Kallikrein genes are associated with lupus and glomerular basement membrane-specific antibody-induced nephritis in mice and humans

Abstract

Immune-mediated nephritis contributes to disease in systemic lupus erythematosus, Goodpasture syndrome (caused by antibodies specific for glomerular basement membrane [anti-GBM antibodies]), and spontaneous lupus nephritis. Inbred mouse strains differ in susceptibility to anti-GBM antibody-induced and spontaneous lupus nephritis. This study sought to clarify the genetic and molecular factors that maybe responsible for enhanced immune-mediated renal disease in these models. When the kidneys of 3 mouse strains sensitive to anti-GBM antibody-induced nephritis were compared with those of 2 control strains using microarray analysis, one-fifth of the underexpressed genes belonged to the kallikrein gene family,which encodes serine esterases. Mouse strains that upregulated renal and urinary kallikreins exhibited less evidence of disease. Antagonizing the kallikrein pathway augmented disease, while agonists dampened the severity of anti-GBM antibody-induced nephritis. In addition, nephritis-sensitive mouse strains had kallikrein haplotypes that were distinct from those of control strains, including several regulatory polymorphisms,some of which were associated with functional consequences. Indeed, increased susceptibility to anti-GBM antibody-induced nephritis and spontaneous lupus nephritis was achieved by breeding mice with a genetic interval harboring the kallikrein genes onto a disease-resistant background. Finally, both human SLE and spontaneous lupus nephritis were found to be associated with kallikrein genes, particularly KLK1 and the KLK3 promoter, when DNA SNPs from independent cohorts of SLE patients and controls were compared. Collectively, these studies suggest that kallikreins are protective disease-associated genes in anti-GBM antibody-induced nephritis and lupus.

Figure 1. Strain-dependent gene expression differences in the renal cortex in AIGN.

(A) Anti-GBM disease was induced in 3 disease-sensitive strains (DBA1, NZW, and 129/SvJ) and 2 control strains (BALB/c and B6), after which renal cortex RNA was analyzed using DNA microarrays on day 10 of disease (i.e., 5 days after injection of anti-GBM Abs). Three biological replicates were included for each strain. The left panel shows that a total of 360 gene transcripts were differentially expressed between the study strains (>2 fold, P < 0.001). The right panel (a higher-magnification view of the boxed region on the left) shows a cluster of gene transcripts that were increased in all control strains but not in the AIGN-sensitive strains (>2-fold difference, P < 0.001), including 10 Klk genes. (B) Renal cortex gene expression differences in Klk1, Klk1b3, Klk1b5, Klk1b26, and Klk1b27 were confirmed by real-time PCR in the indicated strains, before (day 0) and after (day 10) anti-GBM Ab challenge. Each bar represents the mean of 6 samples. Similar changes were seen between B6 and NZW mice (data not shown). P values pertain to comparisons with BALB/c day 10 values. Error bars denote SD.

Figure 2. The differential renal expression of Klk in the AIGN-sensitive versus control strains was confirmed at the protein level.

BALB/c, DBA1, and 129/SvJ mice were subjected to AIGN. (A) Fourteen days after anti-GBM challenge, kallikrein protein expression was assayed in the renal cortex by Western blotting, using a rabbit anti–mouse Klk1 Ab. The bar chart below (n = 3 mice per group) shows Klk expression normalized to GAPDH (AU). *P < 0.001, compared with all other study groups. (B) Twenty-four-hour urine samples collected from these mice on days 0, 10, and 14 after anti-GBM insult were also assayed for kallikrein enzymatic activity, using the synthetic chromogenic substrate HD-Val-Leu-Arg-pNA (S-2266), as detailed in Methods. Similar differences in renal and urinary kallikrein levels were noted between B6 and B6.Sle3^z mice (data not shown). Error bars in A denote SD. In B, each dot represents data from a single mouse, and the horizontal bars denote arithmetic group means.

Figure 3. Impact of BK receptor blockade or activation on the severity of AIGN.

BALB/c mice were treated with BK receptor antagonists (B1 receptor blocker H158 [B1-R] or B2 receptor blocker H157) or PBS vehicle alone as placebo (None), using osmotic pumps, for the 14-day duration of an anti-GBM challenge and phenotyped for proteinuria (A), azotemia (B), and GN (C, B1-R blocked; and D, B2-R blocked). Original magnification, ×400. (E) Conversely, the administration of BK into 129/SvJ mice using osmotic pumps ameliorated disease after anti-GBM challenge, compared with mice treated with vehicle placebo. In A, B, and E, each dot represents data from a single mouse, and the horizontal bars denote arithmetic group means.

Figure 4. The Sle3^z locus, particularly the Sle3^z_157–158 subinterval on chromosome 7, may be responsible for the reduced Klk and enhanced nephritis susceptibility seen in NZW mice.

(A) Shown are the Sle3^z lupus susceptibility interval on chromosome 7 (Chr. 7; black denotes the interval derived from NZM2410/NZW; gray denotes B6 origin); the 4-Mb subinterval spanning D7mit157 to D7mit158 (denoted by the dashed line on right); and the cluster of Klk genes harbored within the indicated subinterval. The numbers on the right refer to the positions of respective microsatellite markers (e.g., 157 represents D7mit157). Shown also are the 24-hour urine protein excretion profiles (B), blood urea nitrogen (BUN) (C), GN pathology score (D), and renal Klk message levels (E), 14 days after anti-GBM challenge of B6, B6.Sle3^z, and B6.Sle3^z_157–158 congenics (n = 5 each). The data shown in B–D were reproduced in at least 2 additional experiments. In the second study, for example, the B6.Sle3^z_157–158 congenics exhibited significantly higher 24-hour protein levels in urine (P < 0.045) and GN score (P < 0.013) and more severe tubulo-intersitial disease (P < 0.001), compared with the B6 control (data not plotted). All statistical comparisons were made with the respective B6 controls, using the Mann-Whitney U test. In B–D, each dot represents data from a single mouse, and the horizontal bars denote arithmetic group means. Error bars in E denote SD.

Figure 5. Sequence analysis of the 5′-regulatory regions of Klk genes reveals nucleotide polymorphisms that distinguish the AIGN-sensitive strains from the control strains.

When 2 kb of the 5′-regulatory regions of Klk1, Klk1b3, Klk1b5, Klk1b26, and Klk1b27 from the indicated strains were sequenced, several SNPs/deletions were identified, as detailed in Table 2 (GenBank accession numbers EU597301–EU597324). (A) Phylogenetic analysis revealed the sequence of the AIGN-sensitive strains to be closely related, compared with the sequences of the 2 control strains. Bars represent the fraction of sequence variation. (B) Part of the nucleotide sequence of the Klk1b3 promoter (up to 200 bp upstream of transcription start site) from the different study strains indicated. Note that the B6.Sle3^z strain bears the NZW allele at Klk. (C) One kilobase of the promoter region of Klk1b3 from both BALB/c and DBA/1 strains was inserted into the pGL4 luciferase vector and transfected into COS-7 cells, and luciferase activity was assayed 24 hours later, as detailed in Methods. Each bar represents the median of triplicate values, and the error bars denote SD. Cells transfected with vectors carrying the BALB/c-derived Klk1b3 promoter showed significantly increased luciferase activity compared with cells with vectors bearing the DBA/1 promoter or vector alone (P < 0.01). Similar differences were noted when the B6 Klk1b3 promoter was compared with the B6.Sle3^z Klk1b3 promoter (data not shown).

Figure 6. KLK association in SLE patients — second validation study.

(A) The Haploview plot shows the genotyped markers in the KLK locus, from KLK1 at the centromeric limit until KLK14 at the telomeric end, as well as the linkage disequilibrium between them measured by the D prime coefficient. Blocks were defined by the solid spine method implemented in Haploview version 4.1. Dataset: UCSF (n = 595 SLE patients) plus SLEGEN (n = 689 SLE patients and n = 3,718 controls). (B) In the indicated numbers of SLE patients and healthy controls, 56 KLK SNPs were tested for disease association using logistic regression analysis, with the phenotype “SLE” as the outcome variable (shown in blue). To examine whether the risks conferred by the KLK polymorphisms were influenced by the presence of nephritis, we tested the KLK SNPs for association, considering the phenotype “nephritis” as the outcome variable. Red indicates significant differences compared with controls; green indicates significant differences between cases with nephritis and cases without nephritis calculated by a metaanalysis, in order to control for heterogeneity among the contributing clinical centers. (C) The observed linkage disequilibrium blocks across the KLK locus were tested for haplotype association, using both omnibus and haplotype-specific association statistics (T test) as implemented in PLINK. Shown are significant haplotype associations when SLE patients were compared with controls (blue), when patients with lupus nephritis were compared with controls (red), and the case-only analysis (green). Besides the SNPs indicated in B and the blocks indicated in C, none of the other SNPs/blocks shown in A showed significant disease associations. For a larger version of this figure, see Supplemental Figure 1.