*2.3. The Impact of CIA and Bone Defect in Circulating Inflammatory Mediators*

To further investigate the systemic inflammatory impact of CIA and the combination with bone injury, 27 chemokines and cytokines were quantified in plasma. From these, epidermal growth factor (EGF), growth-regulated oncogenes/keratinocyte chemoattractant (GRO/KC), Macrophage inflammatory protein-2 (MIP-2, CXCL2), and Granulocyte-macrophage colony-stimulating factor (GM-CSF) were out of the detection range for this assay. From the remaining 23 cytokines/chemokines detected, IL-13 was not detected in the control group, but was detected in the other two animal groups and thus was considered in the analysis. Results obtained are illustrated in Figure 4 and show that, relative to the control group, CIA animals had a significant upregulation of 8 molecules and a downregulation of 5. Among the significantly up-regulated mediators IL-13, IL-2, and IL-6 were also significantly up-regulated in the CIA+FD group, while TNF-α, IL-17 A, IL-4, IL-5, and IL-12 (p70) lost significance in the CIA+FD group. Nonetheless, the tendency for increase was maintained, and in the case of IL-5 it was close to statistical significance (*p* = 0.0534). Also, Monocyte chemoattractant protein-1/C-C motif chemokine ligand 2 (MCP-1/CCL2) and Eotaxin were close to statistical significance in CIA (*p* = 0.0552 and *p* = 0.0620, respectively), and MCP-1 was significantly up-regulated for the CIA+FD group. Regarding the downregulated mediators identified in CIA plasma, Vascular endothelial growth factor (VEGF), Fractalkine, IL-18, C-X-C motif chemokine 5 (CXCL5, LIX), and Leptin, only Fractalkine and Leptin levels were statistically significantly lower in the CIA+FD group, relative to control. The only molecule that showed a significant difference between the CIA and CIA+FD groups was the regulated on activation, normal T cell expressed and secreted (RANTES) chemokine, which was significantly downregulated in CIA when compared to CIA+FD, but neither group had a significant difference to the control group.

**Figure 4.** Changes in plasma cytokine and chemokine profile in CIA and CIA+FD. Quantitative results of the multiplex cytokine/chemokine array performed for all animals in each group. Box plots represent min-to-max distribution of *n* = 4 to 5 animals per group. \* *p* < 0.05, \*\* *p* < 0.01 in relation to control group, and # *p* < 0.05 relatively to CIA group, determined by Kruskal–Wallis test and Dunn's multiple comparisons test.

#### *2.4. CIA Induction Impacts Endogenous MSC Biologic Behavior*

To evaluate the impact of chronic inflammation on endogenous MSC we next isolated and culture bone marrow-derived MSC (BM-MSC) from CIA and control animals. The number of bone marrow cells recovered was similar between the CIA and control animal groups, and after selective culture and expansion, cells obtained were highly positive for the classical stromal markers CD29 and CD90, and did not express the haematopoietic marker CD45 (Supplementary Figure S3). The bone marrow isolated cells showed the ability to differentiate into the chondrogenic, osteogenic, and adipogenic lineages (Supplementary Figure S4). All together, these data confirmed the successful isolation of MSC from bone marrow of both CIA and control animals.

BM-MSC were then used to evaluate the impact of the cells source microenvironment on their biological properties and behavior. CIA and control BM-MSC were metabolic active along 1, 3, and 7 days of culture, but CIA-MSC showed lower metabolic activity that did not increase along time, when compared to control cells (Figure 5A). In line with these results, the Ki-67 fluorescent immuno-staining analysis also showed a tendency for reduced proliferation of CIA-MSC (Figure 5B,C). To access if exposure to the inflammatory microenvironment could be conditioning, MSC proliferation control cells were cultured in the presence of TNF-α or IL-4. Interestingly, even low doses of TNF-α produced significant reductions in the percentage of proliferating cells, while IL-4 had no significant impact (Figure 5D).

**Figure 5.** MSC metabolic activity and proliferation capacity. (**A**) Metabolic activity was measured at 1, 3 and 7 days of MSC culture by resazurin assay. RFU: relative fluorescence units. Statistical differences were evaluated by 2-way ANOVA followed by Turkey's multiple comparison test (**B**) Representative images of nuclei (Dapi, blue) and Ki-67 immunostaining (pink, white arrows) of control and CIA-derived MSC after 2 days in culture. Scale bar: 50 μm. (**C**). Percentage of Ki-67<sup>+</sup> control and CIA-derived MSC, across different experiments. Statistical differences were evaluated by Mann–Whitney test. (**D**) Percentage of Ki-67<sup>+</sup> control- derived MSC after 2 days in absence/presence of 10 ng/mL or 100 ng/mL of TNF-α or IL-4. Box plots represent min-to-max distribution of *n* = 4 to 5 animals per group. \* *p* < 0.05 by Friedman test followed by Dunn's multiple comparisons test.

The endogenous differentiation ability of BM-MSC from CIA and control animals was evaluated at 3 and 7 days of culture, in the absence of chemical and molecular inductors. The results obtained showed that CIA animals-derived MSC had increased expression of osteogenic genes, with significantly higher ALP expression, relative to control cells, at 7 days, and a significant increase in RUNX2 expression from 3 to 7 days in CIA-MSC, which was not observed in control-MSC (Figure 6A). When analyzing chondrogenic differentiation capacity of CIA-MSC in basal conditions, results indicate a significantly higher expression of ACAN at 3 days compared to control cells, and that significantly decreased by day 7, while expression of SOX9 did not show significant differences between the cells of CIA and control animals, or along time (Figure 6B). Then we explored if exposure to cytokines could impact MSC differentiation in the absence of chemical induction. Results obtained show that exposure to TNF-α or IL-4 increased ALP gene expression with IL-4 exposure producing significant increases, even when compared to the positive control, using chemical inductors (Figure 6C). Interestingly, IL-4 exposure led to increased ACAN expression, to levels similar to those obtained with chondrogenic induction and for CIA- MSC, while TNF-α did not lead to the increases observed for MSC from CIA animals (Figure 6D).

**Figure 6.** The capacity of MSC differentiation is affected by arthritis induction. The differentiation of control and CIA-MSC were evaluated in absence of chemical inductors, at 3 and 7 days of culture by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) for (**A**) osteogenic marker genes ALP and RUNX2, and (**B**) chondrogenic differentiation genes SOX9 and ACAN. Box plots represent min-to-max distribution of *n* = 2 to 3 different animals per group, in 2 independent experiments. \* *p* < 0.05 by Kruskal–Wallis test followed by Dunn's multiple comparisons test. The effect of 10 ng/mL or 100 ng/mL of TNF-α and IL-4 in the differentiation of control MSC, were evaluated in basal conditions at 3 days by RT-qPCR for (**C**) osteogenic and (**D**) chondrogenic markers. Cells cultured with osteogenic and chondrogenic inductors were used as positive controls for osteogenic (osteo+) and chondrogenic (chond+) differentiation. \* *p* < 0.05, \*\* *p* < 0.01 determined by Friedman test followed by Dunn's multiple comparisons test.

Taken together, these results indicate that exposure to hallmark arthritis cytokines like TNF-α can partially mimic the CIA impact on MSC.

#### **3. Discussion**

In the current study, arthritis was induced in all animals and reached a plateau in its score by 17 to 21 days after CII-IFA (Collagen type II-Incomplete Freund's adjuvant) immunization. This is in line with previous reports on this model [26]. Histologically, our data shows hind paws swelling and inflammation in digital joints, visible by the infiltration of inflammatory cells and hyperplasia of the synovial membrane, which constitute hallmarks of human RA pathology [27] and have been previously reported in CIA animal model [11,28,29]. The CIA model was combined with a critical size femoral bone defect model in a load-bearing site without the need for fixation material [30] and without significant acute implications for animal welfare and paw swelling. Enlargement of secondary lymphoid organs was similar between CIA and CIA+FD at 3 days post-bone injury, albeit slightly decreased percentages of proliferating cells in CIA+FD were observed. This is in line with previous work in the CIA model, such as in the report by Ibraheem where the injection of different collagen types combined with adjuvant to induce arthritis in rats led to enlarged lymph nodes with histologic alterations [31] and the reports of splenomegaly in the most severe states of adjuvant-induced arthritis [11,32]. Also, a previous report induced a mid-shaft femur fracture with fracture fixation in the CIA rat model and showed delayed healing in CIA animals, but the authors do not analyze inflammation in their study [33]. In fact, the current study is, to the best of our knowledge, the first one addressing the interplay between inflammation and bone injury in these RA models.

To determine the systemic immune cell response to CIA induction and the combination with bone injury, we evaluated the main immune cell populations that are the key cell players involved in the development and maintenance of the inflammation in CIA model [10], in blood and secondary lymphoid organs. Our results show a significant increase in B cells in blood and draining lymph nodes of CIA animals and also increased myeloid cells in blood. This change in proportions may suggest that these cells migrate from the blood to the inflamed tissues. A previous report in the mouse CIA model described similar increases in the percentage of B cells, which they correlate with increased anti-collagen antibodies in serum [34]. Activated B cells have been reported to produce anti-collagen antibodies that bind to cartilage collagen and form an immune complex, generating a localized inflammatory response and attracting monocytes, granulocytes, and T cells to the joint [10,17].

Interestingly, CIA animals responded to the bone injury with significant increases in myeloid cells and their activation, particularly in secondary lymphoid organs. Myeloid cells are important regulators that contribute to the perpetuation of inflammation in RA [35]. Moreover, in a murine model of early RA, increased myeloid dendritic cells were reported in draining lymph nodes before the appearance of joint histological changes, indicating their role in the early stage of murine RA [36]. A significant decrease in T cell proportions was found in lymph nodes of CIA rats, accompanied by a decrease in CD4/CD8 ratios. These data may suggest that T cells are attracted to the joints, but this requires further confirmation. The immune response mediated by T cells is described as crucial for the development of mouse [10,17] and rat CIA, with athymic rats being resistant to CIA induction, but T cell transfer alone was reported to have induced only a mild response [10]. The role of NK cells in RA and CIA is not yet clear, with some evidence that they may contribute to [37,38] or protect against [39] inflammation. Here, we did not observe differences in the percentage of NK cells in peripheral blood as described in a previous study in CIA mice [34], but observed an increase in the CIA+FD group. The same work showed a decrease in splenic NK cells proportion at day 37, different from the increasing tendency that was observed in the current study at day 21 after CIA induction. Interestingly, the changes observed here in systemic immune cell proportions are different from those previously observed in the same model of bone injury, but in healthy animals at day 6 post-surgery [40]. This might be explained by the inflammatory conditions, but also by different times post-injury, as in the same model in healthy animals we observed an acute response at day 3, and resolution of that response at day 14 post-surgery [41].

In the current study, a broad quantitative analysis of plasma cytokines/chemokines comparing control, CIA, and CIA+FD animals was performed. We found increased levels of the hallmark cytokine TNF-α in the plasma of CIA animals. High TNF-α levels in affected joints or in peripheral blood serum are strongly associated with RA disease severity [42,43]. This pro-inflammatory cytokine was described in different RA animal models [43,44], including the transgenic mouse expressing human TNF-α [45] and the CIA model [12,13]. TNF-α has also been reported as accelerating arthritis signals when injected into animals [43]. In response to TNF-α, RA synovial fibroblasts are described to produce MCP-1/CCL2, which was also found increased in the plasma of CIA animals in the current study, particularly after bone injury. MCP-1 is a potent chemokine that mediates monocyte ingress and activation in joints [46]. This was clearly demonstrated in vivo by the injection of a rabbit model with MCP-1/CCL2, resulting in a marked macrophage infiltration of the affected joints [47]. Macrophages have been previously described as a major source of MCP-1/CCL2 [48]. In our results, CD68<sup>+</sup> macrophages were also found in the inflamed synovium.

Levels of IL-17, IL-2, and IL-5 were also increased in CIA animals, and maintained the same tendency (albeit not always significant) in the CIA+FD group. These cytokines are related to the T cell function and response in vivo. Systemic IL-17 levels intensify RA and induce a chronic and erosive form of the disease [49]. A recent report in CIA model in Dark Agouti rats demonstrated that the Th-17 response in draining lymph nodes is greater in females than in males, correlating the Th-17/Treg (regulatory T cell) axis with sexual differences in CIA severity [50]. IL-2 is a positive regulator of T cell proliferation, and administration of IL-2 can mediate a pro-inflammatory effect on CIA mice, increasing the animals' arthritic scores [51]. A recent report shows that IL-2-Anti-IL-2 monoclonal antibody immune complexes can inhibit CIA by increasing Treg cell functions [52]. IL-5 is a cytokine with pleiotropic effects on target cells, including eosinophils and B cells, inducing cell proliferation, survival, and differentiation [53], its role in RA remains unknown, but a recent work found elevated serum levels of IL-5 in 59% of seropositive RA patients [54]. Also, it was reported that lymph node cells isolated from CIA-susceptible mice lines showed a higher number of IL-5 producing cells in culture than cells from resistant mice [55]. Fractalkine was found downregulated in CIA animals with and without bone injury, which is not in agreement with reports indicating amelioration of arthritis symptoms and joint destruction when Fractalkine is suppressed [56,57]. Also, Leptin was found significantly reduced in the plasma of CIA and CIA+FD animals, but its role in arthritis is still controversial. In the mouse CIA model, Leptin injection has been reported to worsen disease via Th-17 cells [58], but has also been reported as anti-inflammatory when collagen-antibody-induced arthritis is induced in mice overexpressing Leptin [59]. Also, its levels have been reported as increased in other models of RA and in patient's plasma, but reduced at the joint level, indicating its consumption [60]. RANTES (or CCL5) was the only molecule that showed a significant difference between the CIA and CIA+FD groups. A recent report on bioinformatic analysis of human RA gene expression data indicates that CCL5 might have a negative impact on the development of RA [61], while another recent report in mouse indicates that peritoneal levels of several CC chemokines, including CCL5, could be related with mouse strain susceptibility to experimental arthritis [62]. However, as in our study, neither group had a significant difference to control, and so the biological meaning of this result remains unclear.

Although the use of MSC as therapeutics for RA is being intensely studied, the effect of CIA induction and its inflammatory milieu on the biological behavior of endogenous bone marrow-derived MSC had not been explored so far. Growing evidence indicates that bone marrow is actively involved in the RA disease process. In RA patient samples, the interaction between synovial inflammatory tissue and bone marrow reportedly results from the disruption of cortical bone leading to bone marrow invasion [63]. Furthermore, previous work described in vivo a significant enlargement of vascularized bone canals that link the bone marrow and the synovium tissue, in joints with synovial hyperplasia [64]. Also, it has been suggested that bone marrow-derived immature mesenchymal cells may migrate through the joint space to replace synovial cells, similar to what happens with inflammatory cells [65,66]. In fact, even if bone marrow is not primarily affected, this interaction, together with the aberrant cytokine production, can potentially affect the biological behavior, namely the survival and proliferation of bone marrow cells [67].

The current work shows that culture-expanded MSC from CIA animals were morphologically and immunophenotypically similar to controls, expressing the characteristic mesenchymal markers (Figure S1C). Nevertheless, the metabolic activity and percentage of proliferative cells were lower in the CIA-MSC cultures. This had not been described in RA animal models thus far, but agrees with what was described in cells derived from RA patients [68,69]. RA-MSC display proliferation defects in culture without significant differences in the percentages of apoptotic cells [68,69], and high levels of the cell cycle inhibitor p21, which can mediate the defects in cell growth and replicative capacity [69]. Mice treated with anti-TNF-α at an early CIA stage showed a reduction in the number of MSC in the bone marrow and synovium [64]. Conversely, a previous report described that the proportion of CD34<sup>+</sup> bone marrow progenitor cells increase after the anti-TNF-α treatment in RA patients, with the number of bone marrow mononuclear cells in clonogenic assays being higher when compared to the pretreatment values [67]. In our study, a significant reduction in the percentage of proliferative MSC was observed in control cells exposed to TNF-α, an important pro-inflammatory cytokine found increased in the plasma of CIA animals. This reduction in healthy MSC proliferation in the presence of TNF-α could have consequences for MSC-based therapies for RA.

Our results show increased osteogenic gene expression of CIA-MSC in basal conditions when compared to control MSC. Contrasting data has been reported on the osteogenic and chondrogenic potential of MSC isolated from RA patients. Previous reports indicate that ALP expression was comparable between the cells from RA and healthy donors [68,69], and that RA MSC do not differ significantly from the normal cells on their chondrogenic differentiation [70]. MSC differentiation in the CIA animal model has been analyzed in response to the chemical/molecular stimulation protocols commonly used in vitro, but not in basal conditions, and further detailed analysis is still required. Previous work showed that control and CIA-derived mouse BM-MSC respond to osteogenic stimulation with similar increases in RUNX2 and ALP gene expression [71]. In our work the cells were isolated from the same site and from animals with the same age and sex. Thus, our results support that the alterations in proliferative and differentiation capacities are not site or age-dependent, but are likely associated with the disease condition of the animals. We also explored if CIA conditions could be mimicked by exposure to TNF-α and observed an increase, albeit not statistically significant in ALP gene expression. This agrees with previous reports showing that exposure to these concentrations of TNF-α impairs MSC proliferation/survival [72]. The same authors also found impaired MSC response to osteogenesis induction upon exposure to these concentrations of TNF-α, but the response of the cells in the absence of osteogenic induction was not analyzed. In the RA model of human TNF-α transgene, previous reports showed the deleterious effects of high levels of this cytokine on fracture repair [45], and in the CIA model fracture healing was also impaired [33]. Further studies are required to analyze if such impairments are related to changes in MSC proliferation and differentiation, and the role of chronic inflammation in bone repair [72].

In conclusion, arthritis induction creates a systemic inflammatory environment, with changes in immune cells and circulating cytokine/chemokine levels. Also, bone marrow-derived MSC isolated from CIA animals display impaired proliferation and increased differentiation in basal conditions. Further studies are needed to clarify the relation between exposure to inflammatory cytokines like TNF-α and MSC phenotype impairments, and their consequences for bone repair in RA. Furthermore, the current work supports the viability of using the CIA model to investigate the mechanisms of bone repair/regeneration in chronic inflammatory conditions, and the development of therapeutic strategies more appropriate to treat bone fractures in RA patients.

#### **4. Materials and Methods**

#### *4.1. Bovine Collagen Type II Emulsion Preparation*

Bovine collagen type II (CII, 2 mg/mL, Chondrex, Redmond, WA, USA) was added in a 1:1 proportion of incomplete Freund's adjuvant (IFA, Chondrex, Redmond WA, USA) and homogenized

in an ice water bath, according to the manufacturer's instructions. Then, the emulsion was kept at 4 ◦C until injection, without this period exceeding 1 h.
