Properdin-deficient and wild-type mice, 20 to 25 g body weight, 10- to 12-week-old, male and female, were used in our experiments. Properdin-deficient mice were genetically engineered through gene-specific targeting to be deficient of properdin and were shown to lack properdin in their serum 29. The novel line is designated Cfp^tm1Cmst, and homo- and hemizygous mice are termed KO. Wild-type C57BL/6 mice were from the same fully backcrossed line. Animals were maintained in a specific pathogen-free environment and had free access to water and standard chow. All experiments were conducted in accordance with the International and National Guidelines for the Care and Use of Laboratory Animals and were approved by the Animal Care Committee at the Institute of Microbiology, Sofia.

A cocktail of four monoclonal antibodies to type II collagen (ArthritoMab; MD Biosciences, Saint Paul, MN, USA; 5 mg/100 μl) was injected intraperitoneally at day 0. The ArtritoMab binds to the epitopes C1^I, J1, U1, and D3 in the full-length CII fragments (CB8, CB10, and CB11). The internal control group of mice received equal volume of sterile phosphate-buffered saline PBS (pH 7.4) (PBS group). At day 3, all animals were intraperitoneally injected with LPS (Escherichia coli 055:B5; MD Biosciences; 50 μg/200 μl endotoxin-free water). Mice were examined for the development of arthritis for 10 days. Clinical score was done blindly by using a system based on the number of inflamed joints in front and hind paws, inflammation being defined by swelling and redness at the scale from 0 (no redness and swelling) to 3 (severe swelling with joint rigidity or deformity; maximal score for four paws, 12).

Synovial fluid was harvested by lavage of the knee cavity with 25 μl of PBS containing 1 mM ethylenediaminetetraacetic acid (EDTA; Sigma-Aldrich, Diesenhofen, Germany). After centrifugation, the cell pellets from all samples per group were pooled, counted, and used for flow-cytometry analyses while supernatants were stored at -70°C and assayed for cytokine and C5a content. Blood was collected in heparin tubes (10 U heparin/ml). Plasma was obtained after centrifugation at 350 g for 15 minutes at 4°C, stored at -70°C, and used for cytokine and C5a measurements.

Tibias and femurs of WT and KO mice were collected in sterile tubes with plastic adapter and then centrifuged at 350 g for 30 seconds. The cell pellet was washed twice with PBS containing 1 mM EDTA, resuspended at concentrations of 1 × 10⁶ cells/ml, and used for flow-cytometry analyses.

Blood was mixed in an equal volume with 6% Dextran T-500 sodium salt (Sigma-Aldrich) diluted in 0.9% NaCl (pH 7.0) and incubated for 40 minutes at room temperature. The upper layer was subjected to gradient centrifugation for 30 minutes at 350 g, 22°C. The layer enriched with mononuclear cells was carefully collected, washed with PBS, resuspended at concentrations of 1 × 10⁶ cells/ml, and used for flow-cytometry analyses or for further purification of CD4⁺ T cells. The erythrocytes in the pellet were lysed with buffer (155 mM NH₄Cl, 10 mM KHCO₃, 2 mM EDTA; pH 7.4) for 3 minutes, neutrophils were washed with PBS, counted, resuspended at concentrations of 1 × 10⁶ cells/ml, and used for flow-cytometry analyses. Exclusion dye staining with 0.05% Trypan blue showed more than 95% viable cells in isolated populations.

Cell suspensions from KO and WT mice were prepared from freshly removed spleens and after erythrocyte lysis 28. The population was passed through a sterile strainer with 70-μm nylon mesh pores (BD Falcon; BD Biosciences) and washed with PBS. Splenocytes were resuspended at concentrations of 1 × 10⁶ cells/ml and then used for flow-cytometry analyses and cytokine assays.

CD4⁺ T cells were purified by indirect panning of spleen and blood cell populations. Petri dishes were coated with Fc specific antibody against rat IgG (10 μg/ml) for 24 hours at 4°C. Cells (1 × 10⁶ cells/ml) in 2% fetal calf serum (FCS)/PBS were incubated with rat antibodies against CD14 (rmC5-3), CD16 (clone 2.4G2), CD19 (clone 1D3), and CD8 (clone OX8) (all from BD Pharmingen, BD Biosciences; 0.2 mg/1 × 10⁶ cells) for 15 minutes at 4°C. After washing with PBS, cell suspension was resuspended in 5 ml 5% FCS/PBS, added to the coated Petri dishes, and incubated for 10 minutes at room temperature. Unbound cells were eluted carefully, centrifuged, and resuspended in sterile complete RPMI-1640 medium (Biowhittaker; Lonza, Basel, Switzerland) containing 5% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 2 mM L-glutamine, and 25 μM β-mercaptoethanol, and counted. The purity of the cell population was 85% to 90%. CD4⁺ T cells (1 × 10⁶ cells/ml) were stimulated in 48-well plates with concanavalin A (ConA; 2 μg/ml; Sigma-Aldrich) and in the presence of murine recombinant IL-2 (10 ng/ml; PeproTech EC, London, UK), or where indicated, with plate-bound anti-CD3 (10 μg/ml; clone 145-2C11; BD Biosciences) and soluble anti-CD28 (2 μg/ml; clone 37.51; BD Biosciences) antibodies. After 48 hours at 5% CO₂, 37°C, plates were centrifuged at 350 g for 10 minutes at 4°C; cells were collected, washed twice, and subjected to flow-cytometry analyses for intracellular production of IL-17 and IFN-γ. Supernatants from splenic CD4⁺ T cultures were used in enzyme-linked immunosorbent assay (ELISA) for determination of IL-6, IFN-γ, and IL-17 secretion.

The levels of IL-17 in synovial fluid, plasma, and culture supernatants and of IL-6 and IFN-γ in the spleen-culture supernatants were determined with ELISA by using commercial kits of PeproTech EC (London, UK) with detection limit of 8 pg/ml and of 20 pg/ml, and of Abcam (Cambridge, UK; detection limit of 16 pg/ml), respectively. The amount of C5a in synovial fluid was evaluated as previously described 30. The samples were assayed in triplicate. The concentrations of the cytokines and C5a were calculated from a standard curve of the respective recombinant mouse protein, by using Gen5 Data Analysis Software (BioTek Instruments, Bad Friedrichshall, Germany).

Freshly isolated cell populations (1 × 10⁵/sample) were washed with 2% FCS/PBS and incubated with antibodies against CD3 (clone 145-2C11; FITC labeled; BD Pharmingen), CD4 (clone L3/T4; PE-labeled; BD Pharmingen), Ly6G (clone 1A8, FITC or APC-labeled; BioLegend, Uithoorn, Netherlands), CD11b (clone M1-70; Alexa-Fluor 647-labeled; BioLegend), CD69 (clone H1.2F3; APC-labeled; BD Pharmingen), CD14 (clone rmC5-3; PerCP-Cy 5.5-labeled; BD Biosciences), RANKL (clone IK22/5; PE-labeled; BioLegend), C5aR (clone 20/70; PE labeled; BD Biosciences or APC-labeled; BioLegend); C5L2 (human reactivity; clone 1D-M12; PE labeled; BioLegend); FcγRIII/II (clone 2.4G2; FITC-labeled; BD Pharmingen), and IgG isotype controls for 15 minutes at 4°C. After washing with 2% FCS/PBS, the samples were analyzed with flow cytometer (BD LSR II) by using BD FACSDiva v6.1.2 Software (Becton Dickinson GmbH, San Jose, CA, USA). In the experiments determining the expression of C5L2 on mouse peripheral neutrophils, purified anti-C5L2/GPR77 (clone 468705; R&D Systems, Wiesbaden-Nordenstadt, Germany) was incubated for 15 minutes at 4°C followed by washing steps and staining with anti-rat IgG-PE (minimal x reactivity; BioLegend). The isotype monoclonal rat IgG2b (R&D Systems) was used to control staining specificity.

CD4⁺ T cells were restimulated with phorbol-12-myristate-13-acetate (PMA; 10 ng/ml; Sigma-Aldrich) and ionomycin (2 μM; Sigma-Aldrich) for 4 hours in the presence of protein-transport inhibitor, monensim (2 μM; Beckton Dickinson). The cells were centrifuged at 350 g for 10 minutes, washed twice with PBS, counted, and resuspended at 2 × 10⁵ cells/ml in PBS. After fixation and permeabilization (BD Cytofix^M buffer; BD Perm/Wash^M buffer; Becton Dickinson), the cells were washed and stained with BD Pharmingen antibodies against murine IL-17 (clone TC11-18H10; PE-labeled; 0.125 μg/1 × 10⁶ cells); IL-4 (clone 11B11; APC-labeled; 0.25 5 μg/1 × 10⁶ cells), and IFN-γ (clone XMG 1.2; FITC labeled; 0.5 μg/1 × 10⁶ cells), or of isotype controls (APC- or FITC-labeled R3-34 or unlabeled TC11-18H10 antibodies). The samples were incubated for 30 minutes at room temperature in the dark, washed, and used for flow-cytometry analysis.

Bone marrow cells were resuspended at 2 × 10⁶/ml in 10% FCS/MEM medium (Lonza). BM cells were incubated for 1 day with medium containing macrophage colony-stimulating factor (M-CSF; 30 ng/ml; PeproTech). Osteoclast precursors were generated in cultures with M-CSF (30 ng/ml) and RANKL (50 ng/ml; PeproTech EC). After 3 days, fresh medium supplemented with growth factors (M-CSF and RANKL) was added to the cultures. In some experiments, blocking anti-RANKL antibody (5 μg/ml; PeproTech EC) was added at this time. The cells were allowed to differentiate for 3 days and the specific tartrate-resistant acid phosphatase (TRAP) staining was performed, as described 31. The number of TRAP-positive cells was determined by light microscopy (Leica Microsystems, Wetzlar, Germany) by two independent observers and additionally, by software analyses (ImageJ 1.42; Research Services Branch, NIH, Bethesda, MD, USA) after photo capturing by a DS-Ri1 Nikon camera (Nikon Instruments Europe, Amstelveen, The Netherlands).

Paws were fixed in 10% paraformaldehyde/PBS (pH 7.4), decalcified for 4 days in 5% nitric acid (Sigma-Aldrich), dehydrated in ethanol series and xylene substitute (Tissue-Clear, Sakura Finetek, Tokyo, Japan) and embedded in paraffin. The sections with thickness 6 μm were cut by rotary microtome (Accu-Cut SRM Sacura Finetek). Safranin O and hematoxylin and eosin (H&E) staining was performed. The sections were examined with a light microscope by using 1 × 100 or 1 × 400 lens. Images were captured with a coupled device camera and exported to Adobe Photoshop 7.0 (Adobe Systems, Munich, Germany).

All histologic assessments were performed in a blinded protocol. The degree of injury was graded for infiltration (score 0, normal; score 1, mild infiltration; score 2, moderate infiltration; score 3, marked infiltration; score 4, severe infiltration). Proteoglycan loss was graded as follows: score 0, normal; score 1, minimal loss; score 2, moderate loss; score 3, marked loss; score 4, severe, diffuse loss). Bone erosion was graded as score 0, no abnormality; score 1, small areas of resorption; score 2, more numerous areas of resorption, not readily apparent on low magnification; 3, obvious resorption in trabecular and cortical bone, and lesions apparent on low magnification; score 4, thickness defects in the cortical bone and trabecular bone loss.

Immunohistochemistry was used to evaluate the expression of STAT1, STAT3, transforming growth factor (TGF)-β3, RANKL, and C5aR in the joints. After blocking of endogenous peroxidase and unspecific binding, the sections were incubated for 1 hour with antibodies against C5aR (1:50 diluted; BD Biosciences), STAT3 (1:100 diluted, Santa-Cruz Biotechnology, Heidelberg, Germany), STAT1 (1:500 diluted, Santa-Cruz Biotechnology), TGF-β3 (1:50 diluted; Abcam, Cambridge, UK), and RANKL (1:50 diluted; PeproTech EC). Isotype antibodies (with rat or rabbit origin, Abcam) were used as specific controls in the experiments. The sections were washed, and HRP/DAB detection kit (Abcam) was used to detect specific staining.

Statistical analysis was accomplished by using InStat3.0 and GraphicPad Prism 5.0 (GraphPad Software, La Jolla, CA, USA). Data were expressed as mean ± standard deviation (SD). The histologic score and the immunohistochemistry data were analyzed with the Mann-Whitney U test. For other data, the differences in the mean values between groups were analyzed with the two-tailed Student t test. Differences were considered significant when P < 0.05.

Properdin-deficient mice developed less severe arthritis. We observed significantly attenuated clinical symptoms of the disease in KO mice compared with WT mice after day 7 (Figure 1A). At the end of the experiment (day 10), the clinical score of KO mice decreased by 52.5% in comparison with WT mice (clinical score, KO CAIA group, 4.2 ± 0.8; WT CAIA group, 8.0 ± 1.2; P < 0.001, Mann-Whitney U test). Histologic examination showed a massive presence of inflammatory cells in the synovial tissue and cartilage of WT mice, which was markedly reduced in KO mice (score for cell infiltration: KO CAIA group, 2.20 ± 0.40; WT CAIA group, 3.00 ± 0.35; n = 5; P < 0.05). Safranin O staining of joint sections showed weak proteoglycan cartilage loss in WT and KO mice with CAIA (score for proteoglycan loss: KO CAIA group, 0.40 ± 0.20; WT CAIA group, 0.60 ± 0.25).

At day 10 of CAIA, we detected higher numbers of SF cells in WT than in KO mice (WT CAIA group, 0.70 × 10⁵ cells/ml ± 0.02; KO CAIA group, 0.10 × 10⁵ ± 0.01 cells/ml; n = 5; P < 0.05). CAIA mice showed increased numbers of Ly6G⁺CD11b⁺ (Figure 1B) and of Ly6G⁺C5aR⁺ (Figure 1C) neutrophils compared with PBS-injected control mice. The lack of properdin significantly decreased the frequencies of Ly6G⁺C5aR⁺ cells (Figure 1C) and C5a levels in SF (Figure 1D). The expression of RANKL was higher on WT CAIA neutrophils in comparison with the KO group (Figure 1E). These data suggest that WT synovial Ly6G⁺ cells can participate more actively in RANKL-dependent processes of osteoclast activation and differentiation. IL-17 enhances inflammation and osteoclastogenesis. Higher amounts of synovial IL-17 were observed in arthritic groups compared with PBS-injected ones (Figure 1F). IL-17 levels in KO CAIA SF were significantly lower than those in WT CAIA mice (Figure 1F). CAIA developed with slightly (nonsignificantly) elevated amounts of plasma IL-17 in WT compared with KO mice (KO CAIA group, 358.8 pg/ml ± 10.2; WT CAIA group, 401.5 pg/ml ± 22.6; n = 5; P = 0.068).

At day 10 of CAIA, we observed increased frequencies of Ly6G^lowCD11b⁺ and Ly6G^highCD11b⁺ cells in BM (Figure 2A) and of Ly6G⁺CD11b⁺ in blood of WT compared with KO CAIA mice (WT PBS group, 5.4% ± 0.9; KO PBS group, 4.2% ± 1.0; WT CAIA group, 35.8% ± 1.2; KO CAIA group, 14.0% ± 5.0; n = 4; P < 0.05). BM and blood Ly6G⁺CD11b⁺ cells were stained with antibodies against RANKL (Figure 2B, D) and C5aR (Figure 2C, E). Surface RANKL was detected on BM WT but not on BM KO CAIA cells (Figure 2B). The molecule was barely expressed on mouse blood KO cells, and we found fewer Ly6G⁺RANKL⁺ cells in KO CAIA mice (Figure 2D). BM cells from WT and KO mice expressed C5aR (Figure 2C) to a similar extent. Interestingly, we found higher C5aR expression and more C5aR-positive cells within the Ly6G⁺ blood population of KO than in that of WT mice (Figure 2E).

Regarding C5L2, another receptor for C5a, we were not able to detect its expression on mouse blood Ly6G⁺ in contrast to human peripheral CD16⁺ cells (Figure 2F). As CAIA is a FcγR-dependent experimental model, we analyzed the expression of FcγR on synovial and blood neutrophils and on monocytes with flow cytometry (see Additional file 1). Additional file 1 shows higher FcγR expression and increased frequencies of Ly6G⁺ FcγR⁺ cells in WT CAIA mice in comparison to the KO CAIA group (Additional file 1A). Circulating CD14⁺ KO CAIA monocytes expressed more FcγR than did those in WT CAIA mice (Additional file 1B). Synovial Ly6G⁺CD11b⁺ cells with elevated FcγR expression were found in WT and KO CAIA mice when compared with PBS-injected control groups (Additional file 1C).

In our study, Ly6G^highCD11b⁺ cells were between 1% and 2%, whereas Ly6G^highCD11b⁺ cells were 5% to 6% in both KO and WT spleen populations (Figure 3A). At day 10 of CAIA, the frequencies of Ly6G^highCD11b⁺ cells increased two times, and fewer Ly6G^lowCD11b⁺ cells were detected in the spleen (Figure 3A). Interestingly, Ly6G^highCD11b⁺ cells upregulated CD86 expression only in the WT CAIA group but not in KO CAIA mice (Figure 3B). CD86 is an important costimulatory molecule for T cells. Flow-cytometry analyses of the splenocyte population demonstrated an increased percentage of CD4⁺ T cells in WT CAIA mice compared with the KO CAIA group at day 10 of disease (Figure 3C). Activation marker CD69 was barely expressed on the arthritic WT and KO CD4⁺ T splenic population (Figure 3D). WT CAIA cells spontaneously release higher amounts of IFN-γ (Figure 3E), IL-6 (Figure 3F), and IL-17 (Figure 3G), whereas KO CAIA cells secrete IFN-γ but not IL-17 and IL-6 (Figure 3F, G). Both WT and KO cells responded to anti-CD3/anti-CD28 stimulation with markedly increased cytokine production. However, the amounts of secreted cytokines by WT CAIA cells were higher than those of the KO CAIA ones. Of note, ConA induced vigorous IL-17 and IL-6 production by WT CAIA but not by KO CAIA cells (Figure 3F, G).

The initial inflammatory process involving complement and blood neutrophils and monocytes results in the activation of peripheral CD4⁺ T cells. Blood CD4⁺ T cells expressing the early activation marker CD69 were fewer in KO CAIA mice in comparison with the WT group (Figure 4A). The lack of properdin resulted in markedly inhibited RANKL expression on CD4⁺ T cells and reduced numbers of CD4⁺ T RANKL⁺ cells in the periphery (Figure 4B).

In RA patients, CD4⁺ T cells in the periphery are prone to produce proinflammatory cytokines. Thus, we determined the intracellular level of IFN-γ and IL-17 in blood CD4⁺ T cells (Figure 4C). Nonstimulated CD4⁺ T cells from control KO and WT mice showed a low production of IFN-γ and IL-17 that was upregulated by Con A stimulation (Figure 4C). However, no significant difference in the frequencies of single IFN-γ- and IL-17-producing cells between Con A-stimulated control KO and WT cells was observed. Nonstimulated control CD4⁺ T lymphocytes showed weaker IFN-γ and IL-17 production than did nonstimulated arthritic cells (Figure 4C). CAIA WT CD4⁺ T cells were more sensitive to Con A stimulation than were those from KO mice, because they had significantly higher intracellular levels of IFN-γ and IL-17 (as shown on Figure 4C and in one representative experiment in Figure 4D).

Concerning Th2 cytokine IL-4, its intracellular level was low in nonstimulated and ConA-stimulated CD4⁺ T KO and WT mice (Figure 4D).

Our data showed that RANKL is expressed to a lesser extent by populations in BM (Ly6G⁺ cells) and in the blood of KO mice (neutrophils and CD4⁺ T cells), suggesting altered RANKL-dependent processes in conditions of properdin deficiency. Thus, we next evaluated the RANKL-mediated osteoclastogenesis of BM cells isolated from control and KO and WT CAIA mice. The specific TRAP staining showed similar numbers of generated osteoclasts in control KO and WT cultures (Figure 5A). Osteoclast differentiation of properdin-deficient BM CAIA cells was inhibited, and fewer TRAP-positive cells were detected in these cultures. We suggest that BM precursors from properdin-deficient CAIA KO mice were less sensitive to the action of RANKL.

We set up an experiment in which WT and KO CAIA BM cells were differentiated in the presence or the absence of a blocking antibody against RANKL (Figure 5B). We observed a reduced number of TRAP-positive cells in WT CAIA cell cultures, similar to that in the KO CAIA group. Osteoclast differentiation of properdin-deficient CAIA BM cells was not changed by blocking RANKL (Figure 5B).

At day 10 of CAIA, the joint sections were analyzed for the expression of C5aR and several markers of bone destruction (Figure 6). The molecule associated with bone resorption RANKL was expressed approximately threefold higher in the cartilage of WT arthritic mice than of KO CAIA mice (Figure 6A). C5aR-positive cells were observed in the infiltration area and cartilage of CAIA groups, with no significant difference between WT and KO animals (Figure 6B). STAT1-positive staining was strong in the WT infiltration areas and cartilage and less obvious in KO joints (Figure 6C). STAT3 expression was similar in the infiltration zones of WT and KO mice, whereas in cartilage, its staining intensity was inhibited in the CAIA KO group (Figure 6D). CAIA development in KO mice was accompanied with suppressed TGF-β expression in the joints (Figure 6E).