C57Bl/10.RIII (B10.RIII) were originally provided by J. Klein (Tübingen, Germany), and kept in the breeding colony at the Department of Medical Inflammation Research, Lund University. RIIIS/J animals were purchased from The Jackson Laboratory (Bar Harbor, ME). The Eae39 congenic mice (C1, Figure 1) were produced by marker selected backcrossing of the RIIIS/J (donor) mice to the B10.RIII (recipient) strain. All experiments were approved by the local ethical authorities in Malmö-Lund, Sweden (permit numbers: M70-04, M75-04, M107-07 and M109-07).

Bovine type II collagen was prepared from calf nasal cartilage by pepsin digestion and was purified as previously described 19. CIA was induced by intra-dermal immunisation at the base of the mouse's tail with 100 μg bovine CII emulsified in incomplete Freund's adjuvant (IFA) (Difco, Detroit, MI, USA). The mice were boosted 35 days later with 50 μg bovine CII emulsified in IFA. The mice, ranging in age between 10 and 24 weeks, were all immunised the same day. Clinical disease was monitored once or twice a week according to a scoring system based on the number of inflamed joints. Each inflamed toe or knuckle was given a score of one and an inflamed wrist or ankle was given five points. Each mouse could in total get 15 points per limb and a maximum score of 60. The area under the curve (AUC) was calculated as the sum of scores for each individual mouse during a defined test period. The mean maximum score, representing disease severity, was calculated as the mean of the maximum score of all sick mice in the respective groups.

Blood was collected on day 14, 21 or day 54 after immunisation. Sera were prepared and stored at -20°C until assayed. ELISA was used to determine levels of antibodies against collagen type II. Plates (Nunc maxisorp, Roskilde, Denmark) were coated with bovine CII (10 μg/ml) in PBS (pH 9) and blocked with 1% BSA. Immunoglobulin (Ig) M, IgG1, IgG2c, IgG3 and total Ig levels were measured using biotinylated secondary antibodies: goat anti-mouse IgM (No. 1020-08); IgG1 (No.1070-08); IgG2c (No. 1079-04); IgG3 (No. 1100-08); and total Ig (No. 1010-08) (Southern Biotechnologies Associates, Birmingham, AL, USA). Binding of biotinylated antibodies was revealed by Extravidin Peroxidase (No. E-2886) (Sigma-Aldrich, St Louis, MO, USA). Plates were developed with ABTS: 2,2'-Azino-di-[3-ethylbenzthiazoline sulfonate (6)] diammonium salt (Roche, Mannheim, Germany). Pooled sera were used as a standard and the antibody levels were measured as arbitrary concentrations.

Genomic DNA was isolated from toe or tail biopsies. The biopsies were dissolved in 500 μl of 50 mM sodium hydroxide for one to two hours at 95°C, and subsequently neutralised with 100 μl 1 mM Tris-HCl (pH 8). To perform a standard 10 μl PCR, 1 μl of the solution was used. The PCR products were analysed on a MegaBACE DNA analysis system 1000 (Amersham Pharmacia Biotech, Little Chalfont, UK), according to the manufacturer's protocol. Fifteen informative fluorescence-labelled micro-satellite markers (Interactiva Biotechnologie, Ulm, Germany and MWG Biotech, Ebersberg, Germany) were used to genotype the Eae39 congenic fragment. Linkage analysis and permutation tests were conducted as previously described 13. Sub-congenic mice were genotyped with additional micro-satellite markers, where some markers are made in-house: D5acacbhm4 (114.42 Mbp, forward primer (FP) 5'-CCCTGTAGAAGACTGGGAATTG-3, reverse primer (RP) 5'-TCCAGGACAGTCAGGGCTAC-3'), D5taokhm12 (117.53 Mbp, FP: 5'-TCAGGGCTCCATGCACTT-3', RP: 5'-CACAAGTGGCTCTCAGTGCT-3), D5sdshm18 (120.74 Mbp, FP: 5'-GGGGAACACAAGGAGTTTGA-3', RP: 5'-ATTCAAGGGCATGTGTGTGA-3').

The Eae39 locus on mouse chromosome 5 (Figure 1) was previously described in a genetic linkage analysis based on a backcross between the RIIIS/J and B10.RIII strains 13. The confidence interval for Eae39 extended from the micro-satellite marker D5Mit259 (90 Mbp) to D5Mit136 (119 Mbp). This locus was shown to control incidence of acute EAE in male mice, with a dominant effect of RIIIS/J alleles on protection from disease development.

To further investigate the Eae39 locus, a 65 Mbp RIIIS/J fragment was bred into the B10.RIII genome to establish a BR. Eae39^RIIIS/Jcongenic strain (Figure 1). This region contains a number of QTLs linked to the development of disease in experimental models for arthritis, and we wanted to investigate whether Eae39 controls the susceptibility to CIA in the B10.RIII and RIIIS/J strain combination. Thus, Eae39 congenic mice were immunised with bovine collagen type II in IFA. As shown in Figure 2 and Tables 1 and 2, genes within Eae39 control CIA. Figure 2 shows the development of disease in mice with the congenic Eae39 fragment shown in Figure 1. Mice with the C1 congenic fragment had higher incidence of disease (64%) and higher accumulated arthritis score (AUC (d50-73) = 125 ± 50) compared with the non-congenic littermates (incidence of disease 24%, AUC (d50-73) = 57 ± 29). (Incidence, p = 0.0271, Chi squared test; AUC (d50-73), p = 0.0379, Mann–Whitney U test).

Next, we intercrossed C1 heterozygous mice in order to get offspring with overlapping congenic fragments (Figure 3) to pinpoint smaller intervals within Eae39 linked to the disease phenotype. The mice were investigated for development of CIA and each individual was genotyped with markers spanning the congenic fragment. We observed that RIIIS/J alleles at marker D5Mit113 (78 Mbp) promoted disease incidence (Table 1). In contrast, one RIIIS/J allele at the marker D5Mit136, in the telomeric part of the fragment, protected against CIA development (Table 1). There was no difference in mean maximum score of the affected mice, except for females with RIIIS/J alleles at marker D5Mit136, which had significantly lower CIA scores compared with littermate controls (Table 1). This is in line with the male mice carrying heterozygous alleles at D5Mit136, where none of the mice developed arthritis.

In Table 2, correlation between the disease severity phenotype AUC, day 50 to 73 after immunisation, and genotype is shown. The AUC is the sum of scores for each individual mouse during a defined test period and describes the development of disease in terms of onset, duration and severity. In line with the disease incidence data (Table 1), RIIIS/J alleles at D5Mit113 promoted disease, whereas one RIIIS/J allele at about 120 Mbp (D5Mit136, D5Mit367) almost completely protected from CIA (Table 2). From these results we conclude that Eae39 harbors genes that, in addition to controlling EAE, are important for susceptibility to CIA, and that the region contains genes operating in different directions in the disease development.

In humans, women are more affected by RA than men. The sex influence on susceptibility to CIA is, however, normally the opposite in mice. In the first investigation of CIA development in mice with overlapping Eae39 sub-interval congenic fragments, we observed that female congenic mice had the same, or slightly higher, incidence of disease compared with male mice (Table 1). Severity of CIA (mean maximum score) was the same, except for females homozygous for B10.RIII alleles (b/b) at marker D5Mit136, in which the severity was higher (p < 0.05) compared with male mice (Table 1). From the original mapping experiment, Eae39 was linked to development of acute EAE in male mice 13. For this reason, and in order to keep the number of mice used to a minimum, we decided to continue the present study with male mice only.

In an F2 cross between the arthritis susceptible DBA/1J and the resistant FVB/N strains, it was recently shown that the Cia27 locus controls anti-collagen type II IgG2a antibody levels 16. To investigate the corresponding region within the Eae39 locus for disease phenotypes, we produced mice with smaller, overlapping congenic fragments (C2 to C5) (Figure 4) and studied the anti-collagen antibody response after immunisation. The IgG1, IgG2c, IgG3 and IgM anti-collagen serum levels at day 14 after immunisation were significantly lower in mice with the C2 congenic fragment (Figure 4) compared with littermate controls (Table 3). By comparing antibody levels in mice with the C3 and C4 congenic fragments, we found that the anti-collagen type II serum titres of the IgG2c isotype were significantly lower in mice with the C4 fragment compared with littermates and to mice with the C3 fragment. Mice with the C5 fragment (spanning from D5Mit317 (112 Mbp) to D5Mit367 (120 Mbp)) had significantly lower IgG1, IgG2c, IgG3 and total Ig serum levels compared with littermate controls (Table 3). This confirms the effect on the antibody response observed with the C2 fragment and shows that genes in this region control antibody responses to type II collagen.

Investigation of CIA development in C5 congenic mice showed that mice with one RIIISJ allele in this interval are protected from disease development compared with littermate controls (C5 congenics, incidence = 19%, mean maximum score = 24 ± 9; littermate controls, incidence = 50%, and mean maximum score = 31 ± 3; Table 4 and Figure 5b).

To further dissect the C5 region within the Eae39 locus, we bred congenic mice with overlapping fragments spanning the C5 region (C6, C9-C11) (Figure 5a). The new congenic mice were investigated for CIA and antibody responses to type II collagen. When splitting up the C5 fragment, we observed two different disease patterns. Mice with the C9 congenic fragment, which in contrast to C5 does not include the D5Mit317 marker, had a similar non-severe disease pattern to mice with the C5 fragment (Table 4, Figure 5d). Mice with the C6 fragment, covering the centromeric part but lacking the most telomeric part of the C5 fragment, developed more severe arthritis compared with mice with the C9 fragment (Table 4, Figures 5c,d). In mice with the C10 and C11 congenic fragments, a different pattern of disease development was observed because these mice were no longer protected from CIA, but instead developed more severe disease compared with littermate controls (Figures 5e and 5f, Table 4).

The anti-collagen type II antibody titre was not significantly lower in mice with the C6 and C9 fragments compared with the controls (Table 5). In the C10 and C11 congenic mice, the collagen type II antibody levels followed the disease course and the antibody concentrations were significantly higher compared with the littermate controls (Table 5).

In conclusion, when splitting up the C5 fragment into smaller intervals, we suggest a disease-controlling gene (or genes) close to the D5Mit317 marker in the upper part of the fragment. This part of C5 is shared with the disease promoting congenic fragments C10 and C11. Although not statistically significant for mice with the C6 fragment, the results from the C6 and C9 congenic mice suggest that a gene conferring protection against CIA development when one RIIIS/J allele is present, is located close to the D5Mit136 marker. In contrast to the C6 fragment, the C9 does not include the promoting gene/genes around the D5Mit317 marker, which could explain why the C9 congenic mice are more protected from disease development. Another possibility would be that there is another protecting gene close to the D5Mit367 marker, which is not present in the C6 fragment.