Mice lacking PARP-1 (kindly provided by G de Murcia, CNRS, Strasbourg, France) have been described previously 22. The mice used in these experiments were of mixed (C57BL/6 × 129Sv) background. More than ten different breeding pairs of parp-1^+/o mice were intercrossed to generate parp-1^+/+, parp-1^+/o and parp-1^o/o mice. Parp-1^o/o mice and control matched littermates (parp-1^+/+, parp-1^+/o) were analyzed.

Genotypes were assessed by PCR of tail DNA. The mice were maintained in the mouse facility of the Facultad de Medicina de Santiago de Compostela. Animal care was in compliance with Spanish regulations on the protection of animals used for experimental and other scientific purposes (Real Decreto 223/1998). The experimental protocols were approved by the Animal Care and Use Committee of the University of Santiago de Compostela.

CAIA was induced in 6-week-old male and female mice by intravenous injection on day 0 of 3 mg/mouse of an arthritogenic cocktail of 4 monoclonal anti-type II collagen antibodies (Arthrogen, Chondrex, Redmond, WA, USA) 23. On day 2, mice were boosted with 50 μg of lipopolysaccharide by intraperitoneal injection. Arthritis was assessed every other day by two blinded observers until day 12, using a semi-quantitative clinical score ranging from 0 to 4: 0, no swelling; 1, slight swelling and erythema of the ankle, wrist or digits; 2, moderate swelling and erythema; 3, severe swelling and erythema; and 4, maximal inflammation with joint rigidity. The maximum possible score was 16 per mouse.

Hind limbs were prepared for histology by dissecting the skin and muscle, and then sectioning knee joints. Specimens were fixed for 24 hours and demineralized in phosphate-buffered saline-0.5 M EDTA for 10 days. Knee joints were embedded in paraffin and sections were cut and stained with hematoxylin and eosin (H&E) for evaluation of inflammation. For analysis of damage to cartilage, knee sections were stained with Toluidine blue, Safranin-O and Masson trichrome following standard methodology. The sections were scored by two blinded observers. Synovial inflammation was scored on a scale of 0 to 3: 0, no inflammation; 1, slight thickening of synovial cell layer and/or some inflammatory cells in the sublining; 2, thickening of synovial lining, infiltration of the sublining; and 3, pannus formation.

Exudate was scored according to the following scale: 0, no detectable neutrophil infiltration in the synovial space; 1, mild infiltration; 2, moderate infiltration; and 3, severe infiltration.

Cartilage damage was evaluated following a scale of 0 to 3: 0, normal cartilage; 1, cartilage surface irregularities and loss of metachromasia adjacent to superficial chondrocytes; 2, fibrillation of cartilage and formation of some chondrocyte clusters, with minor loss of surface cartilage; and 3, gross cartilage abnormalities, including loss of superficial cartilage, extension of fissures close to subchondral bone, and a large number of chondrocyte clusters.

Fibroblast-like synoviocytes (FLSs) were isolated from parp-1^+/+, parp-1^+/o and parp-1^o/o mice. Synovial tissue was minced and incubated with 1 mg/ml collagenase in serum-free DMEM (Gibco, Invitrogen, Barcelona, Spain) for 3 hours at 37°C. After digestion, FLSs were filtered trough a nylon cell strainer (BD Falcon, Franklin Lakes, NJ, USA), washed extensively, and cultured in DMEM supplemented with 10% v/v FCS (Gibco, Invitrogen), penicillin, streptomycin, and L-glutamine (Sigma, St Louis, MO, USA) in a humidified 5% CO₂ atmosphere. After overnight culture, non-adherent cells were removed, and adherent cells were cultured in DMEM supplemented with 10% v/v FCS.

Total proteins (20 μg) were separated by 10% SDS-PAGE, transferred to a PVDF membrane (Hybond-P, Amersham Biosciences, Buckinghamshire, UK) and probed with anti-PARP-1 (VIC-5, kindly provided by G de Murcia, CNRS, Strasbourg, France) and anti-actin (Sigma) antibodies as previously described 24. Bound antibody was revealed with goat anti-rabbit-horseradish peroxidase (Rockland Immunochemicals Inc., Gilbertsville, PA, USA) and the blot was developed using the ECL plus detection system (Amersham Biosciences).

Total RNA was obtained from joints of parp-1^+/+ and parp-1^o/o mice on day 7 following Arthrogen injection, and from joints of parp-1^+/+ and parp-1^o/o control mice without arthritis. We used the RNeasy Kit and RNase-Free DNase Set (Qiagen GmbH, Hilden, Germany) according to the manufacturer's instructions. One microgram of total RNA was subjected to cDNA synthesis using M-MLV reverse transcriptase, random primers and RNaseOUT recombinant ribonuclease inhibitor (Invitrogen). Quantitative real-time PCR was performed in duplicate in a Chromo-4 real-time thermal cycler (MJ Research, Waltham, MA, USA), using a LightCycler DNA Master SYBR Green I kit (Roche Diagnostics, Barcelona, Spain), according to the manufacturers' protocols. The specific primers used in these reactions are listed in Table 1. Relative levels of gene expression were normalized to the β-actin gene using the comparative Ct method, where Ct is the cycle at which the amplification is initially detected. The relative amount of mRNA from the different genes was calculated using the formula 2^-ΔΔCt, where:

ΔΔCt = [Ct_target - Ct_β-actin]_{WT or KO with arthritis} - [Ct_target - Ct_β-actin]_{WT or KO controls}

For wild-type (WT) and PARP-1 deficient samples without arthritis, ΔΔCt equals zero and 2⁰ equals one. For wild-type and knockout (KO) samples with arthritis, the value of 2^-ΔΔCt indicates the fold change in gene expression relative to the wild-type and knockout controls, respectively. Melting curves and agarose gel electrophoresis established the purity of the amplified band.

Knee joints were obtained, frozen in liquid nitrogen and homogenized in 0.5 ml ice-cold 20 mM Hepes buffer supplemented with 1 mM dithiothreitol, 0.1% v/v Triton and a protease inhibitor cocktail. After incubation for 30 minutes at 4°C, the homogenate was centrifuged for 10 minutes at 10,000 × g. Protein concentration was measured in supernatants by the Bradford method and a volume containing 100 μg of proteins was subjected to ELISA for IL-1β, TNF-α, IL-6 and MCP-1 (OptEIA ELISA Sets, BD Pharmingen), according to the manufacture's instructions.

Differences between experimental groups were assessed by ANCOVA, MANCOVA and Mann-Whitney U test. p values <0.05 were considered significant.

Although PARP-1 is found in the majority of the nucleated cells of the body, its expression in joint tissue has never been studied. Here, we have analyzed PARP-1 protein expression by western blot in isolated FLSs from wild-type (parp-1^+/+) mice, mice lacking PARP-1 (parp-1^o/o), and parp-1 heterozygous mice (parp-1^+/o). PARP-1 was highly expressed in FLSs from PARP-1 wild-type mice, moderately expressed in parp-1^+/o and was absent in parp-1^o/o mice (Figure 1a). Comparable results were obtained in the immunohistochemical analysis of joint sections from parp-1^+/+ and parp-1^o/o mice with anti-PARP-1 antibody (Figure 1b, d).

To investigate the contribution of PARP-1 to experimental arthritis, we induced CAIA in control and PARP-1 deficient mice (Figure 2a). In eight separate experiments, male and female parp-1^+/+ (n = 18), parp-1^+/o (n = 8) and parp-1^o/o (n = 19) mice were injected with Arthrogen and lipopolysaccharide 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.

There was no difference in incidence or clinical course of arthritis in parp-1^o/o animals compared with parp-1⁺ control mice. The incidence of disease was 100% in control mice and 94.7% in parp-1^o/o mice. In both groups, arthritis developed rapidly, the signs of disease appearing as soon as three to five days after the injection of antibody, and reached maximum severity around day seven to nine. PARP-1 deficient mice consistently displayed significantly lower severity of arthritis than parp-1⁺control mice (p = 0.03 by repeated measures 1-way ANCOVA test) all through the follow-up (Figure 2b).

These results suggest that PARP-1 has a role in the pathogenesis of this arthritis model.

As parp-1^+/+ and parp-1^+/o control mice had similar clinical phenotypes, for further analysis, they were pooled together and considered as parp-1⁺.

To quantify joint involvement, we assessed synovial inflammation in H&E stained sections of knee joints. Joints were taken from 14 parp-1^o/o and 13 parp-1⁺ mice on days 5, 7 and 12, and histological sections were scored by two blinded observers on a 0 to 3 scale, corresponding to the degree of thickening of the synovial lining, sublining infiltration and pannus formation. On this scale, we observed a clear trend to a lower synovial inflammation score in parp-1^o/o mice compared to parp-1⁺mice (Figures 3 and 4), although the difference was not significant (p = 0.058, by 1-way MANCOVA fixed effects test).

Joint sections were also stained with Toluidine blue, Safranin-O and Masson trichrome to evaluate cartilage damage. The results also showed a trend to less damage in parp-1^o/o mice (Figures 3 and 5), although, again, the difference did not reach statistical significance (p = 0.053, by 1-way MANCOVA fixed effects test). When we considered synovial inflammation and cartilage damage jointly as two facets of the arthritic lesions, the difference between parp-1^o/o and parp-1⁺ mice was significant (p = 0.03).

Thus, PARP-1 protein appeared to be involved in the pathogenesis of the CAIA model, both in synovial inflammation and cartilage damage, although it did not seem to have a pivotal role.

It has been previously described that, in several rodent models of inflammation, PARP-1 activation is involved in neutrophil recruitment 25. Neutrophils have been implicated in arthritis disease; specifically, extensive neutrophil exudate is displayed in CAIA model 23 and neutrophils release elastase and proteases, which degrade proteoglycans 4. It remains possible that the decreased severity of arthritis observed in PARP-1 deficient mice was associated with reduced neutrophil exudate in joints from these mice. To evaluate this possibility, we assessed, at five and seven days, the exudate score on a 0 to 3 scale in H&E stained sections of knee joints. Exudate appeared slightly lower in parp-1^o/o compared to parp-1⁺ mice, although the difference was not significant (p = 0.12, by 1-way MANCOVA fixed effects test) (Figure 6). Thus, lack of PARP-1 does not impair neutrophil exudation in this arthritis model.

To explore the possible mechanisms underlying the reduced arthritis observed in PARP-1 knockout mice compared to control mice, we studied, by quantitative real-time PCR, mRNA levels of IL-1β, IL-6, TNF-α, MCP-1, small inducible cytokine A5 (Cc15; RANTES), inducible nitric oxide synthase (iNOS) and cyclooxygenase (COX)-2 in arthritic joints at day seven after Arthrogen injection. The fold change in mRNA of arthritic versus non-arthritic parp-1^o/o and parp-1⁺ mice is shown in Figure 7. All the inflammatory mediators were detected in both groups of mice and, interestingly, IL-1β and MCP-1 mRNA were significantly less induced in arthritic parp-1^o/o compared to arthritic parp-1⁺ mice. IL-6 mRNA showed a trend towards lower induction in parp-1^o/o compared to parp-1⁺ arthritic mice (p = 0.1, by Mann-Whitney U test). However, mRNA expression of TNF-α and Ccl5 were induced to a similar extend in both groups of arthritic mice (p = 0.9 and p = 0.7, respectively, by Mann-Whitney U test). Transcription of genes encoding iNOS and COX-2, which are involved in the synthesis of nitric oxide and prostaglandin E₂, respectively, were induced to levels that were not significantly different in PARP-1 deficient and sufficient arthritic mice. However, a tendency towards lower induction in the parp-1^o/o mice was noted.

To confirm these findings, we next determined the levels of IL-1β, IL-6, TNF-α and MCP-1 proteins in joint tissues. IL-1β and MCP-1 were significantly reduced in joints from arthritic parp-1^o/o compared to arthritic parp-1⁺ mice (Figure 8); however, there was no difference in the production of TNF-α and IL-6 in both groups of mice (p = 0.4 and p = 0.3, respectively, by Mann-Whitney U test). These results are consistent with those obtained for the mRNA analysis.