Exercise Exacerbates the Transcriptional Profile of Hypoxia, Oxidative Stress and Inflammation in Rats with Adjuvant-Induced Arthritis

Physical exercise (PE) is recommended for Rheumatoid Arthritis (RA), but the molecular and biological mechanisms that impact the inflammatory process and joint destruction in RA remain unknown. The objective of this study was to evaluate the effect of PE on the histological and transcriptional changes in the joints of adjuvant-induced arthritis (AIA) rat model. AIA rats were subjected to PE on a treadmill for eight weeks. The joints were subjected to histological and microarray analysis. The differentially expressed genes (DEGs) by PE in the arthritic rats were obtained from the microarray. The bioinformatic analysis allowed the association of these genes in biological processes and signaling pathways. PE induced the differential expression of 719 genes. The DEGs were significantly associated with pathogenic mechanisms in RA, including HIF-1, VEGF, PI3-Akt, and Jak-STAT signaling pathways, as well as response to oxidative stress and inflammatory response. At a histological level, PE exacerbated joint inflammatory infiltrate and tissue destruction. The PE exacerbated the stressed joint environment aggravating the inflammatory process, the hypoxia, and the oxidative stress, conditions described as detrimental in the RA joints. Research on the effect of PE on the pathogenesis process of RA is still necessary for animal models and human.


Introduction
Rheumatoid arthritis (RA) is one of the most common inflammatory arthritis affecting 0.5-1.0% of the population [1]. RA is a systemic autoimmune disease characterized by inflammation of the synovial membrane and periarticular structures. Joint pain, stiffness, and swelling are the primary symptoms, which eventually lead to joint deformity and functional disability. Extra-articular manifestations, including cardiovascular, pulmonary, psychological and skeletal disorders are common in RA patients [2]. The etiology and pathophysiology of RA remain unresolved; however, genetic and environmental factors have been associated with the inappropriate immunomodulation that triggers the inflammatory process in the joints and the subsequent damage to synovial structures [3]. The treatment of RA aims to reduce pain and inflammation in the joints and also to preserve their structural integrity and the patient's functionality. Currently, treatment includes a variety of pharmacological agents, education programs, joint protection, lifestyle changes, PE and surgical intervention as a final step [4][5][6].

Physical Capacity Test
After the familiarization, a physical capacity test (PCT) was conducted to establish the maximum speed reached by the animals before the intervention of PE. The PCT consisted of a single treadmill session in which the rats, after 5 min of warm-up (3 m/min), ran in the band with an increase in the speed of 3 m/min every 2 min until they fatigued and stopped running. The maximal physical capacity (100%) was defined as the maximum speed reached by each animal. The average speed per group was calculated, and this was used to establish the speed of the exercise sessions.

Exercise Program
The rats were exercised three times a week for eight weeks. The exercise started with a speed of 20% of the PCT and increases of 13% in speed were applied every two weeks until reaching 60%. Each exercise session included four phases: (1) acclimatization: rats were placed on the treadmill switched off for 5 min; (2) warm-up: rats walked for 5 min at lowest speed; (3) exercise: rats run for 25 min at the corresponding speed; and (4) cool down: rats walked at lowest speed for 5 min. The exercise sessions were administered at the same time each day.

Histological Analysis
After exercise intervention, the rats were euthanized, and the tarsal joints were immediately dissected. The ankle, the subtalar and the navicular cuboid joints were extracted and cut proximally in the distal fibula and tibia; and, distally in the metatarsal bones at the diaphysis. The muscle was dissected as much as possible and the synovial and ligament structures were preserved and fixed in 10% formalin phosphate buffer for 48 h and decalcified with 5% nitric acid for 24 h [24]. The tissues were dehydrated in graded ethanol, and embedded in paraffin, sectioned, and stained with hematoxylin-eosin (H&E Merck, Darmstadt, Germany). The histological variables: (a) inflammatory infiltrate, (b) synovial hyperplasia, (c) pannus formation, (d) synovial vascularity, (e) cartilage damage and (f) bone erotion were semi-quantitatively evaluated using a scale of 0 = absent, 1 = mild, 2 = moderate and, 3 = severe on each animal. The maximum score reached per animal for each parameter was 12 when the four joints showed a severe level. The scoring average per study group was estimated. The histological variables were evaluated bu two experienced operators blinded to the different groups.

DNA Microarray and Bioinformatics Analysis
The effect of PE on transcriptome was evaluated in the DNA microarray by comparing the arthritis-exercise group (experimental group) and the arthritis group (reference group). The tarsal joints dissected were immediately placed in liquid nitrogen and disrupted with mortar and pestle. Total RNA was extracted using RNeasy ® Lipid Tissue Mini Kit (QIAGEN, Germantown, MD, USA) according to the manufacturer´s instructions. The RNA quality and integrity were verified using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). For the microarray, the RNA of every group was pooled keeping equimolar quantities of every individual.
The microarray analysis was performed at the Institute of Cellular Physiology of the National Autonomous University of Mexico (México City, México). Briefly, the cDNA was synthesized and labeled for subsequent hybridization in the Rn5K microchip containing 5000 rat genes. Scanning and signal acquisition was performed using the ScanArray 4000 (Packard BioChips Technologies, Billerica, MA, USA). The GenArise software was used for the analysis of the microarray scan, and the lists of DEGs [Z-score ≥ 1.5 standard deviations (SD)] were obtained. To assess the biological relevance of the DEGs, DAVID Bioinformatics Resources 6.8 platform (https://david.ncifcrf.gov/) was used. Gene Ontology (GO) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway mapper with significant associations (p ≤ 0.5) were obtained [25]. Also, the STRING 10.5 database (http://string-db.org/) was used to obtain the analysis and integration of direct and indirect protein-protein interactions (IPP) centered on the functional association [26]. The DEGs identified in the microarray were loaded, and the interactions were selected with minimal confidence (interaction score > 0.4). The obtained IPP network was analyzed more thoroughly to obtain primary clusters of sub-networks using the Cytoscape software version 3.7.0 (Bethesda, Rockville, MD, USA) with the Molecular Complex Detection (MCODE) complement [27,28].

Statistical Analysis
The statistical analysis was made in SPSS statistics v22 software (SPSS Science Inc., Chicago, IL, USA). Measures of central tendency were estimated for each variable. T-test was used to compare the effect of PE on histological parameters. Differences were considered significant when p < 0.05.

Maximal Physical Capacity
The maximal physical capacity obtained in the PCT for the arthritis-exercise group was 21.28 ± 5.96 m/min. The initial speed and its increments along the exercise intervention were based on this value.

Histological Analysis
The effect of exercise in the joints was histologically evaluated using H&E staining. The scores of inflammatory infiltrate, synovial hyperplasia, pannus formation, synovial vascularity, and cartilage and erosions obtained for each study group are shown in Figure 1. The highest scores of all evaluated parameters were reached in the arthritis exercise group. However, only hyperplasia, cartilage damage and bone erosion were statistically different between the groups. Figure 1. Effects of exercise on tarsal bone histological parameters in adjuvant-induced arthritis rats. (A) Representative images of histological findings in the tarsal joints of the study groups at the end of the exercise intervention using H&E staining. (B) Joint involvement was scored by the semi-quantitative scale (showed in E) to describe inflammatory changes and structural remodeling in the tarsal joints (7 rats per group). The t-student test was used to compare histological measurements between groups. * p < 0.01. (C) Exercised rats on a treadmill. (D) Representative images of clinical changes on hind paws of adjuvant-induced arthritis (AIA) rats exercised (left) and non-exercised (right). (E) Representative images of the inflammation and structural joint damage scores in the tarsal joints of AIA rats. The 0 (normal) score was established in healthy rats, where the bone (bo), cartilage (ca) and synovium (sy) did not show alterations. The arthritis scores 1 (mild), 2 (moderate), and 3 (severe) were based on the inflammatory changes: the presence of synovial hyperplasia (sh) and pannus (pa) and structural remodeling: cartilage damage (cd) and bone erosion (be). The images were acquired with a 10× and 40× amplification. AIA: adjuvant-induced arthritis.

Microarray and Bioinformatic Analysis
The transcriptome-wide microarray analysis identified a pool of DEGs in the joints from arthritic exercised rats in comparison to arthritic non-exercised rats. A total of 719 genes were differentially expressed (Z-score ≥ 1.5 SD), 361 up-regulated, and 358 down-regulated. AB000928 (Zp1) and X68101 (Trg) were the most significantly up-and down-expressed genes by PE, respectively.
To understand the related biological functions of the 719 DEGs, we used the DAVID database for the enrichment analysis of GO and KEGG pathways. The top 10 enriched terms of GO and KEGG pathways are shown in Figure 2. Response to organic cyclic compound, response to drug, and response to ethanol were among the top enriched biological processes. The DEGs were enriched in Figure 1. Effects of exercise on tarsal bone histological parameters in adjuvant-induced arthritis rats. (A) Representative images of histological findings in the tarsal joints of the study groups at the end of the exercise intervention using H&E staining. (B) Joint involvement was scored by the semi-quantitative scale (showed in E) to describe inflammatory changes and structural remodeling in the tarsal joints (7 rats per group). The t-student test was used to compare histological measurements between groups. * p < 0.01. (C) Exercised rats on a treadmill. (D) Representative images of clinical changes on hind paws of adjuvant-induced arthritis (AIA) rats exercised (left) and non-exercised (right). (E) Representative images of the inflammation and structural joint damage scores in the tarsal joints of AIA rats. The 0 (normal) score was established in healthy rats, where the bone (bo), cartilage (ca) and synovium (sy) did not show alterations. The arthritis scores 1 (mild), 2 (moderate), and 3 (severe) were based on the inflammatory changes: the presence of synovial hyperplasia (sh) and pannus (pa) and structural remodeling: cartilage damage (cd) and bone erosion (be). The images were acquired with a 10× and 40× amplification. AIA: adjuvant-induced arthritis.

Microarray and Bioinformatic Analysis
The transcriptome-wide microarray analysis identified a pool of DEGs in the joints from arthritic exercised rats in comparison to arthritic non-exercised rats. A total of 719 genes were differentially expressed (Z-score ≥ 1.5 SD), 361 up-regulated (Appendix A), and 358 down-regulated (Appendix B). AB000928 (Zp1) and X68101 (Trg) were the most significantly up-and down-expressed genes by PE, respectively.
To understand the related biological functions of the 719 DEGs, we used the DAVID database for the enrichment analysis of GO and KEGG pathways. The top 10 enriched terms of GO and KEGG pathways are shown in Figure 2. Response to organic cyclic compound, response to drug, and response to ethanol were among the top enriched biological processes. The DEGs were enriched in molecular functions like protein binding. The most enriched KEGG pathways included hypoxia-induced factor-1 (HIF-1), cyclic adenosine monophosphate (cAMP) and mitogen-activated protein kinase (MAPK) signaling pathways. Subsequently, we selected the biological processes and KEGG signaling pathways known to be relevant in the pathogenesis of RA, which included: response to hypoxia, response to oxidative stress, angiogenesis and inflammatory/immune response (Table 1).
Cells 2019, 8,1493 6 of 33 molecular functions like protein binding. The most enriched KEGG pathways included hypoxiainduced factor-1 (HIF-1), cyclic adenosine monophosphate (cAMP) and mitogen-activated protein kinase (MAPK) signaling pathways. Subsequently, we selected the biological processes and KEGG signaling pathways known to be relevant in the pathogenesis of RA, which included: response to After the analysis on DAVID, the list of the 719 DEGs was analyzed on the STRING and Cytoscape-MCODE platforms. The first three clusters obtained are shown in Figures 3-5. The genes from the first three clusters were loaded in STRING to identify the associated KEGG signaling pathways; those relevant in RA were selected and marked with different colors. The pathways considered as relevant included the HIF-1 signaling pathway. The analysis of the protein networks unveiled relevant genes dysregulated by PE. Those presented multiple functional connections included up-regulated genes such as serine-threonine protein kinases Akt1, Akt2, and Akt3; the vascular endothelial growth factor A (Vefga), the mitogen-activated protein kinase kinase 1 (Mapk1), phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit beta isoform (Pik3cb), glucokinase (Gck), fructose-biphosphatase2 (Fbp2); and also down-regulated genes such as interleukin 6 (Il6). After the analysis on DAVID, the list of the 719 DEGs was analyzed on the STRING and Cytoscape-MCODE platforms. The first three clusters obtained are shown in Figures 3-5. The genes from the first three clusters were loaded in STRING to identify the associated KEGG signaling pathways; those relevant in RA were selected and marked with different colors. The pathways considered as relevant included the HIF-1 signaling pathway. The analysis of the protein networks unveiled relevant genes dysregulated by PE. Those presented multiple functional connections included up-regulated genes such as serine-threonine protein kinases Akt1, Akt2, and Akt3; the vascular endothelial growth factor A (Vefga), the mitogen-activated protein kinase kinase 1 (Mapk1), phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit beta isoform (Pik3cb), glucokinase (Gck), fructose-biphosphatase2 (Fbp2); and also down-regulated genes such as interleukin 6 (Il6).

Discussion
Current evidence describing the influence of PE on the pathogenic mechanisms of the arthritides is scarce and limited to studies in animal models. Most of these studies have focused on animal models of osteoarthritis and, in most cases, only clinical and histological parameters have been evaluated, limiting our understanding of the PE in the pathogenesis of active synovitis. To our knowledge, no previous study has assessed the effects of PE on the transcriptome profile in the AIA model.
Our study shows that PE induced several genes linked to RA pathogenesis. Specifically, PE exacerbated hypoxia, oxidative stress, and the immune/inflammatory response in rats with AIA. The histology confirms our transcriptome results; in the joints of the exercised animals, we observed an increase in the inflammatory infiltrate and more severe scores for bone and cartilage destruction.
In our opinion, the connection between PE and its potential detrimental role at active stages of synovitis should be considered as a relevant event in patients with inflammatory arthritides. Current recommendations for physical activity and PE in people with inflammatory and non-inflammatory arthritis [4,5,7,29] have been established mostly on clinical studies in humans, which the intimate effect of PE in the arthritic joints at a molecular level has not been considered. These recommendations have not been specified according to the disease activity; however, as our study

Discussion
Current evidence describing the influence of PE on the pathogenic mechanisms of the arthritides is scarce and limited to studies in animal models. Most of these studies have focused on animal models of osteoarthritis and, in most cases, only clinical and histological parameters have been evaluated, limiting our understanding of the PE in the pathogenesis of active synovitis. To our knowledge, no previous study has assessed the effects of PE on the transcriptome profile in the AIA model.
Our study shows that PE induced several genes linked to RA pathogenesis. Specifically, PE exacerbated hypoxia, oxidative stress, and the immune/inflammatory response in rats with AIA. The histology confirms our transcriptome results; in the joints of the exercised animals, we observed an increase in the inflammatory infiltrate and more severe scores for bone and cartilage destruction.
In our opinion, the connection between PE and its potential detrimental role at active stages of synovitis should be considered as a relevant event in patients with inflammatory arthritides. Current recommendations for physical activity and PE in people with inflammatory and non-inflammatory arthritis [4,5,7,29] have been established mostly on clinical studies in humans, which the intimate effect of PE in the arthritic joints at a molecular level has not been considered. These recommendations have not been specified according to the disease activity; however, as our study shows, likely at very active stages of inflammatory arthritides, PE could increase the inflammatory process. It is unlikely that PE will be prescribed in patients with active arthritis; however, a wide prescription of PE could also include the large proportion of patients whom remain with some degree of inflammatory activity despite therapy. Probably, the PE benefits may not be as clear in those patients with partial disease control. Therefore, the disease activity should be considered as an element for PE prescription.
RA pathogenesis remains a primary research field; we now understand that the rheumatoid pannus is an altered microenvironment that carries a significant array of abnormal immune processes and also several metabolic alterations. The increase in hypoxia and oxidative stress are recognized as relevant inflammation drivers within the rheumatic synovium. Their pathological implications are now sustained by experimental results both: in animal models [30,31] and human disease [32,33]. Synovial hypoxia promotes inflammation, angiogenesis, production of chemokines, and cartilage and bone destruction, mediated in part, through HIF-1 that is a potent pleiotropic mediator [32,34].
Our study confirms that PE up-regulates genes linked to hypoxia and oxidative stress in the arthritic joints; indeed, the HIF-1 signaling was the first enriched pathway in our bioinformatic analysis, suggesting that this pathway was the strongest influenced by the PE. Downstream HIF-linked genes as the vascular endothelial growth factor (VEGF) are also up-regulated. VEGF is critical for the onset and perpetuation of vascular proliferation in synovitis [35][36][37]. Also, it was previously reported the VEGF increase in the serum and synovial fluid of RA patients [35,38] as well as in experimental animal models [39].
Hypoxia could activate the genes encoding nearly every step of glycolysis [32]. Abnormal glucose metabolism was observed in the synovial fluid and synovium of RA, which is evidenced by increased glycolytic enzyme activity [40]. A recent study proposes the inhibition of hexokinases as a potential therapeutic strategy in the treatment of patients with RA [41]. According to this, the glucokinase (Gck) and the hexokinase 1 (Hk1) were up-regulated by PE in our study. Additionally, genes involved in the oxidative stress process were also dysregulated by PE. It is recognized that in an inflamed joint, hypoxia and reperfusion cycles occur, producing an excessive amount of reactive oxygen species (ROS) known as oxidative stress [33,42].
In addition to hypoxia and oxidative stress conditions induced by the effect of PE, other KEGG pathways were dysregulated (Table 1) such as rheumatoid arthritis, T-cell and B-cell receptor signaling, chemokine, Jak-Stat, TNF, Toll-like-receptor, WNT and osteoclast differentiation signaling pathways. This array of dysregulated pathways suggests a widespread effect of PE both in the adaptive and innate immune response.
Other genes participating in several KEGG pathways in our study include Akt1, Akt2, Akt3, Map2k1 and PIk3cb which belong to PI3K/AKT/mTOR and MAPK pathways that are considered central in RA pathogenesis [43,44]. They have been implicated in fibroblast-like synoviocyte proliferation and activation. They stimulate the production of pro-inflammatory and osteoclastogenic cytokines; besides, the inhibition of either pathway has been correlated with the improvement of disease activity parameters. Therefore, the use of agents with the potential to regulate either PI3K/AKT/mTOR or MAPK pathways has been considered as a potential therapeutic strategy [45].
The molecular results of the PE effect are not easily comparable with previous studies due to the low number of reports, which had objectives aimed at particular aspects of the disease. However, we can distinguish that the negative effect of PE has also been reported in the two most recent and complete studies in this field [16,17]. The results of these studies suggest a direct link between the degree of mechanical stress to inflammation and tissue localization, which strengthens previous studies that propose biomechanical factors as potential determinants for the topographic pattern of joint disorders in arthritis [46]. Cambré et al., suggest that the link between mechanical stress and the onset of arthritis is explained by local recruitment of Ly6high inflammatory cells elicited by mechanostress-induced chemokine induction in resident mesenchyme cells [17]. They also provide evidence of the local participation of the complement activators to maintain the progression and chronicity of arthritis through an impaired resolution [16].
On the other hand, it has been shown that in healthy rats, PE has a positive effect in the joints it induces extracellular matrix biosynthesis, cartilage strengthening, and attenuation of inflammatory pathways [47]. This suggests that the gene induction pattern of mechanosensitive genes seems to be distinct between different arthritis models and healthy animals, and likely also in regarding to the degree of arthritis activity.
Consistent with the genetic analysis, our histological findings showed that PE had an adverse effect on increasing the inflammatory infiltrate and the joint destruction. In experimental models of arthritis, it has been confirmed that mechanical stress increase joint inflammation, stimulate the progression and chronicity of arthritis and structural damage [16,17,48]; besides, the mechanical stress decrease the bone quality [16]. In contrast to our findings, other studies in the AIA model have shown an anti-inflammatory effect of PE in some parameters such as synovial hyperplasia [49,50], and joint destruction [51].
Mode, intensity, frequency, timing, and duration of the PE protocol are decisive aspects to the physiological responses and different outcomes [52]. Moreover, the severity of arthritis at the starting time of PE could be definitory. Studies in the AIA model that started the PE at the pre-arthritic stage [49,50] showed beneficial results. However, we started the PE at an arthritic stage because we think this scenario is more representative of RA patients.

Conclusions
The present study explored the influence of PE in the genetic expression and the histology of the actively swollen joints in the AIA model. Our results describe an exploratory and preliminary scenario in which PE increased the expression of genes with a known pathogenic role in RA; these findings were consistent with the histological findings. These results suggest that the PE exacerbated the stressed joint environment, where the inflammation, hypoxia, and the oxidative stress prevail, rendering clear parallels to the RA joints. The molecular effect of exercise on active stages of inflammatory arthropathies requires further studies in animal models and humans, which could contribute with the development of adequate programs for RA patients that can ensure a beneficial effect without negative implications at the tissue level.

Conflicts of Interest:
The authors declare no conflict of interest.