Adipose Tissue Inflammation 2022

Edited by Javier Gómez-Ambrosi

mdpi.com/journal/cells

## **Adipose Tissue Inflammation 2022**

## **Adipose Tissue Inflammation 2022**

Editor

**Javier G ´omez-Ambrosi**

Basel • Beijing • Wuhan • Barcelona • Belgrade • Novi Sad • Cluj • Manchester

*Editor* Javier Gomez-Ambrosi ´ Metabolic Research Lab Department of Endocrinology and Nutrition Cl´ınica Universidad de Navarra Pamplona Spain

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Cells* (ISSN 2073-4409) (available at: www.mdpi.com/journal/cells/special issues/Adipose Tissue Inflammation).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

Lastname, A.A.; Lastname, B.B. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-8675-5 (Hbk) ISBN 978-3-0365-8674-8 (PDF) doi.org/10.3390/books978-3-0365-8674-8**

© 2023 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license. The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons Attribution-NonCommercial-NoDerivs (CC BY-NC-ND) license.

## **Contents**



**Fien Demeulemeester, Karin de Punder, Marloes van Heijningen and Femke van Doesburg** Obesity as a Risk Factor for Severe COVID-19 and Complications: A Review Reprinted from: *Cells* **2021**, *10*, 933, doi:10.3390/cells10040933 . . . . . . . . . . . . . . . . . . . . **137**

## **About the Editor**

## **Javier G ´omez-Ambrosi**

Javier Gomez-Ambrosi is a Researcher at the Metabolic Research Laboratory of the Cl´ınica Universidad de Navarra and Associate Professor at the School of Medicine in the University of Navarra, Pamplona, Spain. His main area of research is obesity and related morbidities, from a clinical and molecular point of view. His research combines basic research in experimental animals and cells with the clinical setting, trying to disentangle the pathophysiological mechanisms responsible of the impact of excess adiposity on the development of comorbidities. He has published more than 200 articles (h-index 58) and has been the PI in more than 20 research projects.

## **Preface**

Over the last few decades, obesity has become one of the most prevalent metabolic disorders. Excess adiposity favors the development of cardiometabolic alterations, such as type 2 diabetes (T2D), cardiovascular disease, dyslipidemia, non-alcoholic fatty liver disease, and cancer. In the last years, adipose tissue inflammation represents one of the major mechanisms underlying adipose tissue dysfunction, contributing to the development of metabolic derangements in other organs. The contribution of the different adipose tissue depots, the discovery of the function of new adipokines, the involvement of the inflammasome, or the dual effect of macrophage polarization have greatly contributed to the improved understating produced in previous years, of the role played by adipose tissue inflammation in the development of metabolic derangements. This reprint presents recent advances in understanding the molecular processes that take place in adipose tissue inflammation; moreover, it discusses the impact of adipose tissue inflammation on systemic metabolic alterations associated with excess adiposity, as well as its repercussion in several pathological conditions.

> **Javier G ´omez-Ambrosi** *Editor*

## *Editorial* **Adipose Tissue Inflammation**

**Javier Gómez-Ambrosi 1,2,3**


In recent decades, obesity has become one of the most common metabolic diseases. Excess adiposity increases the risk of developing type 2 diabetes (T2D), cardiometabolic diseases, dyslipidemia, fatty liver, and several types of cancer [1]. Much progress has been made in understanding the major regulatory pathways underlying adipose tissue inflammation, which represent one the main drivers of adipose tissue dysfunction and, consequently, of obesity-associated metabolic alterations [2].

This Special Issue presents recent advances in understanding the molecular processes that take place in adipose tissue inflammation; moreover, it discusses the impact of adipose tissue inflammation on systemic metabolic alterations associated with excess adiposity, as well as its repercussion in several pathological conditions.

Although obesity has traditionally been considered a single medical entity, in recent years, greater importance has been placed on phenotyping the different obesities in order to improve their clinical management [3,4]. In their cross-sectional and prospective study, Goldstein et al. analyze the usefulness of determining mast cell accumulation in human adipose tissue as a proxy of metabolic phenotyping [5]. They suggest that patients with obesity with high expression of mast cell genes exhibit a healthier metabolic phenotype than individuals with low expression. The authors also find that higher mast cell accumulation in adipose tissue in patients undergoing bariatric surgery is a predictor of greater weight loss after surgery. They conclude that a high number of mast cells defines a clinically favorable obesity phenotype [5].

Several studies discuss the molecular mechanisms that regulate the impact of adipose tissue inflammation. Lempesis et al. show that low physiological oxygen tension decreases the expression and secretion of proinflammatory adipokines in adipocytes obtained from patients with obesity, an effect that is not found in cells derived from donors of normal weight [6]. Kochumon and colleagues report that the adipose tissue expression of steroid receptor RNA activator 1 (SRA1) may represent a novel surrogate marker of metabolic inflammation through its association with Toll-like receptors (TLRs) [7].

Other studies provide interesting information regarding the regulation of inflammation in adipose tissue in mouse models. In one such study, Sandrini et al. study the effects of physical exercise on *BDNF* Val66Met mice, a model of increased adiposity associated with a proinflammatory and prothrombotic profile [8]. They find that four weeks of voluntary wheel running changes epididymal adipose tissue morphology and the expression of proinflammatory genes, inducing reversion of the prothrombotic phenotype; this suggests that a reduction in adipose tissue inflammation is important in promoting the positive effects of physical activity [8]. In another study, Mendes de Farias and colleagues find that daily melatonin supplementation for 10 weeks in mice on a high-fat diet reduces fat accumulation, adipocyte size, and the expression of proinflammatory adipokines in adipose tissue and the circulation; this suggests that melatonin could be considered as a therapeutic molecule for the treatment of obesity [9].

**Citation:** Gómez-Ambrosi, J. Adipose Tissue Inflammation. *Cells* **2023**, *12*, 1484. https://doi.org/ 10.3390/cells12111484

Received: 22 May 2023 Accepted: 24 May 2023 Published: 26 May 2023

**Copyright:** © 2023 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Adipokines and adipose tissue inflammation have been shown to play a role in several physiologic and pathophysiologic conditions [10,11], as stated in several studies included in this Special Issue. Morais et al. show that an adequate balance between adiponectin and leptin concentrations in human milk may regulate colostrum mononuclear cell activity, eliciting a more effective response against neonatal infection in breastfeeding infants [12]. Another adipokine, fatty acid-binding protein 4 (FABP4), implicated in the control of cellular lipid metabolism, is also involved in inflammation and the development of insulin resistance. In an exhaustive review, Trojnar et al. detail the different mechanisms by which FABP4 is involved in inflammation and insulin resistance and the potential role of this adipokine in T2D, gestational diabetes, and fatty liver, among other conditions [13]. Chang and Eibl describe the relevance of adipose tissue inflammation as an important driver of obesity-associated pancreatic ductal adenocarcinoma, and consider strategies aimed at reducing inflammation in this tissue as a weapon against this type of cancer [14]. In another interesting review, Cornide-Petronio and colleagues reinforce current knowledge regarding the interaction between the liver and adipose tissue during liver surgery. The scientific and clinical controversies in this area are reviewed, as are potential therapeutic approaches. The information provided could help to develop protective measures focused on manipulating the liver–visceral adipose tissue axis to enhance the postoperative results of hepatic surgery [15]. Finally, Demeulemeester et al. summarize scientific research on the link between obesity and COVID-19 severity, and analyze probable mechanisms that could help to understand why patients living with obesity exhibit an increased risk of serious consequences during COVID-19 [16].

In recent years, there have been significant advancements in the understanding of the cellular and molecular mechanisms involved in adipose tissue inflammation. Thanks to this progress, more tools and approaches are available for the treatment of obesity and T2D. However, in order to optimize the management of patients with obesity, more research needs to be conducted.

**Funding:** This paper was supported by ISCIII–FEDER (PI20/00080) and CIBEROBN, ISCIII, Spain, and by PC098-099 MEPERTROBE and the Department of Health 58/2021, Gobierno de Navarra-FEDER, Spain.

**Conflicts of Interest:** The author declares no conflict of interest.

## **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

*Article*

## **Higher Mast Cell Accumulation in Human Adipose Tissues Defines Clinically Favorable Obesity Sub-Phenotypes**

**Nir Goldstein 1,2, Yarden Kezerle <sup>3</sup> , Yftach Gepner <sup>2</sup> , Yulia Haim <sup>1</sup> , Tal Pecht <sup>1</sup> , Roi Gazit 4,5 , Vera Polischuk <sup>6</sup> , Idit F. Liberty <sup>6</sup> , Boris Kirshtein <sup>6</sup> , Ruthy Shaco-Levy <sup>3</sup> , Matthias Blüher 7,8 and Assaf Rudich 1,5,\***


Received: 4 April 2020; Accepted: 18 June 2020; Published: 20 June 2020

**Abstract:** The identification of human obesity sub-types may improve the clinical management of patients with obesity and uncover previously unrecognized obesity mechanisms. Here, we hypothesized that adipose tissue (AT) mast cells (MC) estimation could be a mark for human obesity sub-phenotyping beyond current clinical-based stratifications, both cross-sectionally and prospectively. We estimated MC accumulation using immunohistochemistry and gene expression in abdominal visceral AT (VAT) and subcutaneous (SAT) in a human cohort of 65 persons with obesity who underwent elective abdominal (mainly bariatric) surgery, and we validated key results in two clinically similar, independent cohorts (*n* = 33, *n* = 56). AT-MC were readily detectable by immunostaining for either c-kit or tryptase and by assessing the gene expression of KIT (KIT Proto-Oncogene, Receptor Tyrosine Kinase), TPSB2 (tryptase beta 2), and CMA1 (chymase 1). Participants were characterized as VAT-MClow if the expression of both CMA1 and TPSB2 was below the median. Higher expressers of MC genes (MChigh) were metabolically healthier (lower fasting glucose and glycated hemoglobin, with higher pancreatic beta cell reserve (HOMA-β), and lower triglycerides and alkaline-phosphatase) than people with low expression (MClow). Prospectively, higher MC accumulation in VAT or SAT obtained during surgery predicted greater postoperative weight-loss response to bariatric surgery. Jointly, high AT-MC accumulation may be used to clinically define obesity sub-phenotypes, which are associated with a "healthier" cardiometabolic risk profile and a better weight-loss response to bariatric surgery.

**Keywords:** obesity; type 2 diabetes; bariatric surgery; adipose tissue; mast cells

## **1. Introduction**

The inflammation of adipose tissue (AT) may link obesity to its cardiometabolic comorbidities. Although macrophages (CD68+ cells) were the first immune cell type realized to infiltrate adipose tissue (AT) in obesity and to associate with obesity-related metabolic dysfunction [1], the current view in the field engages virtually all immune cell types in obesity-associated AT inflammation, including T and B lymphocytes (and their sub-classes), dendritic cells, neutrophils, and natural killer (NK) cells [2]. Within the changing environment of adipose tissues in obesity, these cell types undergo complex phenotypic alterations, rendering them highly diverse. This challenges the efforts to establish clear contributions of specific cell types to obesity-associated metabolic deterioration, even for adipose tissue macrophages [3] (also recently reviewed in [4,5]). In 2009, elevated numbers of mast cells (MC) were demonstrated in obese AT, located mainly near micro-vessels, both in humans and in mouse models [6]. Mast cells are hematopoietic, bone-marrow-derived immune cells, which mature and differentiate in the tissue in which they eventually reside. Although mainly found in tissues with greater interaction with the outer environment such as the reticular layers of the skin, gastrointestinal system, and the airways, MC can be found in all organs and tissues. In AT of lean mice, MC are more abundant in subcutaneous AT (SAT) than in epididymal, mesenteric, or perirenal fat pads. Interestingly, obesity is associated with increases in MC numbers in all white AT depots besides SAT [7].

The physiological/functional impact of increased numbers of AT-MC in obesity currently remains controversial: Two independent groups demonstrated that MC-deficient mice by mutation in the growth factor receptor KIT (KitW-sh/W-sh), which is required for MC development, are protected from diet-induced obesity and its co-morbidities [6,8]. In contrast, two different, none KIT-dependent MC-deficient mouse models (Cpa3cre/+ (Cre recombinase in the carboxypeptidase A locus) and Mcpt5-Cre+ R-DTA+ (MC specific expression of diphtheria toxin A)) exhibited no phenotypic impact on the development of obesity or on its metabolic consequences, including insulin resistance, hepatic steatosis, or inflammation [9,10].

In humans, the AT of people with obesity exhibited higher MC numbers compared to lean patients, both in SAT [6,11] and VAT [11]. Importantly, this finding was particularly evident among patients with obesity and type 2 diabetes [11], suggesting a link between high AT-MC infiltration and the severity of obesity-related metabolic disturbance. Consistently, a sub-analysis among 20 persons with obesity suggested that higher MC numbers associate with higher fasting glucose and HbA1c [11]. Interestingly, in that study, two sub-populations of AT-MC were assessed based on the expression of their proteases—tryptase+ and tryptase+/chymase+ MC (MCT and MCTC, respectively), rendering tryptase, the gene product of tryptase beta-2 (TPSB2), a common marker for MC. Yet, the ratio between these two sub-populations remained similar in different depots and in leanness versus obesity. Isolated MC from the SAT of patients with obesity and type 2 diabetes were more activated, releasing more inflammatory cytokines and proteases such as tryptase [11]. Correspondingly, serum tryptase concentrations were higher in people with obesity compared to lean [6]. In a different study examining a mixed cohort of persons without or with obesity, the expression of MC-tryptase (TPSB2) in VAT did not associate with parameters of glucose homeostasis or insulin sensitivity [12]. Collectively, obesity seems to associate with increased numbers of MC in AT. Mouse models so far yielded conflicting results. Human studies suggest that higher AT-MC numbers (i.e., accumulation), and perhaps their activation, may associate with poor glycemic control, but assessment of whether and how AT-MC associate within cohorts of patients with obesity, and whether it corresponds to obesity sub-phenotypes, is limited.

In the present study, we aimed to address this current gap of knowledge by hypothesizing that among persons with obesity, higher numbers of AT-MC associate with an obesity phenotype characterized by a poorer metabolic profile. We sought to assess the relevance of such MC-based obesity sup-phenotyping both in cross-sectional analyses of patients with obesity, with and without type 2 diabetes, and for predicting patients' response to bariatric surgery.

## **2. Materials and Methods**

## *2.1. Human Cohorts*

We recruited persons with obesity (Body mass index (BMI) <sup>≥</sup> 30 kg/m<sup>2</sup> ) undergoing elective abdominal surgery (mainly bariatric surgery or elective cholecystectomy) as part of the coordinated human adipose tissue bio-banks in Beer-Sheva, Israel (*n* = 65, main cohort) and in Leipzig, Germany (*n* = 32 and *n* = 56, validation cohorts 1 and 2, respectively) (Table 1). Prior to operation, under overnight fasting conditions, body weight and blood samples were obtained. Both visceral (omental) and superficial-subcutaneous adipose tissues biopsies were obtained during the surgery and processed for histology and gene expression using coordinated methodologies, as we previously described in detail [13,14]. Persons were identified as normoglycemic if fasting plasma glucose (FPG) levels were lower than 5.6 mmol/L, HbA1c ≤ 38 mmol/mol (5.6%), and with no anti-diabetic medications on the day of operation. Prediabetes was defined as FPG 5.6–6.9 mmol/L and HbA1c 39–46 mmol/mol (5.7–6.4%), and type 2 diabetes if glucose ≥ 7.0 mmol/L or HbA1c ≥ 48 mmol/mol (6.5%). In 13 patients (20%), in whom glycemic status was ambivalent, medical records were screened up to 4 months pre-operation for additional FPG and HbA1c measurements, and final categorization was made by co-author IFL, who is a Diabetologist. A homeostatic model assessment of insulin resistance (HOMA-IR) and homeostatic model assessment of beta cells reserve (HOMA-β) were calculated as follows: HOMA-IR: (FPG (mmol/L) × Insulin (µIU/mL)/22.5) and HOMA-β: (20 × Insulin (µIU/mL)/ FPG (mmol/L)-3.5) [15]. For post-operation follow-up sub-study, only people undergoing bariatric surgery for the first time and for whom postoperative information was available were included. In addition to the main Beer-Sheva cohort, we included two independent cohorts—validation cohorts 1 and 2, with *n* = 32 and *n* = 56 individuals, respectively, all with obesity (BMI range: 30–75 kg/m<sup>2</sup> , Table 1), from the University of Leipzig Obesity Treatment Center. Paired abdominal subcutaneous and omental adipose tissue biopsies were taken during elective sleeve gastrectomy, Roux-en-Y gastric bypass, hernia, or cholecystectomy surgeries and processed as previously described [16]. As for validation cohort 2, we included data from 56 patients who underwent a two-step bariatric surgery strategy with laparoscopic gastric sleeve resection as the first step and a Roux-en-Y gastric bypass as the second step 12 ± 2 months later (Table 1). At both time points, serum/plasma samples, omental, and abdominal subcutaneous adipose tissue biopsies were obtained. All patients provided before the study a written informed consent to participate, and all procedures were approved in advance by the local ethical committees and conducted in accordance with the declaration of Helsinki guidelines (0348-15-SOR; for Leipzig cohorts:, 017-12-23012012, and Reg. No. 031-2006).


**Table 1.** Baseline characteristics of obese patient.


Values are mean ± standard deviation. \* *p* < 0.05 different from VAT-MClow by independent *t*-test. † *p* < 0.05 compared visceral adipocyte area, by paired *t*-test. # C- reactive protein (CRP) values are higher compared with validation cohort 1 and the main cohort since inclusion criterion of patients with CRP < 47.6 nmol/L (5.0 mg/L) was not applied in this cohort.

## *2.2. RNA Extraction and Quantification*

mRNA was extracted as described previously [13]. Briefly, 300 mg of tissue were minced and extraction was done with an RNeasy lipid tissue minikit (Qiagen. Hilden Germany). cDNA was produced using the reverse transcriptase kit (Applied Biosystems. Beverly, MA, USA). mRNA was quantified using the Taqman system, where expression levels of selected genes were calculated as a fold from AT-specific endogenous control genes (PGK1 and PPIA) [17] and calculated as 2−∆∆ct [18]. Probes assay IDs are in Table S1. For the Leipzig validation cohort 1, mRNA expression of candidate genes (cKIT, CMA1, TPSB2) was analyzed using Illumina human HT-12 expression chips. RNA integrity and concentration were examined using an Agilent 2100 Bioanalyzer (Agilent Technologies, USA). In validation cohort 2, additional measurements of cytokine serum concentrations and adipose tissue expression (Interleukin -6 (IL-6), Interleukin -1beta (IL-1beta), Tumor necrosis factor alpha (TNFalpha)) were performed as described previously [3].

#### *2.3. Histology and Immunostaining*

Immunostaining for C-Kit (Dako (Agilent), Santa Clara, CA USA) or Tryptase (ThermoFisher. Waltham, MA USA), and CD68 (Dako (Agilent), performed in a sub-cohort of *n* = 30 from the Beer-Sheva cohort), is performed routinely in clinical pathology to stain tissue MC and macrophages, respectively. Immunostaining was performed in 5 microns-thick sections from paraffin embedded visceral AT (VAT) and subcutaneous (SAT) samples as described before [14]. Cell count/field was performed, blindly and independently, by two pathologists (Y.K and R.S-L) using an Olympus BX43 light microscope, in 10 consecutive high-power fields (X400), and the number of C-Kit+ and tryptase+ per 100 adipocytes was calculated.

### *2.4. Statistical Analyses*

Baseline clinical characteristics values are presented as mean ± SD. An independent t-test was used to compare between groups with Levene's test for equality of variances. In cases of non-normal distribution, ln-transformation was made. Comparison between percentages of medications usage was calculated with a chi-square test. An ANOVA test was used to compare between the four groups when sex, age, and diabetes stratification was used in combination with MC high/low stratifications. In order to detect differences between the four groups, Least Significant Difference (LSD) post-hoc analysis was used. A 2 × 2 ANOVA was performed to test the relationship between non- and type 2 diabetes groups and the MC group (MClow/high) using an age-adjusted model. Extreme outliers were excluded using the interquartile range 3 × (Q3-Q1). We also used Spearman's correlation to test the association between c-Kit+, tryptase+, and CD68<sup>+</sup> cells, clinical parameters, and MC-related genes. A linear regression model was used to assess the association between VAT-CMA1 high/low expression and percentage of weight loss. The model was adjusted for baseline BMI or surgery type. Analyses were performed using SPSS Version 20.0 (SPSS, Inc., Chicago, IL, USA) and GraphPad prism (8.4.2., San-Diego, CA, USA). All tests were two-tailed and *p* < 0.05 was considered statistically significant.

## **3. Results**

#### *3.1. Assessing AT-MC Histologically and by Gene Expression*

To test if the degree of AT-MC accumulation may reflect obesity sub-phenotypes, we first investigated a cohort of persons with obesity (BMI <sup>≥</sup> 30 kg/m<sup>2</sup> , *n* = 65) (Table 1, Beer–Sheva cohort). Among the persons with obesity, 38 persons (58.5%) were morbidly obese (Class III, BMI <sup>≥</sup> 40 kg/m<sup>2</sup> ). Compared to those with 30 <sup>≤</sup> BMI <sup>&</sup>lt; 40 kg/m<sup>2</sup> , persons with morbid obesity differed only in weight, BMI, waist circumference, and low-density lipoprotein (LDL).

We contemplated both histological and gene expression approaches to estimate AT-MC accumulation, utilizing several genes/proteins considered as MC-specific, including KIT/c-Kit (c-KIT Proto-Oncogene Receptor Tyrosine Kinase), TPSB2/Tryptase (mast cell tryptase beta II), and CMA1 (mast cell chymase 1 [19]). The specificity of selected genes to MC is presented in Supplemental Figures S1–S3. By immunohistochemistry, c-Kit+ or tryptase+ cells were readily discernable in human visceral AT from obese patients (Figure 1A,B). When assessing serial sections, the number of c-Kit+ cells per 100 adipocytes (percentage of c-Kit+) highly correlated with the percentage of tryptase+ cells (Figure 1C). AT-MC could be detected both in fibrotic areas within AT and dispersed between the adipocytes, being more prevalent in VAT sections rated by Clinical Pathologists (co-authors YK and RSL) as exhibiting more severe fibrosis (Figure 1D, for SAT see Figure S4). Interestingly, AT-MC correlated with CD68<sup>+</sup> macrophages only within fibrotic areas of the tissues (Figure 1E), but not in parenchymal areas between adipocytes (*r*(%) = 0.264, *p* = 0.159, *n* = 30). Using a gene expression approach, the expression of all three MC genes was significantly intercorrelated in both VAT (Figure 1F) and SAT (not shown). Moreover, VAT-KIT gene expression levels correlated with visceral percentage of c-Kit+ cells (log-transformed, *r* = 0.363, *p* = 0.038), and VAT-CMA1 expression correlated with both percentage of c-Kit+ and percentage of tryptase+ (*r*(%) = 0.348, *p* = 0.044; *r*(%) = 0.561, *p* = 0.008, respectively). Consistent with the proposed association between MC and fibrosis, we found significant associations between AT-MC genes and the expression of collagens 1A1, 3A1, and 6A1 in VAT (Table 2), but less so in SAT that exhibited only correlations between SAT-MC genes with collagen 6A1 (Table S3). We contemplated both histological and gene expression approaches to estimate AT-MC accumulation, utilizing several genes/proteins considered as MC-specific, including KIT/c-Kit (c-KIT Proto-Oncogene Receptor Tyrosine Kinase), TPSB2/Tryptase (mast cell tryptase beta II), and CMA1 (mast cell chymase 1 [19]). The specificity of selected genes to MC is presented in Supplemental Figures S1–S3. By immunohistochemistry, c-Kit+ or tryptase+ cells were readily discernable in human visceral AT from obese patients (Figure 1A,B). When assessing serial sections, the number of c-Kit+ cells per 100 adipocytes (percentage of c-Kit+) highly correlated with the percentage of tryptase+ cells (Figure 1C). AT-MC could be detected both in fibrotic areas within AT and dispersed between the adipocytes, being more prevalent in VAT sections rated by Clinical Pathologists (coauthors YK and RSL) as exhibiting more severe fibrosis (Figure 1D, for SAT see Figure S4). Interestingly, AT-MC correlated with CD68+ macrophages only within fibrotic areas of the tissues (Figure 1E), but not in parenchymal areas between adipocytes (*r*(ρ) = 0.264, *p* = 0.159, *n* = 30). Using a gene expression approach, the expression of all three MC genes was significantly intercorrelated in both VAT (Figure 1F) and SAT (not shown). Moreover, VAT-KIT gene expression levels correlated with visceral percentage of c-Kit+ cells (log-transformed, *r* = 0.363, *p* = 0.038), and VAT-CMA1 expression correlated with both percentage of c-Kit+ and percentage of tryptase+ (*r*(ρ) = 0.348, *p* = 0.044; *r*(ρ) = 0.561, *p* = 0.008, respectively). Consistent with the proposed association between MC and fibrosis, we found significant associations between AT-MC genes and the expression of collagens 1A1, 3A1, and 6A1 in VAT (Table 2), but less so in SAT that exhibited only correlations between SAT-MC genes with collagen 6A1 (Table S3).

*Cells* **2020**, *9*, x; doi: FOR PEER REVIEW www.mdpi.com/journal/cells **Figure 1.** Identification of human visceral (omental) adipose tissue mast cells (AT MC). AT stained for c-KIT Proto-Oncogene Receptor Tyrosine Kinase (C-Kit+, white arrows) with (**A**) ×100 and (**B**) ×400 magnification (left), or stained for tryptase+ cells ×400 (**B**, right). (**C**) Spearman's correlation between %C-Kit+ and %tryptase+ cells (percentage being per 100 adipocytes) in serial AT sections (*n* = 14). (**D**) C-Kit+ is discernable both within fibrotic areas (black arrows) and around adipocytes (white arrow), and bar graph—the number of C-Kit+ cells in sections rated by clinical pathologists (co-authors YD and RSL) as exhibiting mild, moderate, or severe degree of fibrosis. (**E**) Representative histological sections from two representative patients—one with low and the second with high MC in fibrotic areas within VAT, stained for either C-Kit (MC) or CD68 (macrophages). Graph below depicts Spearman's correlation between %C-Kit+ and %CD68+ cells (i.e., macrophages, percentage being per 100 adipocytes) in serial VAT sections within the fibrotic areas, *n* = 30. (**F**) Spearman's intercorrelations **Figure 1.** Identification of human visceral (omental) adipose tissue mast cells (AT MC). AT stained for c-KIT Proto-Oncogene Receptor Tyrosine Kinase (C-Kit+, white arrows) with (**A**) ×100 and (**B**) ×400 magnification (left), or stained for tryptase+ cells ×400 (**B**, right). (**C**) Spearman's correlation between %C-Kit+ and %tryptase+ cells (percentage being per 100 adipocytes) in serial AT sections (*n* = 14). (**D**) C-Kit+ is discernable both within fibrotic areas (black arrows) and around adipocytes (white arrow), and bar graph—the number of C-Kit+ cells in sections rated by clinical pathologists (co-authors YD and RSL) as exhibiting mild, moderate, or severe degree of fibrosis. (**E**) Representative histological sections from two representative patients—one with low and the second with high MC in fibrotic areas within VAT, stained for either C-Kit (MC) or CD68 (macrophages). Graph below depicts Spearman's correlation between %C-Kit+ and %CD68+ cells (i.e., macrophages, percentage being per 100 adipocytes) in serial VAT sections within the fibrotic areas, *n* = 30. (**F**) Spearman's intercorrelations (*n* = 65) between VAT MC-related genes (KIT, TPSB2, and CMA1). In each correlation, the upper line indicates *r*(%) value and lower line—*p*-value.

**Table 2.** Intercorrelation between VAT-MC and collagens' gene expression in the Beer-Sheva cohort (*n* = 65). n.s, not significant. Values are *r* and *p* value (upper and lower line respectively) of Spearman's correlations. CMA1: mast cell chymase 1, TPSB2: mast cell tryptase beta II, COLA1A1: collagen type I alpha 1 chain, COLA3A1: collagen type III alpha 1 chain, COLA6A1: collagen type VI alpha 1 chain.


In the independent validation, cohort 1 of *n* = 32 patients with obesity whose VAT gene expression was assessed by microarrays [16], all MC genes probe sets (KIT, TPSB2, CMA1) exhibited significant negative correlations with macrophage gene probes (CD68, Mac2, IGTAX(CD11c)). Most significant were correlations between higher VAT-KIT and lower expression of CD68 (*r*(%) = −0.559, *p* = 0.001, *n* = 32), and between higher VAT-CMA1 and lower expression of Mac2 (*r*(%) = −0.448, *p* = 0.010, *n* = 32). Consistently, AT-MC genes' expression in VAT negatively associated with the number of macrophage crown-like structures (CLS) in validation cohort 2 (*r*(%) = −0.294, *p* = 0.047, *n* = 56). Jointly, these analyses demonstrate (1) that AT-MC accumulation can be estimated confidently by histology and gene expression, (2) that histologically, AT-MC may correlate with the abundance of macrophages only within fibrotic regions. By gene expression, AT-MC do not positively correlate, and may even exhibit a negative association with the total abundance of AT macrophages.

## *3.2. Cross-Sectional Analyses*

Next, we wished to challenge our hypothesis that higher AT-MC accumulation would signify an obese sub-phenotype characterized by worse cardiometabolic risk. For this purpose, we used the larger Beer-Sheva cohort of *n* = 65 patients with obesity to compare those with obesity and high versus low VAT-MC accumulation. Given the suspicion that KIT may not be sufficiently MC-specific [9], but that c-Kit and TPSB2 largely identify the same cells in the tissue (Figure 1D), and that AT-MC express TPSB2−/+ CMA1 [11], we based MC gene expression stratification on the combined VAT expression levels of TPSB2 and CMA1: Patients were categorized as VAT-MClow only if the AT expression of both TPSB2 and CMA1 genes was below the median level of expression in the entire cohort, whereas VAT-MChigh were all those with above-median gene expression in either or both genes.

Contrary to our hypothesis, patients with obesity and low VAT-MC gene expression (below median expression of both TPSB2 and CMA1) were either not significantly different, or in some parameters trended to, or exhibited significantly worse clinical parameters (Table 1, Figure 2A). Significant differences between patients with obesity and VAT-MClow and VAT-MChigh were observed in triglyceride (TG) levels and HDL (high-density lipoprotein) ratio, in circulating alkaline phosphatase and in HOMA-β. Sensitivity analysis in which outliers were excluded rendered the difference in HOMA-β between those with VAT-MClow versus VAT-MChigh insignificant. We could not attribute the differences to the different use of medications (Table S4). Validation cohorts 1 and 2 exhibited similar trends to those observed in the main cohort (Table 1), and such trends were less robust when defining VAT-MC accumulation based on the expression of KIT alone (Table S2). To further explore if this unexpected association between higher VAT-MC expression and seemingly improved clinical markers of cardiometabolic risk are contributed by specific sub-groups of patients, the analysis was repeated after stratification of the main cohort by either sex, age (above/below median of 45 years old), T2DM status, and obesity class (Figure 2 and Figure S5, respectively). Remarkably, significant differences exhibiting improved parameters in VAT-MChigh were more readily observed in females (Figure S5A), in participants whose age was above median (Figure 2B), in patients with type 2 diabetes mellitus

(T2DM) (Figure 2C), and in those with BMI 30.0–39.9 kg/m<sup>2</sup> (Figure S5B). Intriguingly, a significant *p* interaction was observed between diabetes status and VAT-MC in FPG and in HbA1c (*p* = 0.016 and *p* = 0.041, respectively). This could not be attributed to different diabetes duration or different use of medications (Tables S4 and S5). *Cells* **2020**, *9*, x 3 of 18

**Figure 2.** Comparison of clinical parameters between participants with high versus low MC accumulation in VAT. MClow (white bars) were defined as people in whom the expression of both *TPSB2* and *CMA1* were below the median value of the cohort for each of the genes; all others were defined as MChigh (i.e., with either one or two of the genes expressed above median, gray bars). (**A**) Differences in triglycerides (TG)/HDL ratio, alkaline phosphatase, FPG, Hemoglobin A1c (HbA1c), and HOMA-β between VAT- MClow versus VAT- MChigh. (**B**,**C**) Same analysis as in **A**, but the cohort was stratified by age (below/above median = 45 years, B), or type 2 diabetes status (**C**). Age-adjusted 2 by 2 ANOVA was used to test for variance. Age-adjusted Least Significant Difference (LSD) posthoc test was used to compare between groups. Values are expressed as mean ± SD. \* *p* < 0.05, \*\* *p* < **Figure 2.** Comparison of clinical parameters between participants with high versus low MC accumulation in VAT. MClow (white bars) were defined as people in whom the expression of both *TPSB2* and *CMA1* were below the median value of the cohort for each of the genes; all others were defined as MChigh (i.e., with either one or two of the genes expressed above median, gray bars). (**A**) Differences in triglycerides (TG)/HDL ratio, alkaline phosphatase, FPG, Hemoglobin A1c (HbA1c), and HOMA-β between VAT- MClow versus VAT- MChigh. (**B**,**C**) Same analysis as in **A**, but the cohort was stratified by age (below/above median = 45 years, B), or type 2 diabetes status (**C**). Age-adjusted 2 by 2 ANOVA was used to test for variance. Age-adjusted Least Significant Difference (LSD) post-hoc test was used to compare between groups. Values are expressed as mean ± SD. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001.

0.01, \*\*\* *p* < 0.001. Indeed, by correlation (rather than group statistics) analysis, a significant negative association was observed between VAT-CMA1 expression level and FPG or HbA1c among the sub-group of patients with obesity and T2DM (association with TPSB2 exhibited a similar trend) (Figure 3A,B, respectively). Importantly, this association was also evident in the two independent Leipzig cohorts (Figure 3C,D). In SAT, associations between the expression of MC genes and clinical parameters in the 3 cohorts is presented in Table S6, displaying less consistent cross-sectional associations than those observed with VAT-MC. Jointly, these cross-sectional analyses disproved our initial hypothesis that increased VAT-MC accumulation would signify a worse obesity sub-phenotype. Rather. they provided evidence that the reverse association may in fact hold true. This was particularly apparent in certain sub-groups of patients with obesity, such as females, older patients, in those whose obesity was also complicated with T2DM, and in patients with obesity class I + II. Indeed, by correlation (rather than group statistics) analysis, a significant negative association was observed between VAT-CMA1 expression level and FPG or HbA1c among the sub-group of patients with obesity and T2DM (association with TPSB2 exhibited a similar trend) (Figure 3A,B, respectively). Importantly, this association was also evident in the two independent Leipzig cohorts (Figure 3C,D). In SAT, associations between the expression of MC genes and clinical parameters in the 3 cohorts is presented in Table S6, displaying less consistent cross-sectional associations than those observed with VAT-MC. Jointly, these cross-sectional analyses disproved our initial hypothesis that increased VAT-MC accumulation would signify a worse obesity sub-phenotype. Rather. they provided evidence that the reverse association may in fact hold true. This was particularly apparent in certain sub-groups of patients with obesity, such as females, older patients, in those whose obesity was also complicated with T2DM, and in patients with obesity class I + II.

**Figure 3.** Associations between VAT CMA1 gene expression and glycemic parameters in the Beer– Sheva and Leipzig independent cohorts. Spearman's associations between VAT CMA1 gene expression and FPG (**A**) or HbA1c (**B**) among the Beer–Sheva cohort (*n* = 65, all with body mass index ≥ 30 kg/m2), stratified into those with normal glucose homeostasis (normoglycemic, blue circle), prediabetes (orange square), or type 2 diabetes (red triangle). *r* denotes Spearman's rank correlation coefficient (Rho, ρ) among participants with type 2 diabetes. (**C**) Spearman's rank correlation between log-transformed CMA1 expression in VAT assessed by microarray and HbA1c in the Leipzig validation cohort 1 (*n* = 32, all with BMI ≥ 30 kg/m2). (**D**) Spearman's rank correlation between CMA1 expression in VAT and FPG in Leipzig validation cohort 2 (*n* = 56, all with BMI ≥ 30 kg/m2). **Figure 3.** Associations between VAT CMA1 gene expression and glycemic parameters in the Beer–Sheva and Leipzig independent cohorts. Spearman's associations between VAT CMA1 gene expression and FPG (**A**) or HbA1c (**B**) among the Beer–Sheva cohort (*<sup>n</sup>* <sup>=</sup> 65, all with body mass index <sup>≥</sup> 30 kg/m<sup>2</sup> ), stratified into those with normal glucose homeostasis (normoglycemic, blue circle), prediabetes (orange square), or type 2 diabetes (red triangle). *r* denotes Spearman's rank correlation coefficient (Rho, %) among participants with type 2 diabetes. (**C**) Spearman's rank correlation between log-transformed CMA1 expression in VAT assessed by microarray and HbA1c in the Leipzig validation cohort 1 (*n* = 32, all with BMI <sup>≥</sup> 30 kg/m<sup>2</sup> ). (**D**) Spearman's rank correlation between CMA1 expression in VAT and FPG in Leipzig validation cohort 2 (*<sup>n</sup>* <sup>=</sup> 56, all with BMI <sup>≥</sup> 30 kg/m<sup>2</sup> ).

#### *3.3. In Vitro Analyses 3.3. In Vitro Analyses*

The results described so far suggest that increased VAT-MC infiltration may cross-sectionally associate with a more favorable obese sub-phenotype, but the underlying mechanism remains obscure. To this end, given the centrality of the VAT–liver access in metabolic health/dysfunction in obesity, particularly when complicated with diabetes, we hypothesized that secreted factors from VAT-MChigh AT mediate a more metabolically favorable communication with liver cells compared to VAT-MClow AT. To test this hypothesis, we treated human hepatocyte-derived cells with conditioned media (CM) obtained from the VAT of obese people with either high or low MC gene expression and assessed the ensuing acute signaling response to insulin stimulation (Figure 4A). HepG2 cells treated with CM from VAT-MChigh were more insulin-responsive compared to cells treated with CM from VAT-MClow, as determined by the insulin-induced phosphorylation level of GSK3 (Figure 4B,C). The results described so far suggest that increased VAT-MC infiltration may cross-sectionally associate with a more favorable obese sub-phenotype, but the underlying mechanism remains obscure. To this end, given the centrality of the VAT–liver access in metabolic health/dysfunction in obesity, particularly when complicated with diabetes, we hypothesized that secreted factors from VAT-MChigh AT mediate a more metabolically favorable communication with liver cells compared to VAT-MClow AT. To test this hypothesis, we treated human hepatocyte-derived cells with conditioned media (CM) obtained from the VAT of obese people with either high or low MC gene expression and assessed the ensuing acute signaling response to insulin stimulation (Figure 4A). HepG2 cells treated with CM from VAT-MChigh were more insulin-responsive compared to cells treated with CM from VAT-MClow, as determined by the insulin-induced phosphorylation level of GSK3 (Figure 4B,C).

*Cells* **2020**, *9*, x 5 of 18

**Figure 4.** Insulin signaling in HepG2 cells after treatment with condition media from the VAT of obese MC low/high. (**A**) Schematic experimental design: VAT explants were incubated in media for 24 h for the preparation of conditioned media (CM). CM was used to expose human hepatocyte cell line (HepG2) for 24 h, followed by insulin (100 nM) stimulation for 7 min. (**B**) Representative blots of Western blot analysis for the following antibodies: p(Ser473)Akt, tAkt, p(Ser21/9)GSK3, and tGSK3. (**C**) Expression quantification of the proteins above (**B**) from two independent experiments, *n* = 4 for each group. Mann–Whitney non-parametric test was performed in order to compare means. Values are expressed as mean ± standard error of the mean (SEM), \* *p* < 0.05. **Figure 4.** Insulin signaling in HepG2 cells after treatment with condition media from the VAT of obese MClow/high. (**A**) Schematic experimental design: VAT explants were incubated in media for 24 h for the preparation of conditioned media (CM). CM was used to expose human hepatocyte cell line (HepG2) for 24 h, followed by insulin (100 nM) stimulation for 7 min. (**B**) Representative blots of Western blot analysis for the following antibodies: p(Ser473)Akt, tAkt, p(Ser21/9)GSK3, and tGSK3. (**C**) Expression quantification of the proteins above (**B**) from two independent experiments, *n* = 4 for each group. Mann–Whitney non-parametric test was performed in order to compare means. Values are expressed as mean ± standard error of the mean (SEM), \* *p* < 0.05.

#### *3.4. Prospective Analyses 3.4. Prospective Analyses*

Beyond the potential to utilize the estimation of MC accumulation of VAT for sub-typing obese patients in a cross-sectional analysis, we aimed to determine in a subset of patients from the main cohort, for whom postbariatric surgery data was already available, whether the expression level of MC genes in VAT can predict clinically meaningful outcomes of intervention. Using the same approach to assess the VAT-MC accumulation level based on both TPSB2 and CMA1, patients with VAT-MChigh (above median value of the entire cohort) lost significantly more weight 6 months after bariatric surgery compared to VAT-MClow (31.7 ± 1.6% versus 13.4 ± 6.5% of baseline body weight, corresponding to 79.1 ± 5.7% and 30.7 ± 14.5% loss of excess body weight, respectively, *p* = 0.002). For this analysis, we found a yet stronger association when using VAT-CMA1 expression only (Figure 5A,B): People with obesity and high VAT-CMA1 expression achieved 1.74-fold greater weight reduction compared to obese people with low VAT-CMA1 expression 6 months post-surgery (*p* = 0.016). This difference remained significant (*p* = 0.038) even when the two patients that showed minimal weight loss were excluded. Different weight losses corresponded to a 1.9-fold greater loss of excess body weight among the VAT-CMA1High compared to the VAT-CMA1Low, *p* = 0.008, Figure 5C). The difference in weight reduction between the two groups remained significant even when adjusting to either baseline BMI or surgery type (Figure 5D). Beyond the potential to utilize the estimation of MC accumulation of VAT for sub-typing obese patients in a cross-sectional analysis, we aimed to determine in a subset of patients from the main cohort, for whom post-bariatric surgery data was already available, whether the expression level of MC genes in VAT can predict clinically meaningful outcomes of intervention. Using the same approach to assess the VAT-MC accumulation level based on both TPSB2 and CMA1, patients with VAT-MChigh (above median value of the entire cohort) lost significantly more weight 6 months after bariatric surgery compared to VAT-MClow (31.7 <sup>±</sup> 1.6% versus 13.4 <sup>±</sup> 6.5% of baseline body weight, corresponding to 79.1 ± 5.7% and 30.7 ± 14.5% loss of excess body weight, respectively, *p* = 0.002). For this analysis, we found a yet stronger association when using VAT-CMA1 expression only (Figure 5A,B): People with obesity and high VAT-CMA1 expression achieved 1.74-fold greater weight reduction compared to obese people with low VAT-CMA1 expression 6 months post-surgery (*p* = 0.016). This difference remained significant (*p* = 0.038) even when the two patients that showed minimal weight loss were excluded. Different weight losses corresponded to a 1.9-fold greater loss of excess body weight among the VAT-CMA1High compared to the VAT-CMA1Low, *p* = 0.008, Figure 5C). The difference in weight

Intriguingly, people with obesity and low VAT-CMA1 had higher levels of FPG (Figure 5E), HbA1c (Figure 5F), and triglycerides (Figure 5G) pre-operation, and despite lower extent of weight loss, exhibited a greater

reduction between the two groups remained significant even when adjusting to either baseline BMI or surgery type (Figure 5D). Intriguingly, people with obesity and low VAT-CMA1 had higher levels of FPG (Figure 5E), HbA1c (Figure 5F), and triglycerides (Figure 5G) pre-operation, and despite lower extent of weight loss, exhibited a greater reduction, particularly in triglycerides, 6 months postoperatively. Yet, unlike with the weight change, these differences could be explained by their higher baseline values (data not shown). We did not observe any additional difference in the other clinical parameters that reached statistical significance (data not shown). *Cells* **2020**, *9*, x 6 of 18 reduction, particularly in triglycerides, 6 months postoperatively. Yet, unlike with the weight change, these differences could be explained by their higher baseline values (data not shown). We did not observe any additional difference in the other clinical parameters that reached statistical significance (data not shown).

**Figure 5.** Six months postoperative weight reduction and metabolic changes comparison between people with VAT CMA1 high or CMA1 low. (**A**) Body weight change between operation day (0) and 6 months post-surgery in persons with low expression (i.e., below median) of CMA1 (*n* = 6, circles) or high CMA1 mRNA levels (n = 12, squares). Percentage of weight loss (**B**) from preoperative weight and of excess weight loss six months following bariatric surgery (**C**). A paired t-test was used to compare between pre-operation (0) and 6 months post-surgery (6). A Mann–Whitney non-parametric test was performed in order to compare means between CMA1 low and CMA1 high. \* *p* < 0.05, \*\* *p* < 0.01. (**D**) Multi-variant model for association between VAT CMA1 groups and 6 months weight changes as dependent variable, adjusted to baseline BMI and surgery type. Change in FPG (**E**), HbA1c (**F**), and triglycerides (**G**) between operation day (0) and 6 months post-surgery in persons with low expression CMA1 or high CMA1 mRNA levels. The percentage of change in FPG, HbA1c, and triglycerides (insets). Values are expressed as mean ± standard error of the mean (SEM). **Figure 5.** Six months postoperative weight reduction and metabolic changes comparison between people with VAT CMA1 high or CMA1 low. (**A**) Body weight change between operation day (0) and 6 months post-surgery in persons with low expression (i.e., below median) of CMA1 (*n* = 6, circles) or high CMA1 mRNA levels (n = 12, squares). Percentage of weight loss (**B**) from preoperative weight and of excess weight loss six months following bariatric surgery (**C**). A paired t-test was used to compare between pre-operation (0) and 6 months post-surgery (6). A Mann–Whitney non-parametric test was performed in order to compare means between CMA1 low and CMA1 high. \* *p* < 0.05, \*\* *p* < 0.01. (**D**) Multi-variant model for association between VAT CMA1 groups and 6 months weight changes as dependent variable, adjusted to baseline BMI and surgery type. Change in FPG (**E**), HbA1c (**F**), and triglycerides (**G**) between operation day (0) and 6 months post-surgery in persons with low expression CMA1 or high CMA1 mRNA levels. The percentage of change in FPG, HbA1c, and triglycerides (insets). Values are expressed as mean ± standard error of the mean (SEM).

To strengthen the limited observation achievable with our main cohort, we utilized validation cohort 2 of the Leipzig bio-bank, in which 1 year post-laparoscopic sleeve gastrectomy data was available (see details of this 2-step bariatric surgery cohort in the Methods section). As expected, there were significant declines in weight and anthropometric measures, in glycemic and insulin resistance indices, and in lipids (increase in HDL) (Table 3). To strengthen the limited observation achievable with our main cohort, we utilized validation cohort 2 of the Leipzig bio-bank, in which 1 year post-laparoscopic sleeve gastrectomy data was available (see details of this 2-step bariatric surgery cohort in the Methods section). As expected, there were significant declines in weight and anthropometric measures, in glycemic and insulin resistance indices, and in lipids (increase in HDL) (Table 3).


**Table 3.** Changes in clinical and inflammatory markers and AT genes, 12 months following bariatric surgery in validation cohort 2. Values are mean ± SD. CLS—crown-like structures, VAT—visceral adipose tissue, SAT—subcutaneous adipose tissue. ns—not significant (*p* > 0.05).

Among circulating cytokines tested, IL-6 robustly declined (*p* < 0.0001), and a significant reduction was observed in CRP levels (*p* = 0.024). Several inflammatory markers were measured in both VAT and SAT. In VAT, both macrophage CLS number and IL1b expression decreased significantly (*p* < 0.0001 and *p* = 0.003, respectively). Among MC genes, only CMA1 expression exhibited a significant expression reduction (delta of 0.3 ± 1.0, *p* = 0.044). Similar results were achieved in SAT, with significant reductions in macrophages CLS and IL1b (*p* = 0.005, *p* = 0.035, respectively), but with no significant change in CMA1 expression but with a reduction in TGFbeta expression (*p* = 0.024). Next, we looked for associations between baseline VAT and SAT MC genes expression and selected inflammatory markers and changes in anthropometric parameters 1 year following bariatric surgery (Figure 6). In this cohort of extremely obese patients (average: BMI > 50 kg/m<sup>2</sup> , Table 1), the baseline expression of KIT and TPSB2 in VAT seemed to be better predictors of greater weight loss 1 y postoperatively, reaching statistically significant associations (*r*(%) = 0.295, *p* = 0.044, *r*(%) = 0.313, *p* = 0.03, respectively). Baseline VAT-KIT was also associated with BMI loss (*p* = 0.04) and trended with percentage of weight loss (*r*(%) = 0.283, *p* = 0.054). This is somewhat consistent with the results of the main cohort of less obese patients (Table 1); baseline VAT-CMA1 in validation cohort 2 trended to associate with a greater loss in BMI 1 y after surgery (*r*(%) = 0.229, *p* = 0.099). In addition, baseline VAT-TGFbeta was also associated with a greater reduction in WC 1 year postoperatively (*r*(%) = 0.284, *p* = 0.048).

surgery.

weight loss following surgery.

Jointly, these results provide proof-of-principle for VAT and SAT MC and inflammatory gene expression

SAT-CMA1 associated with greater weight, waist circumference (WC), and BMI loss (*r*(ρ) = 0.328, *p* = 0.015, *r*(ρ) = 0.354, *p* = 0.015 and *r*(ρ) = 0.296, *p* = 0.030, respectively). Baseline SAT-KIT similarly associated with greater WC loss (*p* = 0.045). Intriguingly, higher SAT-TNFalpha expression highly associated with greater weight, percentage weight, and BMI losses (*r*(ρ) = 0.400, *p* = 0.004, *r*(ρ) = 0.319, *p* = 0.024 and *r*(ρ) = 0.391, *p* = 0.005, respectively), as did a higher baseline expression of SAT-IL1beta (association with WC decline—*r*(ρ) = 0.321, *p* = 0.028). Baseline macrophages CLS numbers, in both VAT and SAT, and circulating cytokines were not significantly associated with a reduction in any anthropometric parameters. Moreover, although non-significantly, they exhibit negative *r*(ρ) coefficients with 1 year postoperative weight loss measures, which is opposite to the associations (both statistically significant or non-significant) between higher baseline VAT or SAT-MC gene expression and greater

**Figure 6.** Correlation between baseline parameters and 1 year postoperative weight loss response in validation cohort 2. Values are *r*(ρ) of Spearman's correlations. WC: waist circumference. \* *p* < 0.05; \*\* *p* < 0.01. **Figure 6.** Correlation between baseline parameters and 1 year postoperative weight loss response in validation cohort 2. Values are *r*(%) of Spearman's correlations. WC: waist circumference. \* *p* < 0.05; \*\* *p* < 0.01.

**4. Discussion**  In this study, we hypothesized that AT-MC accumulation may aid in characterizing obesity subphenotypes, and we addressed this hypothesis by cross-sectional and prospective/predictive analyses of three independent cohorts of patients, all with obesity. Our results suggest the following: (1) We anticipated that greater AT-MC accumulation of VAT would characterize a more adverse obese phenotype as determined by cardiometabolic risk parameters. This hypothesis could be clearly rejected by the data. Moreover, contrasting our initial hypothesis, we found that MC numbers and MC-related gene expression in omental fat (VAT) associated in 3 independent cohorts with a better cardiometabolic risk profile, associations that are not consistently observed with SAT. Conducting post-hoc exploratory analyses, this association was strengthened when stratifying the larger cohort by basic clinical sub-groups, such as sex, age, and type 2 diabetes status. In the latter group, high VAT-MC gene expression associated with a metabolically healthier phenotype compared to those with low MC gene expression pattern. In some parameters, patients with obesity and diabetes, and with high MC gene expression in VAT were indistinguishable from patients with obesity without diabetes. (2) In vitro studies suggest that VAT with high MC gene expression communicates more favorably with liver-derived cells, Interestingly, unlike the cross-sectional analyses (Section 3.2), in which VAT-MC exhibited more robust associations with patients' clinical characteristics than SAT-MC (Figure 3 and Figure S5 and Table S6), SAT-MC genes and inflammatory markers exhibited more associations with 1 year postoperative outcomes (Figure 6). Higher SAT-CMA1 associated with greater weight, waist circumference (WC), and BMI loss (*r*(%) = 0.328, *p* = 0.015, *r*(%) = 0.354, *p* = 0.015 and *r*(%) = 0.296, *p* = 0.030, respectively). Baseline SAT-KIT similarly associated with greater WC loss (*p* = 0.045). Intriguingly, higher SAT-TNFalpha expression highly associated with greater weight, percentage weight, and BMI losses (*r*(%) = 0.400, *p* = 0.004, *r*(%) = 0.319, *p* = 0.024 and *r*(%) = 0.391, *p* = 0.005, respectively), as did a higher baseline expression of SAT-IL1beta (association with WC decline—*r*(%) = 0.321, *p* = 0.028). Baseline macrophages CLS numbers, in both VAT and SAT, and circulating cytokines were not significantly associated with a reduction in any anthropometric parameters. Moreover, although non-significantly, they exhibit negative *r*(%) coefficients with 1 year postoperative weight loss measures, which is opposite to the associations (both statistically significant or non-significant) between higher baseline VAT or SAT-MC gene expression and greater weight loss following surgery.

rendering them more insulin-responsive compared to cells exposed to VAT with low MC gene expression. (3) We observed that analyzing VAT-MC gene expression in samples obtained during bariatric surgery can provide predictive information on the degree of weight loss and metabolic response to bariatric surgery: high pre-Jointly, these results provide proof-of-principle for VAT and SAT MC and inflammatory gene expression as putative molecular predictors of weight loss and metabolic improvement six months to 1 y following bariatric surgery.

#### operative VAT CMA1 or KIT expression predicted significantly greater weight loss 6 months or 1 year postoperatively, respectively. Unlike in the cross-sectional analyses, higher SAT-MC gene expression also **4. Discussion**

correlated with greater weight-loss response to bariatric surgery. Whether AT MC accumulation plays a role in obesity is still debatable. In mice, results are contradictory depending on the models used to generate MC deficiency [6,8–10]. Moreover, MC depletion via c-kit prevented obesity development [6,8–10], and therefore it may not contribute to uncovering MC's role when obesity is established, and how it relates to the development of obesity-associated cardiometabolic morbidity. In humans, particular interest in obesity sub-typing/sub-phenotyping is emerging [20], since with its high prevalence, there In this study, we hypothesized that AT-MC accumulation may aid in characterizing obesity sub-phenotypes, and we addressed this hypothesis by cross-sectional and prospective/predictive analyses of three independent cohorts of patients, all with obesity. Our results suggest the following: (1) We anticipated that greater AT-MC accumulation of VAT would characterize a more adverse obese phenotype as determined by cardiometabolic risk parameters. This hypothesis could be clearly rejected by the data. Moreover, contrasting our initial hypothesis, we found that MC numbers and MC-related gene expression in omental fat (VAT) associated in 3 independent cohorts with a better cardiometabolic risk profile, associations that are not consistently observed with SAT. Conducting post-hoc exploratory analyses, this association was strengthened when stratifying the larger cohort by basic clinical sub-groups, such as sex, age, and type 2 diabetes status. In the latter group, high VAT-MC gene expression associated with a metabolically healthier phenotype compared to those with low MC gene expression pattern. In some parameters, patients with obesity and diabetes, and with high MC gene expression in VAT were indistinguishable from patients with obesity without diabetes. (2) In vitro studies suggest that VAT with high MC gene expression communicates more favorably with liver-derived cells, rendering them more insulin-responsive compared to cells exposed to VAT with low MC gene expression. (3) We observed that analyzing VAT-MC gene expression in samples

obtained during bariatric surgery can provide predictive information on the degree of weight loss and metabolic response to bariatric surgery: high pre-operative VAT CMA1 or KIT expression predicted significantly greater weight loss 6 months or 1 year postoperatively, respectively. Unlike in the cross-sectional analyses, higher SAT-MC gene expression also correlated with greater weight-loss response to bariatric surgery.

Whether AT MC accumulation plays a role in obesity is still debatable. In mice, results are contradictory depending on the models used to generate MC deficiency [6,8–10]. Moreover, MC depletion via c-kit prevented obesity development [6,8–10], and therefore it may not contribute to uncovering MC's role when obesity is established, and how it relates to the development of obesity-associated cardiometabolic morbidity. In humans, particular interest in obesity sub-typing/sub-phenotyping is emerging [20], since with its high prevalence, there is a pressing need to increase the identification of obesity subtypes that may require more intensive treatment. Compared to lean people, persons with obesity exhibit higher numbers of MC both in VAT and SAT [6,11], especially if obesity was accompanied with type 2 diabetes mellitus [11]. Compared to the study by Divoux et al. that examined persons with obesity with/without type 2 diabetes (albeit with a limited *n* = 10 in each group) [11], we could not detect a clear increase in MC parameters in VAT from persons with diabetes compared to those without (*n* = 40 and 25, respectively). Our findings suggest that a larger cohort than that previously analyzed may have been required to uncover heterogeneity within the group of patients with obesity and type 2 diabetes. Indeed, associations between VAT-MC and metabolic parameters were mainly discernable among those with diabetes and higher expression levels of MC genes in VAT. In those patients, VAT-MC gene expression was associated with a healthier metabolic phenotype, suggesting a possible compensatory increase in VAT-MC in response to metabolic impairment, which, differently than previously suspected, exerts protective effects on metabolic health.

This work has some limitations. Our main cohort's size (*n* = 65), though the largest (to our knowledge) used to examine in humans AT-MC, is still relatively small, and the sub-analysis for predicting response to surgery includes an even smaller sub-cohort (*n* = 18). In addition, the high HOMA-β values may indicate the possibility that we investigated a specific obesity sub-group with relatively high beta cell reserve. Yet, we confirmed the key findings in 2 independent cohorts, using different approaches to estimate AT-MC gene expression. Findings were opposite to our pre-defined hypothesis that increased MC accumulation would associate with a worse metabolic outcome, and subsequent post-hoc analyses are more prone to biases and type 2 error and should best be viewed as hypothesis-generating observations. Yet, the original hypothesis was clearly rejected, so MC accumulation, at least in VAT, is not associated with a worse metabolic phenotype. Regarding the estimation of MC accumulation, we did not use flow-cytometry methods that would have enabled characterizing MC sub-types, and potentially their activation state. Rather, we critically assessed the use of several, nicely intercorrelated, "MC-specific genes", defining the AT-MClow sub-group stringently, using combined thresholds for two genes (TPSB2 and CMA1) that would capture the two MC sub-populations. We also verified gene expression agreement with immunohistochemical assessment of AT-MC accumulation. Therefore, we cannot draw conclusions about MC activation state, including a possible association with LDL particles—which are known regulators of MC activity [21], although most of the associations we observed were between AT-MC accumulation and triglycerides, not with LDL or total cholesterol levels. Limiting the clinical implementation of our results is our use of VAT measures as a biomarker, since this AT is not accessible for sampling in the regular clinical setting. People with obesity have been shown to have higher levels of serum tryptase compared to lean people [6], and a positive association has been noted with BMI [22]. Yet, the source for serum tryptase is not restricted to MC in VAT, and therefore it is likely a poor correlate of visceral AT-MC accumulation. Therefore, future studies are required to assess whether MC-derived, circulating blood biomarkers, such as serum tryptase and/or chymase levels, could be clinically used to estimate visceral AT-MC accumulation and obesity sub-types. Our results suggest that clinically meaningful information may be obtained for those undergoing abdominal surgery for obesity management, by assessing tissues

obtained during bariatric surgery molecularly and/or histopathologically. Although such analyses could be predictive of post-operative endpoints and may guide post-operative care, before obtaining more supportive evidence by other groups, it may still be premature to propose adipose tissue MC accumulation as a clinical predictor for the outcome of a bariatric surgery.

Although speculative, the data suggest putative mechanisms to explain the seemingly-positive effect of VAT-MC. Although our study was not focused on providing a systematic analysis of adipose tissue fibrosis, we report that in VAT, MC also populate fibrotic areas, and their gene expression correlates with several AT collagens (in SAT, correlations were only with COL6A1). Yet, recent studies suggest that in certain contexts, MC can exert anti-fibrotic/collagen-degrading effects that may result in a protective, rather than a pathogenic, role [23,24]. In addition, while AT fibrosis in mice is largely thought to contribute to tissue dysfunction [25,26], in humans, the pathological role of AT fibrosis may be more complex: higher SAT fibrosis may associate with poor metabolic and weight loss response to bariatric surgery [27], but VAT fibrosis associates with smaller adipocytes, potentially by limiting visceral adipocyte hypertrophy, thereby mediating better metabolic profile [28,29]. Additionally, MC secrete prostaglandins including 15-deoxy-delta PGJ2, which is an endogenous ligand for PPARγ, particularly in response to high-glucose conditions, resulting in increased adipogenesis [8]. Such MC-mediated PPARγ activation may support "healthy" AT expansion [30]. Indeed, the expression of CPA3, an MC marker, is associated with Uncoupling Protein 1 (UCP1) in the SAT of lean people, and MC-related histamine and IL-4 induce UCP1 expression in adipocytes, promoting AT beiging [31,32]. MC also contribute to angiogenesis, as demonstrated in cancer [33,34]. Thus, AT-MC adjacent to micro-vessels may imply their involvement in increased angiogenesis, possibly further supporting healthy AT expansion by limiting hypoxia. Finally, we show that AT-MC correlate with macrophages only in fibrotic areas, and at the whole-tissue level, MC genes may in fact inversely correlate with macrophage-specific genes or with macrophage CLS numbers. This suggests that higher MC accumulation in VAT might in fact associate with lower macrophage infiltration/activity, which in turn is indicative of a lower inflammatory burden-related adipose tissue dysfunction. Furthermore, MChigh exhibited greater postoperative weight loss. A putative explanation might be high preoperative fasting insulin that enabled greater postoperative decline in insulin levels, thereby contributing to more substantial weight loss.

## **5. Conclusions**

Clinically used methods for better phenotyping people with obesity are still insufficient and greatly limit stratified or more precision obesity management. Our study suggests that VAT-MC accumulation estimates could be used as a tool for obesity sub-phenotyping. This can rely on relatively available laboratory procedures, such as molecular (real-time PCR) and/or histopathological assays using clinically common, known cell markers. Moreover, we propose that molecular and/or histological examinations of tissues obtained during surgery can uncover clinically important information, such as the prediction of post bariatric-surgery outcomes. This could aid in the post-operative management to optimize patient care in an individualized manner, similar to the common practice in the management of other diseases, such as cancer. Jointly, we propose that adipose tissue composition holds clinically relevant information that should be better exploited for improving the care provided to people with obesity.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/9/6/1508/s1, Figure S1: The expression of KIT in cells that can be found in human AT, Figure S2: The expression of TPSB2 in cells that can be found in human AT, Figure S3: The expression of CMA1 in cells that can be found in human AT, Figure S4: SAT C-Kit+ cells within fibrotic area and fibrosis grading. Figure S5: Comparison of clinical parameters between participants with high versus low MC accumulation in VAT, stratified by sex or obesity category. Table S1: PCR probe ID. Table S2: Clinical characteristics of participants with obesity from the Beer–Sheva and Leipzig cohorts, each also stratified by VAT-MC low/high based on expression of KIT. Table S3: Inter-correlation between SAT-MC gene expression and SAT-collagens. Table S4: Usage of medications in persons with VAT MChigh and MClow among the entire main cohort. Table S5: Diabetes duration and medication among persons with type

2 diabetes stratified to VAT MChigh and MClow, main cohort. Table S6: Association between SAT-MC genes expression and clinical parameters in the three cohorts.

**Author Contributions:** N.G. conceived and designed the study, obtained the data, proposed and performed the statistical analyses, conducted the literature search, drafted the report, and reviewed/edited the manuscript. Y.K. and R.S.-L. obtained the histopathological data and provided technical and professional support. Y.G. proposed and performed the statistical analyses and reviewed the manuscript. Y.H. contributed to the literature search and reviewed the manuscript. T.P. contributed to the literature search and study design. R.G. contributed bioinformatics analyses and central intellectual content. V.P., I.F.L., and B.K. recruited patients, obtained clinical data and samples, and reviewed the manuscript. M.B. provided the clinical data of the Leipzig cohorts, reviewed/edited the manuscript and revised the report for important intellectual content. A.R. conceived the study, reviewed and analyzed the data, and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by grants from the DFG (German Research Foundation)—Projektnummer 209933838—SFB 1052 (specific projects: B2 to A.R., project B1 to M.B.), and the Israel Science Foundation (ISF 928/14 and 2176/19 to A.R.).

**Conflicts of Interest:** The authors declare no conflict of interest

## **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Physiological Oxygen Levels Differentially Regulate Adipokine Production in Abdominal and Femoral Adipocytes from Individuals with Obesity Versus Normal Weight**

**Ioannis G. Lempesis 1,2,3,\* , Nicole Hoebers <sup>2</sup> , Yvonne Essers <sup>2</sup> , Johan W. E. Jocken <sup>2</sup> , Kasper M. A. Rouschop <sup>4</sup> , Ellen E. Blaak <sup>2</sup> , Konstantinos N. Manolopoulos 1,3,† and Gijs H. Goossens 2,\* ,†**


**Abstract:** Adipose tissue (AT) inflammation may increase obesity-related cardiometabolic complications. Altered AT oxygen partial pressure (pO<sup>2</sup> ) may impact the adipocyte inflammatory phenotype. Here, we investigated the effects of *physiological* pO<sup>2</sup> levels on the inflammatory phenotype of abdominal (ABD) and femoral (FEM) adipocytes derived from postmenopausal women with normal weight (NW) or obesity (OB). Biopsies were collected from ABD and FEM subcutaneous AT in eighteen postmenopausal women (aged 50–65 years) with NW (BMI 18–25 kg/m<sup>2</sup> , *n* = 9) or OB (BMI 30–40 kg/m<sup>2</sup> , *n* = 9). We compared the effects of prolonged exposure to different *physiological* pO<sup>2</sup> levels on adipokine expression and secretion in differentiated human multipotent adipose-derived stem cells. Low *physiological* pO<sup>2</sup> (5% O<sup>2</sup> ) significantly increased leptin gene expression/secretion in ABD and FEM adipocytes derived from individuals with NW and OB compared with high *physiological* pO<sup>2</sup> (10% O<sup>2</sup> ) and standard laboratory conditions (21% O<sup>2</sup> ). Gene expression/secretion of IL-6, DPP-4, and MCP-1 was reduced in differentiated ABD and FEM adipocytes from individuals with OB but not NW following exposure to low compared with high *physiological* pO<sup>2</sup> levels. Low *physiological* pO<sup>2</sup> decreases gene expression and secretion of several proinflammatory factors in ABD and FEM adipocytes derived from individuals with OB but not NW.

**Keywords:** adipose tissue; adipokines; inflammation; body fat distribution; obesity pathophysiology; hypoxia

## **1. Introduction**

Excess fat mass in obesity poses a major health risk [1]. The research in the past decades has clearly demonstrated that body fat distribution is a better predictor of cardiometabolic complications than total fat mass, with abdominal obesity increasing and lower-body (gluteofemoral) fat accumulation conferring relative protection against chronic cardiometabolic diseases [2–6]. This seems related to the distinct functional properties of these different AT depots. Many studies in rodents and humans have shown that AT dysfunction in obesity is characterised by adipocyte hypertrophy, mitochondrial dysfunction, reactive oxygen species (ROS) production, impaired lipid metabolism, reduced blood flow, and inflammation, together contributing to an increased risk of developing cardiometabolic diseases and cancer [6–11].

The AT microenvironment impacts metabolic and inflammatory processes [8,9]. We, and others, have previously demonstrated that AT oxygen partial pressure (pO2), which

**Citation:** Lempesis, I.G.; Hoebers, N.; Essers, Y.; Jocken, J.W.E.; Rouschop, K.M.A.; Blaak, E.E.; Manolopoulos, K.N.; Goossens, G.H. Physiological Oxygen Levels Differentially Regulate Adipokine Production in Abdominal and Femoral Adipocytes from Individuals with Obesity Versus Normal Weight. *Cells* **2022**, *11*, 3532. https://doi.org/10.3390/ cells11223532

Academic Editor: Javier Gómez-Ambrosi

Received: 4 October 2022 Accepted: 6 November 2022 Published: 8 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

is determined by the balance between local oxygen supply (determined by adipose tissue blood flow) and consumption (primarily mitochondrial oxygen consumption), may be an important determinant of the AT phenotype and whole-body insulin sensitivity [9,12–14]. Interestingly, differences in adipose tissue blood flow and/or adipose tissue oxygen consumption between individuals with normal weight and obesity, and between upper and lower AT depots, have previously been demonstrated [9,10,12,15,16]. Although AT pO<sup>2</sup> is reduced in rodent models of obesity [17–19], conflicting findings on AT pO<sup>2</sup> have been reported in humans [9,20–24]. We have previously shown that AT pO<sup>2</sup> was higher in individuals with obesity and was positively associated with AT gene expression of proinflammatory markers and whole-body insulin resistance [22,25]. Moreover, we found that AT pO<sup>2</sup> was lower in femoral than in abdominal subcutaneous AT in women with obesity [16].

The normal *physiological* range of AT pO<sup>2</sup> in human AT is ~3–11% O<sup>2</sup> (~23–84 mmHg) [9,21–23,25]. Therefore, the outcomes of experiments comparing the effects of pO<sup>2</sup> below and well above these *physiological* levels should be interpreted with caution, because the results may not directly translate to the human in vivo situation [9]. Several in vitro studies have demonstrated that the expression and secretion of many adipokines are sensitive to changes in pO<sup>2</sup> levels, as extensively reviewed [9,26]. Most of these studies have shown that acute exposure to severe, non-physiological hypoxia (1% O<sup>2</sup> for 1–24 h) induces a proinflammatory expression and secretion profile in (pre)adipocytes, while prolonged exposure to mild *physiological* hypoxia (5% O<sup>2</sup> for 14 days) seems to elicit a different adipokine expression/secretion profile [9,16,27]. Recently, we found that prolonged exposure to low *physiological* hypoxia decreased proinflammatory gene expression in abdominal and femoral adipocytes derived from women with obesity [16]. The metabolic and inflammatory responses to changes in the AT microenvironment may differ between individuals and AT depots. Thus, oxygen levels might exert distinct effects on AT function in people with different adiposity and in different AT depots. Importantly, however, studies investigating the impact of altered pO<sup>2</sup> levels on the inflammatory phenotype of adipocytes derived from people with normal weight and obesity are lacking.

Therefore, the aim of the present study was to investigate the impact of prolonged exposure to various *physiological* oxygen levels on gene expression and secretion of inflammatory factors within upper and lower body differentiated human multipotent adipose-derived stem (hMADS) cells derived from women with normal weight or obesity.

### **2. Materials and Methods**

### *2.1. Upper and Lower Body Adipose Tissue Biopsies*

Paired abdominal (ABD) and femoral (FEM) subcutaneous AT needle biopsies were obtained from eighteen postmenopausal women (aged 50–65 years) with normal weight (NW: BMI 18–25 kg/m<sup>2</sup> , *n* = 9) or obesity (OB: BMI 30–40 kg/m<sup>2</sup> , *n* = 9) (Table 1). The U.K. Health Research Authority National Health System Research Ethics Committee approved the present study (approval no. 18/NW/0392). Briefly, the biopsy specimens (up to ∼1 g) were collected 6 to 8 cm lateral from the umbilicus (ABD AT) and from the anterior aspect of the upper leg (FEM AT) under local anaesthesia (1% lidocaine) after an overnight fast. Samples were immediately rinsed with sterile saline, and visible blood vessels were removed with sterile tweezers. Isolation of hMADS cells followed, as described before [16].

#### *2.2. Human Primary Adipocyte Experiments*

Human multipotent abdominal (ABD) and femoral (FEM) adipose-derived stem cells, an established human white adipocyte model [28], were seeded at a density of 2000 cells/cm<sup>2</sup> and kept in proliferation medium for seven days. Thereafter, these cells were differentiated under different *physiological* O<sup>2</sup> levels (10% O2, high *physiological* pO2; 5% O2, low *physiological* pO2) [9,16,22,29] as well as standard laboratory conditions (room air, 21% O2) for 14 days. Gas mixtures were refreshed every 8 h (to maintain variation <0.1% O2), whereas the medium was refreshed three times per week.


**Table 1.** Subjects' characteristics.

BMI, body mass index; DBP, diastolic blood pressure; HOMA2 IR, Homeostasis Model Assessment 2 Insulin Resistance; SBP, systolic blood pressure. Data are mean ± SEM.

#### *2.3. Adipocyte Gene Expression*

Total RNA was extracted from hMADS cells using TRIzol reagent (Invitrogen, Breda, The Netherlands), and SYBR-Green-based real-time PCRs were performed to assess the gene expression of leptin, dipeptidyl-peptidase (DPP)-4, interleukin (IL)-6, plasminogen activator inhibitor (PAI)-1, adiponectin, tumour necrosis factor (TNF)α, and monocyte chemoattractant protein (MCP)-1; the adipocyte differentiation markers peroxisome proliferatoractivated receptor γ (PPARγ), CCAAT-enhancer binding protein α (C/EBPα), fatty acid synthase (FAS), and perilipin 1 (PLIN1); as well as the hypoxia markers glucose transporter 1 (GLUT1), Bcl-2 interacting protein 3 (BNIP3), and vascular endothelial growth factor A (VEGFA) using an iCycler (Bio-Rad, Veenendaal, The Netherlands). Results were normalised to 18S ribosomal RNA.

## *2.4. Adipokine Secretion*

The medium of the hMADS cells was collected over 24 h, from day 13 (after replacement of medium) to day 14 of differentiation, to determine the secretion of adipokines using high-sensitive ELISAs (leptin and DPP-4 from R&D Systems, Inc., Minneapolis, MN, USA; IL-6 and MCP-1 from Diaclone SAS, Besancon Cedex, France; adiponectin and PAI-1 from BioVendor–Laboratorni medicina a.s., Brno, Czech Republic). If necessary, samples were diluted with the dilution buffer provided by the manufacturer prior to the assay, which was performed in duplicates according to the manufacturer's instructions.

## *2.5. Statistical Analyses*

Data are presented as mean ± SEM. The effects of exposure to different oxygen levels on adipocyte gene expression and adipokine secretion were analysed using oneway ANOVA or the Friedman test when data were not normally distributed, followed by post hoc comparison using Student's paired t-tests or the Wilcoxon signed-rank test in case of skewed data. GraphPad Prism version 8 for Windows (GraphPad Software, San Diego, CA, USA) was used to perform statistical analyses. *p* < 0.05 was considered statistically significant.

### **3. Results**

## *3.1. The Effects of Oxygen Partial Pressure on Adipocyte Gene Expression*

The exposure of differentiated hMADS cells derived from ABD and FEM AT to different pO<sup>2</sup> levels induced distinct gene expression patterns. Specifically, exposure to low *physiological* pO<sup>2</sup> (5% O2) increased *leptin* expression compared with exposure to high *physiological* pO<sup>2</sup> (10% O2) or room air (21% O2) in differentiated ABD and FEM hMADS

derived from individuals with NW as well as OB (all *p* < 0.01, Figure 1A). Furthermore, low *physiological* pO<sup>2</sup> markedly reduced the gene expression of the proinflammatory factors *DPP-4* and *IL-6* in both ABD and FEM differentiated hMADS derived from donors with OB (all *p* < 0.01) but not NW compared with high *physiological* pO<sup>2</sup> (Figure 1B,C). Low *physiological* pO<sup>2</sup> levels did not significantly alter the gene expression of *PAI-1*, *TNFα*, or *MCP-1* in differentiated ABD and FEM hMADS derived from NW and OB individuals (Figure 1D–G), except for a modest but significant (*p* = 0.041) increase in adiponectin gene expression in FEM differentiated hMADS derived from individuals with obesity (Figure 1E). In addition, high *physiological* AT pO<sup>2</sup> (10% O2) increased the *PAI-1* (*p* = 0.005) and reduced the *adiponectin* expression (*p* = 0.010) in FEM differentiated hMADS derived from individuals with OB compared with those at 21% O<sup>2</sup> exposure. As expected, exposure to *physiological* oxygen levels, i.e., lower oxygen levels compared with standard laboratory conditions, increased the gene expression of the classical hypoxia markers *GLUT1* and *VEGFA*, and, to a lesser extent, increased that of *BNIP3* (Figure S1A–C). Furthermore, exposure to low *physiological* oxygen levels (5% O2) did not alter the gene expression of adipocyte differentiation markers compared with room air (21% O2) in differentiated hMADS derived from individuals with NW as well as OB (Supplemental Figure S1D–G). In the differentiated hMADS derived from individuals with OB, the gene expression of *PPARγ*, *C/EBPα*, and FAS was lower, and expression of *PLIN1* higher, following exposure to 5% compared with 10% O2. *Cells* **2022**, *11*, 3532 5 of 9

**Figure 1.** Adipokine and inflammatory markers' gene expression in hMADS cells following differentiation under different pO2s (21% vs. 10% vs. 5% O2) (*n* = 9 paired samples): (**A**) *leptin*, (**B**) *dipeptidyl-peptidase (DPP)-4*, (**C**) *interleukin (IL)-6*, (**D**) *plasminogen activator inhibitor (PAI)-1*, (**E**) *adiponectin*, (**F**) *tumour necrosis factor (TNF)α*, and (**G**) *monocyte chemoattractant protein (MCP)-1*. Data are expressed as mean ± SEM. \* *p* < 0.05. **Figure 1.** Adipokine and inflammatory markers' gene expression in hMADS cells following differentiation under different pO2s (21% vs. 10% vs. 5% O<sup>2</sup> ) (*n* = 9 paired samples): (**A**) *leptin*, (**B**) *dipeptidylpeptidase (DPP)-4*, (**C**) *interleukin (IL)-6*, (**D**) *plasminogen activator inhibitor (PAI)-1*, (**E**) *adiponectin*, (**F**) *tumour necrosis factor (TNF)α*, and (**G**) *monocyte chemoattractant protein (MCP)-1*. Data are expressed as mean ± SEM. \* *p* < 0.05.

#### *3.2. The Effects of Oxygen Partial Pressure on Adipokine Secretion 3.2. The Effects of Oxygen Partial Pressure on Adipokine Secretion*

Next, we investigated whether exposure to different pO2 levels elicited functional changes in adipokine secretion from differentiated ABD and FEM hMADS. We found that adipokine secretion from both differentiated ABD and FEM hMADS was significantly affected by changes in oxygen availability (Figure 2). Specifically, low *physiological* pO2 (5% O2) exposure increased leptin secretion in differentiated ABD and FEM hMADS derived from individuals with OB compared with exposure to high *physiological* pO2 (10% O2: Next, we investigated whether exposure to different pO<sup>2</sup> levels elicited functional changes in adipokine secretion from differentiated ABD and FEM hMADS. We found that adipokine secretion from both differentiated ABD and FEM hMADS was significantly affected by changes in oxygen availability (Figure 2). Specifically, low *physiological* pO<sup>2</sup> (5% O2) exposure increased leptin secretion in differentiated ABD and FEM hMADS derived from individuals with OB compared with exposure to high *physiological* pO<sup>2</sup> (10% O2: ABD,

ABD, *p* = 0.009; FEM, *p* = 0.021), and in differentiated ABD and FEM hMADS derived from individuals with NW compared with exposure to room air (21% O2: ABD, *p* = 0.014; FEM,

ABD (*p* = 0.027) and FEM hMADS (*p* = 0.004) and IL-6 secretion in differentiated FEM hMADS only (*p* = 0.007), derived from donors with OB but not NW (Figure 2B,C). Moreover, low *physiological* pO2 (5% O2) reduced MCP-1 secretion (*p* = 0.030) but did not alter PAI-1 secretion from differentiated ABD hMADS derived from individuals with OB compared with 10% O2 (Figure 2D,E). Finally, low *physiological* pO2 (5% O2) reduced both MCP-1 (*p* = 0.028) and PAI-1 (*p* = 0.003) secretion from differentiated FEM hMADS derived from donors with NW compared with 21% O2 (Figure 2D,E). Adiponectin secretion was

not detectable, and these data are therefore not reported.

*p* = 0.009; FEM, *p* = 0.021), and in differentiated ABD and FEM hMADS derived from individuals with NW compared with exposure to room air (21% O2: ABD, *p* = 0.014; FEM, *p* = 0.006) (Figure 2A). Furthermore, DPP-4 secretion was significantly lower following exposure to low (5% O2) compared with high (10% O2) *physiological* pO<sup>2</sup> in differentiated ABD (*p* = 0.027) and FEM hMADS (*p* = 0.004) and IL-6 secretion in differentiated FEM hMADS only (*p* = 0.007), derived from donors with OB but not NW (Figure 2B,C). Moreover, low *physiological* pO<sup>2</sup> (5% O2) reduced MCP-1 secretion (*p* = 0.030) but did not alter PAI-1 secretion from differentiated ABD hMADS derived from individuals with OB compared with 10% O<sup>2</sup> (Figure 2D,E). Finally, low *physiological* pO<sup>2</sup> (5% O2) reduced both MCP-1 (*p* = 0.028) and PAI-1 (*p* = 0.003) secretion from differentiated FEM hMADS derived from donors with NW compared with 21% O<sup>2</sup> (Figure 2D,E). Adiponectin secretion was not detectable, and these data are therefore not reported. *Cells* **2022**, *11*, 3532 6 of 9

**Figure 2.** Adipokine and inflammatory markers' secretion in hMADS cells following differentiation under different pO2s (21% vs. 10% vs. 5% O2) (*n* = 9 paired samples). (**A**) Leptin, (**B**) dipeptidylpeptidase (DPP)-4, (**C**) interleukin (IL)-6, (**D**) plasminogen activator inhibitor (PAI)-1, and (**E**) monocyte chemoattractant protein (MCP)-1. Data are expressed as mean ± SEM. \* *p* < 0.05. **Figure 2.** Adipokine and inflammatory markers' secretion in hMADS cells following differentiation under different pO2s (21% vs. 10% vs. 5% O<sup>2</sup> ) (*n* = 9 paired samples). (**A**) Leptin, (**B**) dipeptidylpeptidase (DPP)-4, (**C**) interleukin (IL)-6, (**D**) plasminogen activator inhibitor (PAI)-1, and (**E**) monocyte chemoattractant protein (MCP)-1. Data are expressed as mean ± SEM. \* *p* < 0.05.

#### **4. Discussion 4. Discussion**

In the present study, we investigated the impact of oxygen tension on adipokine gene expression and secretion in differentiated human multipotent ABD and FEM adipose-derived stem cells from women with NW or OB. Here, we demonstrate that low *physiological* pO2 decreased gene expression and secretion of the proinflammatory factors *DDP-4* and *IL-6* in both differentiated ABD and FEM hMADS derived from individuals with OB, while these responses were not present in differentiated hMADS cells from NW individuals. Our findings highlight that the changes in pO2 within the human *physiological* range in the adipocyte microenvironment contribute to alterations in the AT inflammatory phenotype and that these effects may differ between individuals with normal weight and In the present study, we investigated the impact of oxygen tension on adipokine gene expression and secretion in differentiated human multipotent ABD and FEM adiposederived stem cells from women with NW or OB. Here, we demonstrate that low *physiological* pO<sup>2</sup> decreased gene expression and secretion of the proinflammatory factors *DDP-4* and *IL-6* in both differentiated ABD and FEM hMADS derived from individuals with OB, while these responses were not present in differentiated hMADS cells from NW individuals. Our findings highlight that the changes in pO<sup>2</sup> within the human *physiological* range in the adipocyte microenvironment contribute to alterations in the AT inflammatory phenotype and that these effects may differ between individuals with normal weight and obesity.

obesity. To determine whether the amount of oxygen present in the AT microenvironment affects the gene expression of adipokines, we exposed differentiating hMADS cells from ABD and FEM AT to low (5%) and high (10%) *physiological* pO2 levels in human AT To determine whether the amount of oxygen present in the AT microenvironment affects the gene expression of adipokines, we exposed differentiating hMADS cells from ABD and FEM AT to low (5%) and high (10%) *physiological* pO<sup>2</sup> levels in human AT [9,16,21–24]. As expected, low *physiological* pO<sup>2</sup> levels increased the gene expression of several hypoxia

[9,16,21–24]. As expected, low *physiological* pO2 levels increased the gene expression of several hypoxia markers. Strikingly, we show for the first time that low *physiological* pO2 dur-

tory markers IL-6 and DPP-4 in both differentiated FEM and ABD hMADS derived from individuals with OB, but not NW. Moreover, the present data suggest that these cells maintain a memory of origin (i.e., a normal-weight or obese microenvironment) in vitro, even after 14 days of exposure to the same experimental conditions. In agreement with our findings, we have previously reported that in vivo ABD AT pO2 was positively associated with the AT gene expression of several proinflammatory markers [22] and that low *physiological* pO2 exposure reduced the gene expression of IL-6 and DPP-4 in adipocytes derived from women with obesity [16]. In addition, the present results show that low *physiological* pO2 levels consistently increased leptin gene expression and secretion in

markers. Strikingly, we show for the first time that low *physiological* pO<sup>2</sup> during adipogenesis consistently decreased the expression and secretion of the proinflammatory markers IL-6 and DPP-4 in both differentiated FEM and ABD hMADS derived from individuals with OB, but not NW. Moreover, the present data suggest that these cells maintain a memory of origin (i.e., a normal-weight or obese microenvironment) in vitro, even after 14 days of exposure to the same experimental conditions. In agreement with our findings, we have previously reported that in vivo ABD AT pO<sup>2</sup> was positively associated with the AT gene expression of several proinflammatory markers [22] and that low *physiological* pO<sup>2</sup> exposure reduced the gene expression of IL-6 and DPP-4 in adipocytes derived from women with obesity [16]. In addition, the present results show that low *physiological* pO<sup>2</sup> levels consistently increased leptin gene expression and secretion in differentiated ABD and FEM hMADS derived from donors with NW or OB. Leptin is an important regulator of appetite and energy expenditure, providing important feedback in relation to energy storage in the body through the hypothalamus, and is involved in multiple *physiological* processes such as the regulation of immunity [9,30–32]. Changes in leptin secretion due to altered oxygen tension in the AT microenvironment may thus affect these processes. Notably, pO2-induced alterations in adipokine gene expression were paralleled by comparable changes in adipokine secretion. Importantly, the modest effects of pO<sup>2</sup> levels on adipocyte differentiation, if present at all, do not seem to explain the observed changes in adipokine expression and secretion, exemplified by the opposing effects of low pO<sup>2</sup> on the expression and secretion of leptin and the proinflammatory markers Il-6 and DPP-4. Famulla et al. [27] previously showed increased DPP-4, adiponectin, and IL-6 secretion following prolonged exposure to high *physiological* pO<sup>2</sup> (10% O2), while low *physiological* pO<sup>2</sup> (5% O2) tended to reduce the secretion of adiponectin. These differences between studies may at least partly be explained by differences in donor characteristics.

A strength of the present study is the paired comparisons between differentiated adipose-derived multipotent stem cells derived from ABD and FEM AT of individuals with NW and OB. Previous studies examining the effects of pO<sup>2</sup> levels on adipocyte inflammation have either used cell lines, adipose-derived multipotent stem cells from a single donor, or a pool of stem cells obtained from different donors. Because our findings demonstrate that the impact of changes in the AT microenvironment (i.e., different *physiological* pO<sup>2</sup> levels) on adipokine expression and secretion depends on the characteristics of the donors, future studies in the field of AT biology should take this '*memory-of-origin effect*' into account. Secondly, in contrast to many studies showing that acute exposure to severe (non-physiological) hypoxia evokes a proinflammatory response in murine and human (pre)adipocytes [13,14], we aimed to mimic *physiological* in vivo conditions in terms of pO<sup>2</sup> levels as well as the prolonged exposure duration in the present study. This study also has some limitations. We examined the effects of various oxygen levels in cells derived from postmenopausal women. Therefore, our findings cannot be translated to other subgroups of the population such as men or individuals of different age. Furthermore, we used a targeted approach to examine the gene expression and secretion of several adipokines. Future studies using an untargeted approach (e.g., microarray analysis, RNA sequencing, or proteomics) are warranted.

#### **5. Conclusions**

In conclusion, the present findings demonstrate that *physiological* oxygen levels regulate adipokine expression and secretion in differentiated ABD and FEM hMADS. Differentiated hMADS cells derived from women with OB display lower expression and secretion of several (proinflammatory) adipokines at low (5% O2) compared with high (10% O2) *physiological* oxygen tension. Except for the effects on leptin expression, no significant effects of low compared with high *physiological* oxygen levels were observed in differentiated hMADS cells derived from individuals with NW. Our findings thus indicate that pO<sup>2</sup> levels alter the expression and secretion of several adipokines in differentiated human ABD and FEM hMADS, and that donor characteristics determine experimental outcomes. This has important implications for future mechanistic in vitro studies in the field of AT biology. For example, the outcomes of studies in which the effects of certain interventions on adipocyte inflammation and related biological mechanisms are investigated may depend on the microenvironmental oxygen tension. Furthermore, our findings highlight that it is important to report detailed the characteristics of the cell donor(s) in studies examining human adipocyte biology.

**Supplementary Materials:** The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/cells11223532/s1, Figure S1: Adipocyte differentiation and hypoxia markers' gene expression in adipose tissue-derived mesenchymal stem cells following differentiation under different pO2s (21% vs. 10% vs. 5% O<sup>2</sup> ).

**Author Contributions:** K.N.M. and G.H.G. acquired funding, conceived and designed research, interpreted data, and revised the manuscript; I.G.L. performed experiments, analysed data, interpreted data, prepared figures, and drafted the manuscript; N.H., Y.E., and J.W.E.J. performed experiments and analysed data. K.M.A.R. and E.E.B. interpreted data and revised the manuscript. All authors approved the final version of the manuscript.

**Funding:** This work was supported by the European Foundation for the Study of Diabetes (EFSD) under an EFSD/Lilly European Diabetes Research Program grant to Gijs H. Goossens and Konstantinos N. Manolopoulos, and Maastricht University (The Netherlands) and the University of Birmingham (U.K.) under a joint Ph.D. scholarship grant to Gijs H. Goossens and Konstantinos N. Manolopoulos.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and the U.K. Health Research Authority National Health System Research Ethics Committee approved the present study (approval no. 18/NW/0392).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Data presented in this manuscript are available from the corresponding author on reasonable request.

**Acknowledgments:** The authors would like to express their gratitude to the study participants, the teams of the NIHa/Wellcome Trust Clinical Research Facility at Queen Elizabeth Hospital Birmingham and the NIHR CRN West Midlands (U.K.), as well as the Department of Human Biology at Maastricht University Medical Center+ (The Netherlands) for excellent practical support.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

## **References**


## *Article* **Expression of Steroid Receptor RNA Activator 1 (SRA1) in the Adipose Tissue Is Associated with TLRs and IRFs in Diabesity**

**Shihab Kochumon <sup>1</sup> , Hossein Arefanian <sup>1</sup> , Sardar Sindhu 1,2 , Reeby Thomas <sup>1</sup> , Texy Jacob <sup>1</sup> , Amnah Al-Sayyar <sup>1</sup> , Steve Shenouda <sup>1</sup> , Fatema Al-Rashed <sup>1</sup> , Heikki A. Koistinen 3,4,5, Fahd Al-Mulla <sup>6</sup> , Jaakko Tuomilehto 4,7,8 and Rasheed Ahmad 1,\***


**Abstract:** Steroid receptor RNA activator gene (SRA1) emerges as a player in pathophysiological responses of adipose tissue (AT) in metabolic disorders such as obesity and type 2 diabetes (T2D). We previously showed association of the AT SRA1 expression with inflammatory cytokines/chemokines involved in metabolic derangement. However, the relationship between altered adipose expression of SRA1 and the innate immune Toll-like receptors (TLRs) as players in nutrient sensing and metabolic inflammation as well as their downstream signaling partners, including interferon regulatory factors (IRFs), remains elusive. Herein, we investigated the association of AT SRA1 expression with TLRs, IRFs, and other TLR-downstream signaling mediators in a cohort of 108 individuals, classified based on their body mass index (BMI) as persons with normal-weight (N = 12), overweight (N = 32), and obesity (N = 64), including 55 with and 53 without T2D. The gene expression of SRA1, TLRs-2,3,4,7,8,9,10 and their downstream signaling mediators including IRFs-3,4,5, myeloid differentiation factor 88 (MyD88), interleukin-1 receptor-associated kinase 1 (IRAK1), and nuclear factor-κB (NF-κB) were determined using qRT-PCR and SRA1 protein expression was determined by immunohistochemistry. AT SRA1 transcripts' expression was significantly correlated with TLRs-3,4,7, MyD88, NF-κB, and IRF5 expression in individuals with T2D, while it associated with TLR9 and TRAF6 expression in all individuals, with/without T2D. SRA1 expression associated with TLR2, IRAK1, and IRF3 expression only in individuals with obesity, regardless of diabetes status. Furthermore, TLR3/TLR7/IRAK1 and TLR3/TLR9 were identified as independent predictors of AT SRA1 expression in individuals with obesity and T2D, respectively. Overall, our data demonstrate a direct association between the AT SRA1 expression and the TLRs together with their downstream signaling partners and IRFs in individuals with obesity and/or T2D.

**Keywords:** steroid receptor RNA activator 1/SRA1; TLRs; IRFs; adipose tissue; obesity; type-2 diabetes; inflammation

## **1. Introduction**

Obesity is known as a complex disease due to an excessive amount of body weight, and mainly the body fat associated with expansion and function of white adipose tissue. The expanded white adipose tissue in individuals with obesity produces a wide range of adipocytokines, such as proinflammatory cytokines, chemokines, hormones, and similar mediators [1–4]. Adipocytes, resident monocytes/macrophages, and other cell populations in the adipose tissue are actively involved in production and secretion of

**Citation:** Kochumon, S.; Arefanian, H.; Sindhu, S.; Thomas, R.; Jacob, T.; Al-Sayyar, A.; Shenouda, S.; Al-Rashed, F.; Koistinen, H.A.; Al-Mulla, F.; et al. Expression of Steroid Receptor RNA Activator 1 (SRA1) in the Adipose Tissue Is Associated with TLRs and IRFs in Diabesity. *Cells* **2022**, *11*, 4007. https://doi.org/10.3390/ cells11244007

Academic Editor: Javier Gómez-Ambrosi

Received: 5 November 2022 Accepted: 9 December 2022 Published: 11 December 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

adipocytokines [5–8]. It is speculated that the mechanisms of obesity-induced insulin resistance are initiated and stimulated by these adipocytokines, mainly by proinflammatory cytokines in the white adipose tissue [9–13].

Toll-like receptors (TLRs) are the surface, innate immune receptors that identify the pathogen-associated molecular patterns (PAMPs) to activate and initiate inflammatory responses. In humans, 11 different TLRs have been so far identified [14,15]. Structurally, TLRs have an extracellular, leucine-rich repeat (LRR) domain which recognizes PAMPs, and a cytoplasmic Toll/IL-1 (TIR) domain that activates TLR-downstream signaling after ligand binding with its cognate TLR. LRR and TIR are involved in the recognition of PAMPs and activation of downstream adaptor proteins and signaling molecules, including myeloid differentiation factor 88 (MyD88), interleukin-1 (IL-1) receptor-associated kinases (IRAKs), and tumor necrosis factor receptor-associated factor (TRAF)-6, respectively [15]. After the activation of these adaptor proteins, stimulation of multiple pathways is initiated such as extracellular signal-regulated kinase (ERK), c-Jun N-terminal kinase (JNK), p38 mitogen-activated protein kinases (MAPK), NF-κB, and interferon regulatory factors (IRFs) pathways. The activation of NF-κB signaling results in the up-regulation of inflammatory markers including cytokines, chemokines, adhesion molecules, type-I interferons (IFNα/β), tumor necrosis factor (TNF)-α, and IL-6) which offset the homeostasis between beneficial host innate immune responses and immunopathology [16–19].

TLRs such as TLR2, TLR3, TLR4, TLR7, TLR9, and TLR10 have been identified in multiple immune cell populations within the adipose tissue namely adipocytes, monocytes, and macrophages. Each of these TLRs has distinct ligands such as free fatty acids (FFAs), lipids, lipoproteins, nucleic acids, and proteins which indicate the potential role of innate immune TLRs as nutrient sensors involved in obesity [20,21]. The important roles of TLR signaling cascades in adipose tissue inflammation and impairment in obesity and type-2 diabetes (T2D) have been addressed and reported by us and others [22–24]. Expression changes in TLR1, TLR2, TLR4, TLR5, TLR6, TLR7, TLR8, TLR9, and TLR10 have been often reported in obesity, metabolic syndrome, inflammation and insulin resistance, T2D, and related complications such as cardiovascular disease, diabetic nephropathy, and atherosclerosis [23–34]. Specifically, TLR4/TLR2 have emerged as metabolic sensors of lipopolysaccharide (LPS) and saturated free fatty acids (sFFAs), both of which are abundantly found in individuals with obesity and T2D [35].

Steroid receptor RNA activator 1 (SRA1) was originally identified as an intergenic long non-coding RNA (lncRNA) which acts as an RNA coactivator of nuclear receptors to enhance steroid receptor-dependent gene expression [36]. It binds with DNA via interactions with other proteins that bind directly or indirectly with the DNA. Therefore, serving as a natural organizer by regulating physiological processes that dictate the epigenetic modifications including changes in chromatin and gene expression [37–39]. Increased SRA1 expression in the human liver, skeletal muscle, and in the white/brown adipose tissues as key organs regulating metabolic homeostasis, compared to other tissues, has been documented [36,40,41]. We recently showed that in individuals without T2D, adipose *SRA1* expression was significantly higher in obese people compared with normal weight people and the adipose tissue *SRA1* expression associated directly with metabolic markers including body mass index (BMI), percentage of body fat (PBF), serum insulin, homeostasis model assessment of insulin resistance (HOMA-IR), proinflammatory cytokines and chemokines or their receptors including C-X-C motif ligand-9 (CXCL9), CXCL10, CXCL11, TNFα, transforming growth factor-β (TGFβ), IL2RA, and IL18, but inversely with CCL19 and CCR2 expression. We further showed that TGFβ and IL18 were independent predictors of *SRA1* expression in individuals without T2D, while TNFα and IL2RA were the independent predictors in individuals with T2D. TNFα also predicted *SRA1* adipose expression in both normal weight and obese populations, regardless of diabetes status. Taken together, this study revealed specific association patterns of the adipose *SRA1* expression with diverse immune markers, most of them being inflammatory by nature [42].

In the current study, we examined whether the human adipose tissue SRA1 expression associated with TLRs expression, their adaptor proteins and TLR-downstream signaling molecules including NF-κB, IRF3, 4, and 5 in obesity and/or T2D.

### **2. Materials and Methods**

## *2.1. Study Population and Anthropometry*

This study included 108 participants who were classified based on BMI as persons with normal weight (NW) (BMI < 25 kg/m<sup>2</sup> , n = 12), overweight (25 <sup>≤</sup> BMI < 30 kg/m<sup>2</sup> , n = 32), or obese (BMI <sup>≥</sup> 30 kg/m<sup>2</sup> , n = 64). Among these participants, 53 were with and 55 were without T2D. Diagnosis of T2D was based on fasting blood glucose and glycated hemoglobin (HbA1c) levels. HbA1c levels of <5.7, 5.7–6.4%, and >6.4% represented normal, prediabetes, and diabetes status, respectively. Height, weight, waist and hip circumferences, BMI, and PBF were measured and calculated as previously described [42]. For the assessment of insulin resistance (HOMA-IR) and insulin sensitivity, fasting blood glucose and insulin levels were used to determine HOMA index [43]. This study was approved by ethics committee of the Dasman Diabetes Institute, Kuwait in line with the ethical guidelines of the Declaration of Helsinki (Grant#: RA 2010-003). The written informed consent was obtained from each participant at the time of enrolment in the study. The individuals with chronic diseases of the heart, liver, kidney, lung, or those with type 1 diabetes, immune dysfunction, hematologic disorders, pregnancy, or malignancy were excluded previously stated [42,44]. Clinical and demographic features of the study cohort are presented in Supplementary Table S1.

## *2.2. Collection of Subcutaneous Adipose Tissue Samples*

Biopsies of the human adipose tissue, about 0.5 g in size, were collected from the abdominal subcutaneous adipose tissue, next to the umbilicus by using sterile surgical technique as described [42,45]. The sample was further cut into smaller pieces, about 50–100 mg in size, and added to RNAlater (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany) and stored at −80 ◦C until use for RNA extraction.

### *2.3. Measurement of Metabolic Markers*

Peripheral blood was collected from the individuals fasting overnight and the samples were analyzed for metabolic markers such as plasma glucose, insulin, and lipid profiles including triglycerides (TGL), low-density/high-density lipoproteins (LDL/HDL), and total cholesterol using Siemens Dimension RXL Analyzer (Diamond Diagnostics, Holliston, MA, USA). Glycated hemoglobin (HbA1c) was measured by Variant device (BioRad, Hercules, CA, USA).

### *2.4. Quantitative, Real-Time, Reverse-Transcription Polymerase Chain Reaction (RT-qPCR)*

Fat samples were used for total RNA collection by using RNeasy kit (Qiagen, Valencia, CA, USA), following the protocol as recommended by the manufacturer. RNA template (0.5 µg) was used for cDNA synthesis using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster, CA, USA) as earlier stated [45]. Real-time qRT-PCR was carried out as per the protocol [42,45,46]. Briefly, 50 ng of cDNA was amplified using TaqMan Gene Expression Master Mix (Applied Biosystems, CA, USA) and gene-specific 20 X TaqMan gene expression assays including forward and reverse primers (Supplementary Table S2), target-specific TaqMan 50FAM-labeled and 30NFQ-labeled MGB probe, using 7500 Fast Real-Time PCR System (Applied Biosystems, CA, USA) as described elsewhere [42]. Target gene expression relative to control (NW fat samples) was determined using comparative Ct method [47] and data were normalized to GAPDH gene expression as described [48–52].

## *2.5. Immunohistochemistry (IHC)*

Expression of SRA1, TLR4, IRAK1 or NF-kB was measured in the fat tissue by IHC as described elsewhere [42,53]. Briefly, subcutaneous fat tissue (paraffin-embedded) was cut in 4µm thick sections, deparaffinized by xylene, and rehydrated by serial immersions in 100%, 95%, and 75% ethanol in water. Antigen was retrieved by retrieval solution (pH 6.0; Dako, Glostrup, Denmark) with 8 min boiling and 15 min cooling steps. After 3 washes with PBS, internal peroxidase was blocked by 30 min treatment with 3% H2O<sup>2</sup> and non-specific antibody binding was blocked by 1 h treatment, each, with 5% non-fat milk and 1% BSA. The primary antibody treatment was carried out overnight at room temperature, using anti-human SRA1 rabbit polyclonal antibody (1:800 dilution) (Thermo-Scientific PA5-62145, pH 6.0), rabbit monoclonal anti-IRAK antibody (ab302554, Abcam*®* Cambridge, UK), 1:400 dilution of rabbit monoclonal anti-IRF5 antibody (ab181553, Abcam*®* Cambridge, UK), 1:1000 dilution of rabbit polyclonal anti-NFkB antibody (ab16502, Abcam*®* Cambridge, UK), and 1:1000 dilution of mouse monoclonal anti-TLR4 antibody (ab13556, Abcam*®* Cambridge, UK). After washing 3 times using PBS (0.5% Tween), samples were incubated for 1 h with secondary antibody (HRP-conjugated goat anti-rabbit; EnVision™ Kit from Dako, Glostrup, Denmark) and the substrate (3,30 -DAB chromogen) was added to develop color. After washing 3 times in running water, samples were counterstained (Harris hematoxylin), dehydrated by immersion in 75%, 95%, and 100% ethanol in water, clarified by xylene, and mounted in DPX. Later, digital photomicrographs were taken (40X magnification) and regional heterogeneity was assessed in 4 different regions of tissue sample (PannoramicScan, 3DHistech, Budapest, Hungry). The data were expressed as the staining intensity measured in arbitrary units (AU) and analyzed using imageJ software (NIH, Bethesda, MD, USA).

In addition, we also performed H&E staining on lean and obese adipose tissue samples, 5 each, to represent sample quality for standard tissue staining (Supplementary Figure S1).

## *2.6. Statistical Analysis*

The data obtained were presented as mean ± SEM values and analyzed using Graph-Pad Prism (GraphPad, San Diego, CA, USA) and SPSS (IBM SPSS Inc., Chicago, IL, USA) software. Means between two groups were compared using unpaired *t*-test, and between more than two groups were compared using One-way ANOVA, Kruskal-Wallis and Mann-Whitney tests. Spearman correlation and multivariate regression analyses were performed to determine associations between variables. All *p*-values ≤ 0.05 were considered significant. For multivariate linear regression, the Enter method was used, selecting the variables that significantly correlated with *SRA1* expression as predictor variables and were entered simultaneously to generate the model. *F*-test was used to test whether the independent variables collectively predicted the dependent variable. *R*-squared evaluated how much variance in the dependent variable was accounted for by the set of independent variables. The *p*-value assessed the significance and the β-value identified the magnitude of prediction for each independent variable.

#### **3. Results**

## *3.1. SRA1 Adipose Expression and Its Association with TLRs, Downstream Signaling Mediators and IRFs Expression in the Study Population*

We sought to determine SRA1 protein expression in adipose tissue samples from NW individuals and those with obesity, 10 individuals each, and the data show that the expression was significantly higher in people with obesity compared to their NW counterparts, whether with or without T2D (*p* < 0.0001) (Figure 1).

We next sought to determine the adipose *SRA1* gene expression and assessed its relationship with typical inflammatory sensing and signaling components such as TLRs, their downstream signaling mediators, and IRFs. The expression of these markers was first compared among populations defined as those with NW, overweight, and obesity. To this end, adipose SRA1 mRNA expression differed significantly only between non-diabetic NW and those with obesity individuals (*p* = 0.015) whereas it differed non-significantly between all other BMI groups, with or without T2D (Table 1). Regarding expression of TLRs, their downstream signaling mediators and IRFs, adipose expression of TLR2

(*p* = 0.018), TLR3 (*p* = 0.010), TLR8 (*p* = 0.048), IRAK1 (*p* = 0.047), and IRF5 (*p* = 0.016) was significantly higher in participants with obesity compared to NW participants. IRF5 expression in both overweight (*p* = 0.031) and those with obesity (*p* = 0.016) differed significantly from that of NW participants. Expression of IRF3 (*p* = 0.047), IRF4 (*p* = 0.020), and IRF5 (*p* = 0.039) was significantly higher in individuals with T2D compared with those without T2D. Furthermore, adipose *SRA1* expression was associated with TLR2 (r = 0.218, *p* = 0.036), TLR3 (r = 0.218, *p* < 0.0001), TLR4 (r = 0.226, *p* = 0.027), TLR7 (r = 0.196, *p* = 0.045), NF-κB (r = 0.297, *p* = 0.002), and IRAK1 (r = 0.201, *p* = 0.044) expression in the total (N = 108) study population (Table 2; Figure 2A–F). *and IRFs Expression in the Study Population* We sought to determine SRA1 protein expression in adipose tissue samples from NW individuals and those with obesity, 10 individuals each, and the data show that the expression was significantly higher in people with obesity compared to their NW counterparts, whether with or without T2D (*p* < 0.0001) (Figure 1).

obesity compared to NW persons, whether with or without T2D (*p* < 0.0001).

*3.1. SRA1 Adipose Expression and Its Association with TLRs, Downstream Signaling Mediators* 

*Cells* **2022**, *11*, x FOR PEER REVIEW 5 of 19

**Figure 1.** Adipose SRA1 protein expression in adipose tissue. SRA1 protein expression was determined in adipose tissue samples from non-diabetic individuals with NW and obesity, as well as from persons with type-2 diabetic (T2D), as NW and those with obesity, 10 each, using by immunohistochemistry (IHC) as described in Materials and Methods. IHC staining intensity was expressed as arbitrary units (AU) and the data (mean ± SEM) were compared between NW and obese populations, with or without T2D, using unpaired *t*-test and *p* < 0.05 was considered significant. (A) The representative IHC images are shown for non-diabetic NW/obese and T2D NW/obese individuals, one each. (B) IHC staining data (AU) show the elevated adipose SRA1 expression in persons with **Figure 1.** Adipose SRA1 protein expression in adipose tissue. SRA1 protein expression was determined in adipose tissue samples from non-diabetic individuals with NW and obesity, as well as from persons with type-2 diabetic (T2D), as NW and those with obesity, 10 each, using by immunohistochemistry (IHC) as described in Materials and Methods. IHC staining intensity was expressed as arbitrary units (AU) and the data (mean ± SEM) were compared between NW and obese populations, with or without T2D, using unpaired *t*-test and *p* < 0.05 was considered significant. (**A**) The representative IHC images are shown for non-diabetic NW/obese and T2D NW/obese individuals, one each. (**B**) IHC staining data (AU) show the elevated adipose SRA1 expression in persons with obesity compared to NW persons, whether with or without T2D (*p* < 0.0001).

We next sought to determine the adipose *SRA1* gene expression and assessed its relationship with typical inflammatory sensing and signaling components such as TLRs, their downstream signaling mediators, and IRFs. The expression of these markers was first compared among populations defined as those with NW, overweight, and obesity. To this end, adipose SRA1 mRNA expression differed significantly only between non-Furthermore, to confirm the expression of protein levels in the adipose tissue, we performed immunohistochemistry analysis on TLR4, IRAK1 and NF-kB as representatives. Immunohistochemistry analysis showed that TLR4 (Figure 3A,B), IRAK1 (Figure 4A,B) and NF-kB (Figure 5A,B) were significantly upregulated in individuals with obesity. Our protein data show that SRA1 positively correlated with TLR4 protein (r<sup>2</sup> = 0.623; *p* = 0.0002; Figure 3C). IRAK1 protein (r<sup>2</sup> = 0.0649; *p* < 0.0001; Figure 4C) and NF-kB protein (r<sup>2</sup> = 0.379; *p* = 0.0085; Figure 5C).

diabetic NW and those with obesity individuals (*p* = 0.015) whereas it differed non-significantly between all other BMI groups, with or without T2D (Table 1). Regarding expression of TLRs, their downstream signaling mediators and IRFs, adipose expression of TLR2 (*p* = 0.018), TLR3 (*p* = 0.010), TLR8 (*p* = 0.048), IRAK1 (*p* = 0.047), and IRF5 (*p* = 0.016) was significantly higher in participants with obesity compared to NW participants. IRF5 expression in both overweight (*p* = 0.031) and those with obesity (*p* = 0.016) differed significantly from that of NW participants. Expression of IRF3 (*p* = 0.047), IRF4 (*p* = 0.020), and IRF5 (*p* = 0.039) was significantly higher in individuals with T2D compared with those without T2D. Furthermore, adipose *SRA1* expression was associated with TLR2 (r = 0.218, *p* = 0.036), TLR3 (r = 0.218, *p* < 0.0001), TLR4 (r = 0.226, *p* = 0.027), TLR7 (r = 0.196, *p* = 0.045),

*Cells* **2022**, *11*, 4007


