High-fat dietary consumption promotes disease progression in a surgical joint destabilisation murine model of osteoarthritis: role of systemic eicosanoids and circulating monocytes

In this study we investigated the contribution of high-fat diet-induced metabolic overload to osteoarthritis (OA) progression originally caused by mild mechanical trauma to the mouse knee joint. We hypothesized that metabolic stress would induce a proinflammatory environment by altering systemic lipid levels and immune cell populations. Twelve-week-old male C57BL/6J mice (n=20) were given a low-fat diet (LFD, 10%kcal from fat) or high-fat diet (HFD, 45%kcal from fat) for 18 weeks. OA was initiated by transecting the medial meniscotibial ligament of the right knee joint at t=10 weeks. OA severity and changes in M1/M2 polarization of synovial macrophage populations were determined in serial coronal FFPE-mounted sections. Eicosanoid levels and monocyte populations were evaluated before and after ligament transection.

Metabolic stress resulted in a proinflammatory environment and aggravated injuryinduced OA progression. Our results suggest that a CYP450-focused eicosanoid metabolism and activated circulating monocytes may be drivers of this metabolic stressinduced OA progression, contributing to the mechanistic understanding and potentially serving as future diagnostic and prognostic biomarkers for metabolic OA.

Background
Obesity is the second most significant risk factor for the development of osteoarthritis (OA), the most common form of arthritis [1]. Besides mechanical overload of the joints due to excess body weight, obesity contributes to OA pathogenesis through metabolic disbalance [2], also termed metabolic stress. Metabolic stress is thought to be a major contributor in a subtype of OA referred to as metabolic OA [3][4][5][6], an OA phenotype that is characterized by a systemic low-grade inflammation -although this classification is still under debate [7]. Metabolic OA prevalence is highest in developed countries, where the incidence of obesity progressively increases due to a sedentary lifestyle, low levels of physical activity, and high-caloric diet [8]. The pathogenesis of metabolic OA has been linked to several features of the metabolic syndrome, such as central obesity, insulin resistance and dyslipidemia [2,9,10]. Even though the association between OA and metabolic stress is demonstrated by many studies, the underlying pathophysiologic mechanisms are still vague.
Animal models of diet-induced OA can help elucidate the mechanisms underlying metabolic OA. It is suggested that high-fat feeding induces an overall primed state and the associated metabolic stress undermines the body's capacity to adequately cope with insults like knee injury [11,12]. The obesity-prone C57BL6/J mouse is widely used to study metabolic OA and applied diets are typically high in caloric content, like Western-type diets or high-fat diets (45-60 kcal% energy from fat) [13]. Diet-induced OA severity is typically mild in animal models provided with a high fat diet only and we found that increased articular cartilage damage compared with low-fat diet (LFD) controls is not assured [14]. Rather, our results suggested that an additional trigger, over and above a high-caloric dietary stressor, is necessary to evoke metabolic OA.
In this study we investigated the contribution of metabolic stress to OA progression induced by mild trauma to the knee joint. We placed C57BL/6J mice on a low-fat diet (LFD, 10kcal% energy from fat) or high-fat diet (HFD, 45kcal% energy from fat) and induced OA onset by surgical destabilization of the medial meniscus (DMM). This post-traumatic OA (PTOA) model is low invasive and sufficiently sensitive to study subtle changes in disease progression by mild triggers like genetic background and diet [15,16]. As obesity is associated with a systemic low-grade inflammation, we hypothesized that diet-induced metabolic stress would induce a proinflammatory environment by altering systemic lipid levels and immune cell populations. We determined systemic eicosanoid levels, lipid mediators best known for their pivotal role in inflammation [17], and circulating monocyte populations. Peripheral blood monocytes are precursor cells for macrophages: an immune cell that plays a role in OA pathogenesis and is known to shift towards the proinflammatory (M1) phenotype during adipose-related inflammation [18,19]. Therefore, we also explored changes in M1/M2 polarization of synovial macrophage populations in PTOA.

Methods
Please refer to the online supplemental file for a detailed methods section.
The experiment was approved by the institutional Animal Care and Use Committee of TNO and was in compliance with ARRIVE guidelines and European Community specifications regarding the use of laboratory animals. The experiment was carried out in wild-type male C57BL/6J mice purchased from Charles River Laboratories (L'Arbresle Cedex, France) that received a 2-week acclimatization period after transfer to a specific pathogen-free (SPF) animal facility. Twelve-week old male C57BL/6J mice (n=10 per group) were given a synthetic high-fat diet (HFD, 45% kcal from fat) or low-fat diet (LFD, 10% kcal from fat) for 18 weeks. Ten weeks after starting the diet, OA was surgically induced in the right knee joint by transecting the medial meniscotibial ligament (MMTL) and destabilizing the medial meniscus (DMM). The left knee joint served as an experimental control and received all surgical procedures except for the MMTL transection (sham surgery). Mice were euthanized 8 weeks post-surgery using gradual-fill CO 2 asphyxiation.
Metabolic stress was assessed throughout the experiment by monitoring changes in body weight, body composition, and systemic total cholesterol, glucose and insulin levels. Serial coronal 5 µm FFPE sections were collected throughout the joint at 60 μm intervals, stained, and evaluated for cartilage degradation, osteophyte formation and synovitis using the OARSI scoring system specifically designed for the mouse[20]. Eicosanoid profiling was performed in fasted plasma samples as described previously [21]. Monocyte subpopulations were analysed in heparin-buffered peripheral blood samples stained with the antibody panel shown in Table 1. Data was obtained using a 3-laser FACSCanto TM II flow cytometer and analysed with FlowJo v10.2 software. FFPE sections from DMM-operated knee joints were triple-stained with primary antibodies targeting F4/80 (pan-macrophage), iNOS (M1) and CD206 (M2). Quantification was performed in the lateral patellofemoral synovial lining (ROI), unmixed colour spectra were obtained with a Nuance multispectral imaging system (40x magnification), and analysed with ImageJ 1.51n software. Statistical analysis was performed using IBM SPSS software. In all analyses, a probability value < 0.05 was considered statistically significant. Unless stated otherwise, data are presented by the median with interquartile range (range between the 25th to 75th percentiles).

Metabolic stress aggravates PTOA development
Diet-induced metabolic stress was observed in the HFD group, as demonstrated by the metabolic profile of the HFD animals at the end of the study (Table 2). Metabolic stress was already present before surgery: body weight, cholesterol levels, and HOMA-IR were significantly increased in HFD mice compared with LFD controls after 9 weeks (data not shown). Body composition analysis revealed that the increase in body weight was completely attributed to an increase in body fat mass. Synovial inflammation was significantly increased and more severe in HFD mice (7 out of 8 8 mice; 6.5 [5.3-8.5]) compared with LFD mice (4 out of 7 mice; 1.0 [0.0-6.0]; p=0.04, Figure 2D). Sham-operated contralateral knees showed no synovial thickening. Increased osteophyte formation was observed in the HFD group compared with the LFD controls ( Figure 2E). Sham-operated contralateral knee joints showed hardly any osteophyte formation.

Systemic changes in eicosanoid levels
Eicosanoids can exert both pro-and anti-inflammatory functions and have been implicated in the pathogenesis of a variety of immunometabolic disorders, like obesity 13 [17].
Because of plasma volume limitations and the volume demands of the assay, we were able to reliably detect the levels of 24 eicosanoids. Overall, minor differences were observed between the diet groups. HFD induced a significant increase in arachidonic acid (AA) levels, a n-6 PUFA precursor, with 947 [765-1257] in LFD versus 1538  in HFD mice (p<0.01) at 14 weeks. To find out how this increase affected total eicosanoid metabolism, we evaluated the changes in eicosanoid profiles of the mice before and after DMM surgery. Pathway analysis revealed that the increase in AA levels mainly translated into increased hydroxylation into the fatty acid diols 5,6-dihydroxyeicosatrienoic acid (DHET) and 8,9-DHET ( Figure 3). In addition, 5-HETE and 13-OxoODE levels were increased in HFD mice compared with LFD mice. Similar plasma eicosanoid profiles were observed between 9 and 14 weeks (data not shown).

Activation of circulating monocyte subpopulations
Monocytes are short-lived mononuclear phagocytes that constitute ∼10% of peripheral leukocytes in humans and ∼4% in mice [23]. Monocyte subpopulation percentages showed subtle differences between the LFD and HFD groups over time ( Figure 4A). The most pronounced effect of long-term HFD was an increase in circulating CD43 + + Ly6C int intermediate monocytes at t=17 weeks, at the expense of the CD43 + + Ly6C low non-classical subpopulation. This increase showed significance within the total CD115 + CD11b + SSC-A low monocyte fraction (p = 0.044, data not shown), but due to interindividual variability did not reach significance within the total leukocyte fraction (p = 0.107, Figure 4B).  Figure 4C). CD11c contributes to monocyte arrest on endothelial cells and, as such, is upregulated by activated monocytes during chemotaxis.
Both high-fat diet and knee injury contributed to the increased expression of CD11c, respectively, as: 1) HFD mice showed significantly more CD11c expression compared with LFD controls at 9 and 17 weeks (p<0.01) and 2) LFD mice showed increased CD11c expression over time (p=0.03). Backgating (i.e. highlighting the final gated population within the population of its ancestors) of the CD11c + cells into the total CD115 + CD11b + SSC-A low monocyte fraction revealed that this integrin was predominantly upregulated by the CD43 + + Ly6C low non-classical monocytes ( Figure 4D). Aggravation of local injury due to metabolic stress was evident from the observed spatial differences in OA features. HFD mice showed more generalized cartilage damage compared to LFD mice, which displayed the medially oriented cartilage damage known to the DMM model [15]. The observed diet-induced disease aggravation confirms results from other PTOA mouse models on a HFD regimen [16,25] and is in line with faster disease progression reported for obese OA patients [26]. The clear increases in lateral cartilage damage, synovial inflammation, and osteophyte formation compared with LFD mice qualifies the HFD DMM model as a suitable preclinical mouse model to identify players and pathways involved in metabolic OA. As both the internal sham control joints and the LFD animals developed considerably smaller OA features for all measured parameters, we conclude that HFD feeding on its own has limited potential to induce OA but clearly creates a proinflammatory environment leading to aggravation of PTOA progression.

High-fat feeding induced no macrophage polarization
To identify which factors might contribute to a diet-induced proinflammatory environment, we explored changes in systemic eicosanoid levels. In addition to regulating a wide range of physiologic processes [27], eicosanoids can exert both pro-and anti-inflammatory functions [28]. Of note, synovial fluid eicosanoid profiles were not reflected systemically in a rat model of HFD-induced OA, which suggests differential roles of eicosanoids in local versus peripheral compartments [29]. Although local eicosanoid involvement has been reported in OA [30][31][32][33], their systemic role in OA pathogenesis is less explored. In our model, HFD sparked a clear increase in systemic arachidonic acid (AA) levels, a n-6 polyunsaturated fatty acid that can trigger a pro-inflammatory cascade. Increased AA hydroxylation by the enzyme soluble epoxide hydrolase (sEH) in the cytochrome P450 pathway generated increased levels of dihydroxyeicosatrienoic acids (DHETs). This probably reflects the diet-induced metabolic stress, as obesity has been shown to induce altered CYP-expression and increased sEH-generated DHET levels in models of metabolic syndrome [34][35][36]. Accordingly, metabolic syndrome was ameliorated by pharmacological inhibition of sEH [37,38]. A similar dysregulation of CYP-mediated eicosanoid metabolism has also been observed in patients with cardiovascular disease, a metabolic condition that shares pathophysiological similarities with OA[39], for which sEH inhibitors (sEHi) have shown promising therapeutic effects [40]. Recently, increased synovial fluid levels of sEHgenerated n-6 PUFAs were positively associated with knee OA [41], demonstrating a link between eicosanoid metabolism and OA progression as well. Although initially considered to be inactive EET degradation products, DHETs have been reported to promote CCL2-mediated monocyte chemotaxis in vivo and to restore sEHi-blocked human monocyte migration in vitro [42]. This is particularly interesting, as the CCL2/CCR2 signalling axis has been shown to preferentially mediate monocyte trafficking and promote inflammation and tissue damage in OA [43]. Because circulating monocytes can replenish tissue-resident macrophage populations, which are important players in OA pathogenesis, high systemic DHET levels may directly or indirectly influence the progression of OA through the activation of peripheral monocyte subsets.
To investigate whether peripheral monocyte subsets were indeed affected in our model, we analysed changes in circulating monocyte subsets before and after knee injury. We observed significantly increased monocytic activation, reflected by the upregulation of β 2integrin CD11c. HFD mice increased monocytic CD11c expression before DMM surgery, demonstrating a dietary effect. It is interesting that LFD mice showed more CD11c + monocytes at end point (8 weeks after DMM) than before DMM surgery. As it seems not likely that CD11c is a marker for ageing alone, this finding suggests that monocytic CD11c expression partly reflects the consequences of the DMM surgery. However, due to the lack of external sham-operated controls we were not able to verify this possibility in our study.
Our finding that CD11c was predominantly expressed by non-classical monocytes is consistent with earlier reports [44] and points to tissue infiltration by this subset. It might be that the increase in the circulating intermediate CD43 + + Ly6C int monocyte subpopulation indicates renewing of the non-classical monocyte pool by maturation of peripheral classical monocytes. Alternatively, the increase of this pro-inflammatory monocyte subset might indicate a role in metabolic OA pathogenesis. In humans, intermediate monocytes appear to be the main population to be perturbed in disease conditions [45], including metabolic syndrome, rheumatoid arthritis and atherosclerosis [46][47][48]. Though preliminary, our observations show consistency with current literature on the role of monocyte subsets in inflammation and OA. Elevated monocytic activation has also been reported in women with knee OA and positively associated with BMI [49]. Similar preclinical findings were made in a mouse model of dietinduced atherosclerosis, in which low-grade inflammation sustained elevated levels of CD11b + Ly6C hi pro-inflammatory monocytes [50]. Here, similar low-grade inflammation instigated significant in vitro expansion of intermediate monocytes in human peripheral blood mononuclear cells [50]. Taken together, our data add to the current understanding that a systemic dysregulated lipid metabolism and immune activation may be independent yet intertwined processes that contribute to the progression of metabolic OA.
We hypothesized that M1 polarization as seen in obesity might drive local metabolic OA pathogenesis, but no significant increases or shift in synovial macrophage subpopulations were observed. The relative increase of synovial F4/80 + macrophages in the HFD group is likely an underestimation, as significant synovial thickening inherently affects surface area percentages in our image analysis. However, despite the low sample numbers with interindividual variability, our novel triple immunostaining provided a clear image of the local macrophage polarization. We observed a mixed pattern of both M1 and M2 macrophage markers together with increased iNOS deposition in the synovia of HFD mice (data not shown). These novel findings are in line with clinical studies showing that activated human macrophages in the OA knee joint express both M1 and M2 markers while displaying a pro-inflammatory cytokine profile [51,52]. Therefore, the increased iNOS secretion in the HFD group might reflect altered macrophage activation (instead of polarization) and indicate a pro-inflammatory microenvironment in the knee joint.
In conclusion, we have demonstrated that metabolic stress as applied in this model on its own does not provide sufficient stress for OA development, but clearly aggravated PTOA progression by promoting cartilage damage, synovial inflammation, and osteophyte formation. HFD instigated a proinflammatory environment through systemic changes in lipid metabolism and activation of non-classical monocytes. Based on these results, we propose that a CYP450-focused eicosanoid metabolism and activated circulating monocytes may be drivers of metabolic OA progression. These systemic factors might have future potential as diagnostic and prognostic markers for metabolic OA, consistent with conclusions from a recent clinical study [53,54]. For the clinical setting, our results imply that pharmacological inhibition of metabolic stress could benefit obese patients with secondary OA. Specifically, inhibition of soluble epoxide hydrolase could be a promising strategy to attenuate metabolic OA.

CD11b
Cluster of differentiation 11b (Integrin alpha M)  Original magnification ×20. Data are presented as group median (indicated by bars) with interquartile range (error bars). *Statistical significance was set to p<0.05.  CD11c-expressing monocytes significantly increased in the HFD group and correlate with osteophytosis and synovial inflammation. A) Peripheral CD115+CD11b+SSC-Alow monocyte subpopulations showed subtle changes in