The TGR5 gene is expressed in human subcutaneous adipose tissue and is associated with obesity, weight loss and resting metabolic rate

Highlights • Human adipose tissue (AT) expresses the bile acid receptor TGR5.• Human AT TGR5 expression is linked to obesity.• Resting metabolic rate and AT TGR5 expression is positively correlated.• TGR5 expression is not higher in brown compared to white human AT.


Introduction
Bile acids are synthesized in the liver and play a central role in dietary lipid emulsification in the intestine . The majority of the secreted bile acids is taken up by the distal parts of the small intestine and return to the liver and the gall bladder via the enterohepat ic circulation. Beyond their well established function in dietary lipid emulsification, bile acids are signaling molecules with metaboli c effects mediated by specific receptors. The farnesoid X receptor (FXR) is the most studied receptor for bile acids. FXR is a nuclear receptor that regulates bile acid synthesis as well as lipid and glucose metabolism [1]. The G protein-cou pled receptor TGR5 (also denoted G protein-coupled bile acid receptor 1 or GBPAR1) has also been found to be a functional bile acid receptor [2]. In 2006, Watanabe et al. showed that bile acid administration augmented energy expenditure in mice [3], and that these effects were mediated via the TGR5 and its effects on the thyroid hormone activating enzyme iodothyronine deiodinase type II (DIO2) in brown adipose tissue (BAT). In addition, they showed that bile acid treatment of human skeletal myocytes increased both DIO2 activity and oxygen consumptio n, indicating that similar mechanisms are relevant also in humans. More recently, TGR5 activation has been shown to induce glucagon-like peptide-1 (GLP-1) release in mice, indicating a direct role for TGR5 in glucose homeostasi s [4].
Despite these advances in understand ing of TGR5, much less in known about TGR5 function in humans and its expression in other metaboli cally relevant organs such as adipose tissue. The aim of this study was therefore to investiga te the potential role of adipose tissue TGR5 in human metabolism.

Materials and methods
The regional ethics committ ee in Gothenbur g approved these studies.

Sib Pair study
The swedish obese subjects (SOS) Sib Pair study consists of 154 nuclear families with BMI discordant sibling pairs (BMI difference P10 kg/m 2 ), resulting in a study population consisting of 732 subjects [5]. The subjects were extensively phenotyped [5], including anthropometr ic measure ments and determination of resting metabolic rate (RMR) in a ventilated hood [6]. Subcutaneo  tissue needle biopsies were obtained and used for gene expression analysis. For the current study, complete data from 353 siblings and 86 parents were available for the analysis.

Very low calorie diet (VLCD) study
The very low calorie diet (VLCD) study was performed to investigate gene expression changes in adipose tissue of obese subjects during weight loss induced by caloric restrictio n. Twenty-eight obese subjects (20 women and 8 men, age 39.7 ± 12.7 years, BMI 36.3 ± 3.7 kg/m 2 ) were treated with VLCD (450 kcal/day) for 12 weeks [7,8]. Subcutaneo us adipose tissue needle biopsies were obtained at the start of the VLCD treatment (day 0) and three times during the VLCD treatment (weeks 2, 6, and 12). After 12 weeks of VLCD treatment, the mean weight loss was 19%.

Perithyroid and perirenal adipose tissue studies
Perthyroid adipose tissue surgical samples were obtained from 24 patients undergoing surgery in the thyroid region for malignancies or endocrine disorders. Clinical characterist ics of the patients have been described previously [9]. Perithyroid adipose tissue biopsies containing BAT was identified by expression analysis of uncoupling protein 1 (UCP1) and histological analysis. Samples with high UCP1 expression was classified as BAT positive (BAT+, n = 9). Paired subcutan eous white adipose tissue biopsies from the surgical incision area were also obtained from the same nine patients (2 men and 7 women, age 47 ± 21 years, BMI 23 ± 2 kg/m 2 ).
Biopsies of perirenal adipose tissue were obtained from 55 healthy kidney donors. The perirenal adipose tissue samples were screened for UCP1 expression and samples with high UCP1 expression (samples from 4 men and 6 women) were classified as BAT positive (BAT+) samples. These samples constituted the BAT+ group. A control group (n = 10) with low perirenal adipose tissue UCP1 expression (classified as BAT À samples) was matched to the BAT+ group based on sex, age and BMI. The subjects in the BAT+ group had an average age of 42 ± 14 years and a BMI of 26 ± 2. The subjects in the BAT À group had an average age of 44 ± 9 years and a BMI of 26 ± 3.

Gene expression analysis
Total RNA was isolated from adipose tissue using the RNeasy lipid tissue midi kit (Qiagen, Chatsworth , CA) or the phenol-chloroform extraction method of Chomczyns ki and Sacchi [10].
Gene expression in adipose tissue from the Sib Pair study and in the perithyro id adipose tissue was analyzed study using Human Genome U133 plus 2.0 arrays (Affymetrix, Santa Clara, CA). Gene expression in the perirenal adipose tissue samples were analyzed by (Affymetrix Gene 1.0 ST arrays at the Uppsala Array Platform, Uppsala, Sweden). All arrays were analyzed accordin g to the manufacturer's instructions. Expression data were analyzed using the RMA algorithm (Affymetrix). TGR5 expression was assessed using probe sets 1552501_a_at and 8048249 for Human Genome U133 plus 2.0 and Expression assay Gene 1.0 ST arrays, respectively.
Adipose tissue total RNA from the VLCD study was reversed transcribed using the High Capacity cDNA RT kit (Life Technologies, Paisley, UK) according to the manufacturer's protocol. Reagents for real-time PCR analysis of TGR5 (Hs00544894_m1) and low-density lipoprotein (LDL) receptor-r elated protein 10 (LPR10) (Hs00204094_m1) were purchase d from Life Technologies and used according to the manufactur er's instructions. cDNA was used for real-time PCR in the Applied Biosystems PRISM 7900HT Sequence Detection System (Life Technologie s) using default cycle parameters. A standard curve was plotted for each primer-p robe set with a serial dilution of cDNA synthesized from pooled RNA.
All samples and standards were analyzed in triplicate and LRP10 gene expression was used as a reference gene [11].

Gene ontology (GO) enrichment analysis
GO enrichment analysis was performed using the DAVID web-accessible program [12,13] (http://davi d.abcc.ncifcrf.gov ). Transcrip ts significantly correlated with TGR5 expression were identified in the Sib Pair study using Spearman correlation. Accounting for multiple testing, a threshold of p < 9.1 Â 10 À7 was used as a cut off for declaring statistical significance (Bonferroni corr ecti on ). Th is yi eld ed a tot al of 17 55 ne gat iv e and 17 67 pos it iv it y correlate d transcripts which were included separately in the GO enri chme nt analy si s. The an aly si s was li mit ed to the GOT ERM _BP_ FAT , GEOTERM_CC_F AT and GEOTERM_MF _FAT options and Human Genome U133 Plus 2.0 was used as a background.

Statistica l analysis
Statistica l analyses were performed using the SAS software package (v. 9.1.3, SAS Institute Inc., Cary, NC) or and PASW Statistics (Chicago, IL, USA). Quantitati ve data were transformed towards normal distribution using Box-Cox power transformat ions. Outliers beyond three standard deviations from the trait mean were excluded. Correlations between clinical traits and TGR5 expression were analyzed using the MIXED procedure in SAS. In the linear mixed models, we used a ''sandwich estimator'' of the covariance matrix to adjust for non-independe nce among family members. This asymptotical ly yields the same parameter as ordinary least squares or regression methods while standard errors and, consequently, hypothesis tests are adjusted for the family relatedne ss. Comparis ons in gene expression between BAT and WAT were performed using Student's t-test (paired or unpaired as appropriate). Gene expression during weight loss was analyzed using a oneway repeated measures analysis of variance (ANOVA) with Greenhouse-Geisser correction. Comparisons between time points were assessed by Bonferroni corrected post-hoc tests.

TGR5 adipose tissue expression
Initially, we investiga ted the link between TGR5 adipose tissue expression and obesity. Among the Sib Pair families, TGR5 expression was positively correlated with BMI (r = 0.40, p < 0.0001), independent of age, sex, family membership and generation. In line with this finding, obese subjects treated with VLCD displayed a drastic reduction of adipose tissue TGR5 expression determined by a repeated measure s ANOVA (F(1.4, 38) = 38, p < 0.0005; Fig. 1). Compared to day 0, adipose tissue TGR5 expression was approximately half as low during the entire dieting period and post-hoc tests using the Bonferroni correction showed that TGR5 levels were reduced compare d with day 0 at all subsequent time points measure d (weeks 2, 6 and 12, p < 0.0005 for each test; Fig. 1).

Gene ontology enrichment analysis
To gain insights into functions in adipose tissue that may be related to TGR5 expression, Gene ontology enrichment analysis was performed using transcripts positivel y or negatively correlated to TGR5 expression in subcutan eous adipose tissue in the Sib Pair offspring study. Transcripts negative correlated to TGR5 expression displayed highly significant (adjusted p-values down to 3 Â 10 À20 ) enrichment of several GO terms relating to mitochondria and ribosome (Table 1). Enrichme nt analysis of transcrip ts positively correlate d to TGR5 expression resulted in identification of GO terms primarily related to endoplasmatic reticulum/G olgi and cytoskeleton function but with less striking p-values (Table 1).

TGR5 adipose tissue expression and energy expenditure
Since mitochondr ia play an important role in energy expenditure and nutrient combustion, we next investigated the relation between adipose tissue TGR5 expression and resting metabolic rate (RMR) in the Sib Pair study. Unexpectedly, given the negative correlation to mitochond rial genes, TGR5 expression was positively correlate d with RMR (r = 0.25, p < 0.0001). In a model including age, polynomials of age, sex, lean body mass, generation and family membership , the statistical significance of this correlation persisted.
Brown adipocytes have, in comparison to white adipocytes, substanti ally more mitochondr ia. In mice TGR5 has been shown to be expressed in brown adipose tissue (BAT) and treatment of brown adipocytes with bile acids has been shown to increase oxygen expenditur e via TGR5-mediarted activation of DIO2 in BAT [3]. The expression of TGR5 was therefore investigated in two sets of human adipose tissue samples (the perithyriod and the perirenal adipose tissue studies) containing islets of BAT (BAT+) and compared to expression in white adipose tissue. However , no significant differenc e in TGR5 expression was observed between the BAT containing samples and white adipose tissue ( Fig. 2A and B).

Discussion
In this study, we have characterized the gene expression of TGR5 in human adipose tissue. Our main findings are that TGR5 adipose tissue expression is reduced during weight loss, and that it is positivel y correlated with obesity as well as with resting metabolic rate.
Our results clearly show that TGR5 expression in adipose tissue is regulated in a manner that supports a functional metabolic role for TGR5 in adipose tissue. However, our results are not easily reconciled with the results from the studies of mice treated with bile acids [3]. Whereas mice treated with bile acids lose weight, high human adipose tissue TGR5 expression is associate d with obesity. In addition, the mice studies also show that the mitochondr ia rich BAT is a major target organ for the effects of bile acids whereas we show a negative correlation of TGR5 to the expression of mitochondrial genes in white adipose tissue. In the light of this, it is not unthinkable that the up regulatio n of TGR5 in human adipose tissue can be viewed as compens atory mechanism to prevent adipose tissue growth. However, one has to consider that regulation studies in one single organ may not capture the more complex regulation of the entire body and that receptor gene or protein expression cannot directly be translated to the amount of receptor signaling.
Expression analysis in human brown fat depots is challenging because, in contrast to mouse, human BAT represents a much smaller fraction of adipose tissue and is usually not located in distinct depots but rather as islets within the white adipose tissue [9,14]. Only the expression of a very limited number of genes has been investigated in human BAT [9,15,16]. Our analysis of TGR5 expression in human BAT-containi ng adipose tissue samples provide no evidence that TGR5 is expressed at higher levels in human BAT compared to WAT. This together with the limited amount of BAT in humans speaks against that the observed association between TGR5 adipose tissue expression and RMR is related to human BAT. A more likely scenario is that TGR5 expression in adipose tissue is co regulated with the TGR5 expression in other metaboli cally more active organs such as skeletal muscle or that it is related to adipose tissue thermogenesis not involving BAT. The correlation between adipose tissue TGR5 expression and RMR is well in line with the down-regul ation of TGR5 during the diet intervention, a situation where RMR is reduced [17]. Interestingl y, the drastic reduction of TGR5 expression seen already after 2 weeks preceded the maximal weight loss which occurred at week 12 [8]. This indicates that TGR5 expression in adipose tissue is regulated by energy intake or metabolic rate rather than the level of obesity. One study has shown that human plasma levels of bile acids are not associated with energy metaboli sm [18], indicating that if these systems play an important role in human energy expenditure future research should probably be directed at the receptor level. Obesity represents an independen t cardiovascu lar risk factor [19]. In addition to its metabolic role, TGR5 has recently been directly implicated in the developmen t of atherosclero sis. Activatio n of TGR5 in mice was associate d with attenuation of atherosclerosis by means of decreased pro-inflammatory cytokine production and reduced oxidized LDL uptake in macroph ages [20]. These findings are of particular interest to our study since adipose tissue is a major source of circulating pro-inflammatory cytokines. However , the relative contributi on of the adipose tissue TGR5 expression by adipocytes and macrophages is currently unknown.
In conclusion, we have shown that adipose tissue expression of human TGR5 is reduced during weight loss, and positivel y correlates to obesity as well as to resting metabolic rate, indicating a functional metaboli c role for TGR5 in adipose tissue. However, more studies are needed to specifically define this role.