Objective The mechanisms linking obesity and osteoarthritis (OA) are not fully understood and have been generally attributed to increased weight, rather than metabolic or inflammatory factors. Here, we examined the influence of fatty acids, adipokines, and body weight on OA following joint injury in an obese mouse model.
Methods Mice were fed high-fat diets rich in various fatty acids (FA) including saturated FAs (SFAs), ω-6 polyunsaturated FAs (PUFAs), and ω-3 PUFAs. OA was induced by destabilising the medial meniscus. Wound healing was evaluated using an ear punch. OA, synovitis and wound healing were determined histologically, while bone changes were measured using microCT. Activity levels and serum cytokines were measured at various time-points. Multivariate models were performed to elucidate the associations of dietary, metabolic and mechanical factors with OA and wound healing.
Results Using weight-matched mice and multivariate models, we found that OA was significantly associated with dietary fatty acid content and serum adipokine levels, but not with body weight. Furthermore, spontaneous activity of the mice was independent of OA development. Small amounts of ω-3 PUFAs (8% by kcal) in a high-fat diet were sufficient to mitigate injury-induced OA, decreasing leptin and resistin levels. ω-3 PUFAs significantly enhanced wound repair, SFAs or ω-6 PUFAs independently increased OA severity, heterotopic ossification and scar tissue formation.
Conclusions Our results indicate that with obesity, dietary FA content regulates wound healing and OA severity following joint injury, independent of body weight, supporting the need for further studies of dietary FA supplements as a potential therapeutic approach for OA.
Statistics from Altmetric.com
Obesity is one of the primary risk factors for osteoarthritis (OA), although the mechanisms linking these conditions are not fully understood.1 While it has been believed that increased loading on the joints due to weight gain is responsible for accelerated OA with obesity, mechanical factors alone do not account for the higher incidence of OA in non-weight-bearing joints, such as the hands.2 Furthermore, studies have shown that morbidly obese mice do not develop OA when fed standard (low-fat) chow.3 These findings suggest that factors other than adiposity or body weight—such as dietary content or the circulating levels of adipokines—must contribute to OA in obesity.
Cellular stress due to obesity induces lipolysis of adipocytes,4 increasing the levels of circulating free fatty acids (FA). These FAs can serve as proinflammatory or anti-inflammatory molecules in metabolic signalling. Saturated FAs (SFA) can activate macrophages to secrete tumour necrosis factor α (TNF-α) and interleukin 1 (IL-1).5 Furthermore, the derivatives of ω-6 polyunsaturated FAs (PUFA) are involved in joint pain.6 ,7 Conversely, ω-3 PUFAs have been reported to reduce spontaneous OA in animals fed a low-fat diet.8 These findings imply that dietary or metabolic factors may play a more direct role in joint degeneration. Furthermore, little is known regarding the effects of dietary FAs on motor function, pain perception and wound healing in obesity.
The goal of this study was to determine the effects of dietary FAs content in injury-induced OA, and to identify the primary factors linking obesity and OA. Mice fed various high-fat diets rich in SFA, ω-6, or ω-3 PUFA underwent surgery to destabilise the medial meniscus (DMM) to induce OA.9 We also investigated the effect of FAs on wound regeneration and behavioural activity.
All procedures were approved by the Duke University IACUC. Beginning at 4 weeks of age, mice were fed a control low-fat diet or one of the three high-fat diets: SFA-rich, ω-6 PUFA-rich, or ω-3 PUFA rich (see online supplementary table S1). At 16 weeks of age, mice underwent DMM surgery to induce knee OA in the left hind limb,9 and ears were punched using 1 mm (right) and 1.5 mm (left) diameter ear punches to examine wound healing responses. To determine how diet affected behaviour and activity levels, mice were monitored at 6, 14 and 24 weeks of age. The study design is presented in online supplementary figure S1.
Evaluation of bone structure, OA and synovitis
Morphometric bone parameters and heterotopic ossification of the limbs was analysed using microCT.10 After standard histological processes, the joints were stained for evaluation of OA and synovitis.
Statistical analyses are described in each figure caption. Analyses were performed using SPSS Statistics, with significance reported at the 95% confidence level.
Detailed methods are provided in the online supplementary document.
At 28 weeks of age, SFA and ω-6 mice were heavier than control and ω-3 mice (figure 1A). Although animals lost weight slightly at 8, 17 and 25 weeks, likely due to the mild stress of behavioural testing, they maintained the trend of gaining weight. To describe the influence of weight on joints over time, the area under the weight-versus-time curves (AUC) was calculated by using a trapezoidal rule (figure 1B).11 All high-fat diet-fed mice had higher AUC4–28 week versus control mice.
Body composition and gene expression
At 28 weeks of age, SFA and ω-6 mice had increased body fat percentage, but decreased bone mineral density (BMD) versus control mice. The ω-3 mice displayed lower BMD but no difference in body fat percentage versus control mice (figure 1C, D). The SFA and ω-6 mice had a higher percentage of epididymal fat relative to body weight, and only ω-6 mice had a significantly higher percentage of inguinal fat than control mice (figure 1E).
F4/80+ macrophages staining in adipose tissue indicated that SFA and ω-6 PUFA diets were associated with massive macrophage infiltration (figure 1F). Gene expression of epididymal fat showed that SFA and ω-6 PUFA diets trended toward higher expression of IL-6 with no difference in MCP-1 expression versus those of the ω-3 and control mice (figure 1G). All the high-fat diet fed mice had higher expression of F4/80 and CD11c versus control mice.
MicroCT imaging of the joints (figure 2A) showed that the ω-3 mice had a lower bone fraction (bone volume/total volume, BV/TV) of the femoral condyle as compared to the other diets (figure 2B). Surgery, but not diet, decreased the BMD of the femoral condyle (figure 2C). For the tibial epiphysis, ω-3 mice had lower BV and BMD versus the mice treated with the other diets. For all mice, surgery increased BV but decreased the BMD of the tibial epiphysis (figure 2C, D). Diet affected the incidence of heterotopic ossification in the DMM-operated joint (see online supplementary table S2). The control and ω-3 mice had a lower incidence of heterotopic ossification. The SFA and ω-6 mice had high BV of heterotopic ossification, while control mice had the lowest (figure 2E).
ω-3 PUFAs mitigate injury-induced OA in obese mice
A modified Mankin grading scheme was used to determine OA severity. Severe cartilage loss was observed in the DMM-operated joints of SFA and ω-6 mice; the operated joints of control and ω-3 mice showed surface fibrillation and moderate loss of proteoglycan (figure 3A). Three out of 14 ω-6 mice had severe subchondral bone erosions (see online supplementary figure S2). All operated joints had higher OA scores than their corresponding non-operated joints. The operated joints of SFA and ω-6 mice exhibited the most severe OA versus the operated joints of control and ω-3 mice. The non-operated joints did not differ among diet groups.
Osteophytes were present primarily in the operated joints (see online supplementary table S3). SFA and ω-6 mice trended toward greater osteophyte severity than control and ω-3 mice (see online supplementary table S4 and figure 3B). Osteophyte score also correlated positively with OA (figure 3C).
ω-3 PUFAs decrease synovitis in obese mice
Synovitis was determined by a previously established grading scheme consisting of assessment of stromal cell density and lining layer thickness.12 Compared to the non-operated joints, the DMM-operated joints of all diet groups had increased synovial lining hyperplasia (figure 3D). Operated joints of SFA and ω-6 mice exhibited a thicker synovium with a higher number of infiltrating cells than those of mice fed control or ω-3 diets. The joints of ω-3 mice had less synovitis versus those of ω-6 mice.
Macrophage distribution in synovial tissue
At 12 weeks postsurgery, macrophages were still present within the synovium of the operated joints from all groups (figure 3E). However, for control and ω-3 mice, macrophages appeared mostly in the synovial lining layer, while for SFA and ω-6 mice, macrophages were either distributed throughout the synovial stroma or were contained in the follicle-like lymphocytic infiltrates. The ω-6 mice exhibited increased frequency of macrophages in the synovium (see online supplementary table S5). The ω-6 mice had the highest macrophage score in the medial side of the joint versus control and ω-3 mice (figure 3E).
Serum cytokines and mediators
The SFA and ω-6 mice had the highest insulin concentrations versus control and ω-3 mice at 23 weeks of age. Although the insulin levels of all mice decreased at 28 weeks of age (potentially due to the time of serum collection on the day of euthanasia was different from that of other time-points), ω-6 mice still exhibited higher insulin concentrations than control mice (figure 3F). At 23 and 28 weeks of age, SFA and ω-6 mice had elevated leptin concentrations versus control and ω-3 mice, while ω-3 mice exhibited higher adiponectin levels (figure 3G,H). The ω-3 mice had lower levels of resistin (figure 3I) but exhibited the highest concentrations of prostaglandin E2 (PGE2) relative to other mice (figure 3J). While there were no differences in the active form of TGF-β1 among the mice, the ω-3 mice showed the highest latent TGF-β1 concentration (see online supplementary figure S3) but the lowest active-to-total TGF-β1 ratio (figure 3K). To investigate the relationships between each cytokine and OA, we performed bivariate models (see online supplementary table S6) and found leptin and resistin had a positive association with OA.
The relationships among mechanical factors, metabolic factors and OA
A potential confounding variable in evaluating the links between metabolic factors and OA is the effect of body weight. To control for differences in body weight among the diet groups, we first examined weight-matched mice from each group. In weight-matched mice, SFA and ω-6 mice still demonstrated significantly higher OA severity than the mice fed the other diets (see online supplementary figure S4). Multivariate models were then performed to further validate these associations (table 1). Only diet and metabolic factors, leptin and resistin, but not body weight (AUC4–28 or AUC17–28) were significantly associated with OA.
The data at 6 and 14 weeks of age were used to examine how diet affected changes in biomechanical and neurobehavioural functions of the mice. Results at 24 weeks of age were used to determine whether diets altered mouse activity levels after OA induction.
At 14 and 24 weeks of age, diet did not significantly influence voluntary activity or rotarod performance (figure 4A, B). To determine the effect of high-fat feeding, we combined all high-fat diet groups together and compared them with the low-fat diet group. Six to 14-week-old mice subjected to high-fat feeding had lower motor function but maintained similar spontaneous activities versus control mice (see online supplementary table S7). With age, high-fat feeding decreased forelimb grip strength, but had no effect on hind-limb grip strength (figure 4C, D). Rotarod performance was positively associated with forelimb grip strength, indicating that musculoskeletal strength is related to motor function (see online supplementary figure S5).13
The effect of diet on thermal hyperalgesia was evaluated using the hot-plate and tail-flick tests, which investigate nociceptive reflexes that are associated with supraspinal and spinal pathways, respectively (figure 4E, F).14 There was no significant difference in the hot-plate latency among the mice at any time-points. However, after 10 weeks of feeding, ω-3 and SFA mice had decreased tail-flick latency versus control mice. At 24 weeks, a significant difference in tail-flick latency was observed between ω-6 and SFA mice, but not between other diet groups. High-fat feeding decreased tail-flick latency in the period from 6 to 14 weeks of age.
Using bivariate models, we examined whether activity levels correlated with weight gain or OA (see online supplementary table S8). Spontaneous locomotion did not correlate with either weight gain or OA, while rotarod latency was negatively associated with weight gain but not with OA. Grip strength was negatively associated with weight gain, but only forelimb grip strength was negatively associated with OA. Neither hot-plate nor tail-flick latency correlated with either weight gain or OA.
ω-3 PUFAs accelerate wound repair in obese mice
ω-3 Mice had significantly improved regeneration (figure 5A, B), characterised by increased epithelial thickness (figure 5C, D). Three out of 11 ω-3 mice had complete epithelial fusion of the proximal and distal wound margins (see online supplementary figure S6). All the mice demonstrated some features of regeneration including re-epithelialisation and formation of sebaceous glands (figure 5E). We next evaluated whether the cells in the wound margins were in a proliferative stage using Ki-67 marker; however, no obvious differences were observed (figure 5F). Picrosirius red staining was used to investigate the wound matrix composition. The ω-3 mice showed less collagen type I (COLI) deposition, while the other mice had densely packed COLI fibres (figure 5G). Ear wound size showed a trend toward positive association with OA in the 1.5 mm model (figure 5H).
The findings of this study showed that a small amount of ω-3 PUFA supplementation was sufficient to mitigate the effects of obesity on injury-induced OA, and to accelerate would repair. Conversely, SFA and ω-6 PUFA independently acted as a detrimental factor in OA following joint injury, increasing osteophyte formation, heterotopic ossification, synovitis as well as increasing infiltration of macrophages into synovial tissue. By examining multivariate models and weight-matched mice from different diet groups, we found that injury-induced OA was only associated with dietary content and serum levels of proinflammatory adipokines, but not with body weight or activity levels. Our results indicate that dietary and metabolic factors may play a more significant role than body weight in the link between obesity and post-traumatic OA.
To investigate the specific effects of SFAs on OA, we maintained the same ω-6 to ω-3 PUFA ratio in the low-fat and SFA-rich high-fat diet. We found that SFAs significantly exacerbated OA as compared to a low-fat diet with the same PUFA ratio. This result is consistent with several animal studies showing that a high-fat diet rich in SFA increases the severity of injury-induced arthritis.10 ,15 Although the specific effect of SFAs on chondrocytes is less well characterised, SFAs can activate synovial macrophages to secrete IL-1 and TNF-α that are involved in cartilage degradation.16 The Western diet is characterised by a high ratio of ω-6 to ω-3 PUFAs.17 While maintaining the same PUFA content, but altering ω-6 and ω-3 ratio in PUFA-rich high-fat diets, we found that the ω-6 mice developed severe OA and synovitis with elevated systemic inflammation. By contrast with ω-6 PUFAs, the beneficial effect of ω-3 PUFAs in OA and rheumatoid arthritis has been reported in animal models.8 ,18 However, most of these studies supplemented ω-3 PUFAs in regular chow (ie, low-fat diet) and, thus, the investigation of ω-3 PUFAs in the context of a high-fat diet is a novel aspect of this study. Here, we discovered that even a relatively small amount of supplementation (only 8% kcal of the energy provided), ω-3 PUFAs provided protective effects on osteoarthritic changes of the joint after injury, while reducing leptin and resistin levels. Additionally, ω-3 mice had high levels of adiponectin.19 The influence of adiponectin on chondrocytes is not fully understood, and some evidence suggests that adiponectin is associated with cartilage matrix breakdown.20 Nevertheless, adiponectin may indirectly benefit cartilage by reducing inflammation through polarising macrophages toward anti-inflammatory phenotypes.21
Despite having relatively higher AUC values than controls, obese mice supplemented with ω-3 PUFAs showed similar OA scores as control mice, suggesting that factors other than body weight are responsible for OA severity in this model. In examining weight-matched mice from different diet groups, or using multivariate models, we showed that injury-induced OA was significantly associated with diet and pro-inflammatory adipokines, but it was not with body mass. These results emphasise the potential significance that systemic metabolic factors may play in exacerbating injury-induced OA.
Another significant finding was that ω-3 obese mice showed superior wound healing capacity. However, the cell proliferation marker did not differ among the diet groups, likely due to the fact that the skin wound of rodents enters the maturation phase 14 days postinjury,22 with reduced Ki-67 expression.23 The ω-3 mice also contained low levels of COLI fibres in the wounding area, suggesting less scar formation. Additionally, adiponectin has been shown to accelerate wound repair by promoting keratinocyte proliferation.24 ,25 Furthermore, low ratio of active/total TGF-β1 in ω-3 mice may further prevent them from developing scar tissue because active TGF-β1 is involved in excessive matrix deposition.26
Interestingly, we observed that the healing capacity of the ear wound tended to be negatively associated with OA. These findings are consistent with a recent study demonstrated that ear wound closure and cartilage regeneration may share a common heritable genetic basis that is associated with OA severity.27 Our findings suggest that the effects of diet may similarly reflect associations between wound healing and OA via epigenetic changes, potentially in the body's stem cell populations.28
The ω-3 mice exhibited a low incidence of mature osteophytes, potentially due to their low active/total TGF-β1 ratio systemically, as TGF-β1 is a potent inducer of osteophyte.29 Furthermore, ω-3 mice showed lower levels of heterotopic ossification, which is associated with surgical trauma during DMM. Tendon mineralisation has been reported in patients with tendon rupture, and could be a cause of chronic pain.30 Studies have indicated that PGE2, a lipid derivative from ω-6 PUFAs, enhances osteogenesis of tendon stem cells, providing a potential explanation for the greater heterotopic ossification in ω-6 mice. As anticipated, PGE2 concentrations were relatively higher in the ω-6 mice than those in control mice, although not statistically significant. To our surprise, ω-3 mice had the highest PGE2 levels. The reason for this phenomenon remains unclear but may be related to their high adiponectin levels, as a recent study reported that adiponectin stimulates PGE2 production in cells in a dose-dependent manner.31
Although the whole body BMD did not differ among the obese mice, ω-3 mice had lower BMD and BV in the epiphyseal region of both hindlimbs. Increased subchondral bone density is a hallmark of OA, and an inverse relationship between osteoporosis and OA has been observed in humans.32 Our observations of lower BMD, BV and less heterotopic ossification in the joints of the ω-3 mice do not support the protective role of ω-3 PUFAs on bone measures in post-traumatic OA,33 but are consistent with a recent study demonstrating that mice with upregulated adiponectin expression had decreased osteocalcin levels and displayed low bone-mass phenotypes.34 Nevertheless, further studies are required to determine the long-term effects of ω-3 PUFAs on bone metabolism in obese mice.
The fact that high-fat feeding did not alter spontaneous locomotor activity is consistent with our prior obese animal studies and those of others,13 ,35 ,36 suggesting that weight gain is not associated with lower voluntary activity or energy expenditure. In humans, a study conducted with lean and obese individuals indicates similar levels of spontaneous physical activity.37 Furthermore, our result that spontaneous locomotion was independent of injury-induced OA is in agreement with recent findings using cruciate ligament transection OA model in mice.38 It has also been reported that recreational activities do not contribute to OA in normal and overweight individuals.39 ,40
Our results indicate that dietary FAs differentially influence the development of injury-induced OA, contributing a more critical role in osteoarthritic changes of the joint than does mechanical factor in obesity. The progress of OA and wound repair could be explained by regulation of obesity-associated inflammation (see online supplementary figure S7). Our findings have significant implications on the mechanisms of injury-induced OA and wound healing, and provide a path toward clinical studies of dietary FA supplements to modify the course of OA.
The authors thank Mr Steve Johnson of the Duke Orthopedic Research Laboratories for assistance with animal handling. The authors also thank Mr Theodore Rhodes and Mr Christopher Means of Duke Mouse Neuroendocrine Analysis Core Facility for their assistance in behavioral testing.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors C-LW designed and performed the experiments, analysed the data, and wrote the paper. DJ, JNM, DL, JAA, JLH and RMR performed the experiments. JLH, VBK, RMR and WCW analysed the data. FG designed the experiments and wrote the paper.
Funding NIH grants AR50245, AG15768, AR48852, AR48182, AG46927, AR059784, Taiwan GSSA graduate fellowship, the North Carolina Biotechnology Center, and the Arthritis Foundation.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.