Increased severity of arthritis in mice lacking MMP-8
To ascertain the role of MMP-8 in experimental arthritis, we induced passive K/BxN arthritis in 12-generation B6-backrossed Mmp8-deficient (Mmp8-/-) mice, and their matched wildtype (Mmp8+/+) and heterozygous (Mmp8+/-) littermate controls.
In a first experimental group, male and female Mmp8+/+ (n = 10), Mmp8+/- (n = 10) and Mmp8-/- (n = 10) mice were injected intraperitoneally at day 0 and 2 with 200 μl K/BxN mice serum and monitored for signs of arthritis. Evolution of arthritis was evaluated by two blinded observers on a 0 to 4 scale, as described in Materials and methods.
An incidence of 100% of arthritis was observed in Mmp8-/-, Mmp8+/+ or Mmp8+/- mice (Figure ). The time course of arthritis was also similar in the three groups of mice. The disease developed rapidly and the maximum of severity was observed between 9 and 12 days. Surprisingly, Mmp8-deficient mice displayed significantly higher severity of arthritis than Mmp8+/+ and Mmp8+/- mice (P = 0.025 by repeated-measures one-way ANCOVA test) all through the follow-up. As the severity of arthritis was similar in Mmp8+/+ and Mmp8+/- mice, these mice were considered a unique control group (Mmp8+).
Figure 1 Increased arthritis severity in Mmp8-deficient mice. (a) Clinical score measured in 10 Mmp8-/- mice, 10 Mmp8+/+ mice and 10 Mmp8+/- mice after intraperitoneal injection of 200 μl K/BxN mice serum at days 0 and 2. Values expressed as mean ± (more ...)
To exclude that the system was overloaded by using 200 μl K/BxN serum and to further evaluate the observed differences between Mmp8 control and deficient mice, a second experimental group composed of male and female Mmp8-/- mice (n = 17) and control mice (n = 17) was injected intraperitoneally at day 0 and 2 with 100 μl K/BxN serum (Figure ). Arthritis was monitored until day 9 and the results confirmed those previously obtained - arthritis severity was significantly higher in Mmp8-deficient mice compared with control mice (P = 0.04 by repeated-measures one-way ANCOVA test; Figure ).
Increased joint inflammation and bone erosion in Mmp8-deficient mice
To quantify joint involvement, we assessed synovial inflammation and bone erosion in hematoxylin and eosin stained sections of ankle joints, and cartilage damage was evaluated in Toluidine blue and Safranin-O stained sections. Right ankles were taken on day 9 after serum transfer from seven Mmp8-/- and seven Mmp8+ male and female mice of the group injected intraperitoneally with 100 μl K/BxN serum, and a blinded observer scored the histological sections. The clinical score of the Mmp8-/- mice was higher than in the Mmp8+ mice (P = 0.027 by Mann-Whitney U test).
Synovial inflammation was scored on a 0 to 4 scale, corresponding to the degree of thickening of the synovial lining and sublining infiltration. A significant increase in synovial inflammation score in Mmp8-/- mice compared with Mmp8+ mice was observed (P = 0.04 by Mann-Whitney U test; Figures and ). Changes in cellular infiltrate composition, however, were not observed in mice lacking Mmp8 compared with wildtype mice. Specifically, a similar rate of neutrophils and mononuclear cells were seen in both groups of mice.
Figure 2 Increased joint inflammation and bone erosion in mice lacking Mmp8. Histologic scores of synovial inflammation (SI), cartilage damage (CD), bone erosion (BE) and tartrate-resistant acid phosphatase (TRAP) staining of ankle sections of Mmp8+ mice (n = (more ...)
Figure 3 Higher synovial inflammation and bone erosion in arthritic joints of Mmp8-deficient mice than control mice. Representative sections of ankle joints of Mmp8-/- mice (a) to (c) and Mmp8+ mice (d) to (f) 9 days after intraperitoneal injection of K/BxN serum. (more ...)
As shown in Figures and , bone erosion was more marked in Mmp8-/- mice than in wildtype mice (P = 0.04 by Mann-Whitney U test). Furthermore, staining sections for TRAP activity revealed a significantly increase of TRAP-positive multinucleated cells in Mmp8-/- mice compared with Mmp8+ mice (P = 0.025 by Mann-Whitney U test). These cells were observed at sites of bone erosion in both groups of mice (Figure ).
A trend to higher cartilage damage in Mmp8-/- mice than control Mmp8+ mice was detected (Figures and ), although the difference was not significant (P = 0.11 by Mann-Whitney U test). Significant correlations between synovial inflammation, cartilage damage, bone erosion and TRAP staining with clinical scores were observed (RS >0.64 and P < 0.017).
Figure 4 Clustering of differentially expressed genes. The DAVID utility [24,25] was used to cluster differentially expressed genes according to a false discovery rate 0.01 threshold, in both Mmp8+/+ and Mmp8-/-arthritic mice, exclusive arthritic Mmp8+/+ mice (more ...)
Overall these results suggest that MMP-8 plays a protective role in inflammatory arthritis.
To explore the mechanisms underlying the increased arthritis severity in Mmp8-deficient mice, we used a genome-wide microarray analysis including probes for more than 28,000 mouse transcripts. Ankle joints from three mice from each of the following groups were studied: Mmp8+/+ and Mmp8-/- mice with and without arthritis. Mice with arthritis were injected intraperitoneally on days 0 and 2 with K/BxN mice serum, and joints were taken 7 days after injection. Comparison of expression levels between arthritic and nonarthritic control mice yielded a list of about 3,200 genes that were differentially expressed according to an FDR of 5% (2,996 genes in the comparison among Mmp8+/+ mice and 3,407 genes in the Mmp8-/- comparison), or about 1,000 genes according to the more stringent FDR 1% threshold (Table ). These lists were largely concordant in the two independent comparisons, as assessed by the fact that most genes differently expressed in Mmp8+/+ mice were also differently expressed in Mmp8-/- mice. In fact, direct comparison of arthritic Mmp8+/+ mice with arthritic Mmp8-/- mice did not show any significant difference.
Number of differentially expressed genes
We therefore conducted other types of analyses. First, we compared the functional groups of differentially expressed genes modified in both groups of arthritic mice, only in arthritic Mmp8+/+ mice and only in Mmp8-/- mice. The 660 genes that were modified both in Mmp8+/+ mice and Mmp8-/- mice, according to a FDR 0.01 threshold, could be grouped into eight clusters with an enrichment score over 3.0 (Figure ). These clusters included some that are more structurally defined and others that are more related with cellular or biological pathways. The same type of analysis was also carried out for the sets of genes that were different only in the Mmp8+/+ arthritic mice according to the same FDR 0.01 criteria. There were 334 genes in this class and they were grouped into five clusters of annotations with an enrichment score over 3.0. These five clusters were a subgroup of the eight clusters that were modified in both groups of arthritic mice. The only three clusters missing here were the cluster of epidermal growth factor-like domain proteins and the two last clusters: the one grouping cell migration and motility genes, and the one containing transmembrane proteins. The genes whose expression was significantly modified in arthritic Mmp8+/+ mice and not in arthritic Mmp8-/- mice were therefore largely from the same functional classes as the genes that were modified in both types of mice.
A similar analysis with the 386 genes that were modified only in arthritic Mmp8-/-mice gave very different results. No single cluster of genes showed an enrichment score over 3.0, and only two clusters showed a score over 2.0. This indicates that the modified genes specific of arthritis in the Mmp8-/- mice are very varied and difficult to group. The pattern of genes that were differentially regulated in Mmp8+/+ mice and Mmp8-/- mice are therefore very different: the genes regulated specifically in arthritic Mmp8-/- mice are similar in number but much more diverse functionally.
The lack of any clearly defined functional class of genes specifically modified in arthritic Mmp8-/- mice made it impossible to focus on them to try to discern important factors in the differential arthritis phenotype. We decided to concentrate instead on the genes that, having a most clearly changed expression with arthritis, were also most differentially affected in Mmp8+/+ and Mmp8-/- mice. We selected the 86 nonredundant genes that were different between arthritic and control mice in the comparison of either Mmp8+/+ mice or Mmp8-/- mice according to the very conservative Bonferroni-corrected threshold of P = 0.05. We obtained the fold change ratios between their respective comparisons. Genes with fold change ratios higher than 1.35 and lower than 0.75 were considered interesting (Table ). That is, differences between arthritic Mmp8-/- mice and their controls were compared with differences between arthritic Mmp8+/+ mice and their controls, and the most extreme fold change ratios were selected. Seven out of 29 genes were chosen for confirmatory real-time PCR experiments given their interest in inflammation, autoimmunity or arthritis.
Genes with the most discordant changes in expression in Mmp8+/+ and Mmp8-/-arthritic mice
The data discussed in this publication have been deposited in NCBI'S Gene Expression Omnibus [26
] and are accessible through the GEO Series accession number [GEO:GSE22971] [27
Induction of IL-1β, PROKR2 and PTX3 in arthritic Mmp8-deficient mice
To corroborate the results obtained by the microarray analysis, real-time RT-PCR experiments were performed in arthritic joints from six other Mmp8-deficient mice and six wildtype mice treated in the same way. Increased mRNA expression of IL-1β, PROKR2 and pentraxin-3 (PTX3) was found in arthritic Mmp8-/- mice compared with wildtype mice (P = 0.035, P = 0.032 and P = 0.028, respectively; Figure ). Real-time PCR did not, however, confirm the expression changes observed in CALPAIN 6, MMP-3, C1QTNF3 and TenascinW in Mmp8-deficient mice compared with wildtype mice (data not shown).
Figure 5 Increased IL-1β, PTX-3 and PROKR2 mRNA and protein levels in mice lacking Mmp8. (a) IL-1β, pentraxin-3 (PTX-3) and prokineticin receptor 2 (PROKR2) mRNA levels were measured by quantitative real-time PCR in arthritic joints of Mmp8-/- (more ...)
Increased production of IL-1β and PTX3 was verified by ELISA assay (Figure ), and results showed a significant increase of both proteins in joints from Mmp8-/- mice compared with Mmp8+/+ mice (P = 0.031 and P = 0.017, respectively). PROKR2 production was assessed by western blot and is shown in Figure . As expected, PROKR2 levels were significantly higher in joints from Mmp8-deficient than in control male mice (P = 0.031).
Figure 6 Increased PROKR2 expression in Mmp8-deficient mice. Protein expression of prokineticin receptor 2 (PROKR2) was determined by western blot in arthritic joints of six Mmp8-/- mice and six Mmp8+/+ mice at 7 days after arthritis induction. Densitometric analysis (more ...)
Accumulated evidence indicates that MMPs are involved in the cartilage destruction observed in RA [1
]; MMP inhibitors are thereby of special interest for the treatment of RA. Results from clinical trials of MMP inhibitors in RA have not been encouraging, however, probably due to lack of specificity of such inhibitors [28
]. In fact, analyses of several MMPs in animal models have shown either [9
] exacerbation or reduction of arthritis severity depending on the MMP analyzed. This indicates that specific MMPs could have either a promoting or a protective role in arthritis pathogenesis. Knowledge of the role of specific MMPs in the pathogenesis of arthritis therefore seems pivotal to obtain successful inhibitors for treatment. In the present study, we have investigated the impact of lack of Mmp8
in the K/BxN serum-transfer arthritis model. The advantages of this transfer model with respect to other arthritis models is its 100% penetrance, early onset, rapid development of osteolytic lesions and its MHC independence. Clinical features and histopathology are very similar to human RA. Another characteristic of the K/BxN transfer model is that it allows us to focus on the effector phase mechanisms of arthritis that are dependent on neutrophils, macrophages, mast cells, and inflammatory mediators, especially IL-1β, but independent of T cells and B cells.
We have found that the absence of Mmp8
increased the severity of arthritis without noticeably affecting its time course, either at its onset or at its spontaneous remission. The aggravated clinical course of arthritis was accompanied by exacerbated synovial inflammation and bone erosion. These effects were associated with modified expression of a varied array of genes, including overexpression of IL-1β
in arthritic joints. Surprisingly, despite the known collagenolytic activity of Mmp-8, its absence did not protect from cartilage damage but a trend to increased damage was observed compared with Mmp8
wildtype mice. This finding may indicate that other Mmps could compensate for its absence. These data indicate that Mmp-8 plays a protector role against arthritis in this model. This effect is consistent with the previously reported effect of Mmp-8 absence in other inflammation models such as OVA-induced airway inflammation [17
], chemically-induced skin carcinomas [19
] and skin wound healing [16
], in which the absence of Mmp-8 increased the severity of these pathologies and delayed wound healing. In these studies, disease aggravation was linked to increased neutrophil accumulation in the mice lacking Mmp8
. In our work, we did not observe differences in cellular infiltrate composition between Mmp8
control and deficient mice, suggesting that mechanisms involved in the Mmp-8 regulation of inflammation are complex and include its effect in other aspects of inflammation as shown by our expression studies. It is possible that differences between models depend on the nature of the inflammatory stimulus or of the tissue affected.
To elucidate the mechanisms behind arthritis aggravation in Mmp8-/- mice, we have investigated the gene expression profile in Mmp8-sufficient and Mmp8-deficient mice with and without arthritis using microarray technology. There was a wide array of genes that changed expression in arthritic mice. Most were coincident in Mmp8-sufficient and Mmp8-deficient mice, and they can be grouped in functional categories that are congruent with current knowledge of arthritis mechanisms. The functional spread of the genes whose expression was only modified in the arthritic Mmp8-/- mice contrasted with the clustering in five functional categories of the genes significantly modified only in the arthritic Mmp8 wildtype mice, despite being similar in number. This result is consistent with the lack of any clearly different phenotype in the histological analysis and has been taken into consideration to interpret the analyses of individual genes. To select genes for detailed analysis, we decided to focus on the genes that with high likelihood were differentially expressed with respect to arthritis and the presence of Mmp8. After selection of a group of seven genes, we found an increased expression of IL-1β, PTX3 and PROKR2 in arthritic joints from Mmp8-deficient mice compared with wildtype mice that were confirmed by real-time PCR assays. The corresponding increase in protein expression was validated by ELISA and western blot.
is highly expressed in the synovium of RA patients and plays a crucial role in production of inflammatory mediators and articular damage [2
]. This cytokine's functional relevance has been demonstrated in several animal models, including the K/BxN model [30
]. Results of these studies indicate that the increased IL-1β
expression observed in Mmp8
-deficient mice can contribute to the higher clinical score, synovial inflammation, osteoclast activity and bone erosion found in these mice.
PTX3 is the prototypic member of the long pentraxin family of acute phase reactants. PTX3 rapidly increases in serum during endotoxic shock, inflammation and infections [37
]. A possible role of this protein in potentiating inflammation has been reported in a model of intestinal injury by ischemia/reperfusion in which PTX3 transgenic mice showed exacerbated inflammatory response and increased lethality [38
]. Also, mice lacking PTX3 displayed reduced tissue inflammation and increased survival rates [39
]. Our results showed an increased PTX3 expression in mice lacking Mmp8
compared with wildtype mice, where it was also increased, indicating that PTX3 upregulation could have contributed to the higher arthritis severity in the knockout mice. This result suggests that the accumulation of PTX3 in the synovial fluid of RA patients after being produced by synoviocytes and synovial endothelial cells [40
] can be also a contributor to the inflammation process.
PROKR2 is a seven-transmembrane coupled G-protein receptor that binds prokineticin-2. PROKR2 is highly expressed in the bone marrow, and in neutrophils, monocytes and dendritic cells [41
]. Signaling through this receptor induces survival, differentiation and activation of granulocytic and monocytic lineages [42
]. The higher expression of PROKR2 found in the arthritic joints from Mmp8
-deficient mice could therefore have contributed to the increased inflammatory infiltration observed in them.
Changes in the expression of these three genes exemplify different ways in which the lack of MMP-8 led to an aggravation of arthritis: promotion of inflammation by IL-1β and other molecules like PTX3, induction of maturation and activation of osteoclasts by IL-1β and PROKR2, and enhanced inflammatory infiltrate by IL-1β and possibly PROKR2 - however, other contributing mechanisms are possible as only a fraction of the genes with possible differential expression were analyzed. Similar analysis in other models of inflammation will help to unravel the many ways in which MMP-8 seems to protect against inflammation.