Machine Learning based histology phenotyping to investigate the epidemiologic and genetic basis of adipocyte morphology and cardiometabolic traits.

Genetic studies have recently highlighted the importance of fat distribution, as well as overall adiposity, in the pathogenesis of obesity-associated diseases. Using a large study (n = 1,288) from 4 independent cohorts, we aimed to investigate the relationship between mean adipocyte area and obesity-related traits, and identify genetic factors associated with adipocyte cell size. To perform the first large-scale study of automatic adipocyte phenotyping using both histological and genetic data, we developed a deep learning-based method, the Adipocyte U-Net, to rapidly derive mean adipocyte area estimates from histology images. We validate our method using three state-of-the-art approaches; CellProfiler, Adiposoft and floating adipocytes fractions, all run blindly on two external cohorts. We observe high concordance between our method and the state-of-the-art approaches (Adipocyte U-net vs. CellProfiler: R2visceral = 0.94, P < 2.2 × 10-16, R2subcutaneous = 0.91, P < 2.2 × 10-16), and faster run times (10,000 images: 6mins vs 3.5hrs). We applied the Adipocyte U-Net to 4 cohorts with histology, genetic, and phenotypic data (total N = 820). After meta-analysis, we found that mean adipocyte area positively correlated with body mass index (BMI) (Psubq = 8.13 × 10-69, ßsubq = 0.45; Pvisc = 2.5 × 10-55, ßvisc = 0.49; average R2 across cohorts = 0.49) and that adipocytes in subcutaneous depots are larger than their visceral counterparts (Pmeta = 9.8 × 10-7). Lastly, we performed the largest GWAS and subsequent meta-analysis of mean adipocyte area and intra-individual adipocyte variation (N = 820). Despite having twice the number of samples than any similar study, we found no genome-wide significant associations, suggesting that larger sample sizes and a homogenous collection of adipose tissue are likely needed to identify robust genetic associations.

GLP-1 analogue-induced weight loss does not improve obesity-induced AT dysfunction.

Glucagon-like peptide-1 (GLP-1) analogues aid weight loss that improves obesity-associated adipose tissue (AT) dysfunction. GLP-1 treatment may however also directly influence AT that expresses the GLP-1 receptor (GLP-1R). The present study aimed to assess the impact of GLP-1 analogue treatment on subcutaneous AT (SCAT) inflammatory and fibrotic responses, compared with weight loss by calorie reduction (control). Among the 39 participants with Type 2 diabetes recruited, 30 age-matched participants were randomized to 4 months treatment with Liraglutide (n=22) or calorie restriction based on dietetic counselling (n=8). Assessments included clinical characteristics and repeated subcutaneous abdominal AT biopsies. Liraglutide resulted in weight loss in most participants (-3.12±1.72 kg, P=0.007) and significant reduction in visceral AT (VAT). It was more effective in lowering fasting glucose, in comparison with weight loss by dieting. However, tumour necrosis factor-α (TNFA) AT-expression (P=0.0005), macrophage chemoattractant protein-1 (MCP-1) expression (P=0.027) and its serum levels (P=0.048) increased with Liraglutide, suggestive of an inflammatory response unlike in the diet arm in which a trend of lower cluster of differentiation 14 (CD14) expression (P=0.09) was found. Liraglutide treatment also increased expression of factors involved in extracellular matrix (ECM) deposition, transforming growth factor-ß (TGFB) and collagen type 1 alpha 1 chain (COL1A1) (TGFB1: before 0.73±0.09 arbitrary units (AU), after 1.00±0.13 AU, P=0.006; COL1A1: 0.84±0.09 AU compared with 1.49±0.26 AU, P=0.026). Liraglutide thus appears to induce an inflammatory response in AT and influences ECM remodelling. Despite its superior effect on glycaemia, Liraglutide does not improve obesity-associated AT dysfunction in subcutaneous tissue. It is yet unclear whether this limits AT storage capacity for lipids. This may be of importance in patients being re-exposed to positive energy balance such as post GLP-1 discontinuation.

Lysyl oxidase and adipose tissue dysfunction.

BACKGROUND/OBJECTIVES: Lysyl oxidase (LOX) is an enzyme crucial for collagen fibre crosslinking and thus for fibrosis development. Fibrosis is characterised by a surplus of collagen fibre accumulation and is amongst others also a feature of obesity-associated dysfunctional adipose tissue (AT) which has been linked with type 2 diabetes. We hypothesised that in type 2 diabetes and obesity LOX expression and activity will be increased as a consequence of worsening AT dysfunction. This study aimed to provide a comprehensive characterisation of LOX in human AT. METHODS: LOX mRNA expression was analysed in omental and abdominal subcutaneous AT obtained during elective surgery from subjects with a wide range of BMI, with and without diabetes. In addition, LOX expression was studied in subcutaneous AT before and 9.5months after bariatric surgery. To study the mechanism of LOX changes, its expression and activity were assessed after either hypoxia, recombinant human leptin or glucose treatment of AT explants. In addition, LOX response to acute inflammation was tested after stimulation by a single injection of lipopolysaccharide versus saline solution (control) in healthy men, in vivo. Quantity of mRNA was measured by RT-qPCR. RESULTS: LOX expression was higher in obesity and correlated with BMI whilst, in vitro, leptin at high concentrations, as a potential feedback mechanism, suppressed its expression. Neither diabetes status, nor hyperglycaemia affected LOX. Hypoxia and lipopolysaccharide-induced acute inflammation increased LOX AT expression, latter was independent of macrophage infiltration. CONCLUSIONS: Whilst LOX may not be affected by obesity-associated complications such as diabetes, our results confirm that LOX is increased by hypoxia and inflammation as underlying mechanism for its upregulation in adipose tissue with obesity.

Aldo-Keto Reductases 1B in Adrenal Cortex Physiology.

Aldose reductase (AKR1B) proteins are monomeric enzymes, belonging to the aldo-keto reductase (AKR) superfamily. They perform oxidoreduction of carbonyl groups from a wide variety of substrates, such as aliphatic and aromatic aldehydes or ketones. Due to the involvement of human aldose reductases in pathologies, such as diabetic complications and cancer, AKR1B subgroup enzymatic properties have been extensively characterized. However, the issue of AKR1B function in non-pathologic conditions remains poorly resolved. Adrenal activities generated large amount of harmful aldehydes from lipid peroxidation and steroidogenesis, including 4-hydroxynonenal (4-HNE) and isocaproaldehyde (4-methylpentanal), which can both be reduced by AKR1B proteins. More recently, some AKR1B isoforms have been shown to be endowed with prostaglandin F synthase (PGFS) activity, suggesting that, in addition to possible scavenger function, they could instigate paracrine signals. Interestingly, the adrenal gland is one of the major sites for human and murine AKR1B expression, suggesting that their detoxifying/signaling activity could be specifically required for the correct handling of adrenal function. Moreover, chronic effects of ACTH result in a coordinated regulation of genes encoding the steroidogenic enzymes and some AKR1B isoforms. This review presents the molecular mechanisms accounting for the adrenal-specific expression of some AKR1B genes. Using data from recent mouse genetic models, we will try to connect their enzymatic properties and regulation with adrenal functions.

Aldose reductases influence prostaglandin F2α levels and adipocyte differentiation in male mouse and human species.

Aldose reductases (AKR1B) are widely expressed oxidoreductases whose physiological function remains elusive. Some isoforms are genuine prostaglandin F2α (PGF2α) synthases, suggesting they might influence adipose homeostasis because PGF2α inhibits adipogenesis. This was shown by Akr1b7 gene ablation in the mouse, which resulted in increased adiposity related to a lower PGF2α content in fat. Yet humans have no ortholog gene for Akr1b7, so the role of aldose reductases in human adipose homeostasis remains to be explored. We analyzed expression of genes encoding human and mouse aldose reductase isoforms in adipose tissues and differentiating adipocytes to assess conserved mechanisms regulating PGF2α synthesis and adipogenesis. The Akr1b3 gene encoded the most abundant isoform in mouse adipose tissue, whereas Akr1b7 encoded the only isoform enriched in the stromal vascular fraction. Most mouse aldose reductase gene expression peaked in early adipogenesis of 3T3-L1 cells and diminished with differentiation. In contrast with its mouse ortholog Akr1b3, AKR1B1 expression increased throughout differentiation of human multipotent adipose-derived stem cells, paralleling PGF2α release, whereas PGF2α receptor (FP) levels collapsed in early differentiation. Pharmacological inhibition of aldose reductase using Statil altered PGF2α production and enhanced human multipotent adipose-derived stem adipocyte differentiation. As expected, the adipogenic effects of Statil were counteracted by an FP agonist (cloprostenol). Thus, in both species aldose reductase-dependent PGF2α production could be important in early differentiation to restrict adipogenesis. PGF2α antiadipogenic signaling could then be toned down through the FP receptor or aldose reductases down-regulation in human and mouse cells, respectively. Our data suggest that aldose reductase inhibitors could have obesogenic potential.

Aldo-Keto Reductases 1B in Endocrinology and Metabolism.

The aldose reductase (AR; human AKR1B1/mouse Akr1b3) has been the focus of many research because of its role in diabetic complications. The starting point of these alterations is the massive entry of glucose in polyol pathway where it is converted into sorbitol by this enzyme. However, the issue of AR function in non-diabetic condition remains unresolved. AR-like enzymes (AKR1B10, Akr1b7, and Akr1b8) are highly related isoforms often co-expressed with bona fide AR, making functional analysis of one or the other isoform a challenging task. AKR1B/Akr1b members share at least 65% protein identity and the general ability to reduce many redundant substrates such as aldehydes provided from lipid peroxidation, steroids and their by-products, and xenobiotics in vitro. Based on these properties, AKR1B/Akr1b are generally considered as detoxifying enzymes. Considering that divergences should be more informative than similarities to help understanding their physiological functions, we chose to review specific hallmarks of each human/mouse isoforms by focusing on tissue distribution and specific mechanisms of gene regulation. Indeed, although the AR shows ubiquitous expression, AR-like proteins exhibit tissue-specific patterns of expression. We focused on three organs where certain isoforms are enriched, the adrenal gland, enterohepatic, and adipose tissues and tried to connect recent enzymatic and regulation data with endocrine and metabolic functions of these organs. We presented recent mouse models showing unsuspected physiological functions in the regulation of glucido-lipidic metabolism and adipose tissue homeostasis. Beyond the widely accepted idea that AKR1B/Akr1b are detoxification enzymes, these recent reports provide growing evidences that they are able to modify or generate signal molecules. This conceptually shifts this class of enzymes from unenviable status of scavenger to upper class of messengers.

Depressed levels of prostaglandin F2α in mice lacking Akr1b7 increase basal adiposity and predispose to diet-induced obesity.

Negative regulators of white adipose tissue (WAT) expansion are poorly documented in vivo. Prostaglandin F(2α) (PGF(2α)) is a potent antiadipogenic factor in cultured preadipocytes, but evidence for its involvement in physiological context is lacking. We previously reported that Akr1b7, an aldo-keto reductase enriched in adipose stromal vascular fraction but absent from mature adipocytes, has antiadipogenic properties possibly supported by PGF(2α) synthase activity. To test whether lack of Akr1b7 could influence WAT homeostasis in vivo, we generated Akr1b7(-/-) mice in 129/Sv background. Akr1b7(-/-) mice displayed excessive basal adiposity resulting from adipocyte hyperplasia/hypertrophy and exhibited greater sensitivity to diet-induced obesity. Following adipose enlargement and irrespective of the diet, they developed liver steatosis and progressive insulin resistance. Akr1b7 loss was associated with decreased PGF(2α) WAT contents. Cloprostenol (PGF(2α) agonist) administration to Akr1b7(-/-) mice normalized WAT expansion by affecting both de novo adipocyte differentiation and size. Treatment of 3T3-L1 adipocytes and Akr1b7(-/-) mice with cloprostenol suggested that decreased adipocyte size resulted from inhibition of lipogenic gene expression. Hence, Akr1b7 is a major regulator of WAT development through at least two PGF(2α)-dependent mechanisms: inhibition of adipogenesis and lipogenesis. These findings provide molecular rationale to explore the status of aldo-keto reductases in dysregulations of adipose tissue homeostasis.