaling). Specific signals were detected using the appropriate HRP-labeled secondary antibody and the ECL Prime Western Blotting Detection Reagent (GE Healthcare). Quantification of Western Blots was performed using the ImageJ/Fiji software (48). The intensities of bands are visualized in bar graphs and represent the protein amount in arbitrary units. To analyze phosphorylation status of STAT6, band intensities of pSTAT6 were normalized to total STAT6 signals. β- Actin served as a loading control.

Total RNA was isolated from mock-, IFN-γ or IL-4-stimulated BMDMs generated from CD83wt or CD83 cKO mice. Cell pellets were lysed in RLT+β-Mercaptoethanol extracted by RNeasy Plus Mini Kit (Qiagen) according to the manufacturer’s instructions. In addition, wound biopsies were collected from CD83wt as well as cKO mice at day 0, day 3 or day 6, stored in RNAlater (Qiagen) at -80°C and subsequently used for further analyses. Homogenization of the tissue was performed in RLT+β mercaptoethanol using innuSPEED lysis Tube W (Analytic Jena). We performed three homogenization cycles à four minutes in the SpeedMill PLUS homogenizator (Analytic Jena).

Total RNA was reversely transcribed using the First strand cDNA synthesis Kit (Thermo Fisher Scientific GmbH), as described by the manufacturer. Briefly, 0.5-1 µg RNA was reversely transcribed and diluted 1:5 after synthesis. qPCR analyses were performed using the Sybr Green Super mix (Biozym) on a CFX96 Real time system (BioRad) and normalized to reference gene transcript Hprt. All primers were designed and validated according to the Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines. For primer sequences, see Table 1 .

To investigate the role of CD83 expressed by Mφ in physiological wound healing responses, we used the full-thickness excisional wound healing model as described previously (21). Briefly, CD83wt mice as well as CD83cKO mice were anesthetized using a mixture of ketamine and xylazine (120 mg/kg and 20 mg/kg body weight, respectively). In order to prevent wound healing by contraction and to give a defined scale for subsequent wound closure assessment, 8mm silicone rings (Thermo scientific) were mounted around the wound area, using vetbond (3M). Imaging was performed on day 0, 3, and 6 and wound diameters length (L) and width (W)) were determined by ImageJ and wound closure was calculated relative to the initial d0 wound dimension. Wound area (WA) on day X (dX) and wound closure (% of baseline) was calculated using the following equations:

On day 3 and 6 biopsies were obtained as 8 mm punches (pfm medicals) around the former wound area. Samples were either fixed in 4% paraformaldehyde for histological assessment or stored in RNAlater (Qiagen) at - 80°C for subsequent RNA analyses.

For the histological assessment of wounds, skin biopsies were obtained on day 3 and day 6 after wound infliction using 8 mm biopsy punches around the wound area. Excised tissue was subsequently fixed in 4% paraformaldehyde and processed by conventional histological techniques, embedded in paraffin wax and sectioned at 5 µm thicknesses. Sections were mounted onto glass slides, de-paraffinized and stained with hematoxylin and eosin (HE).

All statistical analyses were performed using GraphPad Prism 9.3.1 and the two-tailed unpaired student’s t- test or one- or two-way ANOVA for parametric data. Wherever necessary, we used non-parametric tests (Mann-Whitney-U or Kruskal-Wallis) when data was not normally distributed. Data are presented as mean values including the Standard Error Mean (SEM). P-values of *p<.05; **p<.01; ***p<.001; and ****p<.0001 were considered statistically significant.