Autoimmune Epididymo-orchitis is Essential to the Pathogenesis of Male-Specific Spondyloarthritis in HLA-B27 Transgenic Rats

Abstract

Objective

Male rats transgenic for HLA-B27 and human-β2-microglobulin (Hu-β2m) spontaneously develop epididymo-orchitis preceding spondyloarthritis. In the specific B27/Hu-β2m transgenic cross (21-3x382-2)F1, only the males develop spondyloarthritis, and neither sex develops gut inflammation. We asked whether epididymo-orchitis and spondyloarthritis in the (21-3x382-2)F1 males are causally related, and we characterized the epididymo-orchitis.

Methods

B27/Hu-β2m (21-3x382-2)F1 transgenic males underwent bilateral, unilateral, or sham epididymo-orchiectomy between ages 36 and 125 d. Castrated rats were given testosterone replacement. Alternatively, the 21-3 and 283-2 transgene loci were crossed with a transgene inducing aspermatogenesis. Rats were observed for epididymo-orchitis, arthritis, and spondylitis.

Results

In unmanipulated transgenic rats, inflammation was first evident in the ductuli efferentes (DE, ducts linking rete testis to epididymis) as early as age 30 d. The inflammation was initially neutrophilic, and later became granulomatous. Serum anti-sperm and anti-testis cell antibodies appeared after age 70 d. Cells infiltrating the testes were predominantly CD4+ T cells and CD68+ or CD163+ macrophages. Quantitative PCR of DE, epididymis, and testis showed elevations of IFNγ, IL-10, andIL-17A. IL-12A, IL-22, IL-23A, and IL-23R were also examined in DE and found elevated. Remarkably, castration before 91 d of age completely prevented subsequent arthritis and spondylitis, as did transgene-induced azospermia.

Conclusion

In the (21-3x283-2)F1 HLA-B27/Hu-β2m transgenic rats, autoimmune epididymo-orchitis develops spontaneously at 30 d, the age when antigen-positive meiotic germ cells first exit the testis. Persistent testicular inflammation and/or antigenic stimulation are essential prerequisites to subsequent spondyloarthritis. Dysregulated innate immunity at immune privileged sites may be an essential mechanism triggering spondyloarthritis.

INTRODUCTION

The spondyloarthritides are associated with inflammatory eye, intestinal, genital, and skin disease (1). To a variable extent, these associated extra-articular processes can be associated with HLA-B27, which is strongly associated with ankylosing spondylitis (AS) and to a lesser extent with other spondyloarthritides.

Recently, other genes have been found associated with AS (2). Some of these are shared by extra-articular inflammatory disorders associated with spondyloarthritis. These include interleukin (IL)-23R, associated with inflammatory bowel disease (IBD) and psoriasis (3–5); ERAP1, associated with psoriasis (6); and STAT3, TNFSF15, IL12B, CARD9, PTGER4, and KIF21B, associated with IBD (4, 5).

It is not clear to what extent these shared disease and genetic associations reflect interdependent pathogenetic processes. Many of these shared genes are associated with the IL-23/17 or TNF pathways, and the associated disorders respond to anti-TNF agents. On the other hand, untreated AS runs a clinical course with onset and severity largely independent of associated uveitis, IBD, or psoriasis (7).

Urogenital inflammation is also seen in spondyloarthritis. Triggering of reactive arthritis by Chlamydia is well known. An association between chronic prostatitis and AS is extensively documented in older literature and was apparently common in the era before nonsteroidal anti-inflammatory agents (8, 9). Orchitis and epididymitis have been noted in several series of AS patients (8, 10–13). In Paronen’s classic series, 11 of 310 men with post-dysenteric reactive arthritis had inflammation of the testis, epididymis, or both (14). The prevalence of asymptomatic orchitis in spondyloarthritis is unknown, but it is commonly found in male infertility (15).

Rats transgenic for HLA-B27 and Hu-β2m develop spontaneous multi-organ inflammatory disease that resembles B27-associated disease in humans. Three disease phenotypes are correlated with transgene copy number (16, 17) (Table 1). Rats with ≥ 40 copies of the HLA-B27 transgene and ≥ 30 copies of the Hu-β2m transgene develop IBD and arthritis in both sexes, and the males develop epididymo-orchitis (EO). In rats with 20 copies of HLA-B27 and 50 copies of Hu-β2m (F1 cross of the 21-3 and 283-2 lines, here abbreviated F1), all of the males develop EO, which becomes clinically evident as scrotal swelling at ~90 d of age, and 70% of the males develop arthritis, which begins after ~110 d of age (17). Up to half of the F1 males develop clinically apparent tail spondylitis, beginning after ~140 d of age. There is no IBD in the F1 rats, and the females remain completely healthy. In rats with 20 copies of HLA-B27 and 15 copies of Hu-β2m, the 21-3 line, the males develop EO that is milder than in the F1 rats, and there is no IBD, arthritis, or spondylitis.

Table 1.

HLA-B27/Hu-β2m transgenic rat lines

Line	Transgene locus zygosity	Transgene copies		Sex	EO	IBD	Arthritis	Spondylitis	Reference
		HLA-B27	Hu-β2m
21-3	hemi-	20	15	M	+	−	−	−	(16)
				F		−	−	−
	homo-	40	30	M	+	+	+	−	(24)
				F		+	+	−
283-2	hemi-	0	35	both	−	−	−	−	(17)
	homo-	0	70	both	−	−	−	−
(21-3x283-2)F1	hemi- x hemi-	20	50	M	+	−	+	+	(17)
				F		−	−	−

All lines are maintained on the Lewis background

The temporal sequence and male specificity of disease in the F1 rats suggested the possibility that EO is a prerequisite for the development of spondyloarthritis in these rats. Here we report investigation of this hypothesis.

MATERIALS AND METHODS

Rats

Transgenic (21-3x283-2)F1 rats (F1) were bred and maintained as previously described (17). Table 1 summarizes the F1 and parental lines. Severely hypospermatogenic Dazl-deficient rats on a Sprague-Dawley background (18, 19) were crossed with the 21-3 and 283-2 lines by sequentially breeding females carrying the Dazl knockdown transgene locus with males carrying the 283-2 and 21-3 loci, and choosing male offspring carrying all 3 transgene loci or their Dazl wild type littermates carrying the 21-3 and 283-2 transgene loci. All animal experiments were approved by the UTSW IACUC.

Surgery

Epididymo-orchiectomy (EOx) was carried out by standard methods (20). Unilateral EOx was carried out by removing the left testis and epididymis. For sham EOx, a midscrotal incision was made and then closed.

Androgen replacement

Pellets containing either 15 or 50 mg of testosterone (Innovative Research of America, Sarasota, FL) were implanted subcutaneously between the scapulae. Alternatively, rats were given twice-weekly subcutaneous injections of testosterone-17β-cypionate in 1 ml sesame oil. Serum testosterone was determined by ELISA (Oregon Regional Primate Center, Beaverton, OR).

Clinical scoring

Rats were observed at least weekly for signs of scrotal swelling, arthritis, and spondylitis, as described (17). Peripheral arthritis was scored 0–3 for each hind paw and 0–1 for each forepaw (maximum score: 8 per rat).

Adjuvant arthritis

This was induced in rats by injecting 200 μg of pulverized heat-killed M. tuberculosis H37Ra (Difco) in incomplete Freund’s adjuvant (Sigma) intradermally at the tail base.

Histology

Testis and epididymis specimens were fixed in Bouin’s solution, and H&E sections were prepared and scored in a blinded manner by KSKT, as previously described (21).

Immunofluorescence

Frozen sections of DE, epididymis, and testis were processed and examined as previously described (22) with mouse monoclonal antibodies against rat CD4 (W3/25 or OX38), CD8 (OX8), CD3 (G4.18), MHC II (OX6), CD103 (OX62), CD68 (1C7), CD163 (HIS36), CD45RA (OX33) or TCRαβ (R73). Sections were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Antibodies were from BD Biosciences or AbD Serotec. Antibodies to sperm were detected qualitatively by indirect immunofluorescence, as described (22).

In situ hybridization

This was carried out as previously described (23) in testis sections from paraformaldehyde-perfused F1 and LEW rats 50–60 d old. Probe sequences used for HLA-B27, Hu-β2m, and RT1 (rat MHC-I) are available upon request.

ELISA for serum sperm and testis cell antibodies

This was carried out as previously described (22). Pooled sera with high sperm and testis cell antibody titers were used as a standard.

Quantitative PCR

F1, 21-3, and non-transgenic Lewis males were sacrificed in groups of 3 to 5 at ages 30, 45, 60, 75, and 90 d, and unilateral specimens of testis and epididymis were frozen in liquid nitrogen and stored at −80 C. Because of the preponderance of adipocytes, DE specimens were placed in RNALater at room temperature. Tissues were homogenized in TriZol reagent (Invitrogen). RNA content was measured by Nanodrop (Thermo Scientific). One μg RNA per sample was treated with DNAse and then reverse-transcribed with Superscript II transcriptase enzyme and oligo dT (Invitrogen). All kits were used according to manufacturers’ protocols. Equal amounts of cDNAs were used in each quantitative PCR amplification, run in duplicate on the same plate. PCR product was generated and detected using the iCycler real-time PCR detection system (Bio-Rad) using the Power SYBRR® Green PCR or TaqMan Master Mix (Applied Biosystems). Gene expression levels were normalized to GAPDH by the ΔΔCt method (24). SYBRR® Green-based assay with custom designed primer sequences was used for IFN-γ, IL-17A, IL-10, GAPDH, and leptin (sequences available upon request). cDNAs from DE samples were also assayed by qPCR for transcripts of the IL-23 receptor (IL23R), IL-22, IL-12A, and IL-23A and GAPDH using TaqMan assays.

Statistics

Onset of disease manifestations was analyzed by Kaplan-Meier statistics. Effect of castration was assessed by ANOVA, chi-square, and Kruskal-Wallis tests. Serum testesterone levels were analyzed by ANOVA with Tukey’s multiple comparisons test. Cytokine qPCR values were compared by the Kruskal-Wallis test. P values lower than 0.05 were considered statistically significant.

RESULTS

Spontaneous epididymo-orchitis is the earliest clinical manifestation in F1 males

Table 1 summarizes the F1 and parental transgenic lines carrying B27 and Hu-β2m lines used in this study. As previously reported (17), male F1 rats develop severe EO with nearly 100% incidence, and this precedes the development of arthritis and spondylitis. Fig. 1A shows the time course and cumulative incidence of scrotal swelling, peripheral arthritis, and tail spondylitis.

Figure 1.

(A) Cumulative incidence of clinical manifestations in 79 male F1 rats. (B) Serum testosterone levels in nontransgenic controls and 8 groups of F1 rats (mean ± SD of samples taken between ages 64 and 221 d). Significant differences: g vs. each other group (p < 0.02), d vs. a and c (p < 0.05), and f vs. a and c (p < 0.05). The nontransgenic (a), and unmanipulated (b) and sham-castrated (c) F1 rats, showed an age-dependent linear decline in serum testosterone levels, (pooled slope = −0.0077 ng/ml/d). Some rats were sampled more than once. (C, D) Deficiency of the Dazl gene prevents both EO and arthritis in F1 rats. 21-3 x 283-2 males, with (n=26) or without (n=20) the Dazl^def transgene, were followed for scrotal swelling and arthritis. None of the Dazl deficient rats showed any clinical signs of disease. Maximum arthritis score in the Dazl wild type rats was 3.5 ± 0.9, age of onset 148 ± 18 d (mean ± SD). (E) Time course of serum anti-testis cell and antisperm antibodies in F1 male rats, showing increasing titers during the period of increasing testis pathology and onset of arthritis. Antibodies were also found in 21-3 hemizygous males.

Castration prevents the development of spondyloarthritis

In the original high-copy disease-prone B27/Hu-β2m lines, the males developed EO (16, 25). In those lines, arthritis was observed in both sexes but was invariably preceded by colitis (25), and both arthritis and colitis were prevented by rendering rats either germfree or genetically athymic (26, 27). This suggested that the colitis and arthritis shared pathogenic mechanisms, but the subsequent observation that the F1 males develop severe arthritis in the absence of colitis (17) indicates that these two processes are not necessarily interdependent.

Because in the F1 rats the females remain healthy and neither sex develops colitis, we asked whether the EO is a necessary precursor for the arthritis or spondylitis seen in the males. Rats underwent bilateral EOx, unilateral EOx, or sham castration between ages of 36 and 125 d, and were subsequently observed to age > 300 d for the development of arthritis and spondylitis. Androgen replacement was given either by subcutaneous implantation of testosterone pellets or by biweekly injections of testosterone ester.

The results are summarized in Table 2. None of 18 rats castrated before 92 d of age developed arthritis or spondylitis, compared with 5 of 14 rats sham castrated, and 13 of 21 rats hemicastrated before 92 d of age (p=0.0003 for comparison of the 3 groups by chi-square, p = 0.0058 for comparison between castrated and sham castrated groups). Of 11 rats castrated between 92 and 125 d, three rats developed arthritis at ages 172, 214, and 403 d of age. Five of seven rats sham castrated at 92 to 115 d developed arthritis and spondylitis with the expected age of onset and severity.

Table 2.

Castration prevents arthritis and spondylitis in (21-3x283-2) F1 HLA-B27/Hu-β2m transgenic male rats

Procedure*	Age at procedure (d)	Total No.	No. with arthritis†	Age onset arthritis (median, d)‡	Maximum arthritis score (mean ± SD)§	No. with spondylitis¶	Age onset spondylitis (median, d)#
Castration	43–91	18	0	—	—	0	—
Sham castration	43–91	14	5	152	3.9 ± 2.5	1	211
Hemicastration	36–91	21	13	148	4.8 ± 1.5	6	168
Castration	92–125	11	3	214	2.3 ± 2.3	1	290
Sham castration	92–115	7	5	158	5.4 ± 1.8	5	202

^*Castrated rats underwent bilateral epididymo-orchiectomy at the indicated ages. Hemicastrated rats underwent left-sided epididymo-orchiectomy. Sham-castrated rats underwent scrotal incision and closure.

^†Arthritis frequency (number with arthritis/total), p < 0.0004 by chi-square test

^‡Age onset arthritis, p = 0.08 by Kruskal-Wallis test

^§Arthritis scores, p = 0.28 by ANOVA

^¶Spondylitis frequency, p = 0.0004 by chi-square test

^#Age onset spondylitis, p = 0.28 by Kruskal-Wallis test

All p values shown here are for comparisons across all 5 groups. See Results for comparisons of arthritis frequency between groups

Serum testosterone levels were measured at varying ages in all groups. Fig. 1B summarizes the levels between 64 and 200 d of age. Unmanipulated and sham-castrated F1 rats had levels that were not significantly different from non-transgenic controls. Hemicastrated rats showed significantly lower levels, compared with the nontransgenic and sham-castrated groups. Among the castrated rats, testosterone ester replacement produced significantly higher levels than all other groups.

Overall, the data suggest that factors related to the testis and/or epididymis are needed for the subsequent development of spondyloarthritis. It is formally possible that protection from arthritis was related to the nonphysiologic androgen replacement, although the overlap in testosterone levels between the groups with or without arthritis argues against this.

Phenotype in Dazl-deficient rats

To examine the role of EO in rats with normal endocrine function, we crossed the 21-3 and 283-2 transgene loci into rats carrying a transgenic knockdown construct for the autosomal gene Dazl (a rat homolog of the human Y-chromosome encoded gene DAZ [deleted in azospermia]), which encodes a germline specific RNA-binding protein essential for gametogenesis (18, 19). Spermatogenesis in these rats is arrested, and they fail to produce elongated spermatids, but androgen function is unaffected, and they are fertile if the testis is repopulated with normal spermatagonial stem cells (19).

In rats with the (21-3x283-2)F1 genotype, those with wild type Dazl developed EO and arthritis with the predicted time course and frequency. In contrast, the Dazl-deficient littermates showed no signs of either EO or arthritis (Fig. 1C, D). Tail spondylitis was seen in the wild type Dazl F1 rats, but not in the Dazl-deficient F1 littermates (not shown). Testosterone levels in the Dazl-deficient F1 rats were not significantly different from their Dazl wild type littermates, or from the levels in nontrangenic or F1 Lewis rats (Fig. 1B). To confirm that the Dazl knockdown transgene did not itself interfere with joint inflammation, a 66 d old Dazl-deficient (21-3x283-2F1 and two Dazl wild type F1 rats (53 and 74 d old) were each injected with complete Freund’s adjuvant. Typical adjuvant-induced arthritis and spondylitis developed in both Dazl genotypes (not shown). These results support the interpretation that the castrated F1 rats failed to develop arthritis because EO was prevented or interrupted and not because of nonphysiologic androgen replacement.

Histopathologic progression of the lesions

To characterize the development of EO, the testis and epididymis from F1 rats between 30 and 125 d of age were examined histologically. Because scrotal swelling was not seen before age 80 d and castration through day 91 apparently protected against subsequent arthritis, it was surprising that inflammation was found as early as 30 d of age, when some rats showed inflammation in the ductuli efferentes (DE), i.e., the small ducts linking the rete testis to the epididymis (Fig. 2A–C), with focal periductal neutrophils and mononuclear cells (Fig 2D).

Figure 2.

(A) Rat testis and excurrent ductal system. (B) Normal DE at age 36 d, surrounded by adipocytes (x100). Seminiferous tubules of testis at bottom left. (C) Normal DE at 36 d (x480). (D) Earliest pathology in DE at age 30 d; spermatocytes sloughed from testis at puberty are present as the round cells in the ductal lumen. There is heavy infiltration of neutrophils confined to interstitial space between the ducts. (x400). (E) EO pathology at 70 d shows severe monocytic inflammation confined to DE (yellow window, x200), completely sparing the testis (x100) (F) Advanced EO pathology at 77 d consists of interstitial fibrosis in DE with monocytic inflammation and necrosis. The seminiferous tubules are devoid of spermatogenic cells or sperm (x100; window: x400) (G) Severe monocytic inflammation of the testis, including multinuclear inflammatory giant cells (arrows in yellow window, x200) surrounded by aspermatogenic tubules; some tubules have giant spermatogenic cells with multiple apoptotic nuclei (arrows in blue window, x400). (x100) (H) Testis in Dazl^def F1 rat: (a) Hypospermatogenesis due to arrested development of round or elongated spermatids (x400); (b) Empty and atrophic DE (x200); and (c) no sperm in epididymis. Note giant apoptotic spermatocytes in lumen of cauda epididymis (arrows, x400).

More advanced DE lesions were seen by age 46 d, with more extensive infiltration of neutrophils and mononuclear cells. From 56 d to 70 days, mononuclear and granulomatous inflammation was extensive and progressive but remained confined to the DE; the testis and the epididymis appeared normal, with normal sperm production (Fig. 2E). Interestingly, the inflammation in DE usually targeted selectively only one of the 5 to 7 parallel ducts characteristic of rat DE (Figs. 2A, B). By 77 d, aspermatogenesis rapidly developed in some specimens of testis and epididymis (Figs. 2F), coinciding with the appearance of autoantibodies in serum (Figs. 1E). By 91 d, there was granulomatous orchitis with multinuclear giant cells and occasional neutrophils (Fig. 2G); this coincided with the first appearance of scrotal swelling (Fig. 1A).

Although similar lesions in the DE, testis, and epididymis were also seen in rats hemizygous for the 21-3 locus alone (Table 1), scrotal swelling in these rats was milder. The 21-3 males remain fertile to ~110 d of age, and neither sex develops arthritis (16, 17, 25).

Immunohistochemistry

DE were examined in two F1 rats with early lesions at 30 d of age. Testis, DE, and epididymis were examined from six F1 rats ages 93–142 d with well-established lesions. In the early DE lesion, foci of neutrophils were seen in the interstitial spaces (not shown). T cells, dendritic cells (DC) (CD103+), and macrophages staining positively for CD68 and/or CD163 were also seen (Figs 3A, C, D). The DE epithelium stained prominently for MHC class II (Fig. 3B). Neutrophils, T cells, DC, and epithelial staining for MHC class II were much less prominent in 30 d old control DE (Fig. 3E–G and not shown), whereas CD68+ macrophages were similar in the F1 and control DE (Fig. 3D, H. At 93 d, when granulomas with abundant T cells were prominent, the DE was heavily infiltrated by T cells, some invading the ducts (not shown). B cells were not seen in the DE lesions.

Figure 3.

Immunofluorescence staining of age 30 d DE inflammatory lesions in F1 rats (A–D), and 30 d DE in control Lewis rats (E–H). (Blue staining represents DAPI staining of nuclear DNA.) (A, E) αβ T cell receptor (mAb R73), with modest infiltration in the inflammatory lesion. (B, F) MHC class II (mAb OX6), with strong expression on both epithelium and infiltrating cells in the inflamed lesion, but expression only on interstitial cells in the control. (C, G) CD103, primarily expressed on dendritic cells (mAb OX62). Moderate infiltration in the inflamed lesion, compared with control. (D, H) CD68, expressed on type I macrophages (mAb 1C7), with comparable expression in both the inflamed lesion and control DE.

Although macrophages were prominent in both DE and testis with severe EO, the CD68+ subset (mainly classical macrophages) and the CD163+ subset (mainly alternatively activated macrophages) showed very distinct distributions. CD68+ cells predominated in the interstitial space surrounding granulomas, while the CD163+ cells were found almost exclusively within granulomas. Orchitis was associated with extensive infiltration of CD4+, and to a lesser extent CD103+ (presumably dendritic) cells within the granulomas, whereas CD8+ cells were outside the granulomas. In addition, granular deposition of rat IgG, presumably immune complexes, was detected focally at the periphery of the seminiferous tubules (not shown).

Sperm and testis cell autoantibodies

Antisperm and anti-testis cell antibodies were detectable after 70 d of age (Fig. 1E). Indirect IF showed several staining patterns (not shown). The appearance of these antibodies coincided with disappearance of sperm from the seminiferous tubules. However, there was no apparent qualitative or quantitative association of the antibodies with the subsequent development of arthritis, and similar antibodies at somewhat lower levels were also found in 21-3 males (Fig. 1E). High titer anti-sperm antibodies stained normal testis, but failed to bind testis of Dazl deficient rats (not shown). This indicates that testis autoantigens absent in the Dazl deficient rat are essential inductive and target autoantigens of the spontaneous EO.

In situ hybridization

This was carried out to investigate the sites of transgene expression in the testis. The mRNA signal for B27, Hu-β2m, and RT1 was uniform within the periphery of the seminiferous tubules, suggesting expression in Sertoli cells but not germ cells, and was also present in Leydig cells in the interstitium (not shown).

Cytokine analysis of inflammation

In attempt to characterize the cytokine response driving the inflammation, we measured mRNA levels by qPCR for IFN-γ, IL-10, and IL-17A, in DE, epididymis, and testis, in groups of F1, 21-3, and nontransgenic Lewis rats at 15 d intervals, from 30 to 90 d of age. Similar measurements for IL-23R, IL23A, IL-12A, and IL-22 were made in the DE specimens. Data from the measurements in DE are shown in Fig. 4.

Figure 4.

Quantitative PCR assessment of cytokine mRNA in the ductuli efferentes. Total cellular RNA, isolated from the DE of rats at different ages, was subjected to reverse transcription. The resulting cDNA was analyzed for the indicated mRNAs, and the results were normalized to GAPDH. For IFNγ, IL-17A, and IL-22, the 2^^{− ΔΔCt} values were multipled by 1,000 for graphic display (32). Each point represents the mean of duplicate determinations for a single rat. The median of the points in each column for each combination of genotype, cytokine, and time point is indicated by a horizontal bar. For IL-17A and IL-22, the medians are significantly different by the Kruskal-Wallis test (p = 0.049 and 0.038, respectively). The DE samples showed a leptin/GAPDH expression ratio of 0.145 ± 0.10 (mean ± SD, n = 24), consistent with their high adipocyte content. Additional qPCR data for testis and epididymis and for other cytokines in DE available on request.

Compared with nontransgenic rats, significant elevations of all 3 cytokine transcripts, normalized to GAPDH, were seen in the epididymis in some of the F1 and 21-3 rats at 75 or 90 d of age (not shown). A similar pattern was seen in the testis for IL-17A and IL-10 (not shown). In the DE, elevations in IFN-γ, IL-10, IL-17, and 1L-22 were seen in some of the F1 and 21-3 rats, but not in nontransgenics, particularly at 60 and 75 d (Fig. 4 and not shown). For IL-17A and IL-22, the difference in the medians at 75 d was statistically significant. The marked elevation of the related cytokines IL-17A and IL-22 in the same two F1 DE d 75 specimens is noteworthy, since the two cytokines were assayed by different methods. Elevation in IL23R expression was seen in the F1 and 21-3 rats by 60 d, but was also seen in one of 4 nontransgenic specimens at 60 and 90 d (Fig 4). Similar patterns were also seen for IL-12A and IL-23A (not shown).

Overall, the elevations in IL-23R, IL-17A, IL-22, and IL-23A are consistent with activation of the IL-17/23 pathway and the prominent neutrophilic infiltration, although the source of these cytokines is unclear (see Discussion). The elevations of IL-12A, IFN-γ, and IL-10 are consistent with the T cell and macrophage infiltration. That not all specimens showed cytokine elevations, especially in the DE, is consistent with the sporadic distribution of the inflammatory sites noted above.

DISCUSSION

Preventing EO prevents spondyloarthritis

The central insight provided here is that ablation of the testis, even after onset of EO, completely prevented the development of arthritis and spondylitis in the F1 rats. This is not likely due to hormonal alteration, since it was unaffected by testosterone restoration, and since B27/Hu-β2m transgenic Dazl-deficient rats, which have normal endocrine function, did not develop arthritis. These findings indicate that the early onset EO is causally linked to spondyloarthritis in the F1 rats.

Epididymo-orchitis develops early in B27/Hu-β2m rats

The consistent first site of inflammation, occurring as early as 30 d of age, is within the ductuli efferentes. The DE consist of 5–7 parallel ducts that coalesce into a common duct that is continuous with the epididymis (Fig. 2A). The efferent duct is homologous to the proximal tubule of the metanephric kidney, and its epithelium carries out reabsorption of >90% of the water and inorganic electrolytes in testicular fluid (28). The DE are known to be particularly vulnerable to autoimmune injury (21). The EO onset coincides with the physiological desquamation of the first wave of meiotic germ cells and their passage from the testis to the epididymis at the onset of puberty (29).

After d 30, the inflammation intensifies and becomes granulomatous, but remains largely confined to the DE until almost 2 months later. Severe granulomatous inflammation then develops in the testis that coincides with the appearance of anti-sperm and anti-testis germ cell antibodies in serum and is associated with depletion of normal germ cells from the seminiferous tubules.

Immune mechanisms driving EO

It is not yet clear whether the inflammation in the DE is initially driven by a transgene-induced defect in the regulation of innate immunity, or whether the rats have pre-existing effector T cells specific for male germ cell antigens that induce inflammation upon contact with these antigens. The influx of neutrophils in the earliest lesions in the DE suggests the involvement of innate immune signals producing an IL-17 response. Misfolding (30) and homodimer formation (31) of HLA-B27 heavy chains have both been implicated in inducing IL-23/17 activation, and these mechanisms may help explain the role of the transgenes in this process. The in situ hybridization results suggest that expression of B27 itself on germ cells can be excluded as a possible mechanism.

Several innate immune cell lineages expressing the IL-23 receptor have been identified, most of them associated with epithelium, that produce IL-17A, IL-22, and other cytokines and are associated with inflammatory disease (32, 33). It will be of interest to investigate whether similar cell types can be identified in the male rat reproductive tract and whether any are activated in the young F1 rats. The detection of IL23R expression in the DE lesions and even in control DE is consistent with the presence of cells capable of IL-17 production. Similarly, IL-12A and IL-23A elevations were seen in some nontransgenic DE specimens, perhaps related to the abundance of resident macrophages.

The subsequent infiltration by CD4+ T cells, production of T cell cytokines, and development of granulomas and autoantibodies indicate the presence of an antigen specific response. Whether primary or secondary to the pathogenesis of the EO, this is evidently an autoimmune process, since it is associated with autoantibodies and does not develop in rats deficient in haploid germ cells that express the cognate antigens. Moreover, the immunopathology is a phenocopy of experimental autoimmune orchitis induced by testis antigens in several species (34) and by T regulatory cell (Treg) depletion in vasectomized mice (35), and also of spontaneous orchitis in the dark mink (36).

Spermatogenesis commences after the acquisition of the immune repertoire, and testicular tissue is autoantigenic (37). The testis is a classical privileged immune site in which foreign grafts are relatively protected from rejection. The blood-testis barrier is maintained by tight junctions between neighboring Sertoli cells, critical for transport physiology in the seminiferous tubules and for immune protection (37, 38).

Tregs have been implicated in supporting testicular immune privilege. Autoimmune orchitis develops in vasectomized mice with Treg depletion but not by either treatment alone (35). Antigen-specific Tregs that suppress organ-specific autoimmunity show anatomic specificity, being preferentially located in the lymph nodes draining the organs they control (39).

DC can induce T cell differentiation into either Tregs or Th17 cells, depending upon the presence of TGFβ and the presence or absence of proinflammatory signals, including IL-6 or IL-21 (40). Only some B27/Hu-β2m transgenic rat lines are predisposed to inflammatory disease, and the disease predisposition correlates directly with DC dysfunction (41). DC from disease prone B27/Hu-β2m transgenic rats drive cocultured naïve CD4+ T cells toward a preferential expansion of Th17 cells, compared with nontransgenic DC(42). One plausible contribution to loss of tolerance in the excurrent ducts in these rats may therefore involve a skewing by DC away from a regulated state.

Role of EO in the pathogenesis of spondyloarthritis

It remains to be determined how the EO leads to spondyloarthritis. One possibility is molecular mimicry. Theoretically, germ cell antigens might be a potent trigger to autoimmunity through molecular mimicry because they are not protected by systemic tolerance. To our knowledge, however, there are no reports of association of anti-sperm immunity with arthritis.

Several lines of evidence implicate innate immunity in the spondyloarthritis of the B27/Hu-β2m transgenic rats. Neutrophils are prominent within the lesions of both peripheral arthritis and spondylitis in the (21-3x283-2)F1 rats¹, and microarray analysis of a limited number of joints with early arthritis suggests a strong IL-17 signature (unpublished data). Both male and female F1 rats also develop accelerated spondyloarthritis in response to doses of complete Freund’s adjuvant too low to cause arthritis in nontransgenic rats (43).

In rat lines with B27 transgene copy number ≥ 40, arthritis occurs in both males and females, with a higher prevalence in males (16, 25). The rats in those lines develop colitis before arthritis, whereas the (21-3x283-2)F1 rats used here show no evidence of gastrointestinal inflammation (17). The sequential onset of EO or IBD followed by arthritis then spondylitis (Fig. 1A, and Fig. 1 in reference (25)) suggests a process of progressive or spreading innate immune activation, beginning at mucosal or epithelial sites, or from the severely inflamed testes. The B27 and Hu-β2m transgene products evidently strongly influence this process both quantitatively and qualitatively, since the hemizygous 21-3 locus itself is sufficient to induce EO, but not IBD or arthritis, whereas in conjunction with the 283-2 locus it induces arthritis and spondylitis, and in the homozygous state it is associated with IBD and arthritis.

Immune privilege and spondyloarthritis

It is noteworthy that other prominent extraarticular sites of pathology associated with spondyloarthritis are also among those constitutively protected by immune privilege and tolerance. These include the anterior chamber of the eye, gut, nail matrix, and certain sites within the bone marrow. Together with the current findings, this suggests that dysregulated innate immunity at immune privileged sites may be an essential mechanism triggering spondyloarthritis. This hypothesis is discussed in more detail elsewhere (44).

Role of EO in human spondyloarthritis

As noted above, orchitis and epididymitis have been described in AS. Testicular biopsies have revealed chronic asymptomatic orchitis as a common finding in idiopathic male infertility (15), and a prominent Th17 response has been described in association with infertility-related chronic testicular inflammation (45). Whether this condition is associated with spondyloarthritis and/or with HLA-B27 has not, to our knowledge, been investigated. A high rate of sperm abnormalities apparently responsive to anti-TNF treatment has recently been described in SpA patients (46).

One interesting anatomic difference between rats and humans may be relevant. In rats, the individual ducts of the DE are surrounded by fat (see Fig. 2B), as part of the epididymal fat pad embedded within the superior epididymal ligament, whereas in humans they are embedded entirely in connective tissue (28). It is possible that the adipocytes that surround the DE in the rat promote inflammation (47) and autoimmunity (48). Further investigation of the mechanisms of EO and its link to spondyloarthritis in the transgenic rats, combined with clinical investigation, may provide deeper insight into the pathogenesis of spondyloarthritis and help explain the male predominance of AS.