Deconstructing the Lectin Pathway in the Pathogenesis of Experimental Inflammatory Arthritis: Essential Role of the Lectin Ficolin B and Mannose-Binding Protein–Associated Serine Protease 2 Free

Rheumatoid arthritis (RA) is an autoimmune disease found in ∼1% of the adult population (1). The prevalence of RA-related disability and the number of RA patients in the United States has been projected to increase by 40% over the next 25 y (2), suggesting that this disease will continue to adversely impact patients and the health care system (3). Although the etiology of RA remains unknown, many potential risk factors for the development of RA have been identified (4). Among other factors, it is now appreciated that the complement system plays an important role in the pathogenesis of RA. Complement activation is essential for disease progression in passive transfer mouse models of RA, and complement fragments derived from the activation process have been found in the synovium of RA patients (5–7). Furthermore, complement-directed therapeutics have shown excellent efficacy in mouse models of RA, such as collagen Ab–induced arthritis (CAIA) (8, 9). The complement system is activated by three interlinked pathways: the classical pathway (CP), the lectin pathway (LP), and the alternative pathway (AP). All of these pathways generate two potent proinflammatory molecules, C3 anaphylatoxin (C3a) and C5a, via C3 and C5 convertases, respectively, as well as the membrane attack complex, each of which plays an important role in CAIA (10).

Our previous studies in the CAIA model have revealed an essential role for the AP in the development of joint damage, with initial studies suggesting that the AP can function fully independently of the CP and LP in this model (6). In contrast, an essential role for the CP has not been observed, because mice deficient in C4 and C1q are fully susceptible to disease (11). The LP has been proposed to be involved in the development of CAIA, but only through the requirement for mannose-binding lectin (MBL)-associated serine protease (MASP)-1 and/or MASP-3 in cleaving the inactive AP protease pro–factor D (pro-fD) into active fD, thus allowing the AP to function in vivo (12–14). An independent role for the LP recognition and activation mechanisms has not been identified yet in this model but is evaluated in this article.

The LP is activated when specific pattern recognition molecules engage carbohydrate and other ligands either expressed on pathogens or revealed in injured tissue (15, 16). The first pattern recognition molecule described in the LP was MBL (17). In addition to MBL, though, ficolins (FCNs) and collectins (CLs) can also recognize unique ligands and activate the LP. With regard to the mechanism of activation of complement, the function of CLs, FCNs, and MBLs are similar, as are all soluble pattern recognition molecules that interact with MBL-associated serine proteases (MASPs) 1, 2, and 3 (MASP-1, MASP-2, and MASP-3, respectively) and activate these proteins after ligand recognition (18–20).

MBL is a C-type lectin containing a carbohydrate recognition domain (CRD) as well as collagen-like domain (21). MBL binds to mannose-containing molecules as well as N-acetylglucosamine (21). FCNs, which also contain a CRD, consist of collagen-like and fibrinogen-like domains and preferentially bind to N-acetylglucosamine (18, 22, 23). There are two mouse FCNs, FCN A (24) and FCN B (25); in contrast, humans express three FCNs: FCN M (also known as FCN 1), FCN L (also known as FCN 2), and FCN H (also known as FCN 3) (26–30). Mouse FCN A, but not FCN B, exhibit a splice variant known as FCN A variant (31). FCN A is present in the serum and expressed in liver hepatocytes (32). Mouse FCN B was originally found in the lysosomes of macrophages, similar to human FCN M, which is also found in the secretary granules of monocytes and neutrophils (33, 34). We have reported that FCN-B is also present in the circulation of mice, suggesting that it is secreted from macrophages (35).

The third class, designated CLs, are similarly C-type lectins containing CRDs. Three different human CLs have been identified: CL-10 (also known as CL liver 1 or CL-L1), CL-11 (also known as CL kidney 1 or CL-K1), and CL-12 (also known as CL placenta 1 or CL-P1) (36–39). CL-11 is present in the serum, liver, adrenal gland, and kidney and binds l-fucose and d-mannose (36, 40). Recently, it has been shown that CL-11 plays an essential role in ischemia/reperfusion (I/R) injury in the kidney (41). The authors demonstrated that CL-11 bound to cell surface l-fucose and required MASP-2 for subsequent complement activation. Although MBL-A/C–deficient mice are fully susceptive to CAIA, there are no reports to date regarding the roles of FCNs and CLs in inflammatory arthritis and how these LP pattern recognition molecules might interact with MASPs to cause pathogenic complement activation in the joint.

MASPs are the enzymes of the LP that cleave C2 and C4 to create the C3 convertase C4b2a. They are predominately generated in the liver and consist of five distinct proteins: MASP-1, MASP-2, sMAp, MASP-3, and MAp44. The MASP-1 gene generates three of these proteins: MASP-1, MASP-3, and MAp44 (also known as MAP-1), a truncation of MASP-1/3 containing only the pattern recognition domain (42, 43). MASP-1 triggers the LP and promotes the activation of MASP-2 (13). MASP-3 has been shown to cleave inactive pro-fD into active fD (44). Both MASP-1/3^−/− and wild type (WT) mice treated with a mixture of MASP-1/3 small interfering RNAs are protected from the development of CAIA (12, 14). The MASP-2 gene generates MASP-2 and sMAp (also known as MAP-2 or MAp19), a truncation of MASP-2 (45). Unlike MASP-1 and MASP-2, MASP-3 is not autoactivated when complexed with FCNs or CLs (20, 40, 46). MASP-2 cleaves C4 and C2 to generate the C3 convertase of the LP. MASP-2 is the primary effector protein of the LP, because both mice and humans lacking MASP-2 are defective in this pathway (47, 48). MASP-2 has been shown to play an important role in focal cerebral ischemia because MASP-2^−/− mice (in contrast with MASP-1/3^−/− mice) were protected from cerebral ischemia (49). Decreased I/R injury of the myocardium, intestine, and kidney has been observed both in mice lacking MASP-2 and mice treated with MASP-2 inhibitory Abs (47). The role of MASP-2/sMAp in experimental arthritis, however, remains to be determined.

We have shown that the AP is necessary for the development of CAIA (6). Given that MASP-1/3^−/− mice are resistant to CAIA (12) and that MASPs are activators of the LP, it follows that there must exist a regulatory relationship between the LP and the AP. We and others have shown that MASP-1/3 is an essential activator of fD through the cleavage of pro-fD (13, 14). Another relationship can also be observed in the ability of C2- and C4-deficient patients to activate complement in response to infection. The observation led to the discovery of a C2-independent means of activating either the CP or the LP, termed the C4/C2 bypass pathway (50, 51). Although the mechanism of the bypass pathway remains to be elucidated, it has been shown that in the presence of the AP, MBL may activate complement in a C2-independent fashion (52). Thus, the pathways of complement activation are not necessarily independent, but rather may interact and regulate each other.

In this study, to our knowledge, we have made two new observations. First, in contrast with other LP ligands, the presence of FCN B is necessary for the full development of CAIA, suggesting that this lectin plays a unique role in recognition of ligand(s) within the inflamed joint. Second, MASP-2 plays an essential role in the development of CAIA, likely through interactions with FCN-B. In addition, the lack of a role for C4 in development of CAIA in conjunction with the residual inflammation in MASP-2–deficient mice suggests the possibility that the AP activation mechanisms are fully capable of mediating disease, or that MASP-1 and MASP-3 may contribute to the development of disease through the C4/C2 bypass pathway.

All genetically deficient mice, that is, FCN A^−/−, FCN B^−/−, CL-11^−/−, and MASP2^−/−/sMAp^−/−, were bred onto the C57BL/6 background. All gene-deficient mice used for CAIA studies were age- and sex-matched. In parallel, WT littermates were used for CAIA studies with each cohort. All mice were genotyped before use. Sera from WT, FCN B^−/−, and MASP2^−/−/sMAp^−/− C57BL/6 mice before and after disease induction were also used for various ELISA studies. WT C57BL/6, C3^−/−, and NOD mice, purchased from Jackson Laboratories, were also used as controls in various complement-related studies. All mice were kept in a barrier facility and fed breeder’s chow provided by the Center for Laboratory Animal Care, University of Colorado Denver.

CAIA was induced in WT, FCN A^−/−, FCN B^−/−, CL-11^−/−, and MASP2^−/−/sMAp^−/− C57BL/6 mice by using a mixture of five mAbs to bovine collagen type II (Arthrogen-CIA; Chondrex) resuspended in sterile Dulbecco’s PBS according to our previously published methods (14, 53). All WT, FCN A^−/−, FCN B^−/−, CL-11^−/−, and MASP2^−/−/sMAp^−/− mice received i.p. injections of 8 mg of Arthrogen per mouse on day 0. The mice then received 50 μg of LPS from Escherichia coli strain 0111B4 on day 3. All mice, after the induction of CAIA, were sacrificed at day 10. The severity of clinical disease activity (CDA) in all groups of WT and gene-deficient mice was determined every day by two trained laboratory personnel acting independently, and in a blinded fashion, according to our previously published methods (14, 53).

At day 10, both forepaws and the entire right hind limb, including the paw, ankle, and knee, were surgically removed from WT, MASP-2^−/−/sMAp^−/−, FCN A^−/−, FCN B^−/−, and CL-11^−/− mice with CAIA and fixed immediately in 10% buffered formalin (Biochemical Sciences). The preparation of all joint samples and their histological analyses were performed as previously described (6, 14). All sections were read by a trained observer, who was also blinded to the genotype and to the CDA of each mouse. The joint sections for WT, MASP-2^−/−/sMAp^−/−, FCN A^−/−, FCN B^−/−, and CL-11^−/− were scored for the changes in inflammation, pannus, cartilage damage, and bone damage, on a scale of 0 to 5. C3 deposition in the joints (synovium and cartilage) was localized with polyclonal goat anti-mouse C3 antisera (ICN Pharmaceuticals) and scored as previously described (14, 54). Immunohistochemical (IHC) staining for macrophages and neutrophils was performed using knee joint sections from WT, FCN B^−/−, and MASP-2^−/−/sMAp^−/− mice and scored as previously described (14, 54).

At day 10, ankles (right and left) from the WT (littermates) and FCN B^−/− and WT (littermates) and MASP-2^−/−/sMAp^−/− mice with CAIA, from the same cohorts, were also processed for the IHC staining analysis for the C4d, fD, and FCN B deposition. Knee joints were not used because of the limitation in the thickness of the knee joint and also their prior use for the earlier-mentioned immunopathology studies. An Ag retrieval method, by pressure, was used for C4d detection using 10 mM citrate/1× TBS buffer at 95°C for 20 min, for fD detection using 100 mM citrate/1× TBS buffer at 110°C for 10 min, and for FCN B detection using 100 mM citrate/1× TBS buffer at 110°C for 10 min. Immunodetection was performed on the Benchmark stainer (Ventana Medical Systems/Roche Diagnostics, Indianapolis, IN) at an operating temperature of 37°C, with a primary Ab incubation time of 32 min. Abs were detected with a modified I-VIEW DAB (Ventana) detection kit. The I-VIEW secondary Ab and enzyme were replaced with a species-specific rabbit conjugated polymer (full strength in place of the secondary Ab, 50% strength diluted in 1× PBS [pH 7.6] in place of the SA-HRP, 8 min each; Rabbit ImmPress; Vector Labs, Carpinteria, CA). All sections were counterstained in acidified Harris hematoxylin for 1.5 min, blued in 1% ammonium hydroxide (v/v), dehydrated in graded alcohols, cleared in xylene, and cover glass mounted using synthetic resin.

To detect membrane-bound C4d deposition in the synovium, we used cartilage and adipose tissue areas of the ankle, human anti-C4d rabbit polyclonal Ab (American Research Products, Waltham, MA) at 1:200 dilution. Normal mouse liver and sections from rejected human kidneys, after transplantation, were used as controls for C4d deposition assessment. In human kidney, after transplantation rejection, there is substantial C4d deposition, and thus the tissue serves as an appropriate positive control. Normal human kidney sections were also used as negative control. No well-characterized mouse anti-C4d Ab was available to us at the time for IHC analysis of paraffin-embedded tissues; therefore, we used a human anti-C4d Ab that cross-reacted with mouse C4d as shown by positive control staining in the mouse liver. Overall, we stained sections with two different mouse anti-C4d Abs, and both of these yielded no detection for mouse C4d using paraffin-embedded ankle sections.

To detect fD deposition on the membrane or in the cytoplasm, in the synovium areas of the ankle, we used rabbit anti-mouse fD polyclonal Ab (GeneTex, Irvine, CA) at 1:25 dilution. Mouse adipose tissue expressing high levels of fD were used as a positive control. Liver expressing no fD was used as a negative control.

To detect FCN B deposition on the membrane or in the cytoplasm, in the synovium areas of the ankle, we used mouse anti-FCN B (FCN2) rabbit polyclonal Ab (MyBioSource, San Diego, CA) at 1:750 dilution. Mouse skin and brain sections that expressed FCN B were used as the positive controls. All Abs for C4d, fD, and FCN B were visualized by using a universal rabbit modified one-view detection kit (Abcam, Cambridge, MA) as mentioned earlier. All stained sections were scored blindly by a trained histology technician using an arbitrary ordinal scale of negative, no staining of cells, low level = few cells (<15 positive cells), medium = some cell (from 15 to 30), and very high level = many cells (more than 30) in the synovium, cartilage, and adipose tissue of ankle.

Serum samples from mice with CAIA were analyzed for C3 deposition and C5a levels induced by adherent anti-collagen Abs. The 96-well costar ELISA plates were precoated with a mixture of anti-collagen Abs (Arthrogen, 25 μg/well) as previously described (14, 54). Sera from mice before and after CAIA were collected by using a standard recommended method as previously described (55). All serum samples from WT, FCN B^−/−, and MASP-2^−/−/sMAp^−/− mice, before and after disease, were diluted 1:10 in a calcium (Ca²⁺)-sufficient (Gelatin Vernal Buffer [GVB⁺⁺]) buffer or in a calcium (Ca²⁺)-deficient (Mg²⁺ EGTA) buffer such that all pathways will be active or only the AP will be active under these conditions. Diluted sera were then added to the wells and incubated at 37°C for 1 h. After washes in PBS/0.5% Tween, HRP-conjugated goat anti-mouse C3 Ab (MP Biomedicals) was added to the wells. The anti-C3 Ab was diluted 1:4000 in freshly made PBS/0.05% Tween to reduce background. Plates were incubated at room temperature for 1 h. After five more washes, the color reaction was performed and absorbance was determined for C3 deposition adherent to the ELISA plate. C5a generation was examined in the supernatant according to our previously published methods (53). Sera from C3^−/− and NOD mice were used as negative controls for C3 deposition and C5a measurements, respectively. Data were expressed as mean OD ± SEM.

Data from all WT and gene-deficient mice were included in the final analysis of CDA, histology, ELISA, and IHC. ANOVA was used when we compared three cohorts; otherwise, Student t test was used to analyze data. All data were expressed as the mean ± SEM, with p < 0.05 considered significant.

CAIA was induced in WT, FCN A^−/−, FCN B^−/−, CL-11^−/−, and MASP-2^−/−/sMAp^−/− mice, and CDA was examined. This CAIA study was performed in four different cohorts comparing WT litter mates with gene-targeted mice as shown (Fig. 1). In cohort 1, at day 10, there was no significant difference (p < 0.115) in the CDA between WT and FCN A^−/− mice. The CDA was 11.75 ± 0.250 and 10.2 ± 0.734, respectively (Fig. 1A). The prevalence rate of disease was 100%, at day 10, in WT and FCN A^−/− mice. In cohort 2, the CDA in WT and FCN B^−/− mice was compared. Interestingly, the CDA of FCN B^−/− mice was significantly decreased as compared with WT (p < 0.031 on day 10) (Fig. 1B). On day 10, the CDA was 10.75 ± 0.977 and 5.75 ± 1.89 in WT and FCN B^−/−mice, respectively (Fig. 1B). Overall, there was a 47% decrease in the CDA in FCN B^−/− mice compared with WT mice (Fig. 1B). The prevalence rate of disease was 100%, at day 10, in WT and FCN B^−/− mice. In cohort 3, there was no significant difference (p < 0.63) in the CDA between WT and CL-11^−/− mice (Fig. 1C). The CDA was 9.53 ± 0.882 and 10.2 ± 1.04, respectively. The prevalence rate of disease was 100%, at day 10, in WT and CL-11^−/− mice. In cohort 4, on day 10, there was a significant decrease (p < 0.05) in the CDA between WT and MASP-2^−/−/sMAp^−/− mice. The CDA was 10.8 ± 0.734 and 3.2 ± 0.583, respectively (Fig. 1D). The decrease in the CDA in MASP-2^−/−/sMAp^−/− compared with the WT mice was ∼70%. Notably, significant differences (p < 0.05) in the CDA were seen with MASP-2^−/−/sMAp^−/− mice compared with the WT mice from days 4 to 10 (Fig. 1D). The prevalence rate of disease, at day 10, in WT and MASP-2^−/− mice was 100%. These CDA data show that, although all mice tested developed disease, FCN B^−/− and MASP-2^−/−/sMAp^−/− mice were significantly protected.

All four cohorts of mice described earlier were sacrificed on day 10. Both forelimbs and the right hind limb (five joints) were processed for histopathology studies and for the measurement of C3 deposition in the joints (Fig. 2). Data are shown as an all-joint mean (AJM) score. No significant differences were found in inflammation, pannus formation, cartilage damage, bone damage, or C3 deposition in the joints of FCN-A^−/− and CL-11^−/− compared with the WT mice (cohorts 1 and 3; data not shown). These data are consistent with the CDA data shown in Fig. 1. Histopathology scores for cohort 2 show a significant decrease (p < 0.04) of 39% in the AJM score in FCN B^−/− mice compared with the WT mice (Fig. 2A). There was a significant decrease in FCN B^−/− mice compared with WT mice in inflammation (34%, p < 0.04), pannus formation (40%, p < 0.042), cartilage damage (41%, p < 0.04), and bone damage (40%, p < 0.045) (Fig. 2A). Representative pictures of T blue staining related to histopathology from the knee joints of WT and FCN B^−/− mice are shown in Supplemental Fig. 1A and 1B. Histopathology AJM scores for MASP-2^−/−/sMAp^−/− mice (cohort 4) showed a 68% reduction (p < 0.001) (Fig. 2E). Not only the AJM but also individual scores for inflammation, pannus formation, cartilage damage, and bone damage were significantly (p < 0.05) decreased in MASP-2^−/−/sMAp^−/− mice compared with WT mice (Fig. 2E). Representative pictures of T blue staining related to histopathology from the knee joints of WT and MASP-2^−/−/sMAp^−/− mice are shown in Supplemental Fig. 2A and 2B.

C3 deposition was measured as described in 2Materials and Methods. Consistent with CDA and histology measurements, FCN-A^−/− and CL-11^−/− mice showed no significant differences as compared with WT mice (data not shown). C3 deposition was significantly (p < 0.033) reduced by 56% in all joints (synovium and cartilage surface) of FCN B^−/− comparing with the WT mice (Fig. 2B). Individual scores for C3 deposition in the synovium (p < 0.03) and also on the surface of cartilage (p < 0.03) were also significantly reduced in FCN B^−/− mice compared with the WT mice (Fig. 2B). Representative pictures of C3 deposition performed immunohistochemically from the knee joints of WT and FCN B^−/− mice are shown in Supplemental Fig. 1C and 1D. C3 deposition was significantly (p < 0.05) reduced to 81% in all joints (synovium and cartilage surface) of MASP-2^−/−/sMAp^−/− compared with WT mice (Fig. 2F). Individual scores for C3 deposition in the synovium and also on the surface of cartilage were also significantly (p < 0.001) reduced in MASP-2^−/−/sMAp^−/− mice compared with WT mice (Fig. 2F). Representative pictures of C3 deposition performed immunohistochemically from the knee joints of WT and MASP-2^−/−/sMAp^−/− mice are shown in Supplemental Fig. 2C and 2D.

All four cohorts of mice from the CAIA studies were examined via IHC for the infiltration of macrophages and neutrophils using specific cell surface markers according to our previously published studies (9). Again, no significant differences were seen regarding numbers of macrophages and neutrophils in knees from WT and FCN A^−/− or from WT and CL-11^−/− mice (data not shown). In contrast, the percentage of synovial macrophages and neutrophils decreased 42% (p < 0.03) and 43% (p < 0.04), respectively, in FCN B^−/− mice compared with the WT mice with CAIA (Fig. 2C, 2D). Representative IHC pictures of macrophage and neutrophil infiltrates from the knee joints of WT and FCN B^−/− mice are shown in Supplemental Fig. 1E–H. In MASP-2^−/−/sMAp^−/− mice, synovial macrophages and neutrophils were decreased 74% (p < 0.002) and 71% (p < 0.002), respectively, as compared with WT mice with CAIA (Fig. 2G, 2H). Representative IHC pictures of macrophages and neutrophils from the knee joints of WT and MASP-2^−/−/sMAp^−/− mice are shown in Supplemental Fig. 2E–H. Overall, the decrease in infiltration of synovial macrophages and neutrophils in the knee joints was consistent with decrease in CDA observed in FCN B^−/− and MASP-2^−/−/sMAp^−/− mice.

At day 10, ankles from WT and FCN B^−/− and WT and MASP-2^−/−/sMAp^−/− mice with CAIA were processed and tissue sections were subjected to IHC staining for complement C4d deposition, fD deposition, and FCN B deposition (Figs. 3, 4). All staining results have been summarized in Table I. Representative IHC images for C4d, fD, and FCN B deposition in the synovium from the ankles of WT compared with FCN B^−/− mice with CAIA are shown in Fig. 3. First, there were low levels of C4d deposition in the synovium of both FCN B^−/− mice and WT mice with CAIA (Fig. 3A, 3B). Mouse liver was used as a positive control for the presence of C4d (Fig. 3C) along with human kidney negative and positive controls (Fig. 3J, 3K). Second, there were very low levels of fD deposition in the synovium in FCN B^−/− mice compared with the very high levels of deposition in WT mice with CAIA (Fig. 3D, 3E). Mouse adipose tissue (fat) was used as a positive control for the presence of fD (Fig. 3F). Finally, there was no detection of FCN B deposition in the synovium in FCN B^−/− mice as expected, compared with the very high levels of deposition in WT mice with CAIA (Fig. 3G, 3H). Mouse skin was used as a positive control for the presence of FCN B (Fig. 3I). These IHC staining data, for C4d, fD, and FCN B deposition, in the synovium from the ankles of WT and FCN B^−/− mice with CAIA suggest that the FCN B ligand recognition molecule of the LP is directly involved in activating the AP through MASP-1 or MASP-3 proteases, but is possibly independent of C4 because of low levels of C4d in the ankle.

All representative IHC images for C4d, fD, and FCN B deposition in the ankles from WT compared with MASP-2^−/−/sMAp^−/− mice with CAIA have been shown in Fig. 4. There was no detection of C4d deposition in the synovium in MASP-2^−/−/sMAp^−/− mice compared with the low levels of C4d deposition in WT mice with CAIA (Fig. 4A–C) (Table I). We found that fD was present at very high levels in ankles of WT mice with disease compared with the medium levels of fD in MASP-2^−/−/sMAp^−/− mice with disease (Fig 4D–F). Again, FCN B was present, equally at a high level, both in WT mice and in MASP-2^−/−/sMAp^−/− mice with CAIA (Fig. 4G, 4H). Mouse brain, which highly expressed FCN B, was used as another positive control (Fig. 4I). Interestingly, these C4d, fD, and FCN B deposition IHC data in MASP-2^−/−/sMAp^−/− mice versus WT mice with disease showed that MASP-2 appears to contribute to the development of CAIA, likely in a C4-independent manner because of the nondetectable levels of C4d deposition in the synovium from ankles of MASP-2^−/−/sMAp^−/− mice with CAIA. Nonetheless there was activation of the LP and AP because of the deposition of FCN B and fD in the synovium from ankles of MASP-2^−/−/sMAp^−/− mice with CAIA.

ELISAs were performed to find out the relative roles of LP and AP in the C3 activation and generation of C5a in the sera from WT, FCN B^−/−, and MASP-2^−/−/sMAp^−/− mice with CAIA (Figs. 5, 6). Sera from only WT, FCN B^−/−, and MASP-2^−/−/sMAp^−/− mice, at day 10, with CAIA were analyzed because these mice were partially resistant to disease (Fig. 1). Sera from these mice with disease were diluted in a Ca²⁺-sufficient and a Ca²⁺-deficient buffer separately as mentioned in the 2Materials and Methods. C5a generation was examined in parallel in a Ca²⁺-sufficient buffer or in a Ca²⁺-deficient buffer from the supernatants of C3 deposition (Fig. 5). No statistically significant change in C3 activation was seen comparing sera from WT and FCN B^−/− mice diluted in a Ca²⁺-sufficient buffer (Fig. 5A). However, a highly significant (p < 0.003) decrease of 80% was seen in C3 deposition, in a Ca²⁺-deficient buffer in the sera from FCN B^−/− mice compared with the WT mice with CAIA (Fig. 5B). There was a minimal but significant decrease (p < 0.002) of 16% in the levels of C5a, comparing sera diluted in a Ca²⁺-sufficient buffer from WT and FCN B^−/− mice with disease (Fig. 5C). Again, there was a highly significant (p < 0.007) decrease of 60% in the levels of C5a seen in the sera of a Ca²⁺-deficient buffer from FCN B^−/− mice compared with the WT mice with disease (Fig. 5D). Interestingly, although sera from NOD mice were used as a negative control for C5a ELISA, there was a significant (p < 0.05) decrease in the C3 activation compared with the WT mice in a Ca²⁺-deficient buffer (Fig. 5B). These data regarding C3 activation and C5a levels strongly suggest a direct effect on the levels of potential AP activation because of the deficiency of the LP FCN B ligand.

There was also a significant decrease (p < 0.002) in C3 activation, comparing sera diluted in a Ca²⁺-sufficient buffer from WT and MASP-2^−/−/sMAp^−/− mice with disease, but this decrease was a minimal of 6% (Fig. 6A). However, a highly significant (p < 0.002) decrease of 82% in the C3 activation was seen, in a Ca²⁺ deficient buffer, in the sera from MASP-2^−/−/sMAp^−/− mice compared with the WT mice with disease (Fig. 6B). There was a significant decrease (p < 0.002) of 33% in the levels of C5a, comparing sera diluted in a Ca²⁺-sufficient buffer from WT and MASP-2^−/−/sMAp^−/− mice with disease (Fig. 6C). Again, there was a highly significant (p < 0.007) decrease of 55% in the levels of C5a seen in the sera, in a Ca²⁺-deficient buffer, from MASP-2^−/−/sMAp^−/− mice compared with the WT mice with disease (Fig. 6D). Sera from NOD and WT mice were used as negative and positive controls, respectively, for C5a levels in each ELISA experiment (Fig 6C, 6D). Again, we found that there was a significantly decreased (p < 0.05) level of the C3 activation using sera from NOD mice compared with the WT mice in a Ca²⁺-deficient buffer (Fig. 6B). Sera from C3^−/− mice were also used as an additional negative control for C3 activation ELISA (Figs. 5, 6). These results show that there is a relationship between the in vivo CDA and the relative activation of C3 and C5a generation ex vivo in the presence of the same anti-collagen mAbs that were used to induce disease in vivo in WT, FCB B^−/−, and MASP-2^−/−/sMAp^−/− mice with arthritis. The specific decrease in the AP in MASP-2^−/−/sMAp^−/− mice with disease compared with the WT mice with disease shows that there is an activation of the AP in MASP-2^−/−/sMAp^−/− mice that might be via a C4-independent and MASP-1– or MASP-3–dependent pathway, but it is subdued. Nonetheless, this pathway is directly activating C3 and generating C5a through a different C4-indepenent route (Fig. 7). However, the decrease in C3 and C5a levels in FCN B^−/− mice with CAIA indicate that this ligand might be directly or indirectly activating the AP (Fig. 7).

Previously we have shown that mice having only a functional CP plus the FCN and CL pattern recognition molecules of the LP (MBL A/C^−/−/Df^−/−) or only the intact LP (C1q^−/−/Df^−/−) are highly resistant to CAIA (56). In contrast, mice having only the AP plus FCNs and CLs (C1q^−/−/MBL A/C^−/−) are highly susceptible to CAIA (6, 35, 54). Furthermore, mice in which the LP is only able to be activated through FCN-B and CLs, but with a functional CP and AP (MBL A/C^−/−/FCN A^−/− or MBL A/C^−/−), are fully susceptible to CAIA (35). Thus, in these previous studies, the key role of the AP was clear; however, additional roles for the LP acting through FCN-A, FCN-B, and/or CLs were not able to be assessed. In addition, the roles of individual MASPs, proteases that link lectin pattern recognition molecules to downstream complement activation pathways, have not been fully defined in this model. Given that MASP-1 and MASP-3 are alternatively spliced products of a single gene, it has also been difficult to dissect the distinct functions of these two enzymes. We have shown that either disruption or knockdown of the MASP1/3 gene product(s) blocks the development of CAIA (12, 14); however, whether MASP-1 or MASP-3 acts solely through cleavage of pro-fD to fD, or alternately through activation of MASP-2 or direct cleavage of C3, or a combination thereof, is uncertain. Meanwhile, preliminary data from Takahashi et al. (44) have demonstrated that patients with mutations in an MASP-3–specific exon show no discernable AP in serum samples. Numerous studies have shown that MASP-1 activates MASP-2 as an essential step in the activation of the LP (57–59). In addition, other studies have shown that MASP-1 can activate MASP-3 (60, 61). Together, these data suggest that MASP-3 is the primary activator of the AP via cleavage of pro-fD, whereas MASP-1 is the primary activator of the LP via activation of MASP-2. The cleavage of pro-fD by MASP-3 illustrates an important regulatory relationship between the LP and the AP. Although disruption of the MBL A and MBL C genes in conjunction with disruption of the FCN A gene has no effect on disease development in the CAIA model, it is clear that MASP-3 (and thus the LP) plays an important role in the activation of the AP through the generation of active fD.

In this study, we have uncovered an important dependency of CAIA on FCN B, with a 47% reduction in disease intensity in FCN B^−/− mice, whereas FCN A^−/− mice show full susceptibility to CAIA. In mice, FCN A is produced in the liver and spleen, and is readily detectible in the plasma (24, 25). FCN B is expressed primarily in myeloid cells in the bone marrow with very low serum concentrations under normal conditions (32, 33). The location of FCN B protein remains unclear; however, it has been detected in the lysosomes of activated macrophages (34). Lysosomal localization of FCN B is not necessarily surprising. Other pattern recognition receptors such as TLR9 (62) and NOD proteins (63) are also localized in lysosomes, where they will interact with phagocytosed pathogens and other molecules. Interestingly, we have found that upon induction of an inflammatory disease state such as CAIA, FCN B can be readily detected in the serum (35), and now its presence in the inflamed ankle, suggesting that FCN B is secreted from activated macrophages locally. Comparisons with human FCN genes suggest that mouse FCN A is the ortholog of human FCN 2, whereas FCN-B is the ortholog of human FCN-M (64, 65). FCN A^−/− and FCN B^−/− mice both demonstrate decreased survival rates when challenged with intranasal infection of Streptococcus pneumonia, indicating an important role for both FCN-A and FCN-B genes in protection against certain infections (32).

During biosynthesis, FCN proteins initially organize into trimeric structures, which then combine into oligomers. Given that FCN-A and FCN-B are expressed in different cell types, we can assume that they form separate trimers and probably oligomers as well. FCN-A oligomers are found complexed with MASPs in the serum (31). Recombinant FCN-B is capable of interacting with MASP-2 and activating the LP (18). Thus, FCN-B may be contributing to CAIA via its role as a lysosomal sensor of Ag within macrophages or via its ability to interact with specific ligands within the joint, activating MASPs and inducing local complement activation in the joint. This observation regarding local activation of complement was further supported by the evidence; in this study, we have confirmed that FCN B and fD are also deposited in the ankle’s synovium of WT mice with CAIA. Previously we have reported the presence of not only the expression for MASP-1 and MASP-2, but also the deposition of MASP-1 and MASP-2 proteins locally in the knee joint of WT mice with CAIA (8, 12, 66). Macrophage infiltration was decreased in CAIA in FCN B^−/− mice along with decreased activation of the AP based on low levels of fD in the ankles of FCN B^−/− mice with CAIA. These data support our hypothesis that FCN B, compared with MBL, FCN A, and CLs of the LP, plays a key role in the activation of the AP. However, we cannot yet say whether decreased infiltration caused reduced disease or whether reduced disease led to decreased macrophage infiltration.

In addition, it has been shown that MBL-A and FCN-A are capable of inhibiting LPS-mediated inflammatory processes on mast cells (67), raising the possibility that the presence of FCN-A is somehow inhibiting TLR4 signaling. The CAIA model requires LPS activation of TLR4 for disease development. More work must be performed to further unravel the mechanistic basis and the complex relationship of FCN-A and FCN-B in CAIA.

Another important finding from this study is the dependency of CAIA on the presence of MASP-2. MASP-2 cleaves C4 and C2 to create the C3 convertase of the LP. Previously we showed that disruption of C4 has no effect on CAIA (6). This would then suggest that MASP-2 must have additional functions beyond activation of C4. In this context, it is interesting to note that Asgari and colleagues (68) have shown that MASP-2 is an important mediator of renal I/R injury and works in a C4-independent fashion that nevertheless is associated with C3 deposition. We found C3 and fD deposition to be markedly reduced in joints from MASP-2^−/−/sMAp^−/− mice after CAIA (Figs. 2F, 4), whereas C3 deposition in C4^−/− mice was actually increased (6). Interestingly, when only AP was active, in the sera from arthritic MASP-2^−/−/sMAp^−/− mice, there was not only a huge decrease in the activation of C3, but also a substantial decrease in C5a levels. Furthermore, IHC data regarding the negative and medium levels of C4d and fD, respectively, in ankles of MASP-2^−/−/sMAp^−/− and FCN B^−/− mice with CAIA clearly indicate that MASP-2 appears to contribute to the AP activation and the development of CAIA, likely in a C4-independent manner. Overall, these data are consistent with the hypothesis that MASP-2 is contributing to C3 deposition in CAIA; it could be suboptimal initially and later amplified by the AP, but not through activation of a C4/C2 convertase because no detection of C4d deposition was seen in MASP-2^−/−/sMAp^−/− mice with CAIA. Given the critical importance of the AP to CAIA disease development, it is probable that MASP-2 is contributing in a more direct and independent manner to AP activity, as has been proposed by others (69). Notably, there might be two potential reasons that there is a decrease in C3 activation in the sera from NOD mice, in the absence of AP, compared with the WT mice. The first is that NOD mice exhibit a relative deficiency in the ability of the AP to be activated. This could be because of relative decreases in factor B or fD levels. The second possibility is that the “distal positive feedback loop” that we have previously described in vivo (6) is also active in vitro.

Because there was no protection from the CAIA in CL-11^−/− mice, we concluded that this ligand plays no essential role in CAIA. We have not examined the expression of CL-11 in the knee joints, and there is a possibility that this site is impenetrable to the circulating CL-11, and thus the molecule has no local function in the knee joint. Recently it has been shown that CL-11^−/− mice are protected from renal I/R injury (41). Therefore, the requirement of various LP ligands must be target organ and/or context specific.

Our data show, intriguingly, that disruption of the FCN B gene can partially protect mice from CAIA, whereas disruption of FCN A does not. One possible explanation is that there was a decrease in C3 activation followed by C5a generation in FCN B^−/− arthritic mice, indicating this LP ligand directly activates the AP though MASP-1/3 or indirectly activates the AP through the MASP-2–dependent pathway (Fig. 7). Additional work is necessary to define these in vivo. The likely source of FCN-B is a basal infiltrating macrophage population (33, 34). Activation of complement via the LP and amplified by the AP would then result in the generation of C3a and C5a, leading to the massive infiltration of macrophages and neutrophils seen in full disease.

In summary, we have shown an important dependence of CAIA on FCN B and MASP-2. It is tempting to speculate that these two dependencies are related as modeled (Fig. 7). This may be tested with the generation of dual FCN B^−/− and MASP-2^−/−/sMAp^−/− mice. In addition, one can ask what is the driver for the partial disease seen in these MASP-2^−/−/sMAp^−/− mice? It might be wholly independent activity of the AP. However, others have reported the activation of C3 in humans who are missing C2 via a bypass pathway (50). We speculate that MASP-1 may be responsible for this by direct suboptimal cleavage of C3 by a C4-independent pathway, which then may be augmented by the AP, if then it will be consistent with the promiscuous activities of the MASP-1 (70). Further work will be required to generate data supporting this conjecture. There is also a possibility that blocking FCN-M in humans, which is equivalent to mouse FCN-B, using an inhibitory anti–FCN-M Ab, might be therapeutically beneficial for certain inflammatory diseases including arthritis.

We thank the Department of Ophthalmology, University of Colorado Anschutz Medical Campus for allowing the use of their Nikon Eclipse 80i microscope equipped with Nikon DS-Qi1MC camera, and specifically to Dr. David A. Ammar for helping to take excellent images of the IHC sections from the ankles of mice with arthritis.

The authors have no financial conflicts of interest.