Metformin alleviates hyperuricaemia-induced serum FFA elevation and insulin resistance by inhibiting adipocyte hypertrophy and reversing suppressed white adipose tissue beiging

Hyperuricaemia (HUA) significantly increases the risk of metabolic syndrome and is strongly associated with the increased prevalence of high serum free fatty acids (FFAs) and insulin resistance. However, the underlying mechanisms are not well established, especially the effect of uric acid (UA) on adipose tissue, a vital organ in regulating whole-body energy and FFA homeostasis. In this study, we noticed that adipocytes from the white adipose tissue of patients with HUA were hypertrophied and had decreased UCP1 expression. To test the effects of UA on adipose tissue, we built both in vitro and in vivo HUA models and elucidated that a high level of UA could induce hypertrophy of adipocytes, inhibit their hyperplasia, and reduce their beige-like characteristics. According to mRNA-sequencing analysis, UA significantly decreased the expression of leptin in adipocytes, which was closely related to fatty acid metabolism and the AMPK signalling pathway, as indicated by KEGG pathway analysis. Moreover, lowering UA using benzbromarone (a uricosuric agent) or metformin-induced activation of AMPK expression significantly attenuated UA-induced FFA metabolism impairment and adipose beiging suppression, which subsequently alleviated serum FFA elevation and insulin resistance in HUA mice. Taken together, these observations confirm that UA is involved in the aetiology of metabolic abnormalities in adipose tissue by regulating leptin-AMPK pathway ， and metformin could lessen HUA-induced serum FFA elevation and insulin resistance by improving adipose tissue function via AMPK activation. Therefore, metformin could represent a novel treatment strategy for HUA-related metabolic disorders.


Introduction
Uric acid (UA) is a final oxidation product of purine catabolism. The serum UA balance is maintained by dietary uptake, production and excretion of purines [1].
Hyperuricaemia (HUA) is associated with various diseases, such as coronary artery disease, hypertension, diabetes mellitus, cerebrovascular disease, chronic kidney disease and gout [2][3][4]. In addition, clinical trials have shown that a high serum UA level also significantly increases the risk of metabolic syndrome [5,6], such as hypertriglyceridemia and high free fatty acids (FFAs) [7], and results in insulin resistance [8]. However, the effects and exact mechanisms remain unclear.
White adipose tissue (WAT) is a vital organ in maintaining lipid homeostasis through a fine-tuned system of uptake, esterification (lipogenesis) and release of FFAs (lipolysis), the so-called "triacylglycerol cycling" [9,10]. When energy is in surplus, lipid is stored through the enlargement of adipocytes (hypertrophy) and increase in adipocyte numbers (hyperplasia) [11], thereby keeping FFAs and blood glucose below toxic levels [12]. Moreover, WAT could be shifted to beige or so-called brite (brown-like-in-white) adipose tissue, which possesses the brown-like feature of energy dissipation through heat production with the use of FFAs as fuel through activating uncoupling protein 1 (UCP1), the brown adipose tissue-specific protein [13]. In such a way, adipose tissue ensures that the serum level of FFAs, which serve as a key energy source for skeletal muscle, myocardium and other major organs, is within the normal range. However, when adipose tissue is dysfunctional, serum FFAs increase, and associated metabolic complications such as insulin resistance occur [14].
Adipocytes express urate-anion exchanger 1 (URAT1), a transporting protein on cell responsible for the UA reabsorption [15], which indicated that UA could be absorbed by adipocytes and influence the function of adipose tissue. Clinical studies have found that obese individuals have a higher UA level than healthy controls [16], and elevated serum UA is closely associated with the accumulation of visceral fats [17]. William et al. demonstrated that UA could increase the expression of monocyte chemotactic protein-1 (MCP-1) and inhibit the production of adiponectin in cultured Downloaded from https://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20200580/884908/cs-2020-0580.pdf by guest on 23 June 2020 Clinical Science. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/CS20200580 adipocytes, which could be a potential mechanism of the insulin resistance in patients suffering from metabolic syndrome [18]. However, the exact effect of UA on affecting the metabolic function of adipose tissue, as well as the underlying molecular mechanisms, are poorly understood.
Metformin is a widely used oral anti-diabetic agent that can exert a beneficial effect on improving lipid metabolism [19], repress de novo lipogenesis in hepatocytes and prevent hepatic steatosis by AMP-activated protein kinase (AMPK) activation [20]. AMPK is a key regulator in different cellular metabolic pathways. Thus, we hypothesized that metformin could lessen HUA-induced serum FFA elevation and insulin resistance by improving adipose tissue function via AMPK activation. To verify this hypothesis, we built both in vitro and in vivo HUA models to investigate the effects as well as major mechanism of UA on the metabolic function of adipocyte tissue, and to test if AMPK activation by metformin is a novel potential treatment strategy for HUA-related metabolic disorders.

Animals and diets
The animal work was completed in the Animal Laboratory of the First Affiliated Hospital of Harbin Medical University and was approved by the Animal Care and Use Committee of the First Affiliated Hospital.
Six-week-old male C57BL/6J mice weighing 20±2 g (Vital River Laboratories, Beijing, China) were housed under a 12:12-h light-dark cycle at constant 22°C. The mice were randomly divided into two groups: the control group (n=10) was fed a chow diet (10% of kcal from fat) and drank normal water; the HUA group (n=30) was fed a high-fat diet containing excess fat (46%), high fructose (17.5%) and sucrose (17.5%) and drank 10% fructose water (HUA diet) for 16 weeks, as high-fat and high-fructose diets are reported to stimulate the overproduction of UA in systemic circulation, leading to the development of HUA in experimental animals [21]. Then, the HUA group mice were further randomly divided into three groups (n=10 each group): the HUA group was continue fed HUA diet; the HUA+Benz group was given HUA diet with intragastric administration of 20 mg/kg benzbromarone (a uricosuric agent clinically used for the regulation of HUA) for 4 weeks; and the HUA+Met group was given HUA diet with intragastric administration of 300 mg/kg metformin for 4 weeks.
Glucose tolerance test (GTT) and insulin tolerance test (ITT) were performed at the end of experiment. On the second day, all mice were killed to obtain adipose tissue and serum for a trial scheduled. Euthanasia was done under deep anaesthesia using 5% isoflurane inhalation and maintained throughout the surgical procedure.

Culture and differentiation of 3T3-L1 cells
The 3T3-L1 pre-adipocyte cells were purchased from the American Type Culture DMEM containing 10% FBS and 5 µg/ml insulin for another 48 h, and then replaced with normal culture medium for 6-8 days [22]. Subsequently, these cells were incubated with UA for 24 h. Here, 4 mg/dl was set as norm uricemia (approximately 250 mol/l), 8 mg/dl as HUA (500 mol/l), and 12 mg/dl as severe HUA (750 mol/l), which is consistent with the clinical assessment.

Western blot analysis
Total protein from eWAT and 3T3-L1 cells was extracted using RIPA buffer containing 10% phosphatase inhibitor and 1% protease inhibitor. The samples were  [24]. After that, the signals were incubated with ECL light reagent for 3 min (Beyotime, Shanghai, China). Lastly, the blots were imaged using a gel documentation system (BIO-RAD, Hercules, CA, USA), and images of blots were analysed by the Image Lab.

EdU proliferation assay
After PBS washing, the 3T3-L1 cells were incubated in 10 μmol/L EdU (RiboBio, Guangzhou, China) for 2 h. After that, the cells were fixed with 4% paraformaldehyde for 30 min and permeabilized with 0.5% Triton-X for 10 min. The number of cycling cells was detected by DNA staining [25]. Fluorescence microscopy was used for imaging.

H&E staining
Paraffin sections were dewaxed to water, stained in haematoxylin solution for 5 min, and then differentiated by 70% hydrochloric alcohol for 10 s. Next, paraffin sections were stained in eosin solution for 60 s after washing with deionized water. The samples were then cleared in xylene after washing and dehydrating by graded ethanol.
Finally, the sections were mounted in neutral balsam and images were obtained under a laser scanning microscope [27].

Immunohistochemistry staining
Paraffin-embedded eWAT sections were incubated overnight at 4°C with anti-F4/80 The 3T3-L1 cells were collected and sent to Novogene Corporation (Beijing, China) for mRNA sequencing. Briefly, 3 µg RNA from each sample was used for analysis.
Sequencing libraries were generated and each sample were index coded and clustered on a cBot Cluster Generation System. After that, the library preparations were sequenced on an Illumina HiSeq platform, and 125 bp/150 bp paired-end reads were generated. In-house Perl scripts were used to process raw data in fastq format.
Differential expression analysis of two conditions/groups was performed using the DESeq2 R package (1.16.1). Statistical enrichment of deferentially expressed genes in KEGG pathways was tested using the Cluster Profiler R package.

Biochemical analysis
The levels of serum triglyceride (TG), FFAs and UA were determined using assay kits

Glucose tolerance test (GTT)
Mice in each group received an intragastric administration of 50% glucose (5 ml/kg) after a 16-h fast. The sample blood glucose was detected from tail incision by a portable glucometer (Roche, USA) at baseline and 15, 30, 60, 90, and 120 min after glucose administration [29].

Insulin tolerance test (ITT)
Mice in each group were injected intraperitoneally with regular human insulin (1.5 U/kg) after a 6-h fast. The level of blood glucose of samples from tail incision was detected by a portable glucometer (Roche, USA) at baseline and 15, 30, 60, 90, and 120 min after insulin administration [29].

Statistical analyses
The data were presented as the means ± SEM. Differences between two groups were analysed by Student's t test, and multiple-group comparisons were analysed using one-way ANOVA followed by Tukey's tests. Figures were analysed using GraphPad Prism 7. A P value<0.05 was considered statistically significant.

Results
Adipocytes in patients with HUA disease showed hypertrophy and decreased UCP1 expression H&E staining was performed to provide an overview of the specimen and to quantify the size of the adipocytes. As Fig. 1A shows, the adipocyte size in patients with HUA was larger than that in the control group, and many broken adipocytes could be observed in the HUA group. Moreover, immunostaining showed that the expression of UCP1 was lower in the HUA group (Fig. 1B), indicating a reduction in beige-like characteristics in the adipose tissue of patients with HUA.

UA induces adipocyte hypertrophy and inhibits the hyperplasia of pre-adipocytes
In order to determine the impact of UA on adipocyte's morphology and function, we exposed 3T3-L1 adipocytes to gradually increasing concentrations of UA (from 0 to 12 mg/dl) and found a dose-dependent increase in lipid droplet volume (

UA reduces the beige-like characteristics of adipocytes
Subsequently, we measured the effects of UA on the beiging process of adipocytes.
We found that with increasing UA concentration, the fluorescence intensity of UCP1 was decreased (Fig. 2B) and that the expression of the main regulators of beige adipocyte formation, such as UCP1, PGC1α, and PRDM16, was also significantly reduced (Fig. 2C). To further determine whether β-adrenergic agonist-induced browning is also inhibited by UA, adipocytes were given the β-adrenergic agonist CL316,243. As Fig. S1B shows, the expression of UCP1, PGC1α, and PRDM16 was inhibited by UA according to western blot analysis, and the extent of UCP1 in the HUA group was also less than that in the control group, as shown by fluorescence intensity (Fig. S1C). These results indicated that beige adipocyte formation was suppressed by UA regardless of whether it was stimulated by a β-adrenergic agonist.
Adipocytes express URAT1, which is a key mediator transporting UA to cells [15], and previous study has found that expression of URAT1 gene in HUA population was associated with serum UA level [30]. Here, we also evaluated URAT1 expression in adipocyte to check if it is the key factor in regulating adipocyte function by UA. Our results showed that the expression of URAT1 didn't have a strong correlation with UA dosage (Fig. S1A).

Leptin plays a key role in the UA-induced FFA metabolic disorder of adipocytes
To characterize the molecular mechanism of UA in adipocyte metabolism, mRNA-sequencing analysis was performed. A total of 1204 mRNAs were differentially expressed, including 658 upregulated and 546 downregulated mRNAs (Fig. 3A). Among these differentially expressed mRNAs, leptin was found to be significantly downregulated with the highest mean ratio according to heat map representations (Fig. 3B). The KEGG analysis revealed that the "fatty acid metabolism" pathway was highly enriched with differentially expressed mRNAs (Fig.   3C), and leptin is a critical one among the mRNAs enriched in the fatty acid metabolism signalling pathway (Fig. 3D). Moreover, among the leptin-related biological pathways, the "AMPK signalling pathway" was at the forefront (Fig. 3E).
Therefore, we suggest that leptin is a key regulator in the UA-induced FFA metabolic Downloaded from https://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20200580/884908/cs-2020-0580.pdf by guest on 23 June 2020 disorder of adipocytes and that AMPK might be the main target of leptin.

Leptin alleviates the UA-induced metabolic dysfunction of adipocytes
To verify the hypothesis that leptin is a key regulator in the UA-induced FFA metabolic disorder of adipocytes, we tested the leptin levels in 3T3-L1 cells treated with UA (Fig. 4B) and epididymal white adipose tissue (eWAT) of HUA mice (Fig.   4C) and found that they were both significantly decreased, which was consistent with the mRNA-sequencing results (Fig. 4A). Then, we added leptin to the culture medium of UA-treated 3T3-L1 cells and found that leptin decreased the expression of the lipogenesis-related proteins CD36, ACC and FASN (Fig. S2A) while increasing the lipolysis regulators ATGL and HSL (Fig. S2B), which led to a reduction in lipid droplet volume (Fig. S2C) and triglyceride accumulation in adipocytes (Fig. S2D).
Moreover, leptin reversed the hyperplasia reduction of adipocytes caused by UA, manifested as an increased number of EdU-positive cells (Fig. S3A).
In addition, leptin restored the beige adipocyte formation inhibited by UA, as shown by the enhanced fluorescence intensity of UCP1 (Fig. S3B). Furthermore, the expression of the main regulators of beige adipocyte activity, including UCP1, PGC1α, and PRDM16, was also upregulated after leptin administration in 3T3-L1 cells (Fig.   S3C). These results suggest that leptin plays a key role in the UA-induced metabolic dysfunction of adipocytes.

Metformin protects against UA-induced cell hypertrophy, hyperplasia inhibition and suppression of beige-like characteristics via AMPK activation
To further dissect the molecular mechanism of UA-induced metabolic dysfunction in adipocytes, we tried to determine the downstream target of leptin. According to the bioinformatics analysis, we hypothesized that leptin interacted with AMPK in UA-treated adipocytes. Then, we tested the p-AMPK/AMPK ratio in 3T3-L1 cells and found that the expression ratio decreased along with the increase in UA concentration (Fig. 4D), and this can be reversed by the addition of leptin (Fig. 4E).
These results indicated that UA inhibited the phosphorylation of AMPK by reducing the expression of leptin.
To investigate the role of AMPK in UA-induced adipose tissue dysfunction, Downloaded from https://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20200580/884908/cs-2020-0580.pdf by guest on 23 June 2020 3T3-L1 cells were given two AMPK activators, namely, metformin and 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR). AMPK activation rebalanced FFA metabolism in adipocytes by decreasing the expression of the lipogenesis-related proteins CD36, ACC and FASN (Fig. 4F) and increasing the lipolysis regulators ATGL and HSL (Fig. 4G), which protected 3T3-L1 cells from UA-induced hypertrophy and excessive TG accumulation, as reflected by the decreased volume of lipid droplets (Fig. 4H) and cellular TG content (Fig. 4I). AMPK activation also restored the hyperplasia of adipocytes (Fig. 5A) and beige adipocyte formation that was suppressed by UA (Fig. 5B, C). Taken together, these findings demonstrated that metformin protects adipocytes from UA-induced cell hypertrophy and hyperplasia inhibition and helps adipocytes restore beige-like characteristics by activating AMPK.

HUA mice
We also evaluated the effects of benzbromarone (a uricosuric agent) and metformin on improving adipose tissue function in mice. The results were consistent with those in vitro. Both benzbromarone and metformin significantly reduced the UA-stimulated intake of fatty acids and lipogenesis of adipocytes, as evidenced by the decreased expression of CD36, ACC and FASN compared to the HUA group (Fig. 6A), and reversed the inhibitory effect of UA on lipolysis with the increased expression of HSL and ATGL (Fig. 6B). Histologically, both benzbromarone and metformin alleviated the hypertrophic features of adipocytes, with a switch toward smaller adipocytes (Fig.   6C, 6D). Moreover, they decreased triglyceride accumulation in the eWAT of HUA mice (Fig. 6E) and increased the number of adipocytes by facilitating hyperplasia (Fig.   6F). Additionally, UA-induced macrophage aggregation around eWAT was alleviated by benzbromarone and metformin, as shown by F4/80 (macrophage) antibody staining (Fig. 6G).
In addition, UCP1 staining of eWAT showed that the expression of UCP1 in the HUA group was significantly lower compared with the control group, while benzbromarone and metformin increased the UCP1 expression in eWAT (Fig. 7A). Consistently, the main proteins regulating adipose browning, namely, UCP1, PGC1α and PRDM16, were also significantly upregulated in mice from the benzbromarone and metformin treatment groups (Fig. 7B).
In serum, both benzbromarone and metformin treatment decreased the concentrations of FFAs and TG (Fig. 7D, E), enhanced glucose clearance and improved insulin sensitivity (Fig. 7F, G) in HUA mice. Moreover, the level of UA also decreased after metformin treatment (Fig. 7C). Overall, we confirmed that metformin preserved the function of adipose tissue by decreasing adipocyte hypotrophy and increasing adipose beiging and consequently stimulated its favourable FFA metabolic flexibility. Therefore, metformin attenuated HUA-induced serum FFA elevation and insulin resistance.

Discussion
In this study, we demonstrated that UA directly induced adipocyte hypertrophy and inhibited the hyperplasia of pre-adipocytes. Moreover, UA decreased the formation of beige adipocytes. The main mechanism was through leptin-AMPK pathway inhibition. However, the exact mechanism by which UA elevates serum FFAs remains unclear. A high-fat and high-fructose diet-induced HUA model is good for simulating the growth and development process of HUA disease, as it has been reported that obese individuals always have a higher UA level [16]. WAT also contains beige adipocytes that are transdifferentiated from white adipocytes under external stimulation and possess a brown-like feature. The accumulation of beige adipocytes in WAT is called "browning" [41]. It has been found that adipose browning could absorb circulating exogenous FFAs and glucose as fuel to generate heat, helping to improve metabolic health [42,43]. In contrast, losing active beige adipocytes or decreasing the browning capacity of adipose tissue results in a reduction in energy expenditure and contributes to progressive metabolic decline [44,45]. In this study, we found that a high level of UA could suppress beige adipocyte formation by inhibiting the expression of UCP1, PGC1a and PRDM16. Here, we wanted to study whether high UA level could influence adipocyte metabolism through upregulating the expression of URAT1 protein. Our results showed that the expression of URAT1 in adipocytes did not have a strong correlation with UA dosage, which implies that URAT1 is not a key target for UA-induced adipocyte metabolic disorders.
To further determine the impact of UA on adipocyte's morphology and function, mRNA-sequencing analysis was performed, and we found that leptin mRNA decreased sharply in 3T3-L1 cells treated with UA and was closely related to FFA metabolism according to mRNA microarray and KEGG enrichment pathway analysis.
In 3T3-L1 cells, after leptin administration, UA-induced FFA metabolic disorders were prevented, which means that leptin is the functional target of UA.
As leptin therapy is not ideal for patients with hyperleptinemia who show poor responses to exogenous leptin [49,50], we tried to identify the downstream target of leptin, which could be an alternative therapeutic strategy to eliminate metabolic disorders induced by HUA. KEGG enrichment pathway and bioinformatics analysis indicated that leptin was associated with AMPK and its related signalling pathway.
Other studies also showed that leptin could directly stimulate AMPK to reduce ACC activity, which decreases FFA synthesis [31] and increases the oxidation of FFAs [51].
Targeting the AMPK pathway could bypass leptin resistance and therefore represents an potential treatment method for metabolic disease [52]. In our study, we found that as the concentration of UA increased, both leptin and AMPK phosphorylation decreased, and leptin administration ameliorated UA-induced AMPK phosphorylation inhibition, which suggested that leptin-AMPK pathway inhibition existed in UA-affected adipose tissue. Moreover, AMPK activation by metformin also reversed Downloaded from https://portlandpress.com/clinsci/article-pdf/doi/10.1042/CS20200580/884908/cs-2020-0580.pdf by guest on 23 June 2020 UA-induced adipocyte hypertrophy and inhibited beige adipocyte formation both in vitro and in vivo. After metformin treatment, the levels of serum FFAs and TG were all reduced, and insulin sensitivity and impaired glucose metabolism were also improved. All these results proved the concept that the leptin-AMPK pathway plays a key role in UA-induced adipose tissue dysfunction.
Metformin has been used widely in the treating type 2 diabetes. In addition to its hypoglycaemic effect, the drug can also improve other metabolic processes. A study showed that metformin could suppress abnormal extracellular matrix remodelling by repressing TGF-β1-induced fibrogenesis in adipose tissue and ameliorate insulin resistance in obesity via AMPK activation [53]. In addition, Miguel A et al. suggested that metformin-induced AMPK activation can also enhance fat oxidation and reduce lipogenesis in sucrose-fed rat hepatocytes [20]. In line with their findings, we found that metformin could alleviate HUA-induced serum FFA elevation and insulin resistance by improving adipose tissue function. Our results suggest that metformin might be an alternative treatment strategy for HUA-related metabolic disorders, especially for patients with HUA and diabetes mellitus.

Conclusion
In summary, our study elucidated the direct effect of UA on adipose tissue and proved that UA damaged the metabolic and beiging function of adipocytes by inhibiting the leptin-AMPK pathway (Fig. 7H). These changes resulted in the dysfunction of adipose metabolism, which led to elevation of serum FFAs and TG, as well as insulin resistance. Our results suggested that metformin could be a novel therapeutic agent for HUA-related glycolipid metabolic disorder.

Clinical perspectives
 Hyperuricaemia (HUA) significantly increases the risk of metabolic syndrome and is strongly associated with the increased prevalence of high serum free fatty acids (FFAs) and insulin resistance. However, the detailed mechanism remains

Competing Interests
The authors declare that there are no competing interests associated with the manuscript.