MicroRNA-22 Is a Key Regulator of Lipid and Metabolic Homeostasis

Obesity is a growing public health problem associated with increased risk of type 2 diabetes, cardiovascular disease, nonalcoholic fatty liver disease (NAFLD) and cancer. Here, we identify microRNA-22 (miR-22) as an essential rheostat involved in the control of lipid and energy homeostasis as well as the onset and maintenance of obesity. We demonstrate through knockout and transgenic mouse models that miR-22 loss-of-function protects against obesity and hepatic steatosis, while its overexpression promotes both phenotypes even when mice are fed a regular chow diet. Mechanistically, we show that miR-22 controls multiple pathways related to lipid biogenesis and differentiation. Importantly, genetic ablation of miR-22 favors metabolic rewiring towards higher energy expenditure and browning of white adipose tissue, suggesting that modulation of miR-22 could represent a viable therapeutic strategy for treatment of obesity and other metabolic disorders.


Introduction
Obesity is considered by the WHO as the primary preventable cause of death and the fifth leading cause of death in western countries. Nearly 40% of adult Americans and Europeans [1] were described as overweight in 2017, and about 13% of the entire world's adult population was considered obese, numbers that have nearly tripled since 1975 [2]. The epidemic of childhood obesity [3] is particularly worrisome, since in 2016 approximately 42 million children under the age of 5 were considered overweight and over 340 million children and adolescents under the age of 20 were either overweight or obese [2].
Additionally, morbid and often fatal obesity is a central manifestation of genetic disorders such as the Prader-Willi Syndrome [4]. Fat accumulation depends on the balance between anabolic processes such as adipogenesis and catabolic processes such as thermogenesis [5]. It is widely acknowledged that there are two general types of adipose tissue in mammals: white adipose tissue (WAT) and brown adipose tissue (BAT). WAT is important for energy storage and secretion of adipokines that support metabolic homeostasis, while BAT is specialized for energy expenditure and thermogenesis [6]. The balance of these processes is critical for maintaining normal adiposity and regulating lipid homeostasis.
MicroRNAs (miRNAs) are short endogenous non-coding RNAs, typically~22 nucleotides in length, that function as post-transcriptional regulators of gene expression by repressing protein translation and promoting mRNA cleavage [7]. Previous reports provided strong evidence that miR-22 acts as an oncogene in breast cancer and leukemia by promoting metastasis and progression of myelodysplastic syndrome, respectively [8,9]. Metastatic spread is often associated with changes and perturbation in cellular metabolism [10,11], and the metabolic status and activation of lipogenic [12] programs have a strong impact on increasing the metastatic potential of certain tumors [13]. Interestingly, miR-22 is predicted to target several genes in different metabolic pathways, such as 'fatty acid metabolism', 'fatty acid biosynthesis' and 'the central carbon metabolism pathway' (Figure 1a), thus implicating miR-22 in the control of metabolic homeostasis. These findings are consistent with the observed metabolic advantages upon pharmacologic inhibition of miR-22 in a mouse model of diet-induced obesity [14,15]. In the present study, we deployed miR-22 knockout and transgenic mice and provide genetic evidence for an essential role of miR-22 in lipid accumulation, steatotic liver disease (STD) and obesity.

Results
To gain further insights into the role of miR-22 in metabolism, we generated an miR-22 conditional transgenic mouse model in which the miR-22 transgene (Tg) expression was controlled by Cre recombinase. Surprisingly, mice carrying the miR-22 Tg allele rapidly gained weight compared to wildtype (WT) littermates (Supplementary Figure S1b), likely due to the fact that the C2 locus into which the miR-22 cassette was cloned showed signs of leakiness [16][17][18]. The miR-22 transgenic mice displayed increased miR-22 levels in several tissues, such as liver, mammary fat pads and visceral fat (Supplementary Figure S1c). To determine the tissue in which miR-22 overexpression was driving the obese phenotype, we compared different Cre-mediated, tissue-specific miR-22 transgenic mouse lines and observed that mice with high levels of miR-22 in WAT and the liver rapidly gained weight and became obese (Supplementary Figure S1b Figure S1h). UCP-1 is a mitochondrial marker that is usually upregulated in brown adipose tissue (BAT). Thus, the observation that WAT in miR-22 −/− mice stains positive for a BAT marker suggests that miR-22 deficiency induces brownization of WAT, which could at least in part explain why miR-22 −/− KO mice do not gain weight when fed a HFD. Of note, the fat mass was markedly increased in WT mice after 8 weeks on HFD, but showed no significant change in HFD-fed miR-22 KO mice (Supplementary Figure S1i), whereas genetic ablation of miR-22 did not affect the body fat upon feeding the KO mice normal chow (NC). Furthermore, the percentage of lean mass remained stable in miR-22 KO mice (Supplementary Figure S1j).
Next, we asked whether the lack of weight gain in miR-22 −/− mice was due to changes in food intake or energy expenditure. At baseline, WT and miR-22 −/− mice showed no differences in oxygen consumption (VO 2 ), food intake, respiratory exchange ratio or locomotor activity (Figure 2a,c and Supplementary Figure S2a). However, after 8 weeks on HFD, the miR-22 /− mice had a significantly higher VO 2 consumption compared to WT mice ( Figure 2b). This effect was independent of the differences in body weight, as shown by an analysis of covariance (ANCOVA) [19] (mass effect: p = 0.0166; group effect: p < 0.001) (Figure 2d). During the entire length of the study, we observed no changes in food intake between the WT and miR-22 −/− mice (Supplementary Figure S2b), indicating that the effect of miR-22 loss of function on body weight is not related to a decrease in appetite. Data collected from metabolic cages showed no difference in respiratory exchange ratio or locomotor activity (Supplementary Figure S2b) between the two mouse cohorts; however, thermal camera analysis revealed that the body temperature was on average 3 degrees higher in miR-22 −/− mice compared to WT animals ( Figure 2e and Supplementary Figure S2c). The thermal signal was localized in the intra-scapular area, where brown fat accumulates, and was more pronounced in miR-22 −/− mice compared to WT littermates, suggesting that loss of miR-22 function leads to metabolic reprogramming when the miR-22 KO mice are challenged with an HFD. In addition, embryonic fibroblasts (MEFs) isolated from miR-22 −/− mice were much less prone to adipocyte differentiation compared to WT cells (Figure 2f and Supplementary Figure S2d). Taken together, these findings imply that miR-22 is involved in fat accumulation and that its overexpression leads to an obese phenotype. By contrast, fat accumulation is impaired in miR-22 KO mice fed a HFD, likely due to metabolic rewiring resulting in increased energy expenditure and suppression of weight gain.
To investigate if miR-22-3p is associated with human obesity, we analyzed RNA sequencing (RNA-Seq) data of 86 adipose tissue needle biopsies from patients from the METSIM study with BMI > 30 (n = 30) or BMI < 25 (n = 56). In this analysis, miR-22-3p was one of the most highly expressed miRNAs, comprising~4.5% of all miRNA reads (Figure 2h,i) [20]. Comparison of obese and normal-weight individuals showed that miR-22-3p was significantly overexpressed in obese human adipose tissue (p-value: 0.0078; q-value: 0.0373) (Figure 2j) [20]. Its expression also correlated positively with different anthropometric and metabolic characteristics, such as waist-hip ratio, body-mass index, serum triglycerides, serum ApoB, HOMAIR, serum alanine aminotransferase and serum C-reactive protein (Supplementary Table S1). Furthermore, miR-22-3p levels correlated negatively with muscle mass, serum adiponectin and serum HDL cholesterol (Supplementary  Table S1), and several metabolic pathways (Supplementary Figure S2e) were predicted to be under the control of miR-22, indicating that the role of miR-22 in metabolic homeostasis is conserved in mice and humans (Figure 2g). Upon analyzing a cohort of 122 patients with different grades of liver damage, ranging from steatosis to grade 3 fibrosis (F3), we observed that miR-22 levels positively correlate with human disease progression. As shown in Figure 2k, the patients with liver fibrosis have higher levels of miR-22 compared with the sub-group with hepatic steatosis. Moreover, when the fibrosis advanced from grade 1 (F1) to grade 3, the amount of miR-22 also increased with the severity of the disease, providing further evidence of the crucial role of this microRNA in controlling biological aspects connected with liver disease progression and specifically fibrosis in NASH patients.

Discussion
In this study, we provide genetic evidence for miR-22 as a key player in the control of lipid and energy homeostasis as well as the onset and maintenance of obesity using miR-22 knockout and transgenic mice. These findings are consistent with the fact that miR-22 regulates direct mRNA targets implicated in several key metabolic pathways [14,15]. Furthermore, we show that genetic ablation of miR-22 enhances beiging of WAT, thereby increasing energy expenditure and impairing lipid biosynthesis, as showed in the schematic model presented in Figure 2l Finally, we report that miR-22 levels are upregulated in obese human subjects (Figure 2j). Taken together, these data suggest that elevated levels of miR-22 could represent a physiological mechanism favoring energy and lipid accumulation. While this mechanism may be beneficial under fasting or starvation, overexpression of miR-22 could become detrimental under consumption of a high-fat diet and high consumption of food resulting in weight gain and obesity. Our findings are also consistent with the pro-metastatic role of miR-22 [9] and with the fact that a lipogenic switch favors cancer progression and metastasis [13,21]. Importantly, our data suggest that inhibition of miR-22 function could represent a new therapeutic strategy for obesity and liver steatosis.
In a separate study, we developed an antisense oligonucleotide (antimiR-22) for targeting of miR-22 (Panella et al., in preparation). Pharmacologic inhibition of miR-22 deploying the antimiR-22 oligonucleotide validated our findings presented here using miR-22 genetic mouse models. Furthermore, additional pharmacology studies in mice enabled a more detailed dissection of the molecular mechanisms underlying the role of miR-22 in modulating lipid and energy homeostasis, which is a limitation of the present study.
The past decade has witnessed significant advances in the development of new therapeutic approaches for the treatment of obesity, ranging from bariatric surgery to pharmacological therapies designed to decrease appetite, increase satiety, and boost metabolism [22]. However, while effective, surgical treatments are associated with serious side effects [23][24][25]. Approved pharmacological approaches, which are mainly based on GLP1-receptor agonists, have provided robust effects on weight loss in obese subjects; however, there are concerns regarding maintenance of body weight due to the reported adverse effects associated with the long-term use of GLP-1 agonists [26,27]. Furthermore, therapies that reduce food consumption and appetite affect both fat and lean mass, which can cause a reduction in muscle mass with an associated series of unwanted effects. Thus, there is still a high unmet medical need for new effective and safe therapies for treatment of obesity. Here, we present genetic evidence for miR-22 as an important rheostat in controlling lipid homeostasis and the onset and maintenance of obesity. Importantly, genetic ablation of miR-22 protects mice from obesity under HFD conditions without affecting food consumption or lean mass. Taken together, our data suggest that pharmacological inhibition of miR-22 function could represent a viable therapeutic strategy for obesity and liver steatosis, with a mechanism that would be complementary to treatment with GLP1 receptor agonist drugs [22,28,29], enabling the development of novel combination therapies for more effective treatment of obesity.

Data Analysis
miRNA expression data from human abdominal subcutaneous fat tissue needle biopsies were obtained from the METSIM study [20]. A total of 185/200 samples passed QC for having comparable total reads on mature miRNAs (<10 6 and >6 × 10 6 ). Comparison of miRNA expression between BMI > 30 (n = 30) and BMI < 25 (n = 56) individuals was performed using DESeq2 [30] with the first two expression principal components (PCs), the second PC as a factor, sample sequencing depth and the subject's age as covariates. Storey's q-value [31] was calculated per gene to control the type I error rate within the experiment. miRNAs with a q-value < 0.05 were deemed as differentially expressed between the two groups. Correlation analysis was performed by calculating Spearman's rho on librarysize normalized data using limma [32]. The same covariates as in differential expression analysis were incorporated into the model, with subject group (>30 BMI and <25 BMI) as a treatment factor to be preserved. Storey's q-value was calculated to control for the multiple tests conducted. Correlations with a q-value < 0.05 were deemed as statistically significant. As in the METSIM publication, Log10-and Log2-transformed values were included in the correlation analysis. In the relevant table, the results correspond to the analysis of log10 values of BMI, total triglycerides, creatinine clearance rate, serum alanine aminotransferase, serum ApoB, HDL cholesterol, HOMAIR, serum CRP, OGGT fasting plasma insulin, Matsuda Insulin Sensitivity Index, plasma adiponectin, serum apoA1 and insulin AUC. KEGG miRNA enrichment was performed using experimentally supported miRNA interactions from TarBase 8 database [33]. Fisher's exact test was performed against Central Carbon Metabolism in Cancer pathway (hsa05230) genes.

Mouse Models
All experiments were carried out in miR-22 −/− mice (a kind gift from Da-Zhi Wang) or miR-22 Tg +/+ /Albumin-Cre mice, obtained by crossing an miR-22 Tg mouse as described in [8,9,34] with the commercially available Alb-Cre mouse model (The Jackson Laboratory, mouse strain B6.Cg-Speer6-ps1Tg(Alb-Cre)21Mgn/J, Cat. #003574). The same strategy was use to generate miR+22 +/+ /MMTV-Cre. Mice were kept under a 12 h light: 12 h dark cycle and an ambient temperature of 22 ± 2 • C. The mice were given ad libitum access to normal chow or a standard high-fat diet with 60% total caloric intake from lard (TD.06414) and drinking water for the entire duration of the experiment. Mouse weight was monitored once a week using a regular scale for the duration of the experiment.

Indirect Calorimetry, Body Composition and Thermal Imaging
All experiments were carried out in 10-week-old female miR-22 −/− mice and agematched WT siblings. Mice were kept under a 12 h light: 12 h dark cycle and an ambient temperature of 22 ± 2 • C. The mice were given ad libitum access to a standard high-fat diet and drinking water for 8 weeks. The high-fat diet had a composition of 60.3% fat, 18.4% protein and 21.3% carbohydrate (9% sucrose) (Envigo, TD.06414) Body weights were assessed once a week in the morning, between 8:00 and 10:00. Body composition, including lean and fat mass, was assessed using an Echo MRI 3-in-1 quantitative nuclear magnetic resonance (qNMR) system (Echo Medical Systems, Houston, TX, USA). This test is performed in conscious mice that are immobilized for one minute. This test was performed at Week 0 (Baseline, pre-diet) and Week 8 (Test, after 8 weeks on the diet).
Metabolic parameters were measured by indirect calorimetry using the Comprehensive Lab Animal Monitoring System (CLAMS; Columbus Instruments, Columbus, OH, USA). This included VO 2 consumption (a measure of energy expenditure), respiratory exchange ratio (R.E.R), energy intake and ambulatory activity. Mice were individually housed with ad libitum access to food and water. Analysis was performed at 24 • C under a 12:12 h light-dark cycle (light period 06:00-18:00). Mice were acclimated in the metabolic chambers for 48 h before collecting 4 days of measurements used for data analysis. This procedure was performed at Week 0 (Baseline, pre-diet) and Week 8 (Test, after 7 weeks on the diet), following the procedure previously described [35].
The infrared camera T420sc (emissivity of 0.98, FLiR System) was used as previously reported [36] to assess the amount of BAT in mice.
All procedures were in accordance with the National Institutes of Health Guidelines for the Care and Use of Animals and approved by the Institutional Animal Care and Use Committee at Beth Israel Deaconess Medical Center (Boston, MA, USA).

Tissue Harvesting and Manipulation
All tissues were harvested fresh in the first 5 min after sacrifice, washed in PBS and divided in three parts on ice. One part was used for paraffin embedding and subsequently cut and stained for the desired markers, and two portions were frozen by immersion in liquid nitrogen and preserved for future analysis.

H&E and HIC Staining
Tissues were fixed in 4% paraformaldehyde overnight, paraffin-embedded and then sectioned at 5 µm. After deparaffinization and rehydration, antigen retrieval was performed in a pressure cooker with sodium citrate buffer at 95 • C for 25 min. Sections were incubated in a 0.3% H 2 O 2 solution in 1 × PBS and then in a 10% serum solution in 1 × PBS for 30 min. Each solution was used to block endogenous peroxidase and background from the secondary antibody. The sections were stained with the primary antibodies and incubated in a biotinylated secondary antibody in 1 × PBS (1:500-1:1000) at room temperature for 30 min. The Vectastain ABC Elite kit was used to enhance specific staining, and the staining was visualized using a 3 -diaminobenzidine (DAB) substrate. Stained sections were counterstained using hematoxylin and dehydrated before they were sealed with a coverslip with Richard-Allan Scientific ® Cytoseal™ XYL Mounting Medium and then analyzed on a Nikon Eclipse 50i microscope or Olympus BX41 equipped with an Olympus Q Color 5 camera.

Isolation and Manipulation of Primary Mouse Cells
Mouse mesenchymal cells were isolated and cultured as previously described [37].

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijms241612870/s1. Author Contributions: R.P. conceived the project with P.P.P., R.P. designed and performed most experiments, interpreted the results and co-wrote the manuscript. A.P. ran miR-22 target prediction and analyzed RNA-seq data. B.N.D. performed indirect calorimetric studies. I.S.V. analyzed human genetic data and ran predictions on target genes. S.F. and F.A. co-executed with R.P. all the analyses on exosomes. C.A.C., P.K.N. and E.M.L. analyzed samples previously generated. D.-Z.W. provided key reagents. A.W., A.M.N. and E.M.-F. contributed to experimental designs. S.K. contributed to experimental designs, interpreted the results and co-wrote the manuscript. P.P.P. supervised experimental designs, interpreted results and provided input into the manuscript. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported in part by NIH grant R35 CA197529 to P.P.P. and Novo Nordisk Foundation Challenge 2018 Grant to R.P., S.K. and A.N.
Institutional Review Board Statement: The animal study protocol was approved by the Institutional Review Board (or Ethics Committee) of Beth Israel Deaconess Medical Center (protocol code 2175 approval date 2017).

Data Availability Statement:
The data presented in this study are available on request from the last author. The data are not publicly available due to confidentiality related to patent application on going.