Early anti-inflammatory intervention ameliorates axial disease in the proteoglycan-induced spondylitis mouse model of ankylosing spondylitis

All experimental procedures were approved by The University of Queensland animal ethics committee and adhered to by the Australian code for the care and use of animals for scientific purposes. In order to achieve 100% disease penetrance and robust severity, disease was induced as described previously [14] in 12-week-old IL4-deficient female BALB/c mice by intraperitoneal injection of partially purified cartilage extract (equivalent with 100 μg core protein of human cartilage proteoglycan (PG)) emulsified with 2 mg dimethyldioctadecylammonium bromide (DDA, Sigma-Aldrich, St. Louis, MO) in a total volume of 200 ml phosphate-buffered saline (PBS, pH 7.4) at day 0, and then boosted with 150 ml emulsion on weeks 3 and 6.

PGISp mice were treated with vehicle or ETN plus PRD (ETN + PRD). Mice were distributed randomly to treatment groups at first PG priming injection. ETN (provided by Pfizer, NY, USA) was administered at 10 mg/kg by subcutaneous injection 3 times a week. PRD was administered at 1.5 mg/kg/day by a slow-release pellet (Innovative Research of America, Sarasota, FL, USA) implanted subcutaneously (Fig. 1a). Vehicle groups were treated with PBS by subcutaneous injection with or without placebo pellets. All treatments started 6.5 weeks post the initial PG priming injection, so that the immune response was primed and re-activated in the absence of any anti-inflammatory treatment. ETN + PRD treatments were continued for 10 days (short-term, 6.5–8 weeks) or ~6 weeks (long-term, 6.5–12-weeks) (Fig. 1a). Peripheral inflammatory symptoms were scored ranging from 0 to 3 based on the degree of redness and swelling three times weekly. Cumulative arthritis scores of the four paws are presented as per previously [17]. Mice were euthanized at the end of each of the assigned treatment regimens and tissues collected for histological analysis.

Dissected thoracic to lumbar spines were decalcified, processed and sectioned (4 μm) in the sagittal plane. Sections were processed for standard haematoxylin and eosin (H&E) and toluidine blue staining as described previously [16].

A semi-quantitative histological scoring criterion based on H&E stain and toluidine stain was used to assess axial disease progression [16]. Briefly, the score for an intervertebral joint was produced by averaging the scores of ventral (anterior) and dorsal (posterior) aspects. An overall average score for each animal was then generated by averaging the scores for all assessable intervertebral joints in each sample (range 6 to 24 intervertebral joints scored in each sample (median = 13)). Inflammatory score: 0, no evident inflammatory infiltrate; 1, inflammation present at periphery of the IVD; 2, inflammatory pannus smaller than 50% IVD area; 3, inflammatory pannus invading more than 50% IVD area. IVD destruction score: 0, no evident disc destruction; 1, less than 50% disc destruction; 2, more than 50% disc destruction; 3, complete disc destruction. Bone erosion score: 0, no erosion; 1, one or a few small areas of resorption in original vertebral bone; 2, numerous areas of obvious focal resorption in original vertebral bone or several areas of severe destruction. Cartilage damage score: 0, no evident cartilage damage; 1, some loss of endplate cartilage and/or growth plate cartilage; 2, severe loss of endplate cartilage and some growth plate cartilage damage. 3, severe loss of both endplate and growth plate cartilages. Excessive tissue formation score: 0, no evident mesenchymal cells or chondrocytes around IVD; 1, mesenchymal cell invasion/expansion; 2, fibrocartilage formation smaller than 50% of the original disc area; 3. fibrocartilage formation larger than 50% of the original disc area. Intervertebral joints that scored one or greater in any of the criteria described above were defined as affected joints. PGISp mice with at least one involved inter-vertebral joint were considered as affected mice. The incidence of axial disease was the percent of affected mice within the group. The axial disease penetrance was defined as the percent of affected intervertebral joints within an individual mouse. Based on expert biostatistical advice from QFAB Bioinformatics, further subgrouping of mice into low and high disease activity was based on median IVD destruction score (vehicle: 1.05 and ETN + PRD: 0.66).

Mann-Whitney tests were used to determine statistically significant differences in univariate analyses. The relationship between histological features was determined by Spearman correlations. These statistical analyses were performed using PRISM 6 (GraphPad Software, La Jolla, CA). The statistical significant differences between the proportions of outcomes between treatment groups were determined using one-tailed Z-tests (http://​epitools.​ausvet.​com.​au/​content.​php?​page=​home). P values less than 0.05 were considered significant in the present study.

We designed and implemented an early and intensive anti-inflammatory therapeutic approach in the PGISp model. Our intention was to initiate treatment post disease initiation but prior to irreversible IVD destruction [16]. In the absence of useful serum markers of axial disease progression (Additional file 1: Figure S1), we selected the treatment commencement point based on our previously reported histology evidence showing that 60 and 100% of PGISp mice exhibited axial inflammation (Fig. 1a) by weeks 6 and 8 post initial priming, respectively. Clear progression of IVD destruction (Fig. 1a) was seen in more than 50% of mice by 8 weeks post-priming with peak IVD destruction severity reached by 10 weeks with 100% of mice affected [16]. Extensive osteoproliferation (Fig. 1b and c) was observed at the affected intervertebral joints initiating 8 weeks post initial priming [16]. Accordingly, the treatment regimen in the present study was initiated 6.5 weeks post-priming to achieve a “therapeutic” delivery prior to development of irreversible IVD changes. Impacts of both short- and long-term therapies were examined with mice harvested either 2 or 6 weeks post-treatment initiation (Fig. 1d). The proportion of mice with no disease or peripheral and/or axial disease within each therapy arm was determined (Table 1). Clinical scores of peripheral joints were used to longitudinally assess systemic disease activity. Axial disease was assessed by histopathology with detection of one or more of the scored histopathological features being deemed as presence of axial disease. Eight weeks after PG priming, 80% of vehicle treated mice had evidence of peripheral and/or axial disease; in contrast, only 14% of mice in the ETN + PRD treated group had both. All mice developed evidence of disease 12 weeks post-priming irrespective of treatment. In the vehicle-treated group, 100% exhibited both peripheral and axial disease. In contrast, in ETN + PRD-treated mice only 62.5% had peripheral plus axial disease, indicating greater variance in systemic disease after early anti-inflammatory therapy. Taken together, these observations indicate that early treatment with ETN + PRD had robust short-term efficacy.

The onset of peripheral arthritis varied between individual mice, with the earliest onset observed 4 weeks post priming (Fig. 2) and no significant difference was observed in peripheral disease incidence at the time of therapy initiation between any of the treatment groups (Table 1). In vehicle-treated mice (Fig. 2, dashed line), the peripheral arthritis score increased from 4 weeks post-priming followed by an accelerated progression after the third PG injection and then “levelled off” until the end of experimental period. In ETN + PRD-treated mice (Fig. 2, solid line) the initial acute arthritis (between weeks 4 and 6.5) was suppressed immediately after the first treatment, and remained negligible for 2–3 weeks. Within this period (from the beginning of ETN + PRD until week 9), there was a significant difference in peripheral paw inflammation between the vehicle-injected and ETN + PRD-treated groups (Fig. 2, p < 0.05). Hence, the ETN + PRD treatment regimen delayed and suppressed the peripheral disease development.

ETN + PRD treatment impacted the incidence, penetrance, severity and progression of the axial disease. Specific features of axial disease activity in each animal were assessed using semi-quantitative histological scoring of spine sections. Due to disease severity being variable between intervertebral joints within individual animals (Additional file 2: Figure S2), global impact of treatments on the specific disease features was assessed by generating an average histopathological feature score for each spine (Additional file 2: Figure S2). ETN + PRD treatment significantly reduced the penetrance of axial disease (percent of affected intervertebral joints within an individual animal) in both short-term and long-term treatment arms (Fig. 3a).

When compared with vehicle controls, ETN + PRD treatment significantly suppressed axial inflammation in both treatment arms (Fig. 3b) supporting the effectiveness of the ETN + PRD regimen on spondylitis. IVD destruction (Fig. 3c), bone erosion (Fig. 3d), and cartilage damage (Fig. 3e) were evident in both short- and long-term vehicle control groups (Fig. 3f). Within the short-term ETN + PRD treatment, scores for IVD destruction and bone erosion were significantly lower than controls with a similar trend toward reduction in cartilage damage and excessive tissue formation (Fig. 3).

In the long-term treatment arm, scores for disease features in the vehicle treated group were non-parametric and exhibited a wide degree of variation (Fig. 3). When data points for the different histopathological features were paired for individual animals, it was clear that a subset (3 of 8 mice, Fig. 3, data points highlighted by the dashed boxes) of ETN + PRD treated mice had minimal or undetectable axial disease as indicated by no or little evidence of disc destruction, bone erosion, cartilage damage or excessive tissue formation (Fig. 3, boxed area). All 3 of these mice had detectable peripheral disease confirming successful disease induction.

While inflammation is not a robust indicator of axial disease progression with a non-linear disease process, IVD destruction is an irreversible event, and therefore its presence is a constant mark of previous disease activity. In the long-term vehicle treatment arm, IVD destruction scores significantly and positively correlated with bone erosion, cartilage damage and excessive tissue formation scores (Fig. 4a). In ENT + PRD long-term treated mice the correlation of IVD destruction with cartilage damage and excessive tissue formation was upheld, but bone erosion scores no longer correlated (Fig. 4b). Interrogation of the IVD destruction data distribution in the long-term treatment arm supported dichotomization of this data set with mild and severe disease courses defined by low versus high IVD destruction. Consequently sub-groups were segregated based on the median IVD destruction score (vehicle: 1.05 and ETN + PRD: 0.66).

When comparing the low IVD destruction sub-groups, there was no significant difference between treatments across all histopathological criteria (not shown). When comparing high (H)-ETN + PRD with H-Vehicle, the mean IVD destruction score was significantly lowered by ETN + PRD (Fig. 4c). Furthermore, H-ETN + PRD had significantly lowered percent of affected IVD (Fig. 4d), inflammation (Fig. 4e) and bone erosion scores (Fig. 4f). Trends toward treatment-induced lowering of cartilage damage (Fig. 4g) and excessive tissue formation (Fig. 4h) were also noted. Overall, in this sub-group of mice that had a more severe clinical progression, early intervention with ETN + PRD treatment ameliorated axial disease severity.