Spleens from HLA-B27 tg, HLA-B7 tg, or wild-type rats (n = 6 rats/group) were collected and mononuclear cells were isolated using Ficoll. Cells were stimulated with 1, 5, or 25 µg/ml heat-inactivated Mycobacterium tuberculosis, 5 µg/ml zymosan, or 50 ng/ml LPS, using unstimulated samples as controls. Gene expression was measured in duplex reactions using SYBR green primers (sequences are available upon request). The relative expression was calculated with the “2^−ddCt method” (26). For protein analysis, supernatant was collected and analyzed by ELISA for TNF, IL-6, IL-1β (R&D) according to the manufacturer’s protocol.

The transgenic 21-3 (HLA-B27/Huβ2m), 120-4 (HLA-B7/Huβ2m), and 283-2 (Huβ2m) rat lines, all on the inbred Lewis background, were maintained in our breeding facility at UTSWMC (Dallas, TX, USA) or AMC (Amsterdam, The Netherlands). For experiments, F1 rats [21-3 × 283-2] and/or [120-4 × 183-2], both males and females, were used at 6–12 weeks of age. Rats were housed, three to four per cage, either conventionally or in individually ventilated cages with appropriate cage enrichment and ad libitum access to water and chow. All animal experiments were performed either at UTSWMC or at AMC and were approved by the Institutional Animal Care and Use Committee.

6-week-old male and female rats were immunized with 30–90 µg of pulverized heat-inactivated M. tuberculosis (Difco, Detroit, MI, USA) in 100 µl incomplete Freund’s adjuvant (Chondrex) via subcutaneous injection in the tail base.

The presence of arthritis in the paws was determined macroscopically and digital swelling was measured with plethysmometry where stated. Arthritis severity in each paw was graded 0–3: 0 = normal joints, 1 = 1 swollen joint, 2 = 2 or more swollen joints, and 3 = extreme swelling of the entire paw and/or ankylosis. Cumulative clinical scores per rat were calculated and used for severity analysis. Swelling (cubic centimeter) was normalized either to the days before disease onset in case of prophylactic treatment or to the day of start treatment in case of therapeutic treatment. Spondylitis was determined macroscopically (yes/no). In case of severe disease, rats were sacrificed due to ethical considerations, with the last observation carried forward. Severe disease was defined by 15% bodyweight loss or two completely swollen paws for validation experiments or prophylactic treatment; and 20% bodyweight loss or four completely swollen paws for therapeutic treatment. During treatment experiments, scoring was performed by a blinded observer.

Surgery was performed using standard methods (Protocol Envigo, Horst, The Netherlands). In brief, 4-week-old rats received 5 mg/kg rimadyl 15 min prior to surgery, rats were anesthetized with isoflurane + 1 l/min O₂. A medial skin incision of 2–3 cm was made, caudal to rostral and a similar size incision was made in the peritoneum, over the Linea Alba. Both testicles, including epididymis and VAS were removed and the peritoneum and skin were sutured.

Rats received 150 mg/kg thalidomide or PBS daily via oral gavage (n = 9/group) in a prophylactic experiment. In separate experiments, rats received 10 mg/kg etanercept or PBS twice weekly via subcutaneous injection, prophylactically (n = 6/group) or therapeutically (n = 5–6/group). Prophylactic treatment started 7 days after immunization, therapeutic treatment started 1 week after 50% arthritis incidence. Treatment continued for 5 weeks in all experiments. In case of therapeutic treatment, rats were randomized based on arthritis score on the first day of treatment.

Paws and tail tissues were fixed, and sectioned as previously described (27). Sections were stained for hematoxylin and eosin or safranin O/fast green. Stained sections were semi-quantitatively scored by two independent, blinded observers (Melissa N. van Tok and Leonie M. van Duivenvoorde). Sections were scored 0–3 for inflammation, destruction, new bone formation, and hypertrophic chondrocytes (0 = normal, 1 = mildly affected, 2 = moderately affected, and 3 = severely affected). In the validation studies, histological scores from HLA-B27 transgenic males immunized with 30 µg heat-inactivated M. tuberculosis were pooled together with the females immunized with 60 µg heat-inactivated M. tuberculosis (n = 10). From the HLA-B7 transgenic rats, the rats immunized with matching doses (30 µg for males, 60 µg for females) of heat-inactivated M. tuberculosis were used to compare with (n = 6). In the therapeutic treatment experiment with etanercept, tissue from age-matched healthy controls was taken along.

Data were statistically analyzed using GraphPad Prism software. Values of p < 0.05 were considered significant. For comparison between two groups, a Mann–Whitney U-test was performed, for multiple group comparison we used a two-way ANOVA or a Kruskal–Wallis test and for the survival graphs (incidence) we used a log-rank test.

To test the hypothesis that HLA-B27 overexpression increases the sensitivity toward innate immune activation, splenocytes of non-immunized, non-diseased HLA-B27 tg rats, HLA-B7 tg rats, and wild-type control rats (n = 6/group) were stimulated with 1, 5, or 25 µg/ml heat-inactivated M. tuberculosis (Figure 1A). Unstimulated splenocytes had low but similar mRNA expression of all measured cytokines in all three rat genotypes. Splenocytes of HLA-B27 tg rats showed significantly increased tnf, il1a, il1b, and il6 expression upon 5 and 25 µg/ml M. tuberculosis stimulation compared with both controls. Il23a (p19) and il10 expression were not increased. These data were confirmed at the protein level by TNF and IL-6 measurements in the supernatant (Figure 1B), with IL-1β being low to undetectable in all conditions (data not shown). As M. tuberculosis can activate a variety of innate immune receptors (28) as well as antigen-specific adaptive immune responses (29), we confirmed these observations by showing a similar increase in pro-inflammatory cytokine production upon stimulation of HLA-B27 tg rat-derived versus control splenocytes with the TLR2/dectin-1 ligand zymosan, but not with the TLR4 ligand LPS (Figure S1 in Supplementary Material). Taken together, in vitro innate immune activation—in particular through TLR2/dectin-1—causes a significant upregulation of pro-inflammatory cytokines in the HLA-B27 tg rat splenocytes.

Male HLA-B27 tg rats spontaneously develop spondylitis and arthritis with an incidence of 40 and 70%, respectively, and a variable disease onset between 4 and 9 months of age. Female rats on the other hand are not capable of spontaneous spondylitis and arthritis development (19). Based on the upregulation of pro-inflammatory cytokine production in vitro, we hypothesized that in vivo activation of the innate immune system would trigger spondylitis and arthritis development in the HLA-B27 tg rats. Heat-inactivated M. tuberculosis was selected for immunization, based on the arthritis model adjuvant induced arthritis (AIA). It is commonly known that wild-type LEWIS rats do not develop spondylitis after immunization with (low doses of) heat-inactivated M. tuberculosis. Due to ethical considerations, no wild types were included in the in vivo experiments. HLA-B27 and HLA-B7 tg males and females were immunized with low doses of heat-inactivated M. tuberculosis and monitored for clinical spondylitis and arthritis development (Table 1). In male rats, immunization with higher doses of heat-inactivated M. tuberculosis (60 and 90 µg) induced spondylitis and arthritis in both HLA-B27 tg rats and HLA-B7 tg controls. Upon immunization with 30 µg of heat-inactivated M. tuberculosis, however, none of the HLA-B7 tg rats developed disease whereas 100 and 80% of the HLA-B27 tg males developed spondylitis and arthritis, respectively. To confirm and extend these findings, we first performed similar experiments in female HLA-B27 tg rats which, in contrast to male rats, do not develop spontaneous spondylitis and arthritis (19, 20). None of the HLA-B7 tg female control rats developed spondylitis or arthritis upon immunization with heat-inactivated M. tuberculosis up to a dose of 90 µg. In contrast, all female HLA-B27 tg rats immunized with 60 or 90 µg heat-inactivated M. tuberculosis developed both spondylitis and arthritis. Epididymo-orchitis has been shown to be a pre-requisite for the development of spontaneous spondylitis and arthritis in the HLA-B27 tg rat (20). We confirmed that innate immune activation triggers spondylitis and arthritis by demonstrating that immunization with 30 or 60 µg of heat-inactivated M. tuberculosis could overcome the protective effect of orchiectomy in male rats. Immunization with 30 or 60 µg heat-inactivated M. tuberculosis was sufficient to induce both spondylitis and arthritis in all HLA-B27 transgenic males. Collectively, these data indicate that innate immune activation using heat-inactivated M. tuberculosis immunization triggers spondylitis and arthritis development in HLA-B27 transgenic rats, including disease-resistant female and orchiectomized male rats.

The previous in vivo pilot experiments were performed in the animal facility of UTSW Medical Center in Dallas, TX, USA (except for the orchiectomy experiment). To confirm our findings and validate the reproducibility and robustness of the model, the experiments were repeated, independently in the animal facility of the AMC in Amsterdam, The Netherlands. A total of 33 HLA-B27 tg rats were immunized with heat-inactivated M. tuberculosis (30 µg in 24 males and 60 µg in 9 females) in four independent experiments. Due to ethical considerations, wild types or non-immunized HLA-B27 transgenic rats were not included in these experiments as it has repeatedly been shown that low amounts of heat-inactivated M. tuberculosis (30–60 µg) did not induce spondylitis in wild types (30–32). Non-immunized females or orchiectomized males do not develop disease (19, 20). In all four experiments, 80–100% of the rats developed spondylitis (Figure 2A) and arthritis (Figure 2B) with a disease onset between days 14 and 21 after immunization (Figure 2C). Clinical scoring of the arthritis (0–3 per paw with a maximum score of 12 per rat), revealed a similar arthritis severity in all four experiments with an average score of 8 (Figure 2C). These results confirm that immunization with heat-inactivated M. tuberculosis reproducibly and predictably triggers clinical spondylitis and arthritis in HLA-B27 tg rats.

HLA-B27 tg rats that spontaneously develop spondylitis and arthritis show histopathological features of inflammation, cartilage and bone destruction, and in particular new bone formation in axial and peripheral joints (27). Histological analysis of spine and ankles was done to evaluate whether similar histopathology is present upon immunization with heat-inactivated M. tuberculosis. In the spine, we detected inflammatory cell infiltration around the intervertebral discs, vertebral endplate destruction, periosteal new bone formation, and ectopic foci of hypertrophic chondrocytes in HLA-B27 tg rats but not in HLA-B7 tg controls (Figures 3A,B). Similarly, the ankle joints were histologically normal in HLA-B7 tg rats but depicted marked articular and peri-articular inflammation, cortical bone destruction, periosteal new bone formation, and hypertrophic chondrocytes in HLA-B27 tg (Figures 3C,D). The axial and peripheral histopathology was similar to what we previously described in a cross sectional study of spontaneous spondylitis and arthritis in male HLA-B27 tg rats including, unlike what is seen in classical arthritis models, typical spinal involvement, and new bone formation (27).

The high disease incidence of 80–100% in both male and female rats, the predictable onset between days 14 and 21, and the robust clinical and histopathological features provided the opportunity to perform intervention studies in the heat-inactivated M. tuberculosis-induced disease in HLA-B27 tg rats. To validate this concept, rats were prophylactically treated with thalidomide, a small molecule targeting, amongst others, the TNF pathway (33). Treatment with thalidomide or vehicle started 1 week after immunization and continued daily for 5 weeks (Figure 4A). Spondylitis and arthritis development were significantly delayed and arthritis severity had the tendency toward a decrease in the thalidomide treated rats when compared with the vehicle treated rats (Figure 4B). Histopathological analysis of the four disease features (inflammation, destruction, new bone formation, and hypertrophic chondrocytes) confirmed a trend toward lower scores in the spine (Figure 4C) and significantly scores in the ankle joints (Figure 4D) in thalidomide versus vehicle treated rats. These data demonstrated that the model is suitable for intervention studies, allowing us to use targeted therapies to establish the role of key innate pathways in the HLA-B27 associated disease process.

TNF is one of the major pro-inflammatory cytokines in innate immune inflammation and plays a crucial role in human SpA. We therefore tested whether TNF inhibition with etanercept could prevent and/or treat experimental SpA in the heat-inactivated M. tuberculosis-induced disease in HLA-B27 tg rats. Similarly to thalidomide, etanercept treatment was started 1 week after immunization but before onset of clinical disease (Figure 5A). Etanercept treatment delayed the appearance of spondylitis and arthritis and suppressed arthritis severity (Figure 5B). Inflammation, destruction, new bone formation, and hypertrophic chondrocytes were not different between the two groups, both in the spine (Figure 5C) and in the ankle joints (Figure 5D).

In contrast to prophylactic treatment, therapeutic treatment with etanercept, initiated at day 31 (Figure 6A), did not affect incidence or severity of clinical disease (Figure 6B). The absence of therapeutic effect in this setting was confirmed by histological analysis, where in contrast to the age-matched healthy controls both vehicle and etanercept treated rats presented with histopathological processes of SpA in the spine (Figure 6C) and the ankles (Figure 6D), as semi-quantitatively assessed. Collectively, these data provide evidence that experimental SpA in HLA-B27 tg rats is, at least partially, TNF dependent and suggests that the role of this key innate immune cytokine is more important in the induction phase than in established disease in this particular model.