TgA86 tmTNF transgenic mice model key features of human SpA
TgA86 mice have been previously described to develop peripheral arthritis with signs of axial involvement (11,20). Here we analyze in greater detail these mice with a primary focus on the characterization of their axial pathology, thus assessing its similarities to human SpA.
TgA86 mice presented with a significantly reduced body weight and body length compared to wild type (wt) mice of the same age (Fig. 1a, b). As early as 3 weeks of age, all mice, with no gender bias (Additional file 1; Figure S1), developed clinical signs of arthritis in all four limbs, which manifested as joint swelling, digit/limb deformation and joint stiffness ultimately resulting in loss of grip strength (Fig. 1c). Axial pathology was clinically evident as tail bending and eventually tail stiffness also accompanied by hyperkyphosis better evident by CT imaging (Fig. 1c, d). All pathological features worsened over time.
To better characterize the TgA86 axial and peripheral pathology, we performed CT imaging of 40-week-old TgA86 mice and wt littermates focusing on sites known to be important also in human SpA pathology, such as the spine, including the cervical, thoracic, lumbar, sacral and caudal segments, as well as the sacroiliac joints. In addition, we also examined the hind limbs to get a better grasp of the peripheral pathology. Interestingly, CT imaging revealed previously undetected pathological features both in the spine, especially at the sacral vertebrae, and the sacroiliac joints. More specifically, the frontal view of the TgA86 sacrum revealed structural changes in sacral vertebrae S3/S4 involving loss of their intervertebral space and fusion of the vertebrae, evident in 83% of the transgenic mice examined (Fig. 2a; red arrow). The side view of sacrum, also showed loss of the intervertebral space (Fig. 2b; red arrow) and fusion of the spinous processes, evident in 80% of the TgA86 mice examined (Fig. 2b; yellow arrow). Moreover, the superior parts of the sacral ala presented with bilateral structural damages in TgA86 mice (Fig. 2a; yellow boxes).
CT imaging of the upper body revealed extensive hyperkyphosis (Fig. 2c). Cervical vertebrae of TgA86 mice were found to be compressed with reduced intervertebral spaces (Fig. 2d), while lumbar as well as thoracic vertebrae did not show major abnormalities.
CT imaging of the caudal area, where axial pathology of TgA86 mice was originally observed (20), revealed significant changes of the vertebrae structure including loss of their bar bell shape and roughening of their surface accompanied in many cases by vertebrae bridging and fusion (Fig. 2e; red arrow).
CT imaging of the peripheral joints revealed thickening of the metatarsal bones (Fig. 2f; red arrow) and phalanges (Fig. 2f; yellow arrow), roughening of their surface as well as structural deformation of the posterior calcaneus at the site of Achilles entheseal insertion (Fig. 2f; blue arrow), while knee and ankle joints presented also with roughened surfaces and deformed tibia and femur heads (Fig. 2f).
To assess alignment of the TgA86 phenotype, with human SpA pathology features, we have evaluated common SpA-related comorbid pathologies. By examining intestine, eyes and skin tissue during disease progression we did not observe any signs of pathology. However, we demonstrated with functional and immunohistochemical data a comorbid left-sided heart valve pathology confirming and expanding previous findings (21). More specifically, we observed that at 40 weeks of age, TgA86 mice displayed aortic valve thickening, consisting mainly of vimentin-positive mesenchymal cells and sparse Gr1-positive neutrophils (Additional file 1; Figure S2a, b). The aortic valve thickening and fibrosis resulted in functional changes including increased aortic velocity (Additional file 1; Figure S2c) with mild regurgitation which was observed in 10-20% of the tested 40-week old mice. The mitral valve was not affected (Additional file 1; Figure S2b), and hence mitral velocity remained unchanged (Additional file 1; Figure S2d). Importantly, ejection fraction was significantly decreased (Additional file 1; Figure S2e), indicating cardiac function impairment probably caused by aortic insufficiency. Despite their impaired heart valve function, TgA86 mice did not display premature death and their electrocardiogram was normal.
TgA86 axial and peripheral pathologies evolve through an early phase of inflammation and bone erosion gradually leading to new bone formation
The main pathological features as well as the progression of the axial pathology of TgA86 mice was further studied in the tail vertebrae that exhibit consistently the more pronounced pathology. As pathology first becomes apparent in the vertebrae proximal to the body of the mouse, all further analysis was based on evaluation of the first five-six caudal vertebrae.
Histopathological evaluation of H&E stained sections of tail vertebrae revealed signs of inflammation starting from 4 weeks of age and progressively worsening with time. By 20 weeks of age, 100% of the transgenic mice examined exhibited aggravated pathological features (Fig. 3a; H&E) at the intervertebral joint area with signs of severe enthesitis (Fig. 3a; H&E; black arrows) and intervertebral disk degeneration (Fig. 3a; H&E). These pathological features were homogeneous in the different vertebrae of the same mouse when examined at 20 weeks of age, but, as disease progressed, heterogeneity of inflammation between different animals and also between the different vertebrae of the same animal, was evident. Interestingly, while the percentage of vertebrae with high inflammation was increased at the age of 20 weeks in TgA86 mice, at the age of 40 weeks it declined with a concomitant two-fold increase in the percentage of vertebrae with low inflammation (Fig. 3c).
Additional pathological features that were observed as early as 10 weeks of age and progressively became more pronounced, included the roughening of the vertebral surface, the squaring of the vertebral body as well as alterations in the bone marrow cavity, where the formation of ectopic cartilaginous matrix and bone marrow cell aggregates was observed (Fig. 3a; H&E).
The extent of bone erosion was evaluated based on the number of osteoclasts stained with TRAP in tail sections of TgA86 mice at different ages. High numbers of osteoclasts were observed in the vertebral body of the tail of TgA86 mice highlighting an active bone remodeling process occurring in these mice (Fig. 3a; black arrowheads). More interestingly, when focusing on the osteoclasts present at the sites of inflammation (enthesitis), we observed an increase in their numbers between 10 and 20 weeks of age and a subsequent drop by 40 weeks of age (Fig. 3c).
Staining of TgA86 tail sections with Safranin-O allowed the detection of ectopic cartilage, gradually leading to new bone formation which is a main pathological feature of human SpA. Starting at 20 weeks of age, we could observe fibrocartilage at the inflamed edges of the vertebrae (enthesis) (Fig. 3b; black asterisk and Fig. 3d), while at later ages, 30 and 40 weeks of age, specifically at sites where inflammation was reduced or absent, we could detect the presence of ectopic chondrocytes that occasionally bridged adjacent vertebrae (Fig. 3b; black asterisk and Fig. 3d).
Overall, the axial pathology in TgA86 mice appeared to evolve through an initial phase that involved pronounced inflammation and bone erosion, while as the disease progressed, between 20 and 30 weeks of age, the signs of inflammation and bone erosion were reduced and replaced by the formation of new cartilaginous tissue, eventually leading to bridging of the vertebrae detected later on in disease.
As sacroiliac joints are sites of preference for the manifestation of human SpA, we assessed histopathologically inflammation and tissue damage in the sacroiliac joints of TgA86 mice. As early as 10 weeks of age, TgA86 exhibited pannus-like tissue at the sacroiliac articular surface, while by 20 weeks of age, invasion of inflammatory cells to the sacrum and iliac bones was also evident (Fig. 3e; black arrowheads).
Peripheral pathology was studied by histopathological examination of the hind joints of TgA86 mice and was characterized by synovitis, enthesitis and focal subchondral bone erosions. By 10 weeks of age, TgA86 mice exhibited extensive enthesitis (Fig. 4a; arrow 1), pannus formation (Fig. 4a; arrow 2), and bone erosion (Fig. 4a; box), while by 20 weeks of age, pannus evolved to progressive destruction of the articular cartilage and subchondral bone (Fig. 4a; black asterisk) that, by 40 weeks of age, was greatly exacerbated.
Increased numbers of osteoclasts were detected in the hind joints at 10 and 20 weeks of age by TRAP staining which however, similarly to our observations in the tail vertebrae, decreased at 40 weeks of age (Fig. 4b). Interestingly, µCT trabecular analysis of femurs of 20 and 40-week-old TgA86 mice revealed an osteoporotic phenotype further indicating progressive systemic bone loss (Additional file 1; Figure S3).
At 20 weeks of age, we also observed the formation of ectopic safranin-O stained cartilaginous tissue at the inflamed enthesis (Fig. 4c; arrows) that was more pronounced at 40 weeks of age (Fig. 4d), leading to the development of new ectopic bone tissue (Fig. 4c; arrows and Fig. 4d).
New bone formation in tmTNF transgenic mice involves mechanisms of endochondral and membranous ossification
Immunohistochemical characterization of the axial pathology of 20-week-old TgA86 mice revealed that the increased cellularity observed during the inflammatory phase, consisted mainly of increased number of mesenchymal cells and neutrophils. More specifically, we observed accumulation of vimentin-positive mesenchymal cells at the enthesis and along the ligaments surrounding the vertebrae, but also inside the bone marrow cavity (Fig. 5a). Osteopontin staining at sites of cortical and cancellous bone as well as inside the bone marrow cavity appeared also increased, suggesting the accumulation of osteoblastic lineage cells, while increased periostin staining was also observed in affected vertebrae, and particularly at sites where bone erosion was evident, also suggesting increased osteoblastic activity (Fig. 5a).
These stainings provided valuable insight on the mechanisms that eventually lead to new bone formation observed at the advanced ages of TgA86 mice. More specifically, the increased vimentin and periostin staining indicates mesenchymal cell accumulation and increased osteoblastic activity in the periosteum that suggests an ongoing active membranous ossification mechanism. Similarly, the increased vimentin and osteopontin staining in the osseous vertebral cavity indicates vimentin-positive mesenchymal cell accumulation and condensation leading to osteopontin-positive osteoblastic lineage differentiation, suggesting an active endochondral ossification process.
Cell aggregates in the bone marrow cavity of TgA86 mice were found to consist mostly of B220-positive B cells (Fig. 5b). Similar structures have been previously reported in human RA patients as well as in TNF-dependent mouse models involving bone pathologies (22,23), with no evidence however of a critical contribution of bone marrow cells in the disease pathogenesis (22,24). Gr1-positive neutrophils were found to accumulate mainly at the sites of the enthesis as well as loosely scattered in the bone marrow cavity and along the periosteum (Fig. 5b).
In a similar fashion to the axial pathology, peripheral pathology in the hind joints was also characterized by the accumulation of vimentin-positive mesenchymal cells and Gr1-positive neutrophils at the sites of synovitis and enthesitis, while there was only sparse presence of B220-positive B cells in the bone marrow cavity (Additional file 1; Figure S4).
Anti-TNF therapy efficiently ameliorates both axial and peripheral TgA86 pathologies
We further assessed the response of the TgA86 model to anti-TNF treatment, a therapy that has been proven efficacious in the early treatment of human SpA patients (25,26). More specifically, we treated TgA86 mice with Etanercept starting either from 2.5 weeks of age, i.e. early in the inflammation phase, or from 9 weeks of age, i.e. late in pathology when inflammation is more extensive leading to activation of cell types contributing to bone remodeling. Treatment continued up to 20 weeks of age and its effect was evaluated both in vivo and ex vivo.
Clinical monitoring of the TgA86 pathology was performed by assessing key phenotypic features of the disease evident up to 20 weeks of age. More specifically, peripheral pathology was assessed on a scale from 0 to 2 taking into consideration the severity of hind joint swelling, finger and limb deformation as well as grip strength (Additional file 1; Table S1), while axial pathology was assessed on a scale from 0 to 3 taking into consideration the tail pathology, including the number and extent of tail bendings as well as tail ankylosis (Additional file 1; Table S2). Early treatment with Etanercept completely abolished both the axial and peripheral clinical manifestations of the TgA86 pathology, while late treatment only partially ameliorated the peripheral arthritis pathology and had a minimal effect on the severity of the clinical manifestations of the axial pathology (Fig. 6a).
The effect of the early and late treatment protocols was also assessed histopathologically as well as by µCT analysis upon completion of the treatment. Histopathological evaluation of the axial and peripheral pathology involved the assessment of inflammation, since at 20 weeks of age this was the most prominent feature of the pathology. Axial inflammation was assessed in the first 5-6 caudal vertebra starting from the base of the tail, while peripheral inflammation was assessed in at least 3 different joints (ankle and metatarsal in sagittal plane) using a scoring scale from 0-3 (Additional file 1; Tables S3 and S4). Early treatment with Etanercept completely abolished both the axial and peripheral inflammation, while late treatment also greatly ameliorated both the axial and peripheral inflammation although to a slightly lesser extent compared to the early treatment protocol (Fig. 6b and 6c).
The structural changes of the TgA86 tail vertebrae and their improvement following Etanercept treatment were also visualized using 3D µCT modeling. The TgA86 tail vertebrae exhibited reduced length compared to the vertebrae of the wt mice and interestingly, early anti-TNF treatment, and to a lesser degree late treatment, could partially restore this feature (Fig. 6d). Moreover, 3D modeling highlighted the changes in the shape and surface of the TgA86 tail vertebrae that appeared squared with a rough surface and a dense matrix structure inside the bone marrow cavity, characteristics that were respectively restored following both early and late anti-TNF treatment to the bar belled shape and smoother surface with absence of dense matrix in the bone marrow cavity observed in wt mice vertebrae (Fig. 6e).