Early arthritis induces disturbances at bone nanostructural level reflected in decreased tissue hardness in an animal model of arthritis

Introduction Arthritis induces joint erosions and skeletal bone fragility. Objectives The main goal of this work was to analyze the early arthritis induced events at bone architecture and mechanical properties at tissue level. Methods Eighty-eight Wistar rats were randomly housed in experimental groups, as follows: adjuvant induced arthritis (AIA) (N = 47) and a control healthy group (N = 41). Rats were monitored during 22 days for the inflammatory score, ankle perimeter and body weight and sacrificed at different time points (11 and 22 days post disease induction). Bone samples were collected for histology, micro computed tomography (micro-CT), 3-point bending and nanoindentation. Blood samples were also collected for bone turnover markers and systemic cytokine quantification. Results At bone tissue level, measured by nanoindentation, there was a reduction of hardness in the arthritic group, associated with an increase of the ratio of bone concentric to parallel lamellae and of the area of the osteocyte lacuna. In addition, increased bone turnover and changes in the microstructure and mechanical properties were observed in arthritic animals, since the early phase of arthritis, when compared with healthy controls. Conclusion We have shown in an AIA rat model that arthritis induces very early changes at bone turnover, structural degradation and mechanical weakness. Bone tissue level is also affected since the early phase of arthritis, characterized by decreased tissue hardness associated with changes in bone lamella organization and osteocyte lacuna surface. These observations highlight the pertinence of immediate control of inflammation in the initial stages of arthritis.


Introduction
Rheumatoid arthritis (RA) is the most common chronic inflammatory joint disease, affecting about 1% of the world population [1]. RA is characterized by synovial hyperplasia caused by a large proliferative cellular infiltrate of leukocytes and high expression levels of proinflammatory cytokines [2]. As RA progresses there is marked articular destruction and decreased joint mobility with radiological evidence of bone erosion within 2 years of disease onset [3]. In addition, osteoporosis is a common finding in patients with RA [4] and is responsible for increased rates of vertebral and hip fractures in these patients [5,6]. RA is associated with an augmented expression of the receptor activator of RANKL (receptor activator of nuclear factor kappa-B ligand, NF-KB ligand) and low levels of its antagonist, osteoprotegerin (OPG) [7]. RANKL is a crucial activator of osteoclastogenesis [8]. In addition, RA serum and synovial fluid present a cytokine profile, including interleukin (IL)1β, IL6, IL17 and tumor necrosis factor (TNF), which further favors osteoclast differentiation and activation since the early phase of the disease [9][10][11].
Bone is a dynamic tissue organically composed mainly by type I collagen matrix that constitutes the scaffold for calcium hydroxyapatite crystal deposition. Remodeling of bone is a continuous process by which osteoclasts resorb bone tissue and osteoblasts produce new bone matrix that is subsequently mineralized. In this process biochemical markers of bone turnover are produced and released into circulation, providing a read-out of remodeling kinetics. Evidence suggests that bone-remodeling disturbances in RA contribute not only to local bone erosions but also to the development of systemic osteoporosis [12].
We have previously found in the adjuvant-induced rat model of arthritis (AIA) that 22 days of sustained and established inflammatory disease progression directly leads to the degradation of bone biomechanical properties, namely stiffness, ductility and bone strength, which was paralleled by a high collagen bone turnover [13].
Our previous results suggested that the impact of inflammation on bone macro and micro properties occurs very early in arthritis. We now hypothesize that this process starts upon the first inflammatory manifestations and that it directly affects intrinsic bone tissue properties (bone nano properties) [9][10][11].
The main goal of this work was to analyze the effects of the early phase of systemic inflammatory process at bone tissue level, including nanomechanical properties and microarchitecture.

Animal experimental design
Eighty-eight, 8 week-old female Wistar Han AIA rats weighing approximately 230-250gr were housed in European type II standard filter top cages (Tecniplast, Buguggiate, Italy) and transferred into the SPF animal facility at the Instituto de Medicina Molecular, under a 14h light/ 10h dark light cycle, acclimatized to T = 20-22˚C and RH = 50-60%. They were given access to autoclaved rodent breeder chow (Special Diet Service, RM3) and triple filtered water. AIA rats were purchased from Charles River laboratories international (Barcelona, Spain) and arthritis was induced in their laboratories in 47 animals. The transport service takes 3 days to arrive at Instituto de Medicina Molecular.
Upon arrival, animals were randomly housed in two groups, individually identified and cages were labelled according to the experimental groups, as follows: adjuvant induced arthritis model (N = 47) and control healthy group (N = 41). The inflammatory score, ankle perimeter and body weight were daily evaluated, during disease development. Inflammatory signs were evaluated by counting the score of each joint in a scale of 0-3 (0 -absence; 1 -erythema; 2 -erythema and swelling; 3 -deformities and functional impairment). The total score of each animal was defined as the sum of the partial scores of each affected joint.
Rats were sacrificed at day 11 (healthy control N = 11 and arthritic animals N = 16) and 22 (healthy control N = 30 and arthritic animals N = 31) post disease induction by CO2 narcosis and blood, femurs, tibias and paw samples were collected. In the AIA rat model of arthritis, day 11 post disease induction represents the fast raising phase of polyarticular involvement. Maximum disease activity and severity occurs at day 19 and plateaus up to day 22 post disease induction. [14] In addition, to minimize animal discomfort paper shavings were used as bedding material in Double Decker GR1800 cages (Techniplast, UK) with 5 animals per cage.
All experiments were approved by the Animal User and Ethical Committees at the Instituto de Medicina Molecular (Lisbon University), according to the Portuguese law and the European recommendations, Directive 2010/63/EU revising Directive 86/609/EEC.

Histological evaluation of hind paws
Histology was performed to evaluate the effect of inflammation on articular joint synovium and bone structures.
Images were acquired using a Leica DM2500 (Leica Microsystems, Wetzlar, Germany) microscope equipped with a color camera.

Biomarkers quantification
Serum samples were collected at the sacrifice time and stored at -80˚C.
IL-6 is one of the most powerful proinflammatory cytokines and is a key signal for bone destruction. IL-6 was quantified in serum samples using a specific rat ELISA kit (Boster Bio, California, USA). Bone resorption marker CTX I reflects osteoclastic activity as it is a degradation product of type I collagen, the major structural protein of bone. The bone formation marker P1NP is a bio product of type I collagen synthesis and thus is a marker for osteoblastic activity. Bone remodeling markers, CTX-I and P1NP, were quantified by Serum Rat Laps ELISA assay (Immunodiagnostic Systems Ltd, Boldon, UK) in order to study the effects of inflammation on bone turnover.
For all biomarkers standard curves were generated by using reference biomarker concentrations supplied by the manufacturers. Samples were analyzed using a plate reader Infinite M200 (Tecan, Mannedorf, Switzerland).

Micro-computed tomography (micro-CT) analysis
Structural properties of the trabecular and cortical tibiae were determined with a high-resolution micro-CT system (SkyScan 1272, Bruker microCT, Kontich, Belgium) in order to study the effects of inflammation on bone microstructure. Moist bones were wrapped in parafilm and covered with dental wax to prevent drying and movement during the scanning. Xray tube was set to 50kV and beam was filtered with 0.5mm Aluminum filter. Sample position and camera settings were tuned to provide 3.0μm isotropic pixel size and projection images were collected every 0.2˚. Tissue mineral density values were calibrated against hydroxyapatite phantoms with densities of 250mg/cm 3 and 750mg/cm 3 . Reconstructions were done with NRecon (v 1.6.9.8; Bruker microCT, Kontich, Belgium) where appropriate corrections to reduce beam hardening and ring artifacts were applied. Bone was segmented in slices of 3μm thickness. After 200 slices from growth plate, we selected and analyzed 1400 slices of trabecular bone. For cortical bone 300 slices (1800 slices from growth plate) were analyzed.
This evaluation was performed in agreement with guidelines for assessment of bone microstructure in rodents using micro-computed tomography [16]. Trabecular bone morphology was analyzed by applying global threshold and despeckle to provide binary image for 3D analyzes. For cortical bone ROI was refined with ROI-shrink wrap operation. This was followed by segmentation of blood vessels using adaptive thresholding. Blood vessels and porosity were analyzed using 3D morphological analyses.

Bone mechanical tests
Flexural tests were used to evaluate the impact of early inflammation on whole bone mechanical competence of arthritic and healthy controls at day 11 and 22 post disease induction. Femurs were subjected to a 3-point bending test using the universal testing machine (Instron 3366, Instron Corp., Massachusetts, USA). Femurs were placed horizontally anterior side upwards on a support with span length of 5mm. The load was applied with a constant speed of 0.005mm/s until failure occurred. Stiffness was analyzed by fitting first-degree polynomial function to the linear part of recorded load deformation data. A displacement of 0.15μm between fitted slope and measured curve was used as criteria for yield point, whereas the breaking point was defined as set where force reached maximal value. For both yield and breaking points, force, deformation and absorbed energy were defined.

Nanoindentation
Nanoindentation was performed using a CSM-Nano Hardness Tester System (CSM Instruments SA; Switzerland; Indentation v.3.83) equipped with a Berkovich based pyramid diamond indenter in order to evaluate the impact of early inflammation on nano-mechanical properties of bone tissue in arthritic and healthy controls at day 11 and 22 post disease induction. After micro-CT, 0.5mm of the proximal extremity of tibia was discarded and the remaining tibia was embedded to low viscosity epoxy resin (EpoThin, Buehler, Knorring Oy Ab, Helsinki, Finland). Slow speed diamond saw was used to remove 10% of total bone length. The sample surface was polished using silicon carbide sandpaper with a decreasing grid size (800, 1200, 2400 and 4800) and finished with cloth containing 0.05μm γ-alumina particles. Indentation protocol was adopted from previous work [17] and an average of 8 indentations were done on a transversal section of cortical and trabecular bone with quasistatic (so called 'advanced') loading protocol. All indentation regions were selected under an optical microscope to achieve the precise location of indentations. For cortical bone, indentations were placed uniformly through the central cortex and were evenly spaced along bone cross section, avoiding osteocyte lacunae and vascular canals. Our goal was to determine the average properties at mid -diaphysis rather than properties at certain specific locations [18].
In the 'advanced' protocol, a trapezoidal loading waveform was applied with a loading/ unloading rate of 20mN/min, and with an intermediate load-hold-phase lasting 30s hold at a maximum load 10mN. The approximate depth range of indentations varied between 1.2 up to 3.5μm. The hardness (H IT ), indentation modulus (E IT ), indentation creep (C IT ) and elastic part of indentation work (η IT ) were measured by advanced protocol using the Oliver and Pharr (1992) method [19].
Histological images of rat tibiae from diaphyseal cortical region were acquired during the nanoindentation technique, using a CSM instruments (Switzerland) microscope equipped with a color camera.
A histologic score was applied in order to evaluate the lamellar structures of bone tissue. This evaluation was performed in a blind fashion using a semi-quantitative score: • Lamellar bone structure: (1-predominantly parallel-lamella; 2-concentric and parallellamellae in the same proportion; 3 -predominantly concentric lamella).
The ratio of osteocyte lacuna area / total tissue area was also evaluated at x200 magnification in order to analyse the percentage of total tissue area occupied by osteocyte lacunae. The method of acquisition and analysis used was the same applied for the evaluation of bone volume / tissue volume in histomorphometry technique [13]. All variables were expressed and calculated according to the recommendations of the American Society for Bone and Mineral Research [20], using a morphometric program (Image J 1.46R with plugin Bone J).

Statistical analysis
Statistical differences were determined with Mann-Whitney tests using GraphPad Prism (GraphPad, California, USA). The normality distribution was assessed by D'Agostino and Pearson test. Data are expressed as median with interquartile range. Differences were considered statistically significant for Ã p<0.05.

The AIA rat model has a rapid and severe disease progression
Results showed that inflammatory signs (Fig 1) boosted sharply in the arthritic group. The inflammatory score (Fig 1a) increased significantly at day 11 and 22 post disease induction (which correspond to an acute phase and a chronic phase of systemic inflammation, respectively) in arthritic rats when compared to healthy controls (p<0.0001, respectively).
Moreover, arthritic animals at day 11 and 22 post disease induction sharply increased the ankle swelling throughout disease progression (Fig 1b), when compared to healthy rats (p<0.0001, respectively)

Inflammation affects local joints and promotes bone damage in AIA rats since the early stage of arthritis
To evaluate the effect of inflammation in local articular joint synovium and bone structures, paw sections stained with hematoxylin and eosin were performed (illustrative images can be observed in Fig 2).
The histological evaluation using 5 semi-quantitative scores is depicted in Fig 3. Sublining layer infiltration (Fig 3a), number of lining layer cells (Fig 3b) and bone erosion score (Fig 3c) were increased in the arthritic group when compared with healthy controls at day 11 and 22 post disease induction (p<0.0001). Arthritic samples also showed increased cartilage damage surface (Fig 3d) since the early phase of arthritis at day 11 and 22 (p = 0.0403 and p<0.0001 vs healthy controls, respectively). These data contributed to the increased values of severity score (Fig 3e) in arthritic group (p<0.0001 vs healthy controls). Moreover, results also demonstrated a continuous disease progression between day 11 and 22 in arthritic animals, as observed by the increase of the sublining layer infiltration, number of lining layer cells, bone erosion score (p<0.0001), cartilage surface score (p = 0.0001) and global severity score (p = 0.0006).

Systemic inflammation occurs in this model
We observed that IL6 levels were increased in the serum of arthritic rats at day 11 and 22 post disease induction in comparison with healthy controls (p = 0.0003 and p<0.0001, respectively), as observed in Fig 4. Results also revealed that IL6 levels decreased in arthritic rats at day 22 when compared with day 11 (p = 0.0092).

Systemic inflammation promotes high bone turnover
We have observed that both CTX-I (Fig 5a) and P1NP (Fig 5b) were significantly increased in the arthritic group at day 22 in comparison with healthy controls (p<0.0001 and p = 0.0007, respectively), revealing an increase of bone turnover in the arthritic group. Moreover, arthritic rats showed already increased values of CTX-I at day 11 post disease induction (p = 0.0218 vs healthy rats at day 11) but not of P1NP. These results suggest that systemic inflammation promotes skeletal bone turnover disturbances since the early stages of arthritis.

Micro-CT
The effect of systemic inflammation on cortical and trabecular skeletal bone was assessed by micro-CT in bone tibia.
The arthritic group showed at day 22 a dramatic deterioration of bone tibia integrity associated with a reduction in cortical bone area (Fig 6a) and average cortical thickness (Fig 6b) (p<0.0001 vs healthy controls, respectively) with an evident increased endocortical perimeter (Fig 6c) (p = 0.0029 vs healthy control). However, changes promoted by inflammation on bone structure begin at the early stages of arthritis as we can observe by the results obtained in the arthritic group by day 11 with a decreased cortical bone area (Fig 6a) (p = 0.0219 vs healthy   Trabecular bone (Fig 7) also showed increased deterioration promoted by inflammation with decreased trabecular bone volume fraction in arthritic rats at day 11 and 22 post disease induction (Fig 7b) (p = 0.0001 and p<0.0001 vs healthy controls, respectively), thickness  https://doi.org/10.1371/journal.pone.0190920.g006 (Fig 7c) (p<0.0001 vs healthy controls, respectively), and number (Fig 7d) (p = 0.0039 and p<0.0001 vs healthy controls, respectively). Results also demonstrated increased values of trabecular separation in the arthritic group at day 11 and 22 (Fig 7e) (p = 0.0043 and p<0.0001 vs healthy controls) and of porosity (Fig 7f) (p = 0.0001 and p<0.0001 vs healthy controls, respectively). Furthermore, structure model index (Fig 7g) showed increased values in arthritic groups at day 11 and 22 (p = 0.0015 and p<0.0001 vs healthy controls, respectively) indicating that the shape of trabeculae is rather rod-like in the arthritic group as compared to plate-like shape in healthy controls.
Altogether, these results showed that inflammation promote bone structural disturbances, leading to bone loss and consequent bone fragility in arthritic rats (Fig 7a).

Mechanical tests
Classical mechanical properties of rat femurs were evaluated using 3-point bending mechanical tests. The stress at yield point was noted where according to Zioupos et al (1994) the first micro fractures appear and irreversible damage occurs. Another interesting point is maximal load at breaking point (where complete fracture occurs) and toughness can be estimated. As shown in Fig 8, arthritic rats at day 22 revealed biomechanical disturbances with a decrease in mechanical properties at yield point, namely by displacement (Fig 8a) (p = 0.0192 vs healthy control), strength (Fig 8b) (p = 0.0229 vs healthy control) and pre yield energy (Fig 8c) (elastic energy) (p = 0.0161 vs healthy control). These results suggest that arthritic bones at day 22 start to accumulate micro fractures with smaller deformations and loads, leading to a decreased energy absorption capability at yield point. Results also demonstrated that arthritic rats at day 22 have decreased maximum load (Fig 8d) and elastic capabilities at maximum load point (Fig 8e) (p = 0.0017 and p = 0.0134 vs healthy control, respectively), which indicates increased bone fragility. Finally, arthritic rat groups showed a significant decrease in toughness (Fig 8f) (p = 0.0143 vs healthy control), demonstrating that arthritic bone can absorb less energy before fracturing.
Altogether, mechanical data revealed that arthritic groups have significantly lower mechanical properties as compared to healthy controls, meaning that arthritic bones are more fragile and prone to fracture, as highlighted by the significantly lower structural strength and poor biomechanical properties. Early arthritis induces bone nanostructural disturbances

Decreased hardness in arthritic bone associated with an increase of the ratio of bone concentric to parallel lamellae and of the area of the osteocyte lacuna
Nanoindentation was performed in order to assess the quality at tissue matrix level as this technique works at the level of a single trabecula or within a confined submicron area of the cortical bone tissue (Fig 9).
Nano-mechanical tests revealed that arthritic rats have decreased hardness in the cortical aspect of bone at day 22 post disease induction (Fig 9a) (p = 0.0010 vs healthy control) and at trabecular bone at day 11 and 22 post disease induction (Fig 9b) (p = 0.0184 and p = 0.008 vs healthy controls, respectively). Results also demonstrated the continuous decreasing of cortical hardness (Fig 9a) during arthritis development among arthritic groups (p = 0.0043). No differences were observed in the other parameters analysed. Early arthritis induces bone nanostructural disturbances Topographic images gathered during nanoindentation allowed the characterization of histologic features from healthy and arthritic bone at day 11 (Fig 9g) and 22 (Fig 9h) days post disease induction. Concentric lamellas were identified in secondary osteons (SO) and more frequently observed in arthritic animals than in healthy controls (Fig 9c) (p = 0.0022, arthritic group at day 22 vs healthy control rats at day 22). On the contrary, healthy animals at day 11 (Fig 9e) and 22 (Fig 9f) presented more parallel-lamellae (PL) structures than SO structures.
Arthritic animals at day 22 post disease induction showed also an increased area occupied by osteocyte lacunae in the total tissue when compared to healthy animals (p = 0.0067) (Fig  9d). Results also demonstrated a slight tendency towards an increase at day 11 post disease induction (Fig 9d).

Discussion
Arthritic groups presented inflammatory manifestations with synovial tissue inflammation and local bone erosions, as expected. Increased values of serum IL-6 were observed in arthritic rats since the early stages of arthritis, confirming the systemic inflammatory component of this animal model. This cytokine plays a pivotal role in the pathologic processes of arthritis with a special emphasis on its impact on skeletal bone [21][22][23][24]. In accordance with this effect an increased and accelerated bone turnover was shown in arthritic animals, as depicted by increased CTX-I and P1NP levels since the early stages of arthritis. Data already published by our group in the same animal model of arthritis had also shown that P1NP levels were increased at day 22 post disease induction in arthritic animals and so did CTX-I levels [13], reflecting an overall increase in bone turnover [25]. Despite the existing of some variability in human studies, CTX-I and P1NP have been found to be increased in RA patients, revealing the coupled compensatory mechanism of bone turnover [13,26].
Micro-CT confirmed that interference of inflammation with bone metabolism translates into bone micro architectural and geometric changes in trabecular and cortical bone, since the early phase of arthritis-11 days after disease induction. In addition micro-CT provides mechanical insights such as polar moment of inertia, which represents the material resistance to torsional deformation in cylindrical objects or segments of cylindrical objects, suggesting low mechanical competences in arthritic bone [27]. Because of the uncertainty on the direction of the predominant bending moments in-vivo we elected to analyse the results with respect to the polar moment of inertia. We have also considered the maximal (Imax) and minimal (Imin) second moments of area (area moments of inertia), which arguably should correspond to anteroposterior (IAP) and mediolateral (IML) directions in-vivo although there is no way of proving that.
For micromechanical property analysis we have selected the most reliable nanomechanical parameters, which reflect the material property (S1 Fig). The three-point bending test confirmed the previous result, demonstrating decreased yield properties and reduced maximum load bearing capacity, which was also reflected in the energy absorption of all mechanical parameters evaluated at 22 days post disease induction. This mechanical fragility, replicating the clinical observations in RA patients, further reinforces the use of the AIA model as an adequate strategy for a fast insight on the impact of inflammation on bone. The first part of this study sets the stage for using this model for evaluating the effects of the early phase of systemic inflammatory process at bone tissue level, including nanomechanical properties and microarchitecture.
Nanoindentation was performed in order to assess the quality of bone at tissue matrix level, as this technique can be used at the level of a single trabecula or within a confined submicron area of the cortical bone tissue. Results showed decreased cortical and trabecular hardness in arthritic rats since the early phase of arthritis (days 11 and 22).
We also observed at day 11 and 22 post arthritis induction concentric lamellas in secondary osteons (SO) microstructures, resulting from high bone remodelling, as previously described [13,28,29]. Dall'Ara et al. suggested that larger numbers of this younger, less mineralised and less hard structures, could be related to reduced hardness of bone tissue identified by nanoindentation. On the contrary, healthy animals presented more parallel-lamellae (PL) structures than SO structures and this PL structures are 10% more harder than the former, representing the mature bone structure (and normal bone remodelling) [29]. In addition, arthritic animals had an increased area occupied by osteocyte lacunae in total tissue. Osteocytes are responsible for the maintenance of the bone homeostasis, regulating the behaviour of osteoblasts and osteoclasts by communicating through gap junctions [30]. Although no previous data is available in the context of arthritis some studies revealed that osteocytes from osteoarthritis patients have an irregular morphology, with limited ability to reply to mechanical stimuli, leading to significant changes in the structure and mineral density [31]. Despite being still unclear this apparent change of osteocyte morphology in arthritic bone might contribute to the structural and mechanical changes observed in this context.
These results are in line with our previous results where we also demonstrated that inflammation induces bone mineral loss since the early phase of arthritis [13]. Other publications from our group in a chronic model of arthritis, showed a decreased mineral content [13] and also a lower density and organization of collagen fibrils when compared to healthy control bone [32].

Conclusion
Systemic inflammation induces very early changes in bone microarchitecture and macro and mechanical behaviour leading to increased bone fragility. At a tissue level a decreased tissue hardness is observed, associated with changes in bone lamella organization and osteocyte lacuna surface.