Transgenic disruption of endogenous glucocorticoid signaling in osteoblasts does not alter long-term K/BxN serum transfer-induced arthritis

Background Disruption of glucocorticoid (GC) signaling in osteoblasts results in a marked attenuation of acute antibody-induced arthritis. The role of endogenous GCs in chronic inflammatory arthritis is however not fully understood. Here, we investigated the impact of endogenous GC signaling in osteoblasts on inflammation and bone integrity under chronic inflammatory arthritis by inactivating osteoblastic GC signaling in a long-term K/BxN serum transfer-induced induced arthritis (STIA) model. Methods Intracellular GC signaling in osteoblasts was disrupted by transgenic (tg) overexpression of 11beta-hydroxysteroid dehydrogenase type 2 (11ß-HSD2). Inflammatory arthritis was induced in 5-week-old male tg mice and their wild type (WT) littermates by intraperitoneal (i.p.) injection of K/BxN serum while controls (CTRLs) received phosphate-buffered saline (PBS). In a first cohort, K/BxN STIA was allowed to abate until the endpoint of 42 days (STIA). To mimic rheumatic flares, a second cohort was additionally injected on days 14 and 28 with K/BxN serum (STIA boost). Arthritis severity was assessed daily by clinical scoring and ankle size measurements. Ankle joints were assessed histopathologically. Systemic effects of inflammation on long bone metabolism were analyzed in proximal tibiae by micro-computed tomography (μCT) and histomorphometry. Results Acute arthritis developed in both tg and WT mice (STIA and STIA boost) and peaked around day 8. While WT STIA and tg STIA mice showed a steady decline of inflammation until day 42, WT STIA boost and tg STIA boost mice exhibited an arthritic phenotype over a period of 42 days. Clinical arthritis severity did not differ significantly between WT and tg mice, neither in the STIA nor in the STIA boost cohorts. Correspondingly, histological indices of inflammation, cartilage damage, and bone erosion showed no significant difference between WT and tg mice on day 42. Histomorphometry revealed an increased bone turnover in tg CTRL and tg STIA boost compared to WT CTRL and WT STIA boost animals, respectively. Conclusions In contrast to the previously reported modulating effects of endogenous GC signaling in osteoblasts during acute K/BxN STIA, this effect seems to perish during the chronic inflammatory and resolution phase. These findings indicate that endogenous GC signaling in osteoblasts may mainly be relevant during acute and subacute inflammatory processes.

findings indicate that endogenous GC signaling in osteoblasts may mainly be relevant during acute and subacute inflammatory processes.Keywords 11ß-HSD2, Cortisol, Cortisone, Rheumatoid arthritis, Antibody, Joint inflammation, Cartilage, Bone

Background
Due to their potent anti-inflammatory and immunomodulatory properties, glucocorticoids (GCs) play an important therapeutic role in controlling inflammatory autoimmune disease activity [1,2].
Although their pharmacological actions and clinical (side-) effects are well characterized, the role of endogenous GCs in autoimmune arthritis is less well-understood [3,4].We previously showed that endogenous GCs are indispensable for osteoblastic differentiation and skeletal maturation through a Wnt-dependent pathway [5,6].
In rheumatoid arthritis (RA), chronic inflammation leads to structural bone damage locally (in the form of bone erosions) and systemically (in the form of osteoporosis), with osteoblasts representing key immunocompetent players in bone turnover regulation [7].
In response to inflammatory stimuli, the expression of 11ß-hydroxysteroid dehydrogenase type 1 (11β-HSD1), which converts inactive endogenous cortisone to active cortisol, is upregulated in both osteoblasts and synoviocytes [8,9], and high RA disease activity is associated with increased levels of 11β-HSD1 in the synovium [10].
We previously studied the effect of transgenic (tg) disruption of endogenous GC signaling through an overexpression of 11ß-HSD2 (which converts cortisol to cortisone) in mature osteoblasts and osteocytes in murine antibody-mediated arthritis.We showed that the inactivation of GC signaling in osteoblasts attenuated T-cell-independent K/BxN serum transfer-induced (STIA) and collagen II antibody-induced arthritis (CAIA) in the acute and subacute phases following disease induction [11,12].Further studies by our group showed that the T-cell-dependent model of antigen-induced arthritis (AIA) was however not affected by disrupted osteoblastic GC signaling.This suggested that the mechanisms by which osteoblasts and osteocytes are able to regulate inflammation involve the innate rather than the adaptive immune system via a GC-dependent pathway [13].
In our previous work, the effect of endogenous GC disruption in osteoblasts on murine K/BxN STIA was only analyzed during the acute and subacute phases of day 0 to 14.As expected, the observed degree of bone erosions and systemic bone loss was, overall, rather mild [11].
However, RA is a chronic inflammatory disease, where morbidity and bone destruction increase over time [14,15].For experimental murine models of arthritis, data on chronic inflammatory changes are scarce, and the effects of chronic inflammation and arthritis resolution on bone are not well-studied.
In 2012, Matzelle et al. conducted a long-term experiment employing the K/BxN STIA model in C57BL/6 WT mice, to study the chronic and resolution phases of arthritis.As clinical inflammation regressed around day 15, RANKL expression, and osteoclast numbers decreased and anabolic and pro-matrix mineralization factors, including osteoprotegerin (OPG), increased [16].While the study characterized the chronic resolution phase of antibody-mediated arthritis in WT animals, the effects of osteoblastic GC signaling in chronic experimental arthritis remained unclear.
Thus, we investigated whether transgenically disrupted GC signaling in osteoblasts impacts inflammation and bone in (a) the resolution phase, where arthritis gradually abates due to its natural and transient course (STIA), and (b) chronically active K/BxN STIA maintained by repeated boost-injections of the arthritogenic serum (STIA boost ).

Transgenic mouse model
Disruption of GC signaling in mature osteoblasts and osteocytes was achieved by overexpression of the GCinactivating enzyme 11ß-HSD2 under the control of an osteoblast-specific 2.3-kb collagen type Ia1 promotor with a CD1 background (Col2.3-11β-HSD2tg mice).Mice were generated as originally described by Sher et al. [17] and kindly provided as a gift by Dr. Barbara Kream (Department of Medicine, University of Connecticut Health Center, Farmington, CT, USA).In accordance with Institutional Animal Welfare Guidelines and an approved protocol (2012/006) by the Sydney South West Area Health Services (SSWAHS) Animal Welfare Committee, animals were housed in the animal facility of the ANZAC Research Institute (Sydney, NSW, Australia).For data reporting and storage, we followed the internationally established ARRIVE guidelines [18].

Initiation and prolongation of K/BxN STIA
Arthritis was induced in 5-week-old Col2.3-11β-HSD2 tg mice (tg) and their wild type (WT) littermates by intraperitoneal (i.p.) injections of 150 μl K/BxN mouse serum on day 0 and day 2. K/BxN mouse serum was obtained from 60-day-old K/BxN mice, who develop spontaneous arthritis due to being crossbred from KRN-and NOD mice.The sera from K/BxN mice were collected, pooled, stored at − 70 °C and injected as previously described [19].

STIA
In a first cohort, the course of arthritis was allowed to naturally abate with the aim of examining the post-acute phase of arthritis and the regenerative processes involved (WT STIA, n = 8; tg STIA, n = 8).Control (CTRL) animals received 150 μl phosphate-buffered saline (PBS) on days 0 and 2 (WT CTRL, n = 5; tg CTRL, n = 5) (Fig. 1).

STIA boost
To maintain a chronically active prolonged arthritis, a second cohort received repeated i.p. boost injections of 150 μl K/BxN mouse serum on days 14 and 28 (WT STIA boost , n = 11; tg STIA boost , n = 11).CTRLs received 150 μl PBS at respective time points (WT CTRL, n = 9; tg CTRL, n = 4).As there were no clinical, histological, or radiological differences observed between CTRL animals who received PBS injections following the STIA or the STIA boost protocol, the animals were grouped as WT CTRL and tg CTRL.All mice were observed for 42 days (Fig. 1).

Clinical arthritis assessment
Body weight and clinical parameters of arthritis were assessed daily following arthritis induction on day 0 up to the endpoint on day 42.Ankle joint swelling was assessed daily by measuring the maximum medial to the lateral diameter of ankles using a digital caliper.Measurements from both ankles were averaged for each time point.Since K/BxN STIA typically affects all paws, a scoring system for each limb (2 forelimbs and 2 hind limbs) was applied by two independent and blinded observers (EW and TM) as previously described [11,[20][21][22]: 0, no swelling; 1, mild-to-moderate swelling and erythematic ankle and/or 1 swollen digit; 2, moderate swelling and erythematic ankle or swelling in 2 or more digits; and 3, marked Fig. 1 Study design and group distributions according to genotype and arthritis protocol.K/BxN STIA was induced by i.p. injections on days 0 and 2 in WT STIA and tg STIA mice (top row) and boosted with additional i.p. injections on days 14 and 28 in WT STIA boost and tg STIA boost animals to maintain a chronic state of arthritis (middle row).CTRL mice were injected with PBS respectively at the same time points (bottom row).As there was no clinical difference between CTRL animals that were injected on days 0 and 2 and those injected on days 0, 2, 14, and 28, CTRL animals were grouped according to their genotype.CTRL, control; i.p., intraperitoneal; PBS, phosphate-buffered saline; STIA, serum transfer-induced arthritis; tg, transgenic; WT, wild type swelling along all aspects of the paw or all 5 digits swollen.The scores for each paw were added and a maximum score of 12 was possible.

Tissue collection and specimen preparation
On day 42, animals were euthanized, and samples were harvested for histological and histomorphological assessments as well as for micro-computed tomography (μCT) as previously described [11,23].Ankle joints and tibiae were dissected and fixed in paraformaldehyde 4% (PFA)/ PBS for μCT analysis (see below).After μCT analysis, ankles and tibiae were decalcified in 10% ethylene diamine tetraacetic acid (EDTA) and embedded in paraffin.Serial sections of 5 μm thickness were obtained to allow a central lateral view of the ankle joint and stained with (1) hematoxylin and eosin (H&E) for an overview of inflammatory changes as well as a (2) toluidine blue (TB) to interpret the degree of cartilage degradation and (3) tartrate-resistant acid phosphatase (TRAP) for identifying and quantifying osteoclast activity and bone resorption [23,24].Proximal tibial metaphyses were stained with H&E and TRAP for histomorphometric analyses.

Histopathological scoring
Histopathological scoring was performed by 2 independent investigators (EW and TM), both blinded to genotype and experimental groups as previously described [11,23,24].Analyses included inflammatory cell infiltration, edema and synovial hyperplasia (H&E), loss of cartilage, chondrocytes and collagen disruption (TB), osteophyte formation, and areas of bone resorption and number of osteoclasts (TRAP).
H&E: Inflammation Score-0, normal; 1, minimal inflammatory infiltration; 2, mild infiltration with no soft tissue edema or synovial lining cell hyperplasia; 3, moderate infiltration with surrounding soft tissue edema and some synovial lining cell hyperplasia; 4, marked infiltration, edema, and synovial lining cell hyperplasia; and 5, severe infiltration with extended soft tissue edema and marked synovial lining cell hyperplasia.
TB: Cartilage Degradation Score-0, normal; 1, loss of toluidine blue staining only; 2, loss of toluidine blue staining and mild cartilage thinning; 3, moderate loss of toluidine blue staining and moderate multifocal cartilage loss; 4, marked loss of toluidine blue staining with marked multifocal cartilage loss; and 5, severe multifocal cartilage loss.
To quantify systemic bone loss, trabecular morphometry of the proximal tibia was assessed, selecting a 1-mm region of interest (ROI) 0.5 mm below the growth plate within the endosteal borders.ROIs defined the measurements of morphological parameters following an automated algorithm in order to determine the bone's structure, density, and volume: BV/TV, bone volume/ total volume (%); Tb.N, trabecular number (/mm); Tb.Th, trabecular thickness (μm); and Tb.Sp, trabecular separation (μm) [11,23,26].
For assessment of bone erosions of the cortical, intraarticular bone of the ankle, a semi-quantitative score was applied to both two-and three-dimensional images of the tibiotalar joint: 0, no erosions or cortical damage; 1, mild cortical bone transformation and beginning erosive changes; 2, moderate bone damage comprising erosions and vacuoles; and 3, severe damage with large erosions and full-thickness defects of the cortex.The total score from two independent and blinded observers (EW and TM) was averaged.Of note, radiological bone erosion scoring data was only available for WT and tg CTRL and STIA boost groups.This explains why there are fewer animals in the CTRL groups.

Statistical analysis
To evaluate the course of ankle size, the semi-quantitative arthritis score and body weight, linear mixed-effects models with splines of grade 4 were fitted using the lme4 and lmerTest packages in R. Genotype (WT or tg) and treatment (STIA or STIA boost ), both additionally in interaction with time, and binary indicator variables for boosting on day 14 and for boosting on day 28, both set for the entire subsequent observation period, were investigated in the modeling.R package version 4.0.2 was used for analyses.
For remaining secondary endpoint comparisons between groups, the non-parametric two-sided Wilcoxon-Mann-Whitney test was employed.To detect outlier values, the Grubb's test was performed for each group [27].Outliers were displayed as square data points and excluded from statistical analyses.Significance was accepted where p < 0.05.Endpoint comparisons were conducted, and figures were created using GraphPad Prism 9, version 9.5.0 (GraphPad Software, Boston, MA, USA).

Sample size calculations
Sample sizes were calculated using the nQuery software.Group sizes of 8 animals with an allocation ratio of 2:1 in favor of STIA compared to CTRL animals were aimed for.This allows to detect large effect sizes of Cohen's d = 2.0 with 80% power and a two-sided type I error rate of 5%, when comparing STIA vs. CTRL with a t-test for independent groups.For comparisons between the two STIA groups, effect sizes of d = 1.51 could be detected.Yet, the statistical power of the analyses was assumed to be higher, since for each subject, multiple time points of observation (up to 42) were available.Applying appropriate models that account for repeated measurements allowed for larger statistical power, so that smaller effect sizes could also be detected.

Long-term effects of experimental arthritis on clinical arthritis, ankle size, and body weight
Following arthritis induction, WT STIA, tg STIA, WT STIA boost , and tg STIA boost mice developed an arthritic phenotype, while WT CTRL and tg CTRL animals did not show any signs of arthritis.In all arthritic WT and tg mice, the degree of arthritis increased gradually, peaked around approximately day 8, and plateaued until day 14.In the STIA groups, arthritis then gradually declined until day 42, whereas in the STIA boost groups, arthritis was maintained at a stable moderate level until day 42, yet only the boost injection on day 28 seemed to influence recurrent ankle swelling and worsening of the clinical arthritis score.Interestingly, we did not detect statistically significant differences between WT and tg animals over time for ankle size measurements (Fig. 2a and Table 1) and only small differences for the clinical arthritis score (Fig. 2b and Table 1).
Body weight gain was higher in tg than in WT mice, with a lower baseline body weight for tg mice.Regardless of the arthritis protocol (STIA vs. STIA boost ), body weight gain was similar within each group of WT and tg mice following disease induction (Fig. 2c and Table 1).Body weight gain flattened in mice receiving the STIA boost protocol compared to mice receiving the STIA protocol, regardless of genotype.It must be considered that baseline body weight was lower in mice receiving the STIA protocol (Fig. 2c and Table 1).When modeling the course of body weight gain, genotype (WT vs. tg) but not STIA treatment was significant both as a single term and in interaction with time (p < 0.001).Also, boosting on day 14 and 28 was recognized as influential in the model fit.

Intraarticular inflammation and cartilage degradation following long-term arthritis exposure
Fourty-two days following arthritis induction, joint inflammation and cartilage degradation were still present in WT and tg animals, yet no significant differences were detected between genotypes for neither STIA nor STIA boost groups (Fig. 3a, b).

Effects of long-term arthritis on local osteophyte formation and intra-and periarticular bone erosions
A central question to be addressed by the prolonged arthritis model is that of appositional osteophyte formation and bone erosions in a chronic inflammatory setting.On day 42, osteophytes and bone erosions were assessed at the arthritic site of the tibiotalar joint by histology and μCT.Similar to histopathological findings for inflammation, the histopathological Osteophyte Score was significantly higher for WT (STIA, p < 0.0001; STIA boost , p < 0.0001) and tg (STIA, p = 0.0170; STIA boost , p = 0.0067) mice compared to CTRLs (Fig. 3c, d).The Bone Erosion Score was also significantly higher for WT (STIA, p = 0.0117; STIA boost , p < 0.0001) and tg STIA boost (p = 0.040) animals compared to CTRLs, while there was no difference between genotypes for both histopathological scores (Fig. 3e).For the analysis of local bone erosions by μCT, threedimensional reconstructions of the ankles were scored by a semi-quantitative bone erosion score (Fig. 3f ).When compared to CTRLs, bone erosions were evident in WT STIA boost and tg STIA boost animals (Fig. 3g), yet only reaching statistical significance for WT mice (p = 0.002).No significant difference was found between genotypes (Fig. 3f ).

Systemic effects of long-term arthritis on bone morphology
Furthermore, μCT analyses of proximal tibiae were performed to quantify the systemic effects of inflammation on bone morphology in a long-term setting.No significant differences were observed for bone volume (BV/TV, Fig. 3h), trabecular number (Tb.N, Fig. 3i), and trabecular separation (Tb.Sp, Fig. 3j).Compared to CTRLs, trabecular thickness (Tb.Th) was significantly decreased in tg STIA boost compared to CTRL animals (p = 0.0381) without any differences between genotypes (Fig. 3k).
To elucidate whether changes in systemic bone morphology observed in μCT were due to high or low bone turnover, histomorphometric analyses were performed at the proximal tibia.
Osteoblast surface (Ob.S/BS) was significantly larger in WT STIA boost (p = 0.0031) and smaller in tg STIA (p = 0.0016) animals compared to CTRLs.Further significant differences were observed between WT CTRL and tg CTRL (p = 0.0252) animals where tg animals showed greater osteoblast surface areas and between WT STIA and tg STIA animals (p = 0.0379) where WT animals showed higher values (Fig. 3l).

Discussion
In the originally described study by Buttgereit et al. from 2009, K/BxN STIA and joint destruction were significantly attenuated by tg disruption of GC signaling in osteoblasts [11].While these findings were highly significant for ankle size and the clinical arthritis score as well as for histopathological subscores of inflammation and cartilage degradation, they were less conclusive for bone erosion indices on day 14.Since the original experiments were conducted over a period of two weeks only, and bone erosions and systemic inflammatory bone loss, at least in theory, might take longer to evolve, we decided to prolong the course of arthritis using a long-term model with repeated boost injections of K/BxN autoantibodies on days 14 and 28.In a second experiment, we investigated the effects of abrogated endogenous GC signaling in osteoblasts and osteocytes on the underlying resolution process of arthritis by allowing arthritis to gradually abate by its natural transient course until day 42.We were able to show that by repeatedly injecting K/ BxN serum i.p., the course of arthritis was effectively prolonged.However, no significant difference between WT and tg mice was observed, regardless of the arthritis protocol (STIA and STIA boost ), for the clinical course of arthritis over 42 days.
Correspondingly, no significant differences were observed between genotypes for histological indices of inflammation, cartilage degradation, osteophyte formation, and bone loss on day 42.It appears that the effect of abrogated GC signaling in osteoblasts on overall inflammation and subsequent structural alterations has less impact over time.
Similar results were previously reported by us for endpoint comparisons between WT and mutant mice on day 48 following antibody-mediated arthritis.While significant differences in histological scores for inflammation were present on day 10, these could not be seen on day 48, underlining the limitations of transient autoantibodyinduced arthritis models in mimicking a chronic disease state [21,22].
It was previously discussed that endogenous GCs may worsen inflammation and corresponding intraarticular damage through immunostimulatory patterns when present at low tissue concentrations [28,29].Although this was the case in the current study, we did not observe differences between genotypes after 42 days of inflammation.As the original work by Buttgereit et al. showed significantly less intraarticular inflammation and cartilage damage in tg compared to WT animals on day 14 following arthritis induction [11], this effect may disappear over time and only be attributed to acute and subacute arthritis.
Although not statistically significant, histomorphometry results for WT animals were largely similar to those described by Buttgereit et al. [11].Bone formation (Ob.S/BS) was lower, and bone resorption (Oc.S/ BS and N.Oc/BS) was higher in WT STIA animals when compared to CTRLs.This was expected, as arthritis-induced bone loss is caused by a disrupted coupling between bone formation and resorption [30,31].In RA, the existing dysbalance in bone homeostasis is not only a result of inflammation-induced osteoclastogenesis, but also a result of an impaired capacity of osteoblast-lineage cells to form bone [32] resulting in a net bone loss.This is likely to be due to the inhibition of Wnt signaling by Dickkopf-related protein (DKK)-1 and secreted Frizzled-related protein (sFRP) families of Wnt signaling antagonists [33].
We further observed increased signs of bone turnover in tg mice for both, the CTRL and the STIA boost group.In previous experiments, we have observed that tg mice seem to have an intrinsically higher bone turnover compared to WT animals which is particularly reflected in bone formation and resorption markers [6,10].Interactions of endogenous GCs with Wnt signaling have been debated, yet clear mechanistic explanations are currently not available [6,10].
Interestingly, we saw a reversed regulation in animals that received the STIA protocol.The long-term STIA protocol allows arthritis to naturally abate, which leaves animals in a recovery state after 42 days.This state between health and inflammation may explain the reversed results for bone turnover markers.
Although histomorphometric indices were previously reported to be similar in healthy and arthritic WT animals overexpressing osteoblastic and osteoclastic 11ß-HSD2 [34,35], a trend similar to our current results was previously described for the CAIA model [12].There is further evidence of endogenous GCs to impair bone quality in aging mice, although our mice were relatively young compared to the reported study animals by Weinstein et al. [36].Whether the observed effects are mainly driven by the arthritis model or the transgenic genotype remains speculative.
We conclude the effect of GC signaling in osteoblasts on arthritis to be of minor importance in a long-term setting.Bearing in mind the various cytokines and primary inflammatory cells that are involved in orchestrating autoimmune processes, osteoblasts are unlikely to call the tune.Rather pathologically altered structures that are a source of potent drivers of inflammation, such as the synovium and subchondral bone, are likely to dominate the course of inflammation in autoimmune arthritis, as laid out in a summary by Mathiessen et al. [37].
When regarding the dwindling differences between tg and WT arthritic mice over time, another factor that could have influenced the results is the genetic status of tg mice.If expression of the transgene was impaired, a reduced effect in the modulation of arthritis could have occurred.In general, a certain degree of instability for tg models exists, with the expression being prone to decrease over time and with several breeds [38].
Matzelle et al. showed that in the long-term K/BxN STIA model, bone formation was significantly increased, due to changes in Wnt signaling [16].In accordance, we detected an increased osteophyte formation during longterm K/BxN STIA in both genotypes which may counterbalance the apparent systemic bone loss and local bone erosions.
This study has some limitations.Although the K/BxN STIA model exhibits similarities to human RA, its induction through preformed autoantibodies is mostly independent of T-and B-cells, which are highly relevant in human inflammatory joint diseases [39].Moreover, while the murine K/BxN model affects the joints symmetrically similar to human RA, its transient nature differs from the chronic disease progression in affected patients.To address this, we chose the STIA boost protocol, mimicking rheumatic flares, next to the conventional STIA protocol.When comparing our data to the study by Buttgereit et al. [11], differences in animal numbers between the studies may have also had an effect on the reported results, although sample size calculations were followed.To minimize variations in the arthritic potency of the K/ BxN mouse serum, sera from K/BxN mice were pooled, before being injected into experimental mice.It cannot be ruled out that the arthritic potency of the K/BxN serum used in our study was not as high as in the original study.Baseline body weight differences between genotypes may have also influenced the disease course, although these differences were also previously described [11].As we did not detect significant differences in clinical parameters (clinical arthritis score and ankle size), we did not collect and analyze blood samples.In future studies, serum samples could provide information on circulating protein patterns following long-term arthritis exposure.Last, genetic variations in mutant mice are possible, but unlikely in this context, since thorough and standardized genotyping was conducted repetitively.

Conclusions
To summarize, this study advances our knowledge of the GC-dependent role of osteoblasts in the modulation of inflammatory arthritis.In the chronically persistent inflammation as well as in the resolution phase of K/BxN STIA, disrupted GC signaling in osteoblasts seems to only have a minor impact on inflammatory and reparatory processes.These results narrow the role of endogenous GC signaling by osteoblasts down to early inflammatory processes.

Fig. 2
Fig. 2 Longitudinal development of a ankle size, b clinical arthritis score, and c body weight over 42 days of STIA for indicated groups.Data points represent means, error bands indicate 95% confidence intervals, and arrows and gray data points boost injections for WT STIA boost and tg STIA boost animals on days 14 and 28.CTRL, control; STIA, serum transfer-induced arthritis; tg, transgenic; WT, wild type

Table 1
Estimates of the linear mixed-effects models for the course of body weight, ankle size, and clinical arthritis score