Conserved 33-kb haplotype in the MHC class III region regulates chronic arthritis

Significance

The role of the MHC region has been a long-standing issue in chronic inflammatory diseases, such as rheumatoid arthritis, and it has not been possible to identify the underlying specific polymorphism. Here, we provide evidence that some of the MHC association must be explained by how closely linked genes operate together as haplotype blocks. We identified a conserved haplotype, Ltab-Ncr3, comprising five genes (lymphotoxin α and β, Tnf, leukocyte-specific transcript 1, and natural cytotoxicity-triggering receptor 3) within MHC class III, regulating arthritis. We found significant coexpression of the Ltab-Ncr3 genes, indicating how these genes may work together as a haplotype. Furthermore, haplotype-specific differences in Ltab-Ncr3 gene expression and alternative splicing correlate remarkably to susceptibility to arthritis. Our data show that a conserved haplotype within MHC class III regulates arthritis development.

Abstract

Genome-wide association studies have revealed many genetic loci associated with complex autoimmune diseases. In rheumatoid arthritis (RA), the MHC gene HLA-DRB1 is the strongest candidate predicting disease development. It has been suggested that other immune-regulating genes in the MHC contribute to the disease risk, but this contribution has been difficult to show because of the strong linkage disequilibrium within the MHC. We isolated genomic regions in the form of congenic fragments in rats to test whether there are additional susceptibility loci in the MHC. By both congenic mapping in inbred strains and SNP typing in wild rats, we identified a conserved, 33-kb large haplotype Ltab-Ncr3 in the MHC-III region, which regulates the onset, severity, and chronicity of arthritis. The Ltab-Ncr3 haplotype consists of five polymorphic immunoregulatory genes: Lta (lymphotoxin-α), Tnf, Ltb (lymphotoxin-β), Lst1 (leukocyte-specific transcript 1), and Ncr3 (natural cytotoxicity-triggering receptor 3). Significant correlation in the expression of the Ltab-Ncr3 genes suggests that interaction of these genes may be important in keeping these genes clustered together as a conserved haplotype. We studied the arthritis association and the spliceo-transcriptome of four different Ltab-Ncr3 haplotypes and showed that higher Ltb and Ncr3 expression, lower Lst1 expression, and the expression of a shorter splice variant of Lst1 correlate with reduced arthritis severity in rats. Interestingly, patients with mild RA also showed higher NCR3 expression and lower LST1 expression than patients with severe RA. These data demonstrate the importance of a conserved haplotype in the regulation of complex diseases such as arthritis.

The MHC, known as the HLA complex in humans, is the most gene-dense and polymorphic region in the human genome with strong associations to many autoimmune diseases (1, 2). One such example is rheumatoid arthritis (RA). RA is a chronic autoimmune inflammatory joint disorder affecting 0.5–1% of the general population. The genetic association of RA is mainly with MHC polymorphism; non-MHC loci [more than 100 RA-risk loci have been identified in genome-wide association studies (GWAS)] contribute only to a minor extent (3, 4).

Recently it has been suggested that six different amino acids in HLA-DRB1, HLA-A, HLA-B, and HLA-DPB1 explain most of the MHC association with RA in seropositive patients (5, 6). In addition, independent genetic susceptibility for RA has been identified in the MHC class III (MHC-III) (7–10). MHC-III is a region sandwiched between MHC class I (MHC-I) and MHC class II (MHC-II) in humans. This region consists of many genes with important immune functions encoding for complement proteins, cytokines, and heat-shock proteins. Many of these MHC-III genes have been described as candidate genes in RA, including a gene encoding for TNF (11), which now is a therapeutic target in RA and other autoimmune diseases (12, 13). Other candidate genes include allograft-inflammatory factor-1 (8) and lymphotoxin-β (LTB) (10). However, so far there is no evidence that polymorphism in the MHC-III region really contributes to the pathogenesis in RA or in its associated animal models.

MHC-III is challenging to study because of its high gene density and extensive linkage disequilibrium (LD) (9). Animal models for RA are attractive alternatives to human association studies for gene identifications because the use of these models not only overcomes genetic heterogeneity and reduces environmental effects but also allows the isolation of specific genetic regions to study their roles in RA in vivo (14–16). Rat arthritis models such as pristane-induced arthritis (PIA) (17, 18), which meet many of the criteria for the diagnosis of RA (19), have been used to reveal highly significant linkages to unique chromosomal regions, often a much smaller interval with far fewer genes than previously postulated for quantitative traits (16, 20, 21).

We recently established a panel of recombinant strains in the MHC region to study the MHC association with T-cell selection (22) and showed that RT1-B (the human ortholog is HLA-DQ) in MHC-II determines the onset and severity of PIA (21). Here, we constructed a panel of recombinant strains in MHC-III to study the association of MHC-III with arthritis on the DA (RT1^av1) background. By dissecting this gene-dense region, gradually narrowing down the size of the arthritis-regulating quantitative trait locus (QTL), we identified a conserved 33-kb MHC-III haplotype comprising highly polymorphic, differentially expressed, and spliced genes that regulates the onset, severity, and chronicity of autoimmune arthritis.

Results

Identification of a 33-kb Locus in MHC-III That Regulates the Onset and Severity of Arthritis.

We previously identified RT1-B in the MHC-II region as a susceptibility locus for acute PIA and excluded a contribution of the classical MHC-Ia region (21). Although rats of both the MHC congenic strain DA.1HR2, which spans the entire MHC, and the MHC-II congenic strain DA.1HR61 (Fig. 1) were less susceptible to PIA, we observed a small but reproducible difference in arthritis protection between these two strains from day 14 to day 25 (Fig. 2A) (21). This observation suggested a putative second QTL in the nonclassical MHC-Ib and/or the MHC-III region, which was confirmed with the recombinant strain DA.1HR7, which carries a fragment from DA.1HR2 in the MHC-Ib/MHC-III region (Fig. 1 and Table 1).

Fig. 1.

Overview of the genetic mapping of MHC recombinant strains of haplotypes RT1^h, RT1^u, and RT1^f based on the University of California, Santa Cruz (UCSC) Genomic Browser 2004 (Baylor 3.4/rn4) assembly. Genes are depicted at the corresponding position (in megabases) at the top of the figure. The overall organization of class Ia, II, III, and Ib regions is adopted from a previous study (48). Congenic strains are depicted as horizontal bars with horizontal lines indicating intervals of unknown genotype. Congenic fragments with an asterisk exhibit a protective disease phenotype in PIA. The prefixes “HR,” “UR,” and “FR” indicate that the congenic fragments originate from DA.1H (RT1^h haplotype), DA.1U (RT1^u haplotype), and DA.1F (RT1^f haplotype), respectively (Materials and Methods). PIA has been previously reported in DA.1HR2, DA.1HR61, DA.1HR62, DA.1HR83, DA.1FR2, DA.1FR9, and DA.1UR2 rats (21, 22). The MHC arthritis QTLs Ltab-Ncr3 and RT1-B (previously identified) (21), are shaded in brown.

Fig. 2.

The Ltab-Ncr3 region regulates onset, severity, and chronicity of PIA. Development of PIA in DA.1HR2 (n = 13), DA.1HR61 (n = 12), DA.1HR56D (n = 15), and DA (n = 14) rats with mean arthritis score (A), disease incidence (B), day of disease onset (C), percentage of weight change from day 9 to day 20 after pristane immunization (D), and percentage of weight change from day 79 to day 212 after pristane immunization (E). (F) Comparison of serum AGP levels in DA.1HR56D and DA rats on day 21 after pristane immunization. (G) Mean arthritis score of recipient rats after transfer of pristane-primed T cells from DA.1HR56D (n = 12) and DA (n = 10) donor rats. (H) Mean arthritis score of DA.1HR56 (n = 11) and DA (n = 9) rats after transfer of pristane-primed T cells from DA rats. (G and H) Lymph node cells were harvested from rats 8 d after pristane immunization. Data shown in A, G, and H are mean ± SEM; horizontal lines in C–F represent mean values. (A and B) An asterisk denotes a significant difference between DA.1HR56D and DA; a plus sign denotes a significant difference between DA.1HR61 and DA; a carat denotes a significant difference between DA.1HR2 and DA. (A–H) One symbol indicates P < 0.05; two symbols, P < 0.01; three symbols, P < 0.001 compared with DA, unless otherwise specified. Disease incidence was evaluated by Fisher’s exact test. Other statistics were determined with the Mann–Whitney U test.

Table 1.

Arthritis disease phenotype of MHC-III congenic fragments and DA in the acute phase of PIA

Strain†	Origin	Haplotype‡	Arthritis score§,¶	Weight change (%)¶,#	Day of onset¶,||	Incidence (%)††	No.‡‡
DA		a	30.02 ± 2.053	−9.27 ± 0.86	9.56 ± 0.18	100	18
DA.1HR7*	DA.1HR2	h	18.75 ± 4.37**	nd	11.50 ± 0.65**	100	4
DA.1HR7N*	DA.1HR7	h	18.46 ± 2.86**	+0.71 ± 1.38****	14.25 ± 0.64**	100	13
DA.1HR56*	DA.1HR7N	h	16.48 ± 2.151***	−4.18 ± 1.25**	12.17 ± 0.43**	96	25
DA.1HR56T*	DA.1HR56	h	13.58 ± 4.507**	−6.85 ± 2.28**	12.14 ± 0.67	89	9
DA.1HR56A*	DA.1HR56T	h	17.42 ± 2.896***	−2.17 ± 0.95***	12.56 ± 0.58**	100	15
DA.1HR56D*	DA.1HR56A	h	13.10 ± 2.82****	+0.23 ± 1.52****	12.00 ± 0.53****	89	18
DA.1HR4	DA.1HR2	h	24.20 ± 4.43	nd	9.40 ± 0.40	100	5
DA.1HR52C	DA.1HR7	h	26.88 ± 3.83	nd	10.63 ± 0.80	100	8
DA.1HR56C	DA.1HR56A	h	25.63 ± 2.51	−12.06 ± 2.13	10.63 ± 0.205	100	19
DA.1HR56E	DA.1HR56D	h	22.67 ± 4.25	−10.8 ± 1.98	13.75 ± 0.92	100	12
DA.1FR5	DA.1FR2	f	33.62 ± 2.471	nd	9.70 ± 0.62	100	10
DA.1UR2A	DA.1UR2	u	22.50 ± 3.74	−9.79 ± 2.64	10.60 ± 0.64	100	10

^†Congenic strains with disease protection are indicated with an asterisk.

^‡Haplotype names have been abbreviated: h = RT1^h, u = RT1^u, f = RT1^f, a = RT1^av1.

^§Arthritis score indicates the mean disease score at the peak of acute arthritis (approximately day 20).

^¶Arthritis score, weight change, and day of onset are shown in mean values ± SEM. **P < 0.05, ***P < 0.01, ****P < 0.001, relative to DA; nd, not determined. Statistics were determined with the Mann–Whitney U test.

^#Weight change is shown as percentage change on day 20 vs. day 9.

^||Day of disease onset is defined as the first day when any clinical signs of arthritis in the rats could be observed.

^††Denotes the percentage of animals which develop arthritis for at least two consecutive scoring days.

^‡‡Denotes the number of animals in the group.

To identify the underlying genetic regulation in DA.1HR7 rats, we generated ten subcongenic strains spanning different intervals in this 2.8-Mb region (Fig. 1). We mapped all the MHC congenic and subcongenic strains on the 2004 (Baylor 3.4/rn4) assembly which, based on more than 70 widely dispersed SNP and short tandem repeat (STR) markers as described in the current and a previous study (22), offers more accurate annotation of the region than the 2012 (RGSC 5.0/rn5) and 2014 (RGSC 6.0/rn6) assemblies. We assessed acute arthritis in these strains and found that rats of the subcongenic strains DA.1HR7N, DA.1HR56, DA.1HR56T, DA.1HR56A, and DA.1HR56D developed significantly milder acute arthritis than DA rats, having lower disease score, reduced weight loss, and a delayed disease onset, whereas rats of the subcongenic strains DA.1HR4, DA.1HR52C, DA.1HR56C, and DA.1HR56E were not protected (Fig. 1 and Table 1). These results demonstrate the existence of a 32.7-kb arthritis-regulating QTL in the MHC-III region (3.655–3.713 Mb) comprising five genes, lymphotoxin-α (Lta), Tnf, Ltb, leukocyte-specific transcript 1 (Lst1), and natural cytotoxicity-triggering receptor 3 (Ncr3) (hereafter referred to as the Ltab-Ncr3 region).

DA.1HR56D rats, which carried the smallest congenic fragment harboring the Ltab-Ncr3 region, developed milder acute PIA than DA rats, with significantly lower disease scores from day 12 to day 25 after pristane injection (Fig. 2A). In addition, DA.1HR56D rats developed arthritis with a lower disease incidence and delayed disease onset (Fig. 2 B and C). The reduced disease severity in DA.1HR56D rats also was reflected in reduced weight loss and a lower serum level of the acute-phase protein α1-acid glycoprotein (AGP) (Fig. 2 D and F).

DA.1HR56D and DA.1HR61 Have Different Effects on Acute and Chronic PIA.

To compare the effect sizes of MHC-II and MHC-III on PIA, we assessed PIA in DA.1HR56D, DA.1HR61, and DA.1HR2 rats. In the acute phase of PIA (days 10–30) (18), DA.1HR2, DA.1HR61, and DA.1HR56D rats developed milder PIA than DA rats (Fig. 2 A–D). DA.1HR61 had a stronger protective effect than DA.1HR56D; this result was not surprising, considering the association between MHC-II and both experimental arthritis and RA patients (21, 23).

In the chronic phase of PIA (day >80) (18), both DA.HR2 and DA.1HR56D rats developed significantly milder disease than DA rats, whereas DA.1HR61 rats showed no significant disease protection (Fig. 2A); this result is consistent with our previous data (21). Also, DA.1HR2 and DA.1HR56D rats gained significantly more weight than both DA.1HR61 and DA rats in chronic PIA (Fig. 2E). We therefore conclude that DA.1HR56D and DA.1HR61 regulate the acute and chronic phases of arthritis to different extent.

Ltab-Ncr3 Regulates the Priming of Arthritis.

To show whether the Ltab-Ncr3 region regulates the priming or the effector phase of arthritis, we performed reciprocal adoptive transfers of in vivo-primed T cells between the congenic and DA rats (24). When we transferred pristane-primed T cells from congenic and DA donor rats to heterozygous recipient rats, we observed that T cells originating from congenic rats were less arthritogenic than the T cells from DA rats, showing that the Ltab-Ncr3 region regulates the priming of arthritis (Fig. 2G). Consistent with the reduced arthritogenicity of the congenic-derived T cells, CD4⁺ T cells from congenic rats expressed lower levels of activation markers and produced less of the proinflammatory cytokine IFN-γ (Fig. S1), which has been shown to be important in PIA (24). In contrast, when T cells were transferred from pristane-primed DA rats to congenic or DA recipients, there was no difference in disease severity (Fig. 2H), demonstrating that the Ltab-Ncr3 region does not regulate the effector phase of arthritis.

Fig. S1.

Pristane-primed DA.1HR56D T cells expressed lower levels of activation markers, secreted lower levels of IFN-γ, and transferred milder arthritis compared with DA T cells. (A–D) Draining lymph nodes of DA.1HR56D (n = 10) and DA (n = 10) rats were harvested on day 8 after pristane immunization. The frequency of CD25 (A) and CD134 (B) and the mean fluorescence intensity (MFI) of CD27 (C) on gated CD4⁺ αβT cells was determined by flow cytometry. (D) The level of cytokine IFN-γ (in relative florescence units, RFU) in the supernatant after ConA stimulation for 65 h was determined by ELISA. (E) Lymph node cells were harvested from rats 5 d after pristane immunization. Shown are the mean arthritis score of recipient rats after receiving pristane-primed T cells from DA.1HR56D (n = 9) and DA (n = 12) donor rats. Data are presented as mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001 compared with DA. Statistics were determined with the Mann–Whitney U test.

Ltab-Ncr3 Is an Evolutionarily Conserved, Disease-Associated Haplotype.

Studying the genomic sequence of the DA.1HR56D and DA rats by next-generation sequencing (NGS) showed that the 33-kb Ltab-Ncr3 region is highly polymorphic, with 261 SNPs (Table S1). When the polymorphisms in the arthritis-protective DA.1HR56D rats were compared with those in the arthritis-prone DA.1UR2A rats (Table 1 and Table S1), 229 of the 261 SNPs (including five nonsynonymous SNPs in Ltb, Lst1, and Ncr3) correlated with the arthritis-protective phenotype in DA.1HR56D (Table S1). Using PROVEAN software (25), we examined the potential effect of the five nonsynonymous SNPs in Ltb, Lst1, and Ncr3 on their protein structure. All the nonsynonymous SNPs were predicted not to impact the structure and function of the proteins (Table S2).

Table S1.

Summary of genetic variations in the Ltab-Ncr3 locus

MHC-III congenic strain	Number of variants*							Total
	Intergenic	5′ UTR	Synonymous	Nonsynonymous	Intronic	3′ UTR	Splice site
DA.1HR56D	123	4	8	6	109	9	2	261
DA.1UR2A	33	0	4	2	19	1	1	60
DA.1HR56D/DA.1UR2A†	19	0	1	1	9	1	1	32
DA.1HR56D (disease-associated)	104	4	7	5	100	8	1	229

^*Number of sequence variants in the Ltab-Ncr3 locus relative to DA rats.

^†Variants found in both DA.1HR56D and DA.1UR2A rats.

Table S2.

Full summary of coding variants, gene expression, splice variants, and arthritis score of different strains

Variants, gene expression, and arthritis score	Location†	Gene	Amino acid‡	Strain#				PROVEAN score§
				DA.1HR56D	DA.1UR2A	DA.1FR9	DA
Coding variants¶	3667565	Ltb	133	L	P	P	P	1.437
	3667652	Ltb	104	D	G	G	G	0.895
	3699415	Lst1	21	L	P	P	P	1.375
	3702148	Ncr3	135	A	A	V	V	−0.317
	3702600	Ncr3	19	V	I	I	I	0.282
	3707078	Ncr3	7	V	I	I	I	0.646
Gene expression (fold-change)		Lta		1.00*	1.27	0.80**	1.20
		Tnf		1.07	1.01	1.22	1.36
		Ltb		1.57****	1.03	1.05	0.98
		Lst1		1.09****	1.54	2.01	1.77
		Ncr3		3.84****	1.15	0.96	0.97
Splice isoform expression (fold change)		Lst1 full		1.09****	383.21*	604.20	593.94
		Lst1 spliced		19.77****	6.13***	0.40	0.88
		Ncr3 full		6.14****	2.23*	0.89	1.01
		Ncr3 fused		1.06****	1.56*	1.94	2.24
Maximum arthritis score				13.10****	22.50	26.75	30.02

^†Genomic coordinates are obtained from UCSC Genome Browser (3.4/rn 2004 assembly).

^‡Denotes the location of the amino acids in the protein.

^§PROVEAN predicts whether an amino acid substitution has an impact on the biological function of a protein. PROVEAN scores less than or equal to −2.5 are considered deleterious’; PROVEAN scores greater than −2.5 are considered neutral.

^¶Only coding variants in which DA.1HR56D (RT1^h) differs from DA (RT1^av1) are shown.

^#Data are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 compared with DA.

To investigate the degree of conservation in this region, we typed 13 SNPs in the Ltab-Ncr3 region in nonrelated wild rats collected at different locations in three countries. Wild rats are genetically more diverse than inbred laboratory strains, thus allowing us to assess better the haplotype structure in this region. Five of the 48 wild rats shared exactly the same allelic variants as DA.1HR56D rats; 17 shared exactly the same allelic variants as DA.1UR2A, DA.1FR9, and DA rats; and the remaining 26 wild rats were found to be heterozygous at all these sites (Table 2). These findings strongly support the existence of a conserved haplotype in the Ltab-Ncr3 region. In addition, despite genotyping more than 8,000 congenic rats, no further recombination was detected in the Ltab-Ncr3 region. This result further indicates that the Ltab-Ncr3 region is resistant to recombinations (Fig. S2).

Table 2.

SNPs in Ltab-Ncr3 haplotype in wild rats

Location*	SNP type	Sequence variant
3661408	Intronic	A	C	A/C
3661469	Intronic	G	A	A/G
3667565	Nonsynonymous	A	G	A/G
3667652	Nonsynonymous	T	C	C/T
3699415	Nonsynonymous	T	C	C/T
3699425	Synonymous	C	G	C/G
3702451	Synonymous	G	A	A/G
3702600	Nonsynonymous	C	T	C/T
3706976	Intronic	A	G	A/G
3707078	Nonsynonymous	C	T	C/T
3707248	5′ UTR	G	A	A/G
3710003	Intergenic	T	A	A/T
3710030	Intergenic	A	T	A/T
Total number of animals		5	17	26

Wild rats were collected at various locations in Norway, Sweden, and The Netherlands.

^*Genomic coordinates are based on the UCSC Genomic Browser 2004 (Baylor 3.4/rn4) assembly.

Fig. S2.

Recombination frequency in the rat MHC-III region. Recombination activity was assessed over 900 Kb (3.6–4.5 Mb). The width of the bars represents recombination intervals. The height of the bars indicates recombination frequency (%). The area shaded in orange indicates recombination segments identified by sperm recombinants in humans (49). The areas shaded in green indicate HapMap-inferred recombination sites (50) and recombination sites identified in an extended haplotype homozygosity decay study (51) in humans.

Excluding Tnf Polymorphism as a Contributor to the Haplotype Disease Effect.

We considered Tnf as a potential candidate gene in the Ltab-Ncr3 haplotype because of its well-recognized role in the pathogenesis and treatment of RA (12). To investigate the potential contribution of Tnf polymorphism, we compared the coding sequence homology and mRNA expression levels of different Ltab-Ncr3 alleles. We found no nonsynonymous SNPs in the Tnf and also found no differential expression of Tnf between strains in the lymph nodes of immunized rats (Fig. S3 A and B). Comparing congenic and DA rats, we found that similar levels of TNF were secreted upon LPS stimulation of whole blood harvested from naive rats (Fig. 3A) and from rats with active arthritis (Fig. 3B) and also by splenocytes and peritoneal macrophages upon stimulation with LPS, zymosan, polyinosinic-polycytidylic acid (poly I:C), or concanavalin A (ConA) (Fig. 3 C and D and Fig. S3 C–F). Because anti-TNF therapy is often used for treatment of RA and the LTA-TNF region has been associated with the response to treatment with etanercept (a TNF inhibitor) (26), we investigated whether the DA.1HR56D and DA rats responded differently to etanercept treatment. We found that both groups responded effectively to the etanercept treatment, with a significant reduction of arthritis (Fig. 3E). Taken together, these findings suggest that Tnf is not responsible for the disease protection in the DA.1HR56D rats.

Fig. S3.

Excluding Tnf polymorphism as a contributor to the haplotype disease effect. (A) Gene expression of Tnf in naive inguinal lymph nodes from DA.1HR56D and DA rats. (B) Gene expression of Tnf in different cell populations [B cells, CD4⁺ T cells, CD8⁺ T cells, and CD11b/c⁺ cells (SI Materials and Methods)] in inguinal lymph nodes isolated from DA.1R56D and DA rats 5 d after pristane immunization. (C and D) The level of TNF in the supernatant produced by splenocytes (C) and peritoneal macrophages (D) harvested from naive DA.1HR56D and DA rats after stimulation for 12 h with LPS, zymosan, poly I:C, and ConA. (E and F) The level of TNF in the supernatant produced by splenocytes (E) and peritoneal macrophages (F) from naive DA.1HR56D and DA rats after stimulation for 12 and 24 h with different concentrations of LPS. (G) The level of TNF (in relative florescence units, RFU) in the supernatant measured by ELISA after ConA stimulation of inguinal lymph node cells for 65 h. Inguinal lymph node cells were harvested from DA.1HR56T and DA rats 8 d after pristane immunization. Data are presented as mean ± SEM.

Fig. 3.

Excluding Tnf polymorphism as a contributor to the DA.1HR56D arthritis protection effect. (A) Level of TNF in whole blood from naive DA.1HR56D (n = 8) and DA (n = 9) rats after LPS stimulation for 18 h. (B) Level of TNF in whole blood from PIA (day 16) DA.1HR56T (n = 12) and DA (n = 15) rats after LPS stimulation for 18 h. (C and D) Level of TNF in the supernatant produced by splenocytes (C) and peritoneal macrophages (D) harvested from naive DA.1HR56D (n = 8) and DA (n = 9) rats after stimulation for 24 h with LPS, zymosan, poly I:C, and ConA. (E) Development of PIA in DA.1HR56D (etanercept: n = 9; control: n = 9) and DA (etanercept: n = 10; control: n = 13) rats treated with etanercept or PBS on days 2, 4, and 6 after pristane immunization. All data are represented as mean ± SEM; *P < 0.05 (DA.1HR56D) and ***P < 0.001 (DA) comparing cumulative disease scores of the treatment group vs. the control group. Statistics were determined with the Mann–Whitney U test.

The Ltab-Ncr3 Genes Are Differentially Expressed and Undergo Alternative Splicing.

Variation in gene expression is an important mechanism underlying susceptibility to complex diseases (27–30). In addition to Tnf, we also studied the gene-expression levels of the other four genes in the Ltab-Ncr3 region in the lymph nodes from naive and immunized rats [on day 5 after pristane injection, because the arthritis protection could be transferred by congenic T cells as early as day 5 (Fig. S1E)].

Gene expression in arthritis-protective DA.1HR56D rats was compared with the expression levels in DA.1UR2A, DA.1FR9, and DA rats. Both DA.1HR56D and DA.1FR9 rats showed lower Lta expression (Fig. 4A); however, because DA.1FR9 is not associated with any MHC-III arthritis regulation (21), it is unlikely that the differential Lta expression is associated with the disease protection in DA.1HR56D rats. This conclusion is supported by the finding that Lta was not differentially expressed on day 5 after pristane immunization (Fig. S4A). More compelling was the finding that lymph node cells from DA.1HR56D rats showed higher expression of Ltb and Ncr3 and lower expression of Lst1 in a strain-specific manner (i.e., relative to DA.1UR2A, DA.1FR9, and DA rats) in both naive (Fig. 4A) and immunized rats (shown for DA.1HR56D vs. DA rats in Figs. S4 A–D and S5).

Fig. 4.

Ltab-Ncr3 genes are differentially expressed and undergo alternative splicing. Gene expression in lymph nodes from naive DA.1HR56D (n = 17), DA.1UR2A (n = 8), DA.1FR9 (n = 9), and DA (n = 14) rats. (A) Expression of genes Lta, Ltb, Lst1, and Ncr3. (B, Left) Expression of the Lst1 full-length transcript (NM_022634.2) and Lst1 splice variant without exon 2 (XM_006256080.2). (Right) Gel photo illustrating the expression of the Lst1 full-length transcript and Lst1 splice variant without exon 2. (C) Expression of the Ncr3 full-length transcript with separate exon 3 and 4 (NM_181822.2) and the variant with fused exon 3 and 4 (XM_006256054.2). The reference genes Actb, Arbp, and Hmbs were used for normalization. Data are represented as mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Statistics were determined with the Mann–Whitney U test.

Fig. S4.

Genes in the Ltab-Ncr3 haplotype are differentially expressed and undergo alternative splicing. Gene expression measured in different lymph node cell populations [B cells, CD4⁺ T cells, CD8⁺ T cells, and CD11b/c⁺ cells (SI Materials and Methods)] on day 5 after pristane immunization in DA.1HR56D (n = 8) and DA rats (n = 8). (A–D) Expression of genes in the Ltab-Ncr3 region: Lta (A), Ltb (B), Lst1 (C), and Ncr3 (D). (E–H) Expression of different isoforms: Lst1 full-length transcript (NM_022634.2) (E), Lst1 splice variant without exon 2 (XM_006256080.2) (F), Ncr3 full-length transcript with separate exons 3 and 4 (NM_181822.2) (G), and Ncr3 variant with fused exons 3 and 4 (XM_006256054.2) (H). *P < 0.05; **P < 0.01, ***P < 0.001 compared with DA. Data are presented as mean ± SEM. Statistics were determined with the Mann–Whitney U test.

Fig. S5.

The level of protein expression of LTB (A) and LST1 (B) in the draining lymph nodes of DA.1HR56D (n = 9) and DA (n = 9) rats on day 8 after pristane immunization, shown as mean fluorescence intensity (MFI) determined by flow cytometry. Data are presented as mean ± SEM; **P < 0.01 compared with DA. Statistics were determined with the Mann–Whitney U test.

Because MHC-III genes such as LST1 and NCR3 are known to exhibit alternatively spliced isoforms in humans (31, 32), we assessed the expression of different isoforms in DA.1HR56D, DA.1UR2A, DA.1FR9, and DA rats. We found that different MHC-III congenic strains expressed different Lst1 and Ncr3 isoforms. Although DA.1UR2A, DA.1FR9, and DA rats expressed the full-length Lst1 transcript (NM_022634.2), DA.1HR56D rats expressed only a shorter spliced variant (XM_006256080.2) lacking exon 2, which leads to a shorter extracellular domain (Fig. 4B). NCR3 is a functional activation receptor on a subset of rat natural killer (NK) cells (33), and the potential importance of NK cells in experimental arthritis is highlighted by our finding that NK cell depletion reduced the severity of arthritis (Fig. S6). We found that although DA.1HR56D rats showed the highest expression of Ncr3 full-length transcript (NM_181822.2) (Fig. 4A), these rats showed the lowest expression of a Ncr3 splice variant formed through the fusion of exons 3 and 4 that results in a longer intracellular domain (XM_006256054.2) (Fig. 4C). Similar splicing patterns of both Lst1 and Ncr3 also were observed in immunized rats (Fig. S4 E–H).

Fig. S6.

The development of PIA with NK cell depletion in DA.1HR56D (n = 8) and DA (n = 8) rats shown as (A) mean arthritis score ± SEM (*P < 0.05; **P < 0.01 comparing cumulative disease scores of the NK cell-depletion group vs. the control group) and (B) percentage of weight change from day 9 to day 20 (**P < 0.01 comparing weight change in the NK cell-depletion group: DA.1HR56D vs. DA). Statistics were determined with the Mann–Whitney U test.

Taken together, our results suggest that reduced arthritis severity in DA.1HR56D rats (RT1^h Ltab-Ncr3 haplotype) is associated with higher Ltb and Ncr3 expression and lower Lst1 expression and with the expression of a shorter splice variant of Lst1.

One possible advantage of the genes in the Ltab-Ncr3 haplotype clustering together as a conserved haplotype could be enhanced coordination of gene expression (34, 35). We therefore determined the correlation between the expression of different genes in the Ltab-Ncr3 haplotype and found a significant positive correlation in the expression of different Ltab-Ncr3 genes, especially Lta-Lst1 and Lst1-Ncr3 (Spearman’s ρ 0.60 and 0.71, respectively) (Fig. 5). This finding indicates that genes in the region may directly or indirectly interact with each other in the haplotype.

Fig. 5.

Ltab-Ncr3 gene expressions are significantly and positively correlated. Pink circles, purple triangles, and filled black circles denote Ltab-Ncr3 gene expression of DA.1UR2A, DA.1FR9, and DA rats, respectively. (A) Correlation analysis of Lta and Lst1 gene expression. (B) Correlation analysis of Tnf and Ltb gene expression. (C) Correlation analysis of Lst1 and Ncr3 gene expression. In A–C each dot represents the fold-change of the corresponding genes of each individual rat (n = 31). Spearman’s correlation coefficient ρ and the P value for each gene pair are indicated in each figure. Reference genes Actb, Arbp, and Hmbs were used as controls for normalization to calculate the fold-change. (D) Summary of correlation showing Spearman’s correlation coefficient ρ between different Ltab-Ncr3 genes. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

LTB, LST1, and NCR3 Gene Expression in RA Patients.

We studied the level of expression of LTB, LST1, and NCR3 in whole-blood samples from a cohort of 32 RA patients and 92 healthy controls. Overall, the expression of these genes was increased significantly in blood cells of RA patients compared with blood cells from healthy controls, indicating that these genes could be important for the development of RA (Fig. 6), as is consistent with previous findings (10, 36).

Fig. 6.

The level of expression of genes LTB (A), LST1 (B), and NCR3 (C) in patients with mild RA (DAS28 ≤ 3.2, n = 10) and severe RA (DAS28 > 5.1, n = 22) and in healthy controls (n = 92). The reference gene ZNF592 was used for normalization. Data are represented as mean ± SEM; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 for comparison between groups. Statistics were determined with the Mann–Whitney U test.

Having shown that the increased expression of Ltb and Ncr3 and reduced expression of Lst1 correlated with reduced arthritis severity in PIA, we next investigated whether the expression of these genes also differed between two extremes, groups of patients with relatively mild and severe RA. On the basis of the disease activity score (DAS28), we used clinically accepted thresholds to define mild RA (DAS28 ≤ 3.2) and severe RA (DAS28 > 5.1). The group of patients with mild RA showed lower expression of LST1 and higher expression of NCR3 than the group of patients with severe RA (Fig. 6 B and C), similar to our observations in both naive and immunized DA.1HR56D rats (Fig. 4A and Fig. S4 B–D). However, LTB gene expression did not differ in the groups of patients with mild and severe RA (Fig. 6A).

Discussion

Although the association of the classical MHC with RA and experimental arthritis is well known (5, 21, 37), we show, for the first time to our knowledge, that MHC-III also controls arthritis. In the MHC, we showed that, in addition to the arthritis-regulatory RT1-B locus in MHC-II (21), there is a second arthritis-regulatory QTL in MHC-III that determines the onset, severity, and also the chronicity of arthritis. We further narrowed down this QTL to a conserved Ltab-Ncr3 haplotype with five polymorphic, tightly linked genes.

MHC-III has been a particularly challenging region to study for association with complex diseases such as RA (9, 38): The strong LD in the MHC makes it difficult to distinguish causal variants from linked neutral variants, and therefore studying human arthritis association within the MHC requires controlling for the effects of HLA-DRB1 and other classical MHC alleles (5). An alternative approach is to study arthritis in animals in a congenic setting. Because congenic fragments of MHC-III are isolated on a fixed genetic background, stratification of the effect of classical MHC alleles is no longer needed. One possible reason an association with arthritis was identified in the Ltab-Ncr3 region in congenic rats but not (in different GWAS) in humans (5, 6) is the difference in the genomic structure of the MHC region between humans and rats. In humans, the LTAB-NCR3 region is very close to HLA-B, a classical MHC-I gene associated with RA risk (5), and therefore disease association from LTAB-NCR3 region is hard to detect because of LD; in rats the Ltab-Ncr3 region is much further away from classical MHC-I.

Despite the identification of thousands of QTLs controlling different phenotypes in both humans and rodents, rather few genes have been identified from these QTLs (39). In the rat, the only genes/loci that have been positionally mapped with high resolution are Ncf1 (40), Igl (20), and RT1-B (21). Instead of identifying an association between a specific gene/locus and arthritis, here we identified the association between a conserved haplotype, Ltab-Ncr3, and arthritis and showed that this haplotype exists in not only laboratory inbred rat strains but also in the wild rat population. In fact, comparative analysis of MHC-III gene sequences between different vertebrates showed that the Ltab-Ncr3 genes have clustered together for millions of years (41). The Ltab-Ncr3 genes encode proteins involved in inflammation, and it is possible that selection pressure has driven the conservation of this haplotype structure so that allelic variants of Ltab-Ncr3 genes of similar functions could operate in cis. This theory is supported by our data showing significant correlation in the expression of the genes in the Ltab-Ncr3 haplotype. For example, we showed significant correlation in the expression of Lta, Tnf, and Ltb in rats. LTA, LTB, and TNF are members of the TNF superfamily important in the organogenesis of lymphoid organs and immune responses (42). In addition, we showed a significant correlation in the expression between Lst1 and Ncr3 in rats, and that correlation also was found in the blood and synovium of RA patients (36). LST1 and NCR3 genes are located extremely close to each other in both human and rats and are transcribed in the opposite orientation. It is possible they share a common regulatory element at the 3′ end of their genes controlling their gene expression, which would explain the strong correlation in Lst1 and Ncr3 gene expression. Furthermore, we showed significant correlation in the expression of Lta and Lst1, Lta and Ncr3, and Tnf and Lst1, illustrating the complexity of the gene network in the Ltab-Ncr3 haplotype.

Haplotype-specific differences in gene expression are, in fact, common across the MHC in humans (43). In the rat, we showed further that higher Ltb and Ncr3 expression and lower Lst1 expression are associated with reduced arthritis severity in the RT1^h Ltab-Ncr3 haplotype. In humans, we showed that RA patients expressed significantly higher levels of LTB, LST1, and NCR3 genes than healthy controls. This finding, together with other RA and animal studies (10, 36, 44, 45), indicates that these genes could be important for the development of arthritis. One recent study demonstrated the linkage between the Lymphotoxin/LIGHT axis and IFN responses in RA patients and suggested that Lymphotoxin/LIGHT could act as an upstream modulator of IFN-signaling genes in RA (46). Furthermore, similar to our observations in DA.1HR56D rats, we found that the group of patients with mild RA showed lower expression of LST1 and higher expression of NCR3 than the group of patients with severe RA, indicating that LST1 and NCR3 could be associated with the severity of arthritis. However, no difference in LTB gene expression, such as we had observed in rats, was observed between groups of patients with mild RA and severe RA. All these data should be taken with caution because various factors can complicate comparisons between experimental arthritis models and RA. For instance, subtle genetic differences identified in inbred rodents can be difficult to show in humans because of the genetic heterogeneity and environmental interactions. In addition, different RA treatments could alter the gene-expression profile of the RA patients and increase the variation in the groups, thus making the comparison between rat and human data even more difficult. Future studies are needed to demonstrate the roles of these differentially expressed genes in RA pathogenesis.

Alternative splicing could be evolutionarily selected to enrich transcript diversity in the MHC for a better and more diverse immune response against pathogens. In humans, specific LST1 splice variants are differentially expressed in RA patients and non-RA controls (36), and here we show that the arthritis-protective congenic rats expressed only the Lst1 splice variant without exon 2 but not the full-length Lst1 transcript with exon 2, which was expressed by other rat strains. In addition, we showed that Ncr3 transcripts with different lengths of the intracellular domain were differentially expressed in the arthritis-protective Ltab-Ncr3 haplotype. In humans, three constitutively expressed NCR3 isoforms characterized by intracellular domains of different lengths are functionally described as either immunostimulatory or immunosuppressive (32). Therefore it is possible that the differential expression of the Ncr3 isoforms in rats might affect the balance of these functions.

In conclusion, using a panel of intra-MHC recombinant congenic strains, we identified a 33-kb conserved haplotype consisting of five tightly linked immunoregulatory genes in the telomeric end of MHC-III regulating chronic arthritis. The Ltab-Ncr3 haplotype showed strong haplotypic-specific differences in both gene expression and alternative splicing, and these differences correlated remarkably with arthritis susceptibility. Here, we demonstrate that congenic mapping in inbred strains and SNP typing in wild rats remain powerful tools in identifying and characterizing conserved haplotypes regulating complex diseases such as arthritis.

Materials and Methods

Animals.

Inbred DA/Ztm rats were obtained from the Zentralinstitut für Versuchstierzucht, and DA/OlaHsd rats were obtained from Harlan Europe. Rats were maintained by sister–brother mating in a barrier facility at Scheele Laboratory, Karolinska Institutet, and were specific pathogen-free according to the guidelines of the Federation of European Animal Laboratory Science Associations. All animals were housed in individually ventilated microisolator cages (Allentown) containing wood shavings (Tapvei) in a climate-controlled environment with 14-h light/10-h dark cycles and were fed standard rodent chow (R70; Lantmännen) with free access to water.

The generation of the MHC congenic fragments DA.1FR2 (RT1^f), DA.1HR2 (RT1^h), and DA.1UR2 (RT1^u) spanning 1.62–5.98 Mb has been described previously (21, 22, 37). Briefly, congenic strains were originally established on the DA/Ztm background (n >20) and thereafter were backcrossed further (n >5) to DA/OlaHsd rats. Congenic strains with MHC haplotype RT1^f were derived from congenic DA.1F (DA.LEW-RT1^f) rats (22). Congenic strains with MHC haplotype RT1^u were derived from DA.1U, generated by introgression of the corresponding E3/ZtmRhd fragment on chromosome 20. Congenic strains with MHC haplotype RT1^h were derived from DA.1H rats (established at the Zentralinstitut für Versuchstierzucht) originating from the KHW strain. Further recombinant congenic strains were generated by crossing F1 hybrid rats and were screened for recombination using different microsatellite markers. Unless otherwise specified, 8- to 12-wk-old, age- and sex-matched congenic rats and wild-type control rats have been used in all experiments. Wild rat samples were collected in different locations in Norway, Sweden, and the Netherlands. Tail biopsies were stored in ethanol until analysis.

Disease Induction and Evaluation.

PIA was induced by intradermal injection of 100 μL pristane (2,6,10,14-tetramethylpentadecane; Acros Organics) at the base of the tail in age-matched rats, and arthritis severity was monitored blindly using a macroscopic scoring system (18). One point was given for each inflamed knuckle or toe, and up to five points were given for an affected ankle (for a maximum score of 15 per limb and 60 per rat). Rats are described as “protected” if the mean arthritis score is significantly lower than that of the control. Besides scoring, another arthritis evaluation criterion is the day of disease onset, defined as the first day when any clinical signs of arthritis in the rats could be observed. For DA rats, the typical day of arthritis onset is day 9 after pristane injection. Weight change, depicted as the percentage change in weight relative to the weight at disease onset, was used as an objective measurement of disease severity. For acute arthritis, the percentage change in weight between day 20 (the peak of acute PIA) and day 9 was calculated. For chronic arthritis, the percentage change in weight between day 212 (the peak of chronic PIA) and day 79 (start of chronic PIA) was calculated. The disease marker AGP in serum diluted 1:20,000 was measured by the AGP ELISA kit (Life Diagnostics). All experiments were approved and performed in accordance with the guidelines of the Swedish National Board for Laboratory Animals and the European Community Council Directive (86/609/EEC).

Patients and Controls.

The RNA-expression experiments used samples from 32 RA patients and 92 healthy controls from the Swedish population; controls were selected with consideration of the gender, age, and ethnicity of the patient group. RA patients were selected at the Rheumatology Clinic at Karolinska University Hospital, and all met American College of Rheumatology 1987 criteria for RA (19). Informed consent was obtained from all the participants. The Stockholm Ethical Review Board approved the study.

RNA and DNA Extraction.

For rat samples, total RNA was extracted using the RNeasy Mini kit (Qiagen) and was treated with DNase I (Roche). cDNA was synthesized using the High Capacity cDNA Reverse Transcription kit (ABI). For human samples, blood was collected into PAXgene Blood RNA Tubes, and total RNA was extracted with the PAXgene Blood RNA kit (PreAnalytiX) according to the manufacturer’s protocol. Samples were treated with DNase (PreAnalytiX ) for 20 min at room temperature to avoid contamination with genomic DNA. RNA was converted into cDNA using the iScript cDNA Synthesis Kit (Bio-Rad). Rat genomic DNA was isolated by proteinase K digestion (AquaPure Genomic DNA kit; Bio-Rad).

Sequencing.

DNA was amplified using specific primers (Eurofins MWG Operon) as previously described (22) and was purified using Bio-Gel P-100 polyacrylamide beads. For sequencing of different splice variants, PCR products were first run on a 1.5% agarose gel at 120 V, and bands of specific size were cut out from the gel and purified using the Gel Extraction Kit (Qiagen). Sequencing was performed with BigDye Terminator 3.1 using 0.4 μM of primer according to the manufacturer’s instructions (Applied Biosystems). Sequencing products were purified by the standard ethanol precipitation method, resuspended in 10 μL Hi-Di formamide (Applied Biosystems), and analyzed on a 3730 DNA analyzer (Applied Biosystems). Genomic DNA for the NGS was obtained from a liver sample harvested from a male DA.1HR2 rat using the standard phenol/chloroform extraction method, followed by ethanol precipitation of the DNA. NGS was performed at the Science for Life Laboratory, Sweden, using Illumina HiSeq. 2500 to generate pair-ended 2 × 100-bp reads. See Table S3 for primers for genotyping recombinant strains and wild rats.

Table S3.

Primers for genotyping recombinant strains and wild rats

Type	Location, kbp*	Primer 1	Primer 2
STR	3430	AACTTCCTGGTCCAGCCTG	TGTGGAAGTGAGGAGCTGC
STR	3452	CGTTCGTGCGTCTGTCTTTA	AGCCCCAAATCCTCTGTTTT
STR	3628	TCCCTAGAGGCAGAAGGACA	GAGGGTAGTGATCCCACGAA
STR	3635	ACAGAGCCTGTGGGAAGATG	CTTCTGCCAGAGCAACAGTG
STR	3649	AGATGGGGATGGGGTACTTG	TGCAGGCATACTCCACCAC
STR	3655	CAGGAGGCAGTGGTGGTGA	AGTGGAAGTTCAAAGCCAACGA
STR	3656	TCCCTCCCCTTCTCTCTGTC	TTAACGGGTAGCTGATGGGC
STR	3658	CCTGTCTGTCCCCTTTTCTG	ATCTCTCTTGCTGCCTCACC
STR	3661	GACAGAGAGGAAATGGGTTT	AATCTCCTCTCCGTCCTCAC
SNP	3661	AGAACTCCAGGCGGTGTCT	CCCATTTGGGAACTTCTGTG
STR	3662	GTTTCAGTTCTCAGGGTCCTA	CAGGATTCTGTGGCAATCTGG
SNP	3667	TAGCCGAATCTGACCTCTGG	AGATCCCGACCGGTTTTTAC
SNP	3699	ACCCAGTGAGTTCCAGGACA	TCTCCCATCAGTTTCCCAAA
SNP	3702	GCATATTGCAGGGGTCAAGT	GTCATGGCTTTGGATGTCCT
SNP	3706	AAATCGGACACGAGGACATC	CCAGCCCTTTCTTTCTACCC
SNP	3710	TCACCATGCACACTCTGACA	CCATGATGAGCAAGCCAGTA
STR	3711	CTCCTGCTTCAGGGTATCCA	CTCGAAGCCAGTGACACTTTT
STR	3713	GCAGTACAAGGGGACATTGC	GTCGGAGCCACTTCCTGTC
STR	3716	TCCCTGCCTTCTCCATAGC	TTGGGGATTTAGCTCAGTGG
STR	3736	GTGTCAGTGAAGCCCTCTCC	AGACCTGCTGGAGAAAAGCA
STR	3771	TGGGGATTTAGCTCAGTGGTAG	AAATGAACTCGCTCGCTCAC
SNP	3801	AGGTAAGGGGCAGGAGGTG	CCACATTCAAGGACGCTTTT
SNP	3806	GTGTGTGGTCACTGCCTGTC	GGGCCAAGTGGATGTTCTC
STR	3812	TATGCAGGTTGGCCTTGAAC	CGGAGGATCCAATGTCTTTTC
STR	3845	CCAGGTCACCTCACACCATC	GAATTCCAGATTGTCCTTTGTCCT
SNP	3850	TTCTGGCTGTACGACATTGC	AAAACCGAGGTGAACAGTGG
SNP	3863	TAAGCCAGCTGCAGATGATG	GCTGGAAACAGGTCCTTCTG
STR	3864	AAGCCTTTGGTTTGGTTCC	GCCTCTGCGTGTCTTCCTAC
STR	3957	CTACATGAGCCCCTATTTGA	TCCACAGCAGGGAGTAATTG
STR	4047	GCGAGACTTCCTTTCACCTG	GGAGAGAGCCAGCTTCTCAAT
STR	4103	CAATTGATCGGTTAGCAGC	GAAAGAGAAAGACAGGGTAAGC

^*Genomic coordinates are obtained from UCSC Genome Browser (3.4/rn 2004 assembly).

Quantitative Real-Time PCR.

For rat gene expression, quantitative real-time PCR (qRT-PCR) was performed on an ABI 7900 HT system (Applied Biosystems) using SYBR Green (Applied Biosystems) and a two-step PCR protocol (95 °C for 10 min followed by 40 cycles for 95 °C for 10 s and 60 °C for 30 s). Primers were designed using Primer-BLAST (National Center for Biotechnology Information) with sequences retrieved from public databases; only primers binding at nonpolymorphic locations were used. The expression fold-change of each gene was determined by the standard comparative cycle threshold (C_T) method after normalization to the geometric mean of the reference genes Actb (actin, beta), Arbp (attachment region-binding protein), and Hmbs (hydroxymethylbilane synthase). See Table S4 for primers used in qRT-PCR.

Table S4.

Primers for qRT-PCR

Gene	Primer 1	Primer 2	Comments
Tnf	TCGGGGTGATCGGTCCCAACAA	GTGGTTTGCTACGACGTGGGCT
Lta	CAGGGACTCTCTGGTGTTCG	GGGGTACCCAACAAGGTGAG
Ltb	ACCTCATAGGCGCTTGGATG	ACCCGACATGGCAGTAGAGA
Lst1	AAAGAAACGCCCAGGTCTCG	CGCCTTCACCTTCTCCTCTG	Amplifies both Lst1 transcripts, amplicon located on exons 4 and 5
Lst1	TGGGGAGAGGCACTCAAGA	GCAGGCGAACAGGATGATGA	Amplifies Lst1 full-length transcripts (with exon 2), amplicon located on exons 1, 2, and 3
Lst1	TCAGGCCTGTTCTCAGTCCC	GTAACACTGTTTCTCCTCTCCCC	Amplifies Lst1 splice variant (without exon 2), amplicon located on exons 1 and 3
Ncr3	GGCTGGTGGTGGAGAAAGAGCC	GGCATAGACGCCAGCGCGAA	Amplifies both Ncr3 transcripts, amplicon located on exons 2 and 3
Ncr3	ACCAGGGCAAATGTCTCTGT	GGCCTCTTTGAGTGCTGGAT	Amplifies Ncr3 transcripts with separate exons 3 and 4, amplicon located on exons 3 and 4
Ncr3	GGAGAAAGAGCCTCCTCAACAAG	CCTTCTCTGTTCCCATGCCC	Amplifies Ncr3 transcripts with fused exons 3 and 4, amplicon located on exon 2 and fused exons 3 and 4
Actb	GGGAAATCGTGCGTGACATT	GCGGCAGTGGCCATCTC
Arbp	GCTTCATTGTGGGAGCAGACA	CATGGTGTTCTTGCCCATCAG
Hmbs	TCTAGATGGCTCAGATAGCATGCA	TGGACCATCTTCTTGCTGAACA

For human gene expression, qRT-PCR was performed on CFX96 system (Bio-Rad) using TaqMan assays (Applied Biosystems) and a protocol of 50 °C for 2 min, 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The assays used to study expression of LTB, LST1, and NCR3 were Hs00242739_m1, Hs00705788_s1, and Hs00394809_m1 respectively. The expression fold-change of each gene also was determined by the standard comparative C_T method using reference gene ZNF592 (Hs00206029_m1) (47).

T-Cell Transfer.

Complete medium was prepared consisting of the following ingredients: DMEM (Gibco) supplemented with 5% (vol/vol) FCS (Gibco), 50 μM β-mercaptoethanol (Gibco), 10 mM Hepes (Gibco), 10 U/mL penicillin, and 100 μg/mL streptomycin (both from Invitrogen Life Technologies). Draining lymph node cells from donor rats were harvested and cultured at a cell density of 3 × 10⁶/mL in complete medium with 3 μg/mL ConA (Sigma-Aldrich) at 37 °C and 5% CO₂. After 65 h, cells were washed, resuspended in PBS, and injected intravenously into recipient rats. In the pooled cell transfer, lymph node cells from different DA rats were pooled together, and 2 × 10⁷ cells were injected into each recipient rat (DA or congenic). In the individual transfer experiment, cells were not pooled; 3 × 10⁷ cells from each donor rat (congenic or DA) were injected into each recipient rat (heterozygous).

In Vitro Stimulation Assays and TNF Measurement.

Peritoneal macrophages were collected from killed rats by lavage with sterile PBS into the peritoneal cavity and were cultured in complete medium (defined above in T-Cell Transfer). Cells (3 × 10⁵) were cultured in a 24-well tissue culture plate. After 2 h, supernatant was removed, leaving adhered macrophages on the culture plate. Splenocytes were collected from spleen by mechanical separation through a mesh followed by erythrocyte lysis. Cells were stimulated for 12 or 24 h with 1–100 ng/mL LPS (Sigma-Aldrich), 100 μg/mL zymosan (Sigma-Aldrich), 200 μg/mL poly I:C, or 3.0 μg/mL ConA (Sigma-Aldrich) at 37 °C and 5% CO₂. For whole-blood LPS stimulation, 100 μL heparinized blood in 400 μL complete medium (defined above in T-Cell Transfer) was cultured in a 48-well cell culture plate with 62.5 μg/mL LPS for 18 h. The level of TNF secretion in supernatant was determined using a Rat TNF ELISA set (BD OptEIA; BD Biosciences). Eu³⁺-conjugated streptavidin (DELFIA; PerkinElmer) was used as a secondary reagent, and detection was performed on a Synergy 2 multimode plate reader (BioTek).

TNF Depletion.

Etanercept (kindly provided by H. Burkhardt, University Hospital Frankfurt and Goethe University, Frankfurt am Main, Germany) was administered subcutaneously in 0.2 mL PBS (12.5 mg/mL) on days 2, 4, and 6 after pristane immunization as previously described (18). Control rats received an equal volume of PBS.

Bioinformatics.

PROVEAN (Protein Variation Effect Analyzer) software was used to predict whether an amino acid substitution has an impact on the biological function of a protein (25). PROVEAN is available online at provean.jcvi.org/.

Statistical Analysis.

The significance of differences in arthritis disease incidence was evaluated by Fisher’s exact test. For correlation of gene expression, in every dataset we calculated a Spearman rank correlation coefficient, ρ, for each gene pair and its corresponding P value. All other statistical analyses were evaluated by Mann–Whitney U test and performed on Prism 6.0.

SI Materials and Methods

Genotyping.

Genomic DNA was extracted from progeny biopsies as previously described (37). PCR primers for STRs and SNPs were designed based on the Baylor 3.4/rn4 2004 genome sequence assembly using the Primer-Select 8.1.3 software (Lasergene; DNASTAR) (see Table S3 for a list of primer sequences). Forward primers were fluorescently labeled, and STRs were amplified with PCR and analyzed on a 48-capillary 3730 DNA analyzer (Applied Biosystems) as previously described (22).

Flow Cytometry.

Single-cell suspensions were prepared in ice-cold FACS buffer (Ca²⁺- and Mg²⁺-free Dulbecco’s PBS supplemented with 1% FCS, 10 mM EDTA) and stained with a saturating concentration of mAbs on 96-well v-bottom polypropylene plates (BD Falcon). The following antibodies were used: CD4 (OX35) was purchased from BioLegend; CD25 (OX39) and CD134 (OX40) were purchased from BD Pharmingen; αβTCR (R73) and CD27 (LG.7F9) were purchased from eBioscience; and Ltb (FL-244) and Lst1 (T-12) were purchased from Santa Cruz Biotechnology. Both LIVE/DEAD Violet (Invitrogen) and the forward scatter vs. side scatter plot were used to include only nonnecrotic cells. A SORP BD LSR II analytic flow cytometer (BD Biosciences) was used for acquisition, and the data were analyzed with FlowJo (Tree Star).

To prepare individual cell populations for gene-expression analysis, lymph nodes were harvested from rats on day 5 after pristane immunization. Lymph node tissue was cut into small cubes, digested with collagenase IV and DNase I, filtered, and washed with cold FACS buffer to obtain single-cell suspensions as previously described (22). Cells were stained with a saturating concentration of the following mAbs: CD3 (1F4), CD4 (OX35), CD11b/c (OX42), and CD45RA (OX33), all purchased from BioLegend, and CD8a (OX8) purchased from BD Pharmingen. Nonnecrotic cells were gated as described above. CD4⁺ T cells (CD3⁺CD4⁺), CD8⁺ T cells (CD3⁺CD8a⁺), B cells (CD45RA⁺CD3⁻), and CD11b/c⁺ cells (CD11b/c⁺ CD45RA⁻) were sorted out into tubes filled with medium containing 10% (vol/vol) FCS (Gibco Laboratories). Cell sorting was performed at BD Influx Cell Sorter at the flow cytometry facility at the Center for Molecular Medicine at Karolinska Institutet.

Cytokine ELISA.

IFN-γ in the culture supernatants was determined by ELISA using 2 μg/mL anti-rat IFN-γ (DB1; BioLegend) as capture antibodies and biotinylated 0.5 μg/mL anti-rat IFN-γ (Poly5109; BioLegend) for detection. Eu3⁺-conjugated streptavidin (DELFIA) was used for detection, performed on a Synergy 2 multimode plate reader (BioTek).

NK Cell Depletion.

For NK cell depletion, 300 μg of anti-rat NKR-P1A/B (10/78) antibody in PBS was injected into the rats subcutaneously at the base of the tail 2 d before pristane immunization. Control rats received an equal volume of PBS.