Genetic Dissection of Vasculitis, Myeloperoxidase-Specific Antineutrophil Cytoplasmic Autoantibody Production, and Related Traits in Spontaneous Crescentic Glomerulonephritis-Forming/Kinjoh Mice¹

The spontaneous crescentic glomerulonephritis (CrGN)³-forming/Kinjoh (SCG/Kj) mouse has been described as an animal model of rapidly progressive CrGN and systemic vasculitis (1). This strain consists of recombinant inbred mice derived from (BXSB/Mp × MRL/Mp-lpr/lpr) F₂ and is characterized by early development of nephritis and systemic vasculitis. The characteristics of renal histology are extreme formation of crescents in most of Bowman’s capsules with diffuse endocapillary and mesangial proliferation (1, 2). Vasculitis occurs in multiple organs, including kidney, ovary, uterus, spleen, heart, and stomach (1).

The primary immunological defects responsible for the initiation and progression of disease in SCG/Kj mice are unclear. Recent studies show that the antineutrophil cytoplasmic autoantibody (ANCA) is associated with both CrGN and small-vessel vasculitis, and myeloperoxidase (MPO)-specific ANCA (MPO-ANCA) is particularly important for pathogenesis of pauci-immune vasculitis and glomerulonephritis (GN) (3). Our previous study has demonstrated that aberrant MPO-ANCA production and consequent hyperfunction of neutrophils are involved in the manifestation of CrGN and vasculitis in SCG/Kj mice (4). However, genetic predisposition, the key element in susceptibility, remains to be understood. Several mouse models of autoimmune diseases, such as systemic lupus erythematosus (SLE), have collectively contributed toward understanding the disease and have led to the definition of susceptible quantitative trait loci (QTLs) (5, 6, 7).

In this study, we identified QTLs susceptible for GN, crescent formation, vasculitis, and the production of autoantibodies, including MPO-ANCA, by establishing (B₆× SCG/Kj) F₂ intercross mice. Using a genome-wide scan, we found multiple QTLs from both parental strains that predispose to disease, including aberrant production of MPO-ANCA.

Female C57BL/6 (B₆) mice were purchased from the Shizuoka Laboratory Animal Center. Male SCG/Kj mice were bred and maintained at the animal facility of Nippon Kayaku under specific pathogen-free conditions. Female B₆ and male SCG/Kj were crossed to obtain (B₆× SCG/Kj) F₁ (BSF1) mice, and brother-sister mating of F₁ produced a total of 420 female (B₆× SCG/Kj) F₂ (BSF2) animals. These F₁ and F₂ mice, as well as parental strains, were maintained in our own animal facility. Only female mice were investigated. All procedures were approved by the Juntendo University Subcommittee on Animal Research (Tokyo, Japan), and the animal care methods and experimental protocols were based on the guidelines for animal experiments set by the National Institute of Infectious Diseases (Tokyo, Japan).

Peripheral blood was obtained from periorbital sinus. The white blood cell count was done by the MEK-6158 automatic blood cell counter (Nihon Koden). Urine was tested for proteinuria biweekly as described by Knight et al. (8) with minor modifications. BUN was measured by a kit using the urease-indophenol method (Wako Junyaku). Hematuria was tested with Uropaper II urine dipsticks (Eiken Kagaku). Serum collection was done at 12 and 24 wk of age, except for the SCG/Kj mice. Because severe CrGN was manifested in most of the female SCG/Kj mice, who died at 12–16 wk of age, their sera at 24 wk of age were not taken. Mice were sacrificed at 24 wk of age or when proteinuria was >200 mg/dl and hematuria was >20 RBC/μl (1+).

Serum levels of total Ig and IgG-class autoantibodies to DNA and chromatin were determined by ELISA. Ninety-six-well flat-bottom plates (Immulon 2; Dynatech Laboratories) were coated with 0.001% protamine sulfate followed by DNA or chromatin. dsDNA was obtained by digestion of calf thymus DNA (Sigma-Aldrich) with S1 nuclease (Seikagaku Kogyo), followed by fractionation on a hydroxylapatite column. ssDNA was obtained by heat denaturation of calf thymus DNA. Chromatin was prepared as described (9). Briefly, nucleosomes were isolated by solubilizing chromatin from purified chicken erythrocyte nuclei with micrococcal nuclease. The solubilized chromatin was fractionated into sucrose gradients that were analyzed for monomers using electrophoresis, and the appropriated fractions were dialyzed and pooled. To measure the amount of immunoglobulins, plates were coated with goat anti-mouse IgM or IgG Abs. Wells were blocked by 50% FCS for 1 h. After washing, the diluted sera were applied to the plates. After a 1-h incubation at room temperature, plates were washed in PBS with 0.05% Tween 20. After washing, appropriately diluted peroxidase-conjugated goat anti-mouse γ Abs were added. The preparations were then incubated for 1 h at room temperature and washed, and the substrate (o-phenylenediamine dihydrochloride) diluted in 0.1 M citrate phosphate buffer (pH 5.0) with 0.5 μg/mg H₂O₂ was added. The reaction was stopped by adding 2.5 N H₂SO₄, and the OD was measured at 490 nm using a microplate reader (Viento; Dainippon Pharmaceutical). The DNA- and chromatin-binding activities were expressed in units referring to a standard curve obtained by serial dilutions of a standard serum pool from 7- to 9-mo-old (New Zealand Black (NZB) × New Zealand White (NZW)) F₁ mice containing 1000 U/ml. The amounts of total IgM and IgG were calculated using a standard curve obtained by affinity-purified serum IgM and IgG derived from (NZB × NZW) F₁ mice.

MPO-ANCA levels were measured as described (4). Briefly, recombinant mouse MPO was coated onto an ELISA plate (TS plate; Toyoshima) overnight at 4°C. The plate was blocked, and mouse serum (×50 dilution) was added for 1.5 h at room temperature. Alkaline phosphatase-labeled anti-mouse IgG Ab (×1000 dilution) was added and allowed to react for 2 h at room temperature. Afterward, p-nitrophenyl phosphate as the alkaline phosphatase substrate was added at a concentration of 1 mg/ml and incubation at room temperature, the absorbance at 405 nm was measured by a model LFA-096 automatic analyzer (Jasco). The titer of MPO-ANCA in mouse sera was described as rabbit anti-mouse MPO IgG (μg/ml).

At autopsy, the spleen was weighed and the kidneys were fixed in 10% formalin in 0.01 mol/L phosphate buffer (pH 7.2) and embedded in paraffin. They were stained with H&E and periodic acid-Schiff for histopathological examinations by light microscopy.

Kidneys from one individual were dissected into more than four sections. Four independent kidney sections were observed, and 30 glomeruli in each section were evaluated so that 120 glomeruli were examined for one individual. Among these, 20 glomeruli with both vascular and tubular poles were evaluated for endocapillary and mesangial proliferative lesions and sclerotic lesions of mesangial areas. The trait for GN was expressed as the percentage of glomeruli with more than one of the following three glomerular lesions: endocapillary proliferation, mesangial proliferation, and mesangial sclerosis.

The trait for crescentic formation was expressed as numbers of glomeruli with cellular and/or fibrous crescents among 120 glomeruli. Vasculitis was expressed as numbers of small vessels with granulomatous vasculitis in four independent kidney sections. Granulomatous vasculitis in this study is defined as vascular lesions with at least one of following findings: perivascular infiltration of lymphocytes, destruction of the vascular wall, and/or myointimal thickening of the vascular wall.

Genotypes were determined by PCR using selected simple sequence length polymorphism markers purchased from either Research Genetics or Invitrogen Life Technologies. D19MIT87 was used for the genotyping of the Fas gene, because it was located within 1 cM of chromosome 19 in the mouse genome database consensus map (The Jackson Laboratory; http://www.informatics.jax.org). Genotyping for the polymorphic FcγRIIB promoter region was done by PCR using a previously described primer pair (10). Genomic DNA of mice were extracted from tail samples stored at −70. Primers flanking chromosomal microsatellite markers (forward primer labeled on the 5′ end with the fluorescent dyes FAM, VIC, PET, or NED and reverse primers) were purchased from Applied Biosystems. A PCR mixture (7 μl) contained 110 nM each primer, 0.23 mM each dNTP, 16 mM Tris-HCl (pH 8.3), 41 mM KCl, 2.7 mM MgCl₂, 2.0 μg/ml genomic DNA, and 0.03 U/μl Taq polymerase (Invitrogen Life Technologies). The PCR mixtures were distributed on MicroAmp optical 96-well reaction plates (Applied Biosystems) using a PT-100 Molecular Biology Station (System Biotics), and amplifications were conducted using PCR System 9700 thermal cyclers (Applied Biosystems). The reaction consisted of initial denaturation at 92°C for 5 min followed by 35 cycles of 92°C for 1 min, 56–58°C for 1.5 min, and 72°C for 2 min, and final incubation at 72°C for 7 min. PCR products were analyzed using an Applied Biosystems 3100 genetic analyzer and genotyped with GENESCAN and GENOTYPER software (Applied Biosystems).

Comparison of renal function (BUN), serum levels of IgM and IgG class immunoglobulins, serum levels of IgG-class autoantibodies (anti-ssDNA, anti-dsDNA and antichromatin antibodies), MPO-ANCA, three histopathological traits in kidneys and spleen weight, and survival weeks among strains were performed with the Mann-Whitney U test or ANOVA. Associations between traits in F₂ mice were determined by correlation coefficients with p values derived from Fisher’s transformation. For detection of QTLs for crescent formation, χ² tests were conducted. In χ² tests, F₂ progenies that had more than three glomeruli with crescents were regarded as positive. All analyses were performed using StatView, version 4.0 (Abacus Concepts) or Microsoft Excel.

The linkage map for the BSF2 intercross was created using MapManager QTX (Dr. K. F. Manly, University of Tennessee Health Science Center (Memphis, TN); http://www.mapmanager.org/mmQTX.html). Interval mapping for QTL detection was done by MapManager QTX in a free regression model. Likelihood ratio statistics were converted to conventional base-10 logarithm of odds (LOD) scores. To establish suggestive and significant threshold values, permutation tests were performed using MapManager QTX as previously described (11) with 1000 permutations of the data. Only loci with suggestive or significant linkages were shown in the figures and tables, and only those that were significant were named.

As summarized in Table I, parental SCG/Kj and B₆ mice, as well as BSF1 and BSF2 mice, were examined for levels of BUN as a marker of renal function and autoantibodies at 12 and 24 wk of age. SCG/Kj presented significantly higher levels of total IgM and IgG class Ig (p < 0.05 and p < 0.0001, respectively), anti-ssDNA Ab (p < 0.0001), anti-dsDNA Ab (p < 0.0001), antichromatin Ab (p < 0.0001), and MPO-ANCA (p < 0.0001) than B₆ at 12 wk of age. Although comparison of BUN between these two parental strains was not significant, marked renal dysfunction (127 mg/dl in BUN) was already observed in one SCG/Kj individual that was not observed in B₆ mice until the end of experiment. Levels of total IgG, anti-ssDNA Abs, and anti-dsDNA Abs were significantly higher in BSF1 mice than in B₆ mice (in all comparisons, p < 0.005), but these levels in BSF1 mice were much lower than those in SCG/Kj mice. Mean levels of antichromatin Ab and MPO-ANCA were higher in BSF1 mice than in B₆ mice but not significant. As for total IgM, the level was slightly lower in BSF1 mice than in SCG/Kj mice but was not significant. These facts suggest that the mode of inheritance of Ig and autoantibody production is incomplete dominance with dominance effects of various degrees.

Kidneys of each strain were evaluated in terms of endocapillary and mesangial proliferation, mesangial sclerosis, crescent formation, and vasculitis. Endocapillary and mesangial proliferation and mesangial sclerosis were recognized as findings of GN; GN and crescent formation were regarded as independent traits. Crescent formation was expressed as number of glomeruli with cellular and/or fibrous crescent because this trait had less penetrance than GN. As summarized in Table II, GN findings were manifested in 12 of 13 (92%) individuals of SCG/Kj mice, and mean glomeruli with GN amounted to 62.3% of total glomeruli in one individual. Kidneys of B₆ mice presented virtually no GN findings (p < 0.0001). There was a tendency for the kidneys of F₁ mice to present more severe GN and more crescents than those of B₆ mice, but the degree was about one-third of that of SCG/Kj. These facts suggest that the mode of inheritance of GN and crescent formation is incomplete dominance, as seen in autoantibody production.

Histopathological findings of vasculitis in SCG/Kj, BSF1, and BSF2 mice were all similar to that of MRL/lpr. As described in the studies of Nose et al. (12), vasculitis in these strains was characterized by granulomatous changes with marked perivascular infiltration of lymphocytes, destruction of external lamina, and myointimal thickening of vascular wall. It has been pointed out (6) that the grading systems for evaluation of murine vasculitis are not ideal parameters in QTL analyses, even after conversion to various mathematical functions. For better suitability as a variate for QTL analyses, we expressed vasculitis as the number of inflammatory lesions observed in four independent kidney sections. Twenty small vessels were observed in each of four sections for a total of 80 vessels in a single individual. The mean value of vasculitis in SCG/Kj was significantly higher than that in B₆ mice (p < 0.0001). Vasculitis in BSF1 mice was significantly more severe than in B₆ mice (p < 0.05) but much less severe than in SCG/Kj mice. Again, this finding indicates that vasculitis is inherited in the mode of incomplete dominance.

Spleen weight seemed to be inherited in the same mode as other traits. Spleens were significantly heavier in SCG/Kj and BSF1 than in B₆ mice (p < 0.0001 and p < 0.01, respectively), but spleen weights in BSF1 were about one-third of those in SCG/Kj mice (p < 0.0001).

Disease traits in BSF2 mice were distributed in a wide range and were generally between parental extremes. A minority of BSF2 mice exhibited trait values outside parental ranges. Presumably, this resulted from the recombination of genetic components in F₂ mice, that is, the inheritance of unique combinations of SCG/Kj and B₆ alleles.

In BSF2 mice, serological and histopathological traits were statistically examined by association study. Histopathological traits were also tested to elucidate correlations (Table III). All Ig and autoantibody levels were significantly (p < 0.005) correlated with histopathological traits, with a positive correlation coefficient (r) ranging from 0.139 to 0.617 (data not shown). MPO-ANCA, considered one of the major causal factors for pathogenesis of human polyarteritis and pauci-immune-type CrGN (3), significantly correlated with crescent formation (r = 0.159). MPO-ANCA exhibited better correlation with vasculitis (r = 0.413).

Correlations among three histopathological traits were all statistically significant (p < 0.0001). All correlation coefficients were >0.403. GN and vasculitis correlated most strongly (r = 0.628) among all association studies. Correlations between crescent and GN and between crescent and vasculitis were less strong (r = 0.445 and 0.403, respectively).

Survival of mice was expressed as total lifespan by the week. All five serological traits, as well as three histopathological ones, significantly correlated with lifespan (p < 0.0001 for all correlations, r = −0.187 to −601; Table III and data not shown). The trait that exhibited the strongest negative correlation with lifespan among histopathological traits was crescent formation (r = −0.601).

We genotyped 420 BSF2 intercross mice to identify QTLs contributing to disease phenotypes. All intercross mice analyzed were female. Seven hundred twenty-seven microsatellite markers were analyzed for size polymorphism between B₆ and SCG/Kj mice, and 158 markers were informative (21.1% of total markers). A total of 102 polymorphic microsatellite markers were used to construct a linkage map designed to cover the whole mouse autosomal genome. Because of a failure in some cases to identify informative markers, gaps of >20 cM were present in four chromosomal regions. These four gaps were located on chromosome 7 between D7MIT80 and D7MIT207 with a distance of 20 cM, on chromosome 9 between D9MIT42 and D9MIT53 with a distance of 29 cM, on chromosome 14 between the centromere and D14MIT5 with a distance of 21.7 cM, and on chromosome 15 between D15MIT26 and D15MIT15 with a distance of 21.5 cM. Two hundred seventy-two markers were investigated to shorten these four and other minor gaps, but they were all noninformative. As a result, markers were distributed throughout the autosomes such that 85% of the genome was within 20 cM of an informative marker. The low frequency of informative markers in BSF2 is probably because SCG/Kj originated from crossing BXSB and MRL/lpr (1). Because ∼50% of the BXSB genome (13) is from B₆ mice, markers on such a common chromosomal segment are expected to be nonpolymorphic.

We observed that the Fas locus on chromosome 19 was the major gene controlling all disease phenotypes (range of LOD from 9.7 for 12-wk MPO-ANCA to 105.0 for 24-wk ssDNA Ab, and range of percentage variance from 11% for 12-wk antichromatin Ab to 70% for 24-wk anti-ssDNA Ab; Table IV). In these analyses, mice homozygous for the lpr alleles of the Fas gene (lpr/lpr) exhibited much higher or more severe traits compared with those of mice heterozygous (+/lpr) or homozygous for the B₆ alleles (+/+). Several traits were revealed to be significantly higher in mice heterozygous for Fas than in mice homozygous for the B₆ alleles (p < 0.05 for the 12-wk anti-dsDNA Ab level; p < 0.005 for the 12-wk and 24-wk total IgM and 24-wk anti-dsDNA Ab levels; and p < 0.0005 for the 24-wk total IgG and 24-wk MPO-ANCA levels and splenomegaly). These facts suggest that the lpr mutation of the Fas is not completely recessive; therefore, we subdivided the BSF2 mice into seven “Fas cohorts” according to the Fas genotypes and their combination (Table V). As shown in Table V, the cohort symbolized by S consists of F₂ individuals homozygous for alleles from SCG/Kj (lpr/lpr). The B cohort consists of F₂ individuals homozygous for alleles from B₆ (+/+). The F cohort consists of F₂ individuals heterozygous for B₆ and SCG/Kj alleles (+/lpr), the same Fas genotype as (B₆× SCG/Kj) F₁. Cohorts symbolized as SF, BF, and SB (Table V) are those in which two of the three cohorts above are combined, and the SFB cohort consists of all F₂ individuals. The practical meanings of the subcohorts are as follows. First, because cohorts are based on the alleles of the Fas gene, QTLs that demonstrate some of the epistatic effects with certain genotype(s) of Fas will be revealed in specific Fas cohort(s). Second, combined cohorts lead to be a sort of “extreme-phenotype analysis” (14) and are good for discovering minor QTLs. Third, combined cohorts possibly have better resolution and power for finding minor QTLs because they contain more progenies.

There were two multifunctional QTLs on chromosome 1. As shown in Table IV and Fig. 1, one of these QTLs was the region between D1MIT11 and D1MIT102 (an interval of ∼14 cM, the position represented by D1MIT191), and the other was the region between D1MIT14 and D1MIT166 (an interval of 18 cM, the position represented by D1MIT15).

The distal QTL on chromosome 1 adjacent to D1MIT15 exhibited significant linkage to the 12-wk total IgG level, the 12-wk anti-ssDNA Ab production, the 12-wk anti-dsDNA Ab production, and the shortened lifespan in the S cohort (QTL exhibited LOD of 4.4–6.5 and accounted for 16–22% of the variance). This QTL was also significantly linked to the 24-wk total IgG level, the 12-wk antichromatin Ab level, GN, vasculitis, and splenomegaly in the BF cohort (LOD of 3.5–7.0 and 5–20% of the variance). This area on chromosome 1 is designated Scg-1; it also exhibitedsuggestive linkage to the 24-wk anti-ssDNA Ab level, the 24-wk anti-dsDNA Ab level, and the 24-wk antichromatin level. These traits were all significantly more severe in mice homozygous for the D1MIT15 allele from SCG/Kj, but mice with heterozygous alleles also exhibited significantly stronger traits than mice homozygous for the allele from B₆ (Fig. 2, D and F, and data not shown), meaning that this QTL is from SCG/Kj and its effect is additive. Because interval mapping analyses did not identify any QTLs for crescent formation other than the Fas gene, we tried χ² analysis to find loci for the crescent. In the S cohort, F₂ individuals that had more than three crescents in their kidneys were defined as positive for crescent formation. χ² tests for independence on the contingency table by crescent formation and genotypes of markers on the whole genome were performed. As a result, a suggestive or significant linkage to crescent formation was observed at D1MIT15 (χ² = 18.9; p = 7.7 × 10⁻⁵) on chromosome 1 (Table VI).

The second QTL that controlled traits including the aberrant production of MPO-ANCA was on the middle of chromosome 1 around D1MIT191, 25 cM from D1MIT15. This QTL was significantly linked to the 12-wk total IgG level, 12-wk anti-ssDNA Ab production, 12-wk anti-dsDNA Ab production, and 12-wk MPO-ANCA production in the SB cohort (LOD of 4.8–6.1 and 11–13% of the variance). It also presented significant linkage to splenomegaly in the B cohort (LOD of 4.6 and 23% of the variance). Suggestive linkage to GN and lifespan was also observed. This area on chromosome 1 was designated Scg-2. A one-log support interval was determined based on the likelihood of 12-wk anti-ssDNA Ab. It is noteworthy that splenomegaly presented significant linkage to Scg-2 in the cohort without the Fas^lpr/lpr or Fas^+/lpr individual (B cohort), suggesting that Scg-2 exerted some effect, if only partial, independently of the Fas gene. Although Scg-1 and Scg-2 were both multifunctional genes with partially common functions, their spectra of effects were different in the following way: Scg-1 was suggested as influencing the antichromatin Ab level, vasculitis, and crescent formation, but these traits were not linked to Scg-2 even at suggestive levels. More surprisingly, the production of MPO-ANCA at 12 wk of age was exclusively controlled by Scg-2 (Fig. 1,A). The 12-wk MPO-ANCA presented significant linkage to Scg-2 in the SFB, S, and SF cohorts, as well as in the SB cohort, but did not present even suggestive linkage to Scg-1 in any of these cohorts. A genome-wide scan did not identify any significant QTLs predisposing 12-wk MPO-ANCA except Fas and Scg-2. These facts suggest that Scg-1 and Scg-2 are different genes, although both are on chromosome 1. A one-log support interval of MPO-ANCA analysis defined about one-half of the Scg-2 interval, designated Man-1 from MPO-ANCA. Scg-2 was inherited from SCG/Kj. Scg-2 was inherited recessively, because mice with the ss genotype for D1MIT191 exhibited much strongertraits than those of the bs or bb genotypes. This result is shown in Fig. 2,A for the 12-wk MPO-ANCA level and is also the case for the 12-wk total IgG, the 12-wk ssDNA Ab and 12-wk dsDNA Ab levels, and GN (Fig. 2, C and E, and data not shown). The difference in modes of inheritance also supports the notion that Scg-1 and Scg-2 are distinct.

Seven to eight non-Fas QTLs were identified for aberrant production of IgM- and/or IgG-class immunoglobulins on chromosomes other than chromosome 1 (chromosomes 2, 4, 6, 7, 11, 13, and 17; Table IV). Loci on chromosomes 6 and 13 were linked to the hyperproduction of both IgM and IgG, and the rest were linked to that of either IgM or IgG.

One significant interval, linked to the 24-wk total IgG level, was identified on chromosome 2 between D2MIT26 and D2MIT213 (LOD of 4.8 in the F cohort). It was derived from SCG/Kj, accounted for 10% of the variance (Table IV), and was designated Scg-3 (Fig. 3,A). The mode of inheritance of Scg-3 appeared to be recessive (Fig. 4 A).

Total IgM levels at 12 and 24 wk of age were significantly linked to the telomeric region on chromosome 4 in the BF cohort (LOD of 5.7–6.8 and 8–10% of the variance; Table IV). A broad area distal to D4MIT178 exhibited a significant level of LOD score, with a peak LOD score at D4MIT71, the most telomeric marker in this study (Fig. 3,B). This area is designated Sxb-1 (SCG/Kj cross B₆-1), inherited from B₆ (Table IV). It is noteworthy that Sxb-1 was associated only to IgM-class Ig levels but not to IgG class levels or to specific autoantibody levels. As shown in Fig. 4 B, Sxb-1 was additively inherited.

On chromosome 6 there was an area of suggestive linkage (LOD of 3.4 in the F cohort) to the 12-wk total IgM level and an area of significant linkage (LOD of 4.6 in the F cohort) to the 12-wk total IgG level (Table IV and Fig. 3,C). The patterns of LOD curves for these traits were similar, suggesting that the QTL was identical for each. It was from B₆ and accounted for 7 and 9% of variances of total IgM and IgG, respectively. The area from D6MIT88 to D6MIT16 is designated Sxb-2 (Fig. 3,C). Sxb-2 inherited additively (Fig. 4 C).

The 12-wk total IgM levels were significantly (LOD of 5.4 in the F cohort) linked to a SCG/Kj-derived QTL on mid-proximal chromosome 7 (designated Scg-4; Table IV and Fig. 3,D), accounting for 11% of the variance. The mode of inheritance appeared to be recessive (Fig. 4 D).

On the mid-proximal region of chromosome 13 (Table IV and Fig. 3,F) there was an interval that significantly linked the12-wk total IgM level (LOD of 4.8 and 17% of the variance in the S cohort), the 24-wk total IgM level (LOD of 6.7 and 27% of the variance in the S cohort), and the 24-wk total IgG level (LOD of 4.4 and 6% of the variance in the BF cohort). This was a SCG/Kj-derived recessive QTL (Fig. 4,F). Values of the LOD score over a significant level ranged widely from D13MIT60 to D13MIT144, an interval of ∼32 cM (designated Scg-5; Fig. 3 F).

On chromosome 17, significant linkages to 12-wk total IgG (LOD of 5.8 and 8% of the variance in the BF cohort) and 24-wk total IgG (LOD of 5.5 and 8% of the variance in the BF cohort) were observed for the region adjacent to the H2 gene, a murine MHC (Fig. 3,G). D17MIT21, located between H2-K and H2-A, represented this region. F₂ progenies with bb and bs genotypes for D17MIT21 exhibited significantly higher levels of 12- and 24-wk total IgG than those of mice with the ss genotype. This is compatible with the dominant mode of inheritance (Fig. 4,H). The region around D17MIT21 is designated Sxb-3 (Fig. 3 G).

One minor QTL of suggestive linkage to the 12-wk total IgG level was detected on chromosome 11 around D11MIT58 (Table IV). It was from SCG/Kj and was inherited in an additive manner (data not shown).

QTLs for aberrant production of autoantibodies and histopathological phenotypes were searched on the genomes, except chromosomes 1 and 19. On chromosome 10 there was an interesting QTL that controlled the production of anti-ssDNA Ab, MPO-ANCA, and the manifestation of GN. Significant linkages were shown for 12- and 24-wk anti-ssDNA Ab (LOD of 3.8–4.5 and 9–10% of the variance in the SB cohort; Table IV and Fig. 3,E) and for the 24-wk MPO-ANCA level (LOD of 3.6 and 9% of the variance in the SB cohort; Table IV and Fig. 3,E). Suggestive linkage was also observed for GN (LOD of 3.2 and 8% of the variance in the SB cohort). All of these traits shared their patterns of LOD curves, and peaks of LOD curves coincided at the position adjacent to D10MIT3 (Fig. 3,E). All linkages were observed in the SB cohort, suggesting that this QTL needs the lpr/lpr allele of the Fas gene for its exertion. Intriguingly, F₂ mice with the bs genotype for D10MIT3 presented stronger traits than mice with ss or bb. This was the case for all three traits, so that some epistatic effect was suggested between SCG/Kj and the B₆ alleles of this QTL (Fig. 4 E). The region around D10MIT3 is designated Sxb-4 and Man-2, because it is one of only two non-Fas QTLs for MPO-ANCA production.

A QTL for vasculitis was identified on chromosome 17 (D17MIT21; LOD of 3.2 and 5% of the variance in the BF cohort), encompassing a region overlapping Sxb-3. As opposed to serological traits, F₂ mice with bs genotype for D17MIT21 were most severe in vasculitis (Fig. 4 G). Another QTL for GN was around D17MIT88 (LOD of 2.9 in the F cohort), ∼11 cM distal to D17MIT21. This QTL was derived from B₆ and appeared to be inherited additively (data not shown).

Two loci associated with antichromatin Ab production were identified on chromosomes 8 and 12. The chromosome 8 locus mapped to the mid-proximal portion around D8MIT189 and was linked to 12-wk antichromatin Ab (LOD of 3.5 and 13% of the variance in S cohort; Table IV). The susceptible allele was inherited from SCG/Kj in a recessive manner (data not shown). Another QTL of possible association with 24-wk antichromatin Ab production (LOD of 3.1 and 5% of the variance in the BF cohort; Table IV) was detected on mid-chromosome 12 around D12MIT214. This QTL was from B₆ and was inherited recessively (data not shown).

For GN, a locus of significant linkage was mapped to mid-chromosome 16 (D16MIT4; LOD of 3.9 and 10% of the variance in the SB cohort). Because the ss and bs genotypes equally conferred higher susceptibility to the GN than the bb genotype, the pathological allele was derived from SCG/Kj in a dominant manner (data not shown).

In this study, we used a panel of (B₆× SCG/Kj) F₂ intercross mice to define non-Fas^lpr susceptibility loci for hypergammaglobulinemia, the aberrant production of autoantibodies including MPO-ANCA, GN, vasculitis, splenomegaly, and shortened lifespan. Fourteen to fifteen QTLs were linked to those traits. Ten significant QTLs were identified on chromosomes 1, 2, 4, 6, 7, 10, 13, 16, and 17. Susceptibility loci inherited from SCG/Kj mice were designated Scg-1 to Scg-5, and loci inherited from B₆ mice were designated Sxb-1 to Sxb-4. Two loci on chromosomes 1 and 10 predisposing production of MPO-ANCA were designated Man-1 and Man-2. To our knowledge, this is the first paper defining QTLs as susceptible for aberrant production of MPO-ANCA.

There were two SCG/Kj-derived pleiotropic QTLs on chromosome 1, Scg-1 and Scg-2. Scg-1 was linked to hyperproduction of IgG, anti-DNA Ab, antichromatin Ab, GN, crescent formation, vasculitis, splenomegaly, and shortened lifespan. It was mapped to the previously reported positions of Sle1 (NZW derived; see Ref.5), Lbw7, Nba2 (NZB derived; see Refs.15, 16), and Bxs3 (BXSB derived; see Ref.17). Those QTLs are susceptible to various immunological traits, including antinuclear Ab production, splenomegaly, GN, and hypocomplementemia (5, 15, 16). This locus has been defined (18) for initial spontaneous loss of immunologic tolerance because of its wide variety of traits. One characteristic of Scg-1 is that it was not associated with the total IgM level at a young (12 wk) or old (24 wk) age. This observation suggests that Scg-1 influences not only activation but also differentiation of B cells. Candidate genes for the Scg-1-interval are Tnfsf6, Cd3z, and Fcgr2b/3. The Fas ligand encoded by Tnfsf6 is a type II membrane protein on activated T cells and binds to Fas (19). The Fas ligand in gld mice carries a point mutation and causes lymphadenopathy and autoimmune diseases similar to those of MRL/lpr mice (20). Cd3z is the gene for the CD3 ζ-chain, the principal signal transduction element of the TCR (21). Abnormal expression of the TCR/CD3 complex on T cells in human SLE patients has been reported (22). Polymorphism(s) of all these genes possibly result in alteration of T cell function and survival of autoreactive lymphocytes. Fc receptors for IgG (FcγR) have important functions in the activation and down-modulation of immune responses. Balanced signaling through activating (FcγRIII) and inhibitory (FcγRIIb) FcγRs intimately regulates the activity of various cells in immune system (23, 24). There is evidence that reduced expression of inhibitory FcγR on B cells is observed in SLE-prone mouse strains (NZB, BXSB, SB/Le, and MRL) due to the polymorphic promoter regions of the Fcgr2b gene (10, 25). Moreover, expression of activating FcγRIII is abnormally high in SLE-prone mice such as MRL mice (25). Down-regulation of FcγRII on B cells affects the extent of IgG-class Ab response to T cell-dependent Ags (10). FcγRIIb-deficient Fas^lpr/lpr is reported to be sufficient for the development of murine SLE even in a B₆ background (26). Single nucleotide polymorphisms of the FCGR2B gene associate with human SLE (27). Hence, Fcgr2b and Fcgr3 are attractive candidate genes for autoimmunity in SCG/Kj mice.

The second SCG/Kj-inherited chromosome 1 locus, Scg-2, was linked to total IgG level, the production of IgG-class autoantibodies, GN, splenomegaly, and shortened lifespan. Scg-2 is located in the middle of chromosome 1 and is proximal to Sbw1 (NZB derived; see Ref.15), which is linked to splenomegaly, and Bxs2 (BXSB derived; see Ref.17), which is linked to GN and autoantibody production. Scg-1 and Scg-2 are ∼25 cM apart and are likely to be two distinct genes, because their spectra of controlling traits and mode of inheritance were different. The most striking difference is that only Scg-2 was linked to aberrant production of MPO-ANCA. This finding allows us to postulate that in the Scg-2 interval are two different genes, one of which controls propagation and differentiation of lymphocyte; the other (Man-1) is related to granulocytes, the major origin of released MPO. Candidate genes for Man-1 are Daf1/2 and Serpinb2. Daf1/2 are genes encoding decay-accelerating factor (DAF), a C3 convertase inhibitor preventing complement-mediated autologous attack. DAF deficiency in human is known as paroxysmal nocturnal hemoglobinuria, a disease of spontaneous hemolysis due to an uncontrolled complement system. Human neutrophils, as well as erythrocytes, express DAF on their surface, and the expression is doubled when activated (28, 29). The lysis of neutrophils from paroxysmal nocturnal hemoglobinuria patients is increased at least in vitro (30). Taken together, it is possible that polymorphism(s) of Daf genes result in the fragility of neutrophils, the increased release of MPO, and more opportunity for MPO to be presented as an autoantigen. Serpinb2 is also an intriguing candidate. It is the gene encoding plasminogen activator inhibitor type 2 (PAI-2), a member of the serine proteinase inhibitor (serpin) genes that exhibit inhibition toward the urokinase-type plasminogen activator (31). PAI-2 is detected mainly in monocyte/macrophage and modulates their functions such as inhibition of apoptosis and altered expression of the adhesion molecule and DAF (31, 32, 33). Because neutrophils express PAI-2 as well as monocytes in the inflammatory state and contribute to the persistence of fibrin and localization of infection (34), these roles of PAI-2 in inflammation possibly result in a more efficient presentation of MPO as an autoantigen in this cytokine milieu.

Scg-3 (chromosome 2), Scg-4 (chromosome 7), and Scg–5 (chromosome 13) are SCG/Kj-derived QTLs and are linked not to autoantibody production but to elevated total Ig levels, suggesting their influence on polyclonal B cell activation. Scg-3 on chromosome 2 may overlap with the MRL-derived QTL reported by Gu et al. (35). A possible candidate gene is Il1, because it can influence IgG production in human SLE (36). Several reports of murine QTLs for lupus are on proximal-mid chromosome 7: NZB-derived Sle3 (5), Lbw5 (15), Nba5 (37), and MRL-derived Lmb3 (38). The positional candidate gene for Scg-4 is Cd22, a negative regulator for the BCR signal, and Bax, a proapoptotic protein expressed in lymphocytes. Both of these molecules are described as related to lupus (39, 40). Several authors (37, 41) have mapped QTLs as controlling the production of nephritogenic gp70 autoantigen and its immune complex on proximal-mid chromosome 13. Yoshiki et al. (42) reported that gp70 was related to the pathogenesis of the immune complex GN of New Zealand mice. Scg-5 may represent these QTLs, because it is linked to GN and Ig production (Table IV and Fig. 3 F), and both BXSB and MRL mice are strains of high gp70 production (43).

Sxb-1 (chromosome 4), Sxb-2 (chromosome 6), Sxb-3 (chromosome 17), and Sxb-4 (chromosome 10) are B₆-derived QTLs and were linked to various traits, including hypergammaglobulinemia, autoantibody production, GN, and vasculitis. NZB-derived loci (Sle2 (5), Lbw2 (15) and Nba1 (44)), BXSB-derived loci (Acla-2 (45) and Lxw1 (46)), and MRL-derived loci (Arvm2 (47) and Asm2 (48)) were previously mapped to chromosome 4. Of interest, Lmb1 (38) is an additively inherited B₆-derived locus linked to lymphadenopathy and anti-DNA Ab production defined by using (MRL-Fas^lpr× B₆-Fas^lpr) F₂. Sxb-1 and Lmb1 share their mode of inheritance and maximal likelihood locations on chromosome 4. Thus, Sxb-1 may represent some of these QTLs, especially Lmb1. Multiple candidate genes are to be considered, such as Ifa, C1q, and Tnfr-2. IFN-α can break tolerance and initiate autoimmunity under prolonged expression (49). C1q deficiency causes autoimmunity by impairment of the clearance of apoptotic cells (50). Polymorphism of TNFR type II is causative of augmented IL-6 production, and such a polymorphism is associated with human SLE (51). These findings suggest that Ifa, C1q, and Tnfr-2 on chromosome 4 are possible candidates for Sxb-1. Sxb-2 on chromosome 6 is linked to both IgM and IgG levels, suggesting harbored T cell-activating candidate(s) as well as B cell activators. Tcrb is within the Sxb-2 interval and can be a candidate, because Vβ repertoire skewing of TCR is observed in both murine and human SLE (52, 53). Sxb-3 is such a multifunctional QTL that mice of heterozygous alleles and/or mice homozygous for B₆ allele exhibit stronger traits than those of SCG/Kj homozygous alleles. The candidate gene for Sxb-3 is H-2, the murine MHC. B₆, one of parental strains used in this study, carries the H-2^b haplotype. Because of a defect in the Ea gene, mice with the H-2^b haplotype do not express class II E molecules, the importance of which is shown in the SLE-prone BXSB strain carrying the H-2^b haplotype. The development of disease is more accelerated in BXSB.H-2^b mice than in BXSB.H-2^d mice (54). The transgene of E molecules in animals with H-2^k/b lessens the disease phenotypes (55). These facts are consistent with the result that heterozygous H-2^b/k and/or homozygous H-2^b mice exhibited more severe autoimmune phenotypes in our study.

Sxb-4 (Man-2) is also a multifunctional QTL, such that mice of the heterozygous genotype exhibit stronger traits than those with either homozygous genotype. This locus mapped centromeric to previously defined MRL-derived Lmb4 (38) and Asm1 (48) and is possibly distinct from these. Candidate genes for Sxb-4 are Ifngr1 and Myb; Ifngr1 encodes IFN-γ receptor 1. The frequency of heterozygous amino acid polymorphism (V14M) of IFNGR1 in human SLE patients is significantly higher than in the control population (56). c-myb expression and aberrant lymphocyte proliferation are associated in MRL/lpr mice (57). Each of these genes possibly contributes to the development of disease through abnormal lymphocyte proliferation or altered cytokine milieu.

The localization of disease-susceptibility genes in the murine model provides important information for the prediction of loci contributing to human disease, because linked genes on chromosomes are conserved between mice and humans. The further characterization of the disease susceptibility loci identified in this work will be helpful in focusing future human studies on specific syntenic intervals, leading to the definition of human susceptibility genes.

We thank Dr. Okio Hino, Department of Pathology, and Dr. Yasuhiko Tomino, Department of Nephrology, Juntendo University School of Medicine (Tokyo, Japan). We also thank N. Ohtsuji, M. Ohtsuji, K. Sasahara, and Dr. Y. Shida for their skillful technical assistance and Dr. Kyogoku for the use of the SCG/Kj mice.

The authors have no financial conflict of interest.