Adipose tissue function and insulin sensitivity in syndromic obesity of Bardet-Biedl syndrome

Bardet–Biedl syndrome (BBS) is a rare autosomal recessive syndromic obesity of childhood onset among many other features. To date, the excess risk of metabolic complications of severe early-onset obesity in BBS remains controversial. In-depth investigation of adipose tissue structure and function with detailed metabolic phenotype has not been investigated yet. To investigate adipose tissue function in BBS. A prospective cross-sectional study. To determine if there are differences in insulin resistance, metabolic profile, adipose tissue function and gene expression in patients with BBS compared to BMI-matched polygenic obese controls. 9 adults with BBS and 10 controls were recruited from the national centre for BBS, Birmingham, UK. An in-depth study of adipose tissue structure and function along with insulin sensitivity was performed using hyperinsulinemic-euglycemic clamp studies, adipose tissue microdialysis, histology and RNA sequencing, and measurement of circulating adipokines and inflammatory biomarkers. Adipose tissue structure, gene expression and in vivo functional analysis between BBS and polygenic obesity cohorts were similar. Using hyperinsulinemic-euglycemic clamp and surrogate markers of insulin resistance, we found no significant differences in insulin sensitivity between BBS and obese controls. Furthermore, no significant changes were noted in an array of adipokines, cytokines, pro-inflammatory markers and adipose tissue RNA transcriptomic. Although childhood-onset extreme obesity is a feature of BBS, detailed studies of insulin sensitivity and adipose tissue structure and function are similar to common polygenic obesity. This study adds to the literature by suggesting that it is the quality and quantity of adiposity not the duration that drives the metabolic phenotype.


INTRODUCTION
Bardet-Biedl syndrome (BBS) is a rare autosomal recessive multisystem disorder. It is caused by a mutation in genes encoding for proteins in primary cilium/basal body complex, which is ubiquitously expressed and involved in cell-to-cell signalling. To date, 21 diseasecausing genes have been identified (BBS1-BBS21) [1] which account for 80% of the cases [2]. BBS1 (~23%) and BBS10 (~20%) are the commonest genotypes in Europe and North America [2]. In Europe and North America, BBS has a prevalence of 1:100,000 but it is more common in certain communities, for example Newfoundland (1:17,000) [3] and Kuwaiti Bedouins (1: 13,400) [2,4]. BBS is characterized by retinal degeneration, polydactyly, genital malformations, renal dysfunction, and severe obesity.
BBS is considered a genetic syndromic obesity characterised by childhood-onset weight gain. Despite normal birth weight, most individuals with BBS experience rapid weight gain in early childhood, with high rates of severe obesity sustained through adolescence [5]. Children with bi-allelic loss of function of genetic variants have significantly higher BMI z-scores compared to missense variants [5]. This trend usually continues throughout the lifetime of an individual. Although severe obesity is an important manifestation of BBS, the aetiology and the metabolic consequences remain unclear. Previous studies have reported that insulin resistance is an early feature of BBS and that childhood prevalence of diabetes is 2-6% [6][7][8][9][10], with adult prevalence as high as 32% (although data are variable) [11,12]. A case-control study involving 152 adults has shown the rate of metabolic syndrome was significantly higher in the BBS group (54.3%) compared with control subjects (26%, P < 0.001) [13]. Other studies, however, have given a mixed picture, with no significant difference in mean fasting glucose, glycosylated haemoglobin and fasting insulin when compared to age, sex and BMI-Z matched controls [14,15].
To date, the excess risk of metabolic complications of severe obesity in BBS remains controversial. It is unclear if adiposity has pathogenic implications beyond what may be proportionate for the degree of obesity. In addition, although obesity substantially increases the risk of insulin resistance and its attendant cardiometabolic complications in the general population, 10-24% of individuals with obesity are insulin sensitive [15]. Variation in body fat distribution, physical activity, and muscle mass partly explain this discrepancy, but the pathophysiological mechanisms linking obesity to insulin resistance remains a matter of intense research. Studying humans with defined genetic causes of obesity offers important insights.
In this study we aim to define the structure and function of adipose tissue, together with gene expression in patients with BBS compared to controls with polygenic obesity using an integrative physiological approach.

MATERIALS AND METHODS Subjects
Participants with BBS were recruited from the nationally commissioned highly specialist service for BBS at the University Hospitals Birmingham NHS Trust. Participants with common obesity were recruited through local advertising. The research project was conducted in accordance with the ethical standards of the Helsinki declaration and was approved by an NHS Research Ethics Committee (IRAS project ID: 204444, REC reference: 16/ WM/0400). Written informed consent was obtained from all participants.

Clinical and laboratory assessment
Weight was measured using the same digital weighing scale in all participants. Height was measured using a stadiometer, and BMI was calculated as weight in kg/height in metres 2 . Waist circumference was measured with the participant in an erect posture at a level just above the iliac crest. Blood samples were collected after a 12-hour fast, and plasma or serum were separated immediately and stored at −80C. All samples were analysed as described previously [16].

Transient elastography
Liver transient elastography was performed using fibroscan (Echosens, France). All fibroscans were performed using an M-(3.4 Hz frequency) or XL-probe (2.4 Hz frequency) in a dorsal decubitus position with the tip of the ultrasound probe overlying right lobe of the liver. Validity of readings was classified according to the manufacturer's recommendations [17].

Hyperinsulinemic-euglycemic clamp
Hyperinsulinemic-euglycemic clamping (HEC) was commenced at 09:00 h after an overnight fast. Glucose concentrations were measured in arterialized blood using a hot box. Participants were given a bolus (2 mg/kg) of D-Glucose (U-13 C6) (Cambridge isotope laboratories, UK) followed by a continuous infusion (0.02 mg/kg/min) for 4 h. During the first 2 h (basal phase), steady-state blood samples were taken after 90, 104, 120 min for measurement of insulin, free fatty acids (FFA), and stable isotopes. At 120 min, an insulin infusion (Actrapid; Novo Nordisk, Copenhagen, Denmark) was commenced at a rate of 40 mU/m 2 /min for 2 h. At 124 min, 20% dextrose solution with [ 13 C]-glucose (enriched to 4%) was infused. Further blood samples were taken after 210, 224, 240 min for measurement of insulin, FFA, and stable isotopes. Serum U-13 C glucose enrichment was determined by gas chromatography-mass spectrometry (GC-MS: GC, Agilent 6890 N; MS, Agilent 4973 N; Agilent Technologies).

Adipose tissue microdialysis
A microdialysis catheter (CMA 63, M Dialysis AB, Sweden) was inserted approximately 10 cm lateral to the umbilicus. The catheter was connected to a microdialysis pump containing an isotonic fluid, Perfusion Fluid T1 (M Dialysis AB, Sweden), which was infused at a fixed rate of 0.3 μl/min. A micro vial was connected to the exit end and samples were collected every 30 min throughout the clamp procedure. Glycerol concentrations were measured in micro-dialysate using an enzyme-kinetic analyser (CMA, Iscus Flex).

Adipose tissue biopsy and analysis
Subcutaneous adipose tissue (SAT) biopsies were collected from the periumbilical region in the abdomen. Samples were preserved in 4% formalin solution and subsequently embedded in paraffin. Fixed samples were sectioned in 4-micron thick cuts and stained with hematoxylin and eosin (H&E) for determination of adipocyte morphology. Picrosirius-red staining was used to assess fibrosis, and CD68 and CD163 immunostaining to assess macrophage infiltration.
Analysis of fibrosis was undertaken by histo-morphometry using Calopix (TRIBVN Healthcare, France) with content colour thresholds. Adipocyte images were obtained at 40× magnification and morphological analyses were performed using Adiposoft software (CIMA, University of Navarra). Adipocyte size and number were measured in 2 fields of 1000 × 1000 microns using previously validated optimised sequence of image analysis steps in Adiposoft as described by Galarraga et al. [18]. Immunohistochemical staining of M1 and M2 macrophages was undertaken using mouse anti-human CD68 and CD163 (Bio-Rad, UK), respectively. Slides were scanned at x20 magnification and CD68 and CD163 positive cells were counted in 6 areas measuring 400 × 400 microns randomly selected using Calopix software (Tribvn, France). Macrophage numbers were expressed per 100 adipocytes. The number of crown-like structures (CLS) was counted across all AT scanned at ×10 magnification and normalized by the total tissue area.

RNA extraction and sequencing
RNA was extracted from 40 to 100 mg adipose tissue using the Qiagen Tissue Lyser II and RNeasy Lipid Tissue Mini Kits (Qiagen) according to the manufacturer's protocol followed by DNAse treatment and cleaning using the Qiagen RNeasy RNA clean up protocol. RNA with RNA integrity number (RIN) > 7.0 was subjected to next generation sequencing (NGS). In brief, 100 ng total RNA was processed using the Illumina Truseq Stranded mRNA library preparation kit (Illumina, San Diego, CA, USA). Sequencing reads were aligned to hg19 human genome using STAR RNA-Seq aligner software [19]. Reads mapping to transcripts were counted by the same software. Differential expression analysis was performed using DESeq2 R Bioconductor package24 and gene set enrichment analysis was performed using GAGE R Bioconductor package26 with Gene Ontology, KEGG and MSigDB databases.

Statistical analysis
The data are presented as mean ± standard deviation (SD) when normally distributed or where relevant as median with interquartile range (IQR). Categorical variables are reported as numbers and percentages. Visual assessment of normality curves along with the Shapiro-Wilk test was used to test the normality of the distribution. Independent Sample t test (or where appropriate its non-parametric equivalent Mann-Whitney U test) was used to compare between independent groups, i.e. BBS to controls. The Fishers exact test was used to compare categorical variables. Spearman's correlation was calculated for non-parametric data. The significance level for this study was set at p < 0.05.
The required sample size per each group to detect a statistically significant difference between groups with a two-tailed student t test was calculated based on the sample size determination method. To detect a difference in insulin resistance based on an effect size of 60% [13], (through 60% decrease in glucose disposal following hyperinsulinaemic euglycaemic clamp) between the two groups at 80% power with Type I error probability of 0.05, 18 participants (9 in each group) were required [20]. All analyses were performed using SPSS statistical software version 24.

RESULTS
Baseline characteristics of study participants All participants with BBS had disease-causing pathogenic mutations and met clinical diagnostic criteria. As expected, the predominant mutation was in the BBS1 gene (Supplementary Table 1). Baseline demographic, anthropometric, clinical and metabolic characteristic data are summarised in Table 1. The patients' age ranged from 30 to 52 years, and most were males (56%). There was no statistical significant difference in weight (120 vs. 101, p = 0.06), BMI (41 vs. 34, p = 0.08), and waist circumference (124 vs 113, p = 0.3) between the BBS and obese controls, respectively. In line with syndromic feature, BBS participants had significantly reduced eGFR (80 vs 107 ml/min/m2, p < 0.01), and higher systolic blood pressure (132 vs 118 mmHg, p = 0.04) compared to controls, although this was still within the normal range. Lipid profiles and parameters for non-alcoholic fatty liver disease were similar in both groups. One BBS patient had a confirmed diagnosis of diabetes mellitus.    seen in endogenous (hepatic) glucose production (EGP) rates nor in the hepatic insulin sensitivity disposition index (p > 0.05 for both, Fig. 1B, D and E). Both BBS and control groups showed a statistically significant correlation between hyperinsulinemic-euglycemic clamp studies and surrogate markers of insulin sensitivity. In BBS, HE-clamp insulin sensitivity index showed positive correlation with HOMA-IR (r 2 = 0.88, p < 0.005), HOMA-beta (r 2 = 0.85, p < 0.005) and Insulin (r 2 = 0.93, p < 0.001), but negative correlation with QUICUK (r 2 = 0.88, p < 0.005). Similarly, there was a positive correlation between HEclamp insulin sensitivity index and HOMA-IR (r 2 = 0.75, p < 0.01), HOMA-beta (r 2 = 0.88, p < 0.001) and Insulin (r 2 = 0.84, p < 0.001) while a negative correlation with QUICUK (r 2 = 0.75, p < 0.01). (Supplementary Table 3)

Subcutaneous AT function in BBS is comparable to the BMImatched controls
Systemic non-esterified fatty acids (NEFA) concentrations were similar in BBS and controls in the fasting state and after insulin infusion (0.71 ± 0.11 vs. 0.67 ± 0.21 mmol/L; p = 0.59) (Fig. 1F-H). When the level of fasting insulin was taken into account to derive the AT insulin resistance index (ADIPO-IR = fasting NEFA × fasting insulin) [21] there was a trend towards higher levels in patients with BBS but this did not reach statistical significance (65.6 ± 58.9 vs. controls 31.3 ± 14, p = 0.09). The responsiveness of SAT lipolysis to insulin action was quantified by direct measurement of interstitial glycerol concentration using microdialysis during the HEC (Fig. 2). In the fasting state, the rate of interstitial glycerol release was similar in BBS and controls (basal AUC: BBS 284 ± 71 vs. controls 266 ± 74 μmol/l; p = 0.6). On commencement of insulin infusion, the rate of interstitial glycerol release as a reflection of adipose tissue lipolysis, was suppressed to a similar extent in both groups (post insulin AUC: BBS 161 ± 48 vs. controls 200 ± 48 μmol/l; p = 0.24).
Adipose tissue histology SAT histology was performed to determine whether there were structural differences in AT between the two group. There was no difference in average SAT cell numbers in two 1000 × 1000 micron fields between the subjects with BBS and controls (266 cells vs. 287 cells, p = 0.23) (Fig. 3A/B). Likewise, the size distribution of adipocytes in BBS was similar to controls (Fig. 3C).
As macrophage infiltration into AT under caloric stress may contribute to the pathogenesis of insulin resistance, the two main resident AT macrophages, namely M1 (CD68) or "classically activated" and M2 (CD163) or "alternatively activated" were then quantified (Supplementary Table 2). CD68 macrophages were increased in BBS but similar levels of CD163 macrophage infiltration were noted in both groups, with a shift towards a more inflammatory profile of adipose tissue macrophages in BBS as reflected by an elevated ratio of M1/M2 (CD68/CD163) phenotype. No difference was found in the amount of crown like structures, a marker of dead adipocytes and macrophage infiltration, between groups.

SAT transcriptomic signature
The SAT transcriptomic signature among the two groups was then compared. As expected, sexual dimorphism was noted (data not shown) but in both sexes there was no significant differences in the gene expression analysis with only 2 genes being differentially expressed at adjusted P value (q value) of <0.04. These were KIF24 (kinesin family member, EntrezID 3834) and NNAT (neuronatin, EntrezID 4826).

DISCUSSION
This is the first study to compare adipose tissue structure, gene expression and function in extreme obesity in both monogenic and polygenic cohorts with a view to elucidating the relationship between insulin resistance and obesity. Using the hyperinsulinemic euglycemic clamp as the gold standard for assessing insulin sensitivity, we were able to show the specific contribution of adipose tissue, skeletal muscle and liver to systemic insulin resis tance in both cohorts. Surprisingly, we found that patients with obesity due to BBS had similar glucose tolerance and insulin sensitivity to polygenic BMI-matched controls irrespective of the duration of fat mass accumulation. Furthermore, their adipose tissue structure, transcriptomic finger prints and functions were similar.
There was no significant difference between BBS and controls with common obesity in most parameters of insulin sensitivity in our study, although it is important to note that the predominant phenotype in our study was BBS1. The results remained unchanged when dedicated subgroup analysis using BBS1 was performed. Insulin resistance and diabetes mellitus have historically been considered an important feature of BBS, but a number of recent studies have paradoxically suggested improved glucose tolerance with certain BBS subtypes highlighting simultaneous inhibition of antiadipogenic, and activation of proadipogenic, pathways in certain subtypes [23]. In addition, although reduced hypothalamic response to leptin may be driven for BBS obesity, adipocyte biology studies explained the contribution of peripheral defects for its aetiology including the involvement of BBS genes in enhanced adipogenesis through repression of anti-adipogenic signalling pathways [24][25][26]. In a study by Feuillan et al., carried out in children and young adults with a mean age of 14 years, a similar degree of impaired glucose tolerance was noted in 43 patients with BBS and 100 BMI-matched controls [14]. This finding was more apparent in the patients with BBS1 mutation who were similar to BMI-matched controls for all variables whereas the differences in certain variables between patients with BBS10 mutations and controls persisted on subgroup analysis [14]. Another study in the BBS12 phenotype showed normal glucose tolerance by OGTT. Further in vitro analysis showed improved glucose absorption and insulin sensitivity in BBS12 inactivated human adipocytes. These findings were reproduced in a Bbs12 knockout mouse (Bbs12 − /−) model which showed improved insulin sensitivity [23]. In another study using zebrafish models, suppression of the BBS1 or BBS4 gene was associated with increased β-cell mass during development and a compensatory increased proliferation of β-cell when exposed to high glucose conditions [27]. Taken together, most mutations in BBS genesis is associated with improved or equal insulin sensitivity relative to the degree of their obesity, which is in concordance to our in-depth IR analysis of BBS cohort. In present study, patient with BBS had significantly high Adipsin level which was reported to be low in the patient with insulin resistance irrespective their BMI [28]. Similarly, Vaspin levels, which is overexpressed in human adipose tissue in IR condition as a compensatory mechanism, in patients BBS was considerably low compared to the control [29]. However, age and BMI matched large sample size may provide better understanding of these supportive findings.
The adipose tissue in BBS showed both hyperplasia and hypertrophy, both of which phenomenon are a common adaptive response to obesity. Both structurally and functionally, the adipose tissue behaved similarly to controls with common obesity. Despite extreme obesity being an important phenotype, there is a paucity of studies assessing the structure and function of adipose tissue in BBS. Studies in pre-adipocytes have shown that BBS genes play a role in adipogenesis [25,26]. BBS12 inactivated human adipocytes and Bbs12 − /− mice showed improved insulin sensitivity and enhanced adipogenesis [23]. These studies implicate this BBS gene in adipogenesis but direct evidence that this translates to impaired adipose tissue function is lacking. Similarly, there was a shift in the inflammatory profile of adipose tissue macrophages as reflected by an elevated CD63/CD168 ratio. This was not supported by other findings. Similar levels of crownlike macrophage structure and macrophage infiltration with CD63 and CD 168 labelling were noted in both groups. Furthermore, no significant changes were noted in an array of cytokines and proinflammatory markers. Studies have suggested that patients with BBS have increased leptin at systemic levels for their degree of adiposity with, BBS1 (27% (12/44) of participants) had lower leptin than BBS10 (30% (13/44) of participants) [14]. Our study with 7/9 of BBS1 and 1/9 of BBS10 show a positive trend in leptin level compared to control, although this did not reach statistical significance. However, our cohort predominantly represents the BBS1 phenotype. BBS proteins are involved in leptin receptor (long LepR) signalling in the hypothalamus [30]. BBS knockout mice models are obese, recapitulating the human phenotype [31]. In previous animal studies, BBS knockout mice were hyperphagic but continued to develop increased adiposity despite pair-feeding, suggesting that reduced energy expenditure may play an important role in the development of obesity [32]. Disruption of the BBSome by deleting the BBS1 gene from the nervous system causes obesity in mice. This phenotype is reproduced by ablation of the BBS1 gene selectively in the LRb (long LepR) -expressing cells [33]. Disruptive leptin signalling is a critical determinant of energy balance through its role in the regulation of the trafficking of the long signalling form of the leptin receptor (long LepR) [33]. How ciliary dysfunction causes obesity in BBS remains unknown [34] but imbalances in energy intake and energy expenditure continue to be the main hypothesis for obesity in BBS, mechanisms that have long been implicated in common polygenic obesity.
In general, based on the aetiology, obesity is divided in two broad categories: monogenic obesity, which is inherited in a Mendelian pattern and polygenic (common) obesity, which is the result of numerous polymorphisms that each has a small effect. Although often considered to be two distinct forms, gene discovery studies for both forms of obesity show that they have shared genetic and similar underlying biology, pointing to a key role for the brain in the control of energy homoeostasis and ultimately body weight [35]. The findings in our study would be in keeping with this notion, showing no major structural or functional difference in adipose tissue in monogenic BBS and polygenic obese controls.
Limitations of the present study should be acknowledged. Firstly, even though our national service is one of the largest Fig. 3 Subcutaneous adipose tissue histology. A-C No significant heterogeneity observed in adipocyte size in BBS when compared with obese controls as seen on H&E staining of subcutaneous adipose tissue. D, E Total and pericellular fibrosis quantification of picro-Sirius red staining. Data are presented as mean ± SD. Significance was set at p < 0.05. ns, p value non-significant. centres in the UK for BBS, the rarity of the syndrome and the demands of this functional study of insulin resistance precluded our ability to recruit large numbers across currently known BBS genotypes. We believe however, that BBS1, the majority in our study as the most common form in Europe and North America is a representative model to study. In addition, our in-depth study using a gold standard of insulin sensitivity along with the in vivo and in vitro adipose tissue studies allowed us sufficient power to compare the two cohorts. Nevertheless, our study is the only study in the literature to perform in-depth functional and structural adipose tissue study comparing extremely monogenic and polygenic forms of obesity in subjects matched for obesity.
In conclusion, our data shed new lights on the relationship between the duration of fat mass, genetic phenotype, insulin sensitivity and AT function. This study adds to the literature by suggesting that both polygenic and early-onset monogenic obesity share biological underpinning and lie on a spectrum rather than discrete entities. Hence, the metabolic dysfunction in BBS is proportionate to the degree of adiposity rather than the cause or duration of obesity.