Continuous infusion of dexamethasone aggravates the damage of cartilage via upregulating p-AKT and impairing articular autophagy in experimental OA model

Objective To explore the effect of dexamethasone (Dex) infusion on articular cartilage and the underlying mechanisms in vitro and in vivo. Methods Destabilization of medial meniscus (DMM)-induced OA mouse model was used in this study. The mice with Dex treatment were sacriced and then the knee joint samples were obtained for pathological analysis. Mouse primary chondrocytes were isolated and cultured in the presence or absence of Dex, which were used for calcication analysis and western blot assay. Results Dex accelerated the loss of articular cartilage matrix in mice, while it aggravated the damage of cartilage in DMM-induced OA model at the late stage. The calcium content in calcied cartilage layer in the joints from Dex treated OA mice was signicantly higher than that from control mice. Dex treatment enhanced mineralization of articular cartilage matrix and leaded to massive apoptosis of chondrocytes in OA model. In addition, Dex caused autophagy of chondrocytes in the early stage, which was decreased at the late stage of Dex treatment. Moreover, we found that the effect of Dex on the mineralization of articular cartilage matrix in mice was related to AKT activation.

Osteoarthritis (OA) is the most common joint disease and leading cause of disability worldwide [1].
Through decades of research, OA is currently accepted as a whole joint disease [2] and even wholeperson disease [3] developing along a continuum from early to late stages. The epidemiology of this disorder is complex and multifactorial, with genetic [4], biological [5], and biomechanical [6] components. However, the therapeutic effect is still unsatisfactory in clinic until now.
Corticosteroids [7] have been widely used as a rst-line anti-in ammatory and immune-modulating drug for many immune-mediated conditions or as an adjunctive therapy for some infectious or malignant diseases. However, high doses or prolonged use leads to a number of side-effects including joint injury [8][9][10][11]. In the past 10 years, several professional societies including Osteoarthritis Research Society International (OARSI), National Institute for Health and Care Excellence (NICE) in 2014 and American College of Rheumatology (ACR) in 2012 recommended glucocorticoids for patients with knee OA [12]. Long acting corticosteroids including dexamethasone (Dex) are recommended for the treatment of are of knee pain in clinic mainly due to its anti-in ammation effect [13]. However, the American Academy of Orthopedic Surgeons (AAOS) found a lack of compelling evidence to support the use of glucocorticoids for the treatment of OA, as well as an unclear balance between the bene ts and potential harms of this treatment [14]. Moreover, more and more studies have suggested that corticosteroids such as Dex could aggravate the damage of articular cartilage [15][16][17][18][19]. One study showed that patients who had been treated with 2 years of intra-articular triamcinolone suffer from signi cantly greater cartilage volume loss and no signi cant improvement of knee pain can be found [18]. Yihui Tu et al. reported that Dex can induce the apoptosis and signi cantly upregulate the expression of apoptotic gene Fas/FasL in human articular chondrocytes [19]. However, the detailed mechanisms of Dex-mediated cartilage damage are still not well known.
In this study, we explored the effect of dexamethasone infusion on articular cartilage in vivo and the potential mechanisms. Our study revealed that continuous infusion of Dex can enhance the calci cation of cartilage via AKT activation and increase chondrocyte apoptosis through inhibiting autophagy, which aggravates the damage of articular cartilage and accelerate the progression of OA in vivo. Our study provided a new perspective to understand the mechanisms of Dex related side effect on joint, which will be bene cial for future treatment in OA.

Sample collection and processing
This study was conducted in compliance with the regulations of the ethics committee of the Daping Hospital (Chongqing, China). In OA group, human articular cartilage samples were collected from 3 primary knee OA (grade IV in The Kellgren Lawrence grading system) patients with multiple intra-articular corticosteroid injections and Varus deformity receiving knee arthroplasty. In control group, human articular cartilage samples were collected from 3 patients who had amputations due to trauma. All of them were con rmed without in ammatory arthritis or prior knee surgery history. All samples of full thickness articular cartilage were cut from patient's tibial plateau and xed in 4% PFA. Half of them were decalci ed in 15% EDTA for wax embedding, and remaining samples underwent hard tissue embedding.

Primary chondrocyte isolation and culture
Mouse primary chondrocytes were isolated and cultured according to our previous report [20]. Brie y, the cartilage was isolated from knee joints of 3 ~ 5-day-old C57BL/6J mice and treated with 0.25% trypsinase/DMEM at 37℃ for 15 min to remove the soft tissues (including muscles, ligaments and bone tissues). After further incubation with 0.1% collagenase II/DMEM overnight in a CO2 incubator at 37 ℃, the chondrocytes were collected and cultured in DMEM/F12 supplemented with 10% FBS.

Methylmethacrylate embedding and sectioning
The undecalci ed specimens were dehydrated in ascending concentrations of ethanol (from 70-100%) and then embedded in methylmethacrylate according to the instructions of manufacturer. The embedded specimens were trimmed with a hard tissue cutting device (Leica RM2265; Germany) to expose the target area. Finally, the cut specimens were sanded down with sequential #2000, #4000, #8000 and #10000 grit lapping plastic sandpaper (3M; Japan). Each specimen was placed in an ultrasonic bath after the polishing steps.
Scanning Electron Microscope Imaging and Energy Dispersive Spectrometer detection After vacuum drying, the samples were sputter-coated with gold and palladium and then observed by Hitachi SU8010 scanning electron microscope (SEM) under 5 keV accelerating voltage, 10 µA probe current, a 10-mm working distance, and an image resolution of 1560 × 1920. The surface calcium element analysis of the target area was analyzed with line scan by Oxford X-max50 Energy Dispersive Spectrometer (EDS).

Surgical model of OA in mice and Dex deliver
Adult male C57BL/6J mice (10 ~ 12 weeks) were purchased from purchased from the Beijing HFK Bioscience Co.Ltd. and maintained in the animal facility (speci c pathogen free) of the Daping Hospital (Chongqing, China). The mice were randomly divided into six groups: Control groups (vehicle group and Dex group), Sham groups (Sham + vehicle group, Sham + Dex group), and DMM groups (DMM + vehicle group, DMM + Dex group) (n = 36 ~ 45 per group). DMM surgery was performed on the right knee joints to establish experimental OA model according to the described methods [21]. In short, after anesthetization (1% pentobarbital sodium), joint capsule was incised, then the medial meniscotibial ligament was sectioned with microsurgical scissors. As a control, sham operation was performed on the left knee joint, but the ligaments remained intact. After 2 days, dissolved dexamethasone with saline was injected intraperitoneally (5 mg/Kg, 3 times per week). The vehicle group was injected intraperitoneally with saline. Then the effect of dexamethasone on the progression of osteoarthritis was observed at week 4, week 8, and week 12. All experiments were performed according to protocols approved by the Laboratory Animal Welfare and Ethics Committee of Army Medical University (Chongqing, China).

Histological analysis
The knee joints were xed with 4% paraformaldehyde for 24 h, decalci ed with 0.5M EDTA at pH 7.4 for 2 weeks and embedded in para n. Five-micrometer-thick sagittal serial sections were made across the knee joints. Sections from the medial tibial plateau and medial femoral condyle of knee joints were stained with Safranin-O/Fast Green and scored on a scale of 0-6 according to the recommendation of the Osteoarthritis Research Society International (OARSI) [22]. Each section was assessed by two blinded, independent graders (Jinfan Zhang, Junlan Huang) and the average score was used for statistical analysis.
Western Blot assay Cells lysate were lysed with RIPA lysis buffer (Beyotime, P0013B). Equal amount of protein samples were resolved on 10-12% SDS-PAGE gel and transferred onto polyvinylidene di uoride membrane. After blocking with 5% nonfat milk, the membrane was probed with indicated primary (CST, LC3B Antibody #2775) and secondary antibodies. The signals were detected using chemiluminescence (Pierce, NCI4106) according to the manufacturer's instruction. Images were captured by ChemiScope (Clinx, Shanghai, China).

Calci cation analysis
For chondrocyte calci cation analysis, cells were cultured for 7 days in complete Fitton-Jackson Modi ed (BJGb) medium (Gibco) (10% FBS, 50 mg/mL ascorbic acid, 20 mM β-glycerol phosphate), stimulated with Dex as well as AKT inhibitor: LY294002 (10 nM, CST). Medium was changed for the last 4 days. Alizarin red staining and Alizarin red absorbance at 405 nm were used to evaluate the degree of calci cation.

Tunel assay
Apoptosis of articular chondrocytes in cartilage tissues was determined by TUNEL assay using a kit from Invitrogen (Thermo Fisher, USA) according to the manufacturer's instructions. Specimens were visualized under a uorescence microscope. The number of apoptotic chondrocytes in relation to the total number of cells was quanti ed in tissue sections. More than three elds of microscopic view in each section, and multiple sections (more than three) from 4 different experimental animals in each experimental group were used.

Statistical analysis
Statistical analysis was performed using GraphPad Prism (GraphPad, La Jolla, CA). All numerical values were presented as mean values ± the standard deviation (SD). Student's t-test or two-way ANOVA analysis was used to determine signi cance. Difference was considered signi cant when p 0.05.
Continued use of Dex leads to the loss of articular cartilage matrix in mice Firstly, we study the effect of Dex on the articular cartilage of control mice. The mice were sacri ced at the time points of 4 weeks, 8 weeks and 12 weeks after continuous intraperitoneal Dex injection. Then the knee joint samples were obtained and stained with Safranin O-fast green (Fig. 1A) to assess the extent of articular cartilage degeneration by OARSI scoring system (Fig. 1B). The results showed that intraperitoneal Dex injection led to signi cant progressive loss of extracellular matrix (Fig. 1A) of knee joint cartilage in the middle and late stage (8 weeks, 12 weeks) of OA, including the femur side and the tibial side. The sum score of medial femoral condyle (MFC) and medial tibial plateau (MTP) (Fig. 1B) was also decreased after Dex treatment. These data indicate that continued intraperitoneal use of Dex lead to the loss of articular cartilage matrix in normal adult mice.

Long-term continuous Dex application aggravates DMMinduced cartilage damage
Next, we further evaluated the effect of long-term continuous use of Dex on the cartilage damage in an experimental OA model. Surgical DMM in mice is a well-established OA model, which is commonly used to study and evaluate the effects of drugs on OA. DMM surgery was applied in adult mice to create a mechanically unstable OA model. Safranin O-fast green staining and OARSI scoring system were taken to assess the extent of articular cartilage degeneration. The data in Fig. 2A showed that continuous injection of Dex for 12 weeks signi cantly aggravated cartilage destruction in mice leading to de ciency of large areas of surface articular cartilage in the calci ed cartilage layer. Meanwhile, the total score and maximum score of MTP and MFC were signi cantly increased in Dex-treated group (Fig. 2F-I). These data suggest that long-term Dex treatment can accelerate cartilage damage in OA mice.
Continuous Dex results in decreased subchondral bone mass and bone density and alleviated synovial in ammation OA is considered as a disease of the whole joint. In recent years, a large number of studies have shown that dysregulated subchondral bone remodeling and synovial in ammation are also involved in OA process [23]. Therefore, we further observed the effect of Dex on subchondral bone and synovial in ammation in DMM OA model. Continuous Dex injection resulted in decreased bone mineral density (BMD) and bone mass/volume ratio (BV/TV) of subchondral bone in both DMM and Sham groups ( Fig. 3C and D) at 12 weeks. These results showed that the local Dex application could lead to loss of subchondral bone mass and decreased bone density, which was consistent with previous study [24]. Meanwhile, the data from synovitis scoring showed that Dex attenuated the severity of synovitis following DMM surgery compared with the sham mice ( Fig. 3A-B),suggesting that Dex, as a classical anti-in ammatory drug, has a remarkable inhibitory effect on synovial in ammation in DMM-induced OA model.

Dex increases calcium content in calci ed cartilage layer
The calci cation of cartilage is highly involved in the pathological changes of OA [25]. To further investigate the mechanisms of Dex-worsened cartilage damage, we detect the calcium content in each layer of articular cartilage of mouse knee joint by SEM and EDS. There were signi cant calcium concentration gradients in each layer of cartilage in the knee joint of normal adult mice (Fig. 4A-B). The calcium content in the non-calci ed cartilage layer was extremely low, while it increased successively in the calci ed cartilage and subchondral bone plate (Fig. 4B). After use of Dex, the abnormally increased calcium content in the calci ed cartilage layer (CCL) was detected from 4 weeks to 12 weeks (Fig. 4B), which disturbed the original calcium distribution pattern. Next, we assessed the changes of cartilage calci cation in DMM model ( Fig. 4C-D). We found that the calcium content in CCL was gradually increased during the OA process (from 8 to 12 weeks). Dex group, however, showed an abnormally sudden increase of calcium content in the CCL at the early period (after 4 weeks). Moreover, the samples from OA patients with a history of Dex treatment presented a high calcium content in calci ed cartilage ( Fig. 5B and D) and a large number of horizontal clefts in the junction between calci ed and non-calci ed cartilage (arrowheads in Fig. 5), which suggested to be closely related to the rapid degeneration and peeling of articular cartilage [26].

Dex induces calci cation of ECM by partially activating AKT
Previous study showed that AKT could positively regulate the calci cation of chondrocytes in vitro, and calci ed osteophyte formation was prevented in the Akt1-/-joints in mice with surgically induced OA. Therefore, we suspect that AKT signal contributes to Dex-mediated cartilage calci cation. As shown in Fig. 6A, the effects of Dex at different concentration gradients (0, 1 nM, 10 nM and 100 nM) on extracellular matrix calci cation were observed, and the OD value of alizarin red was detected at 405 nm ( Fig. 6B). As shown in Fig. 6B, Dex dose-dependently promoted the calci cation of extracellular matrix of primary chondrocytes. Next, we detected the change of AKT signal after Dex treatment. The AKT pathway inhibitor LY294002 reversed Dex-induced increase of extracellular matrix calci cation ( Fig. 6C and D), suggesting that AKT pathway plays an important role in Dex-mediated chondrocyte calci cation in vitro. In addition, the protein expression of AKT in Dex group was slightly increased, while the expression of P-AKT 308 rather than PAKT 473 was signi cantly higher (Fig. 6E), suggesting that Dex may partially lead to calci cation of extracellular matrix of chondrocytes through activation of P-AKT 308 . As the activation of AKT contributes to Dex-induced chondrocyte calci cation in vitro, we further examined the activation state of AKT in the joint sample from Dex-treated mice. After 4 weeks of Dex treatment, the percentage of P-AKT 308 positive cells was signi cantly increased in articular cartilage of DMM mice (Fig. 6F). The percentage of P-AKT 308 positive cells was almost 90%, which mainly appeared in the deep layer of cartilage in Dex-treated group (Fig. 6G). These results suggest that Dex may partially, at the early stage, activate the AKT pathway to cause abnormal calcium deposition in calci ed articular cartilage layer of OA.

Dex promotes the apoptosis of articular chondrocytes in vivo
The increase of chondrocyte apoptosis is an important reason for the destruction and degeneration of articular cartilage [27]. Previous studies have revealed that Dex could enhance apoptosis in multiple types of cells [28,29]. Therefore, we further estimated the effect of Dex on the apoptosis of articular chondrocytes in sham and DMM mice. As shown in Fig. 7C, the number of apoptotic cells in Dex group was signi cantly increased in the non-calci ed layer in sham group. For the sham group, the numbers of cells in the non-calci ed cartilage layer, calci ed cartilage layer and subchondral bone plate were not changed signi cantly in the Vehicle group and the Dex group (Fig. 7B, E, H). The number of apoptotic cells in Dex group was signi cantly increased in the non-calci ed layer (Fig. 7C), while the number and proportion of the apoptotic cells in the calci ed layer was signi cantly increased (Fig. 7F, G). It is suggested that long-term use of Dex promotes apoptosis of chondrocytes, especially in calci ed chondrocytes. Similarly, in the DMM + Vehicle group, we also found that apoptotic cells were concentrated in the calci ed cartilage layer (Fig. 7F). In the DMM group, Dex signi cantly decreased the cell number of chondrocytes in both non-calci ed and calci ed cartilage layers ( Fig. 7B and E). Both the number and proportion of apoptotic cells were increased in the non-calci ed cartilage layer ( Fig. 7C and   D). While the number of apoptotic cells in the calci ed cartilage layer was not signi cantly changed, but the proportion of apoptotic cells was signi cantly increased due to the decreased total number of cells ( Fig. 7F and G). These results suggest that continued use of Dex promotes apoptosis of articular chondrocytes in both OA and non-OA models, while the apoptotic chondrocytes were mainly located in the calci ed and non-calci ed cartilage layer in OA models.

Continuous Dex impaires chondrocyte autophagy at the late stage of DMM surgery
As autophagy could protect the cells from apoptosis in various conditions [30], and Dex was reported to induce autophagy in cultured chondrocytes [31], we deduced whether autophagy is involved in the Dexinduced chondrocyte apoptosis in DMM model. Our results showed that the LC3-positive cells in articular cartilage at 4 weeks and 8 weeks after DMM operation were signi cantly higher in Dex group than those in the Vehicle group ( Fig. 8A-C). At 12 weeks after DMM, the numbers of LC3-positive cells were few in Dex-treated groups and there is no signi cant difference between control and Dex-treated groups (Fig. 8D), indicating the stress-response autophagy was eliminated after long time processing of Dex. As autophagy could either promote cell survival or cell death, we further investigated the effect of Dexinduced autophagy on chondrocyte apoptosis using autophagy inhibitor Baf1. We found that Baf1 could enhance the apoptosis induced by Dex alone (Fig. 8E), suggesting that Dex could induce a protective autophagy of primary chondrocytes to prevent their apoptosis.
Collectively, our data revealed that Dex could enhance the calci cation of cartilage via AKT activation and increase chondrocyte apoptosis through inhibiting autophagy, which may further promote the lesion of articular cartilage and accelerate OA progression (Fig. 8F).

Discussion
Corticosteroids has been wildly used in the management of osteoarthritis, rheumatoid arthritis and some sports injuries for decades [32][33][34]. However, there has been more and more evidence showed that corticosteroids had a degenerative-effect on several collagen-producing tissues, such as bone, tendons [35,36]. Accordingly, interests have focused on how corticosteroids led to degeneration in these tissues. However, the mechanism of corticosteroids induced degeneration of cartilage is not well unknown.
Chrysis and his colleagues revealed that Dex induces apoptosis of chondrocytes in a caspasesdependent manner [37]. In addition, Dex signi cantly increased ATP-induced mineralization in articular chondrocytes in vitro [38], suggesting a potential role of Dex in the pathologic mineralization and loss of cartilage in OA.
The calci ed cartilage layer (CCL) forms an important interface between compliant cartilage and stiff bone for transmitting force, attaching cartilage to bone, and limiting material diffusion [39]. It is separated from the other zones of cartilage by the tidemark, and the CCL borders the subchondral bone with the cement line, which forms a highly interdigitated interface with subchondral bone [40].
Contributing to the stiffness gradient in the soft-hard tissue junction, the CCL is about 100 times stiffer than the overlying hyaline cartilage and 10 times softer than the underlying bone [41]. However, this stiffness gradient is altered during the OA process, during which the stiffness of the calci ed cartilage zone is gradully increased from super cial layers of cartilage to subchondral bone [42]. The mechanical properties of cartilage could be affected by the extent of its mineralization. For calci ed cartilage, the nanoindentation modulus is positively related to the local mineral content [43,44]. In healthy human knee specimens, the percentage of inorganic compound in CCL is less than that of subchondral bone [45]. More detailed analyses of bovine tibiofemoral joints nd that the articular cartilage zone was mineral free, whereas the mineral content of calci ed cartilage zone increases exponentially but is still signi cantly lower than that of the normal bone [46]. Interestingly, horizontal splits at the interface between the uncalci ed and the calci ed layers of the articular cartilage have previously been described in degenerative joint disease of humans, mice and hamsters, and have been suggested to be related with the shearing damage at the uncalci ed calci ed cartilage interface [47]. In OA, extremely hypermineralization was found in calci ed cartilage zone [44]. The above studies suggest that changes in the mineral content of calci ed cartilage zone may lead to changes in its mechanical properties, which damages the role of calci ed cartilage zone as the middle buffering zone between cartilage and bone.
AKT plays an important role in the maintenance of cartilage homeostasis and the progression of OA. In mice with surgically induced OA, calci ed osteophyte formation was prevented in the Akt1 −/− joints.
Calci cation was suppressed in cultured Akt1 −/− chondrocytes or ATDC5 cells with Akt1 knockdown, but enhanced in ATDC5 cells overexpressing constitutively active Akt1 [48]. The forced expression of constitutively active AKT rescued the expression of phenotypic markers and the apoptosis induced by CXCR2 blockade, indicating CXCR2-dependent chondrocyte homeostasis was mediated by AKT signaling [49]. IL-1β-mediated activation of NF-κB and apoptosis in chondrocytes was inhibited by IGF-1 and PDGFbb, which could be related to the suppression of Src/PI-3K/AKT pathway [50]. Similarly, Tenuigenin inhibits IL-1β-induced in ammation in human osteoarthritic chondrocytes by suppressing PI3K/AKT/NF-κB pathway [51]. The AKT is necessary for the synergistic induction of MMP-1 and MMP-13 and the cartilage breakdown stimulated by IL-1 in combination with oncostatin M. Moreover, C.Shen et al. reported that Dex increased the expression of Akt in human chondrocytes, which was related to the degenerative process in cartilage [52]. However, some studies reported that the AKT signaling was inhibited by Dex in other models, indicating the complicated mechanisms of the effects of Dex on different cells. More studies are needed to investigate the details.
Autophagy contributes to the maintenance of homeostasis of chondrocytes, whose impairment greatly aggravates OA. Autophagy is constitutively active in normal cartilage, which would be compromised with aging and precedes cartilage cell death and structural damage [53]. Rapamycin could improve severity of cartilage degradation as well as synovitis in mouse OA model via inducing autophagy, indicating pharmacological activation of autophagy may be an effective therapeutic approach for OA [54]. Besides, cartilage-speci c ablation of mTOR results in increased autophagy level and reduced articular cartilage degradation, apoptosis and synovial brosis in DMM OA model [55]. In addition, Bouderlique, T. et al reported that targeted deletion of Atg5 in chondrocytes promotes age-related OA by facilitating chondrocyte survival, suggesting that autophagy is bene cial to the age-related OA [56]. In this study, Dex treatment increased the autophagy activity of chondrocytes at the early stage, which was gradually decreased with the extension of processing time. We speculate that autophagy may be one of the adaptive protective responses for chondrocytes under the stimuli of Dex. The long-time application of Dex, however, weakened the autophagy-mediated protective effect and ultimately aggravated the damage of cartilage. More research are needed to investigate the details.
In brief, our present study revealed that Dex could enhance the calci cation of cartilage via AKT activation and increase chondrocyte apoptosis through inhibiting autophagy, which aggravates the damage of articular cartilage and accelerate the progression of OA in vivo. Our data provided a new perspective to understand the effect of Dex on cartilage maintenance and degeneration, which may be bene cial for the clinical use of Dex for OA treatment in future.

Conclusions
Continuous infusion of Dex can enhance the calci cation of cartilage via AKT activation and increase chondrocyte apoptosis through inhibiting autophagy, which aggravates the damage of articular cartilage and accelerates the progression of OA in vivo.

Declarations
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.  Scale bar: 100μm. Data were expressed as the mean ± 95% con dence intervals. *=P 0.05, **=P 0.01, ***=P 0.001, ****=P 0.0001.

Figure 6
The calci cation was obviously increased by Dex partially dependent on AKT (A and C) Alizarin red staining was used to observe the effect of different concentration gradient Dex(A) and AKT signaling inhibitor LY294002(C) on extracellular matrix calci cation. (B and D) Alizarin red absorbance was detected by microplate reader at 405nm (n=3). (E) Cell lysates of primary chondrocytes were analyzed by western blotting using antibodies of AKT, P-AKT308, P-AKT473 (n=3). (F)Immunohistochemistry of P-