Dysregulation of Inflammation, Oxidative Stress, and Glucose Metabolism-Related Genes and miRNAs in Visceral Adipose Tissue of Women with Type 2 Diabetes Mellitus

Background

The burden of obesity and type 2 diabetes (T2DM) is indisputable. Projections dramatic and suggest that at least 642 million people will suffer from T2DM by 2040, thus providing a rationale for extensive research aimed at identifying biomolecules that allow for early prediction of T2DM development [1]. As evidenced by multiple studies, visceral adiposity is strongly associated with risk of prediabetes, T2DM, and gestational diabetes mellitus (GDM) [2–4]. Reduction of visceral adipose tissue (VAT) improves insulin resistance (IR) and fasting plasma glucose (FPG) more than reduction of subcutaneous adipose tissue (SAT) [5]. Reduction of VAT is also a desirable and positive side effect of a few anti-diabetic drugs under extensive investigation [6,7]. Considering the connection between T2DM and VAT, glucose metabolism abnormalities, including prediabetes (impaired glucose tolerance (IGT) and/or impaired fasting glucose (IFG)) and T2DM, are interrelated with altered (in hypertrophy and hyperplasia) adipogenesis (ADG). The latter also leads to hypoxia of white adipose tissue (AT), including VAT, both being recognized as tissues with an increasingly recognized endocrine role. While overgrown and hypoxic, VAT independently contributes to the development of IR by producing a detrimental profile of cytokines (eg, increased TNF-α, IL-6, resistin, and decreased adiponectin) into the bloodstream. Additionally, paracrine secretion of pro-inflammatory cytokines stimulates M1 macrophage infiltration and supports low-grade chronic inflammation of VAT, often induced by nuclear factor kappa β (NF-κB) signaling, which is also a target of the anti-diabetic drug metformin, which is considered to be involved in IR [8]. Importantly, inflamed VAT produces and secretes potent epigenetic modifiers – microRNAs (miRNAs). The latter are capable of affecting the expression of target genes even in distant tissues that are critical for insulin sensitivity (eg, utilizing adipocyte derived-extracellular vesicles), as observed for miR-155 and miR-34a [9,10]. All of these aforementioned aspects make VAT an especially interesting organ for epigenetic-based studies in terms of metabolic disorders.

Each fat depot is composed of inflammatory cells, adipocytes (Ads), and a reservoir of stem cells and preadipocytes [11]. Preadipocytes can replace dysfunctional mature adipocytes via starting differentiation on biological request. In this way, stem cells and preadipocytes of subjects with disturbed glucose metabolism are exposed to the detrimental impact of fluctuating and elevated glucose levels, most commonly before prediabetes or T2DM are diagnosed. Such preadipocytes differentiate and give rise to mature adipocytes with a history of exposure to hyperglycemia (HG), which is a strong inducer of epigenetic modifications. Indeed, there is great interest in finding molecules that could help estimate the risk of transition from a healthy to hyperglycemic phenotype [12].

The results of our previous investigation suggested that expression changes of some miRNAs are highly similar in female subjects with T2DM and IFG, and are predicted to be associated with inflammation, IR, and oxidative stress [13]. It was proven that untreated HG in prediabetes leads to further endothelial dysfunction and cardiovascular complications (in addition to faster progression to T2DM) [14]. Therefore, recognition of biomolecules (miRNA or mRNA) related to low-grade chronic inflammation, IR, and oxidative stress deregulated in VAT in early stages of T2DM development could be a promising solution for early diagnosis, improving global efforts in prevention of T2DM and its complications. There is broad interest in identifying biomarker molecules of among miRNAs in various diseases, including diabetic retinopathy, coronary artery disease, osteoarthritis, and multiple sclerosis [15–18].

On the other hand, understanding of epigenetic modulation of genes related to inflammation, insulin sensitivity, oxidative stress, and glucose metabolism may be a basis for development of precise epigenetic therapy in the future. Several miRNA therapeutics are already in II phase clinical trials (eg, miR-103/107 in nonalcoholic steatohepatitis and miR-132 in heart failure) or have completed phase II (eg, miR-29 in keloid disorder) [19].

Hence, we speculate that the development of IR (eg, during the IFG state, and thereafter T2DM) could be contributed to by the altered expression of miRNAs and genes associated with oxidative stress, glucose metabolism, and inflammation. To examine this role, firstly, we aimed to investigate the expression of several interconnected genes: an antioxidative gene (superoxide dismutase 2 – SOD2), a kinase sensitive to DNA damage that activate the DNA damage response pathway (ataxia-telangiectasia mutated – ATM), the receptor for insulin (INSR), the repressor of glycolysis and positive regulator of apoptosis (tumor protein p53 (TP53)-induced glycolysis and apoptosis regulator – TIGAR), and an inflammatory transcription factor (nuclear factor kappa B subunit 1 – NFKB1) in differentiating human visceral preadipocytes (HPA-vs) to examine whether chronic or intermittent HG affect the next generations of adipocytes. We also aimed to investigate HG-induced epigenetic stigma via miRNAs either negatively regulating the expression of target genes or those participating in NF-κB-dependent downstream signaling (Table 1), in both chronic and intermittent HG-exposed differentiating HPA-vs Moreover, we considered evaluating the results obtained in vitro in the VAT of subjects with normal glucose metabolism (NGM), IFG, and T2DM [20,23,29,32,37]. Additionally, in the ex vivo study we evaluated the expression changes of miRNAs, in which expression in HPA-vs were previously reported (miR-26a-5p and miR-26b-5p) [36].

Material and Methods

Studies on HPA-v Cells

CELL CULTURE AND TREATMENT: The HPA-v cell culture was managed as previously described [38]. Cell culture reagents and HPA-v cells were obtained from ScienCell Research Laboratories (Carlsbad, CA, USA). ADG of preadipocytes was performed in accordance with the specific protocol designed and validated by the manufacturer in 3 stages, including (Figure 1) proliferation (5 days), differentiation (12 days), and maturation (6 days). Preadipocytes were cultured in the following media: preadipocyte medium (PAM), preadipocyte differentiation medium (PADM), and adipocyte medium (AdM). After completion of every stage of the cell culture, HPA-v cells were harvested and processed for further analyses. The process of ADG was performed in 3 independent experiments. HPA-v cells were cultured in HG conditions during the whole ADG (chronic exposure) or during a single stage of ADG, producing 11 variants at different stages of adipocyte development (Figure 1). In HG-treated variants, a full medium (5.5 mM of glucose) was supplemented with d-(+)-Glucose (Sigma-Aldrich, Saint Louis, MO, USA) to obtain a concentration of 30 mM of glucose. In our research model, variants were denoted with letters ‘N’ or ‘H’, which reflected the glycemic conditions applied in subsequent stages of cell culture, normoglycemic (NG) conditions, or HG conditions. Variants of HG-treated HPA-v cells could be chronically (eg, H, HH, HHH) or intermittently treated cells (eg, NNH - HPA-v cells were cultured in NG conditions during preadipocyte proliferation and differentiation, whereas the cells were HG-treated during adipocyte maturation). The latter variants (with a single hit of HG) were designed to measure the impact of HG treatment on modulation of expression of miRNAs and their target genes related to inflammation, insulin sensitivity, glycolysis, oxidative stress defense, and DNA repair at a particular stage of ADG, along with the influence of glycemic normalization.

Overly normoglycemic variants (N, NN, NNN) served as a control and depicted physiological ADG. Overly hyperglycemic variants (H, HH, HHH) were designed to show changes during ADG in diabetic subjects. Finally, the research on HPA-v cells sought to investigate the impact of chronic (mimicking diabetes) and intermittent (mimicking IFG) HG, as compared to NG at a particular stage of HPA-v cell culture.

The HPA-v cell culture protocol was validated by the manufacturer and previously by our team in terms of successful ADG of cultured Ads on the basis of: i) BODIPY 505/515 staining, ii) Raman spectroscopy of lipid content, and iii) analysis of expression of the adipogenic markers (PPARγ and C/EBPα) at mRNA and protein level; showing functional competence of cultured Ads for lipid accumulation [38].

RNA ISOLATION AND RT-QPCR: Trypsinized HPA-v cells were counted for RNA isolation and approximately 6×106 cells were used for RNA isolation in every studied variant. Firstly, RLT buffer and β-mercaptoethanol were used for cell lysis, and subsequently, homogenization was done by 20-gauge (0.9 mm diameter) needle pipetting (with attached RNAse-free syringe). Next, RNA was isolated from HPA-v cells using the AllPrep DNA/RNA/Protein Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s instructions, and its concentration and purity were measured with a Synergy HT microplate reader (Biotek, Winooski, VT, USA). Isolated RNA was stored frozen at −80°C for further processing. Reverse transcription of obtained RNA was performed in a Tgradient Thermocycler (Biometra, Göttingen, Germany) using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions. Expression of the studied genes was estimated by RT-qPCR method, using TaqMan Gene Expression MasterMix and TaqMan Gene Expression Assays, as shown in Table 2 (Applied Biosystems, Foster City, CA, USA). The RT-qPCR protocol included the following steps: UNG incubation (50°C, 2 min), polymerase activation (95°C, 10 min), and 40 PCR cycles comprising of denaturation (95°C, 15 s) and annealing/elongation, (60°C, 1 min). Expression profiling of the examined genes was preceded with an experimental selection of 2 out of 16 tested reference genes using TaqMan Array Human Endogenous Control Plates (Applied Biosystems, Foster City, CA, USA) and RefFinder [39], as described previously [38]. Gene expression data were normalized to the average Ct values obtained for RPLP0 (ribosomal protein lateral stalk subunit P0) and UBC (ubiquitin C).

MIRNAS SELECTION, MIRNA ISOLATION, AND RT-QPCR: Total RNA was isolated from HPA-v cells with a miRVANA Isolation Kit (Applied Biosystems, Vilnius, Lithuania) and reverse transcribed with TaqMan™ MicroRNA Reverse Transcription Kit and Custom RT Primer pool (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions in a Tgradient Thermocycler (Biometra, Göttingen, Germany). To select differentially expressed miRNAs in the studied model, we performed screening with TaqMan Low-Density Arrays (TLDA) cards involving 754 miRNAs (including an internal negative control assay; Applied Biosystems, Foster City, CA, USA) in the 7900HT Fast Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc., Waltham, MA, USA) using TaqMan® Universal PCR Master Mix II, no UNG (Applied Biosystems, Foster City, CA, USA), as described previously [40]. Based on the obtained results as well as the bioinformatics analysis and literature research, we selected the miRNAs that either target the studied genes or participate in their downstream signaling. For evaluation in HPA-v cells we selected: let-7g-5p, miR-10b-5p, miR-132-3p, miR-145-5p, miR-16-5p, miR-20a-5p, miR-21-5p, miR-214-3p, miR-222-3p, miR-374a-5p, miR-93-3p, miR-99b-5p, all being depicted with their respective targets in Table 1. Expression levels of the selected miRNAs were evaluated in duplicate with custom TLDA cards and TaqMan Gene Expression Assays (as shown in Table 3). Expression of miRNA was normalized to the arithmetic average of Ct values for U6 and let-7b-5p, as identified by RefFinder, considering multiple candidate reference miRNAs [39].

STATISTICAL ANALYSES:

Statistical analyses were performed on −ΔCt values in GraphPad 6.0 software and data were presented as mean±SD. To evaluate the differences in expression of the studied genes and miRNAs, we used several statistical approaches: one-way ANOVA with Tukey’s post hoc test to detect changes during ADG in NG (‘N’ vs ‘NN’ vs ‘NNN’); chronic HG (‘H’ vs ‘HH’ vs ‘HHH’) and after exposure to HG during the preadipocyte proliferation (‘H’ vs ‘HN’ vs ‘HNN’); two-tailed t test (to compare expression changes between ‘N’ and ‘H’ variants); and finally, one-way ANOVA with Dunnett’s post hoc test (to evaluate the impact of chronic and single stimuli of HG in relation to NG (‘NN’ vs ‘NH’, ‘HN’, ‘HH’ and ‘NNN’ vs ‘NNH’, ‘NHN’, ‘HNN’, ‘HHH’). P<0.05 was considered as significant.

STUDY SUBJECTS AND ETHICAL ISSUES: Eighty-three unrelated Polish subjects who were at least 50 years old were enrolled into the study from among the patients admitted to the Clinical Department of General and Oncological Surgery at the Central Teaching Hospital of the Medical University of Łódź (Poland) for planned laparoscopic cholecystectomy between October 2017 and January 2020. To classify probands into particular study groups, we used additional inclusion criteria in accordance with the 2021 American Diabetes Association criteria. Subjects without history of glucose metabolism disorders, of FPG below 100 mg/dL, glycated hemoglobin (HbA1c) below 5.7% were assigned to NGM group. For the IFG group: FPG in the range of 100–125 mg/dL and HbA1c ranging from 5.7% to 6.4%. The earlier diagnosis of T2DM in medical history of the subjects, the ongoing therapy (insulin and/or metformin and/or sulfonylurea derivatives and/or flozins), and currently measured FPG and HbA1c were the basis of assignment to the T2DM group. Exclusion criteria were: neoplasms or their medical history along with acute inflammation. The participants included to the study were carefully selected to preclude differences between the 3 studied groups in terms of age, body mass index (BMI), body mass, waist-hip ratio (WHR), and waist circumference. Equal distribution of male and female subjects in each of the studied groups was not provided, possibly due to a higher frequency of cholecystectomy in women [41]. Importantly, evaluation of mRNA expression was finally conducted on 52 subjects (34 females and 18 males) (Figure 2), while miRNAs changes were measured only in 40 study subjects (13 male and 27 female probands) due to the scarcity of RNA material isolated from VAT samples in comparison to the number of planned analyses. The NGM group for miRNAs expression profiling had 5 males and 9 females, the IFG group had 5 males and 8 females, and the T2DM group had 10 females and 3 males.

The protocol of the study was approved by the Ethics Committee of the Medical University of Łódź, Poland (no. RNN/287/13/KE). All study participants were informed about the study protocol and provided written consent for the collection of a sample of visceral adipose tissue from the visceral peritoneum as well as a sample of peripheral venous blood (collected after a minimum of 8 h of overnight fasting).

BIOCHEMICAL AND ANTHROPOMETRIC MEASUREMENTS: Data collection included: height, weight, hip, and waist circumference, which were used to calculate BMI and WHR. Analysis of blood samples involved the evaluation of FPG, HbA1c, fasting insulin (FI), triglycerides (TG), total cholesterol (TC), and high-density lipoprotein cholesterol (HDL-C) levels. The Friedewald formula was used to calculate low-density lipoprotein cholesterol (LDL-C) levels. Glucose metabolism-associated parameters were used for the calculation of insulin resistance estimates such as TG to HDL-C ratio (TG/HDL) and homeostasis model assessment for insulin resistance (HOMA-IR) [42].

VAT SAMPLE COLLECTION AND ISOLATION OF RNA:

VAT samples collection took place after cholecystectomy, in an appropriate location in the visceral peritoneum, to avoid contact with the intestines and vital vessels, which minimized the risk of complications, while any observed complications were rather associated with cholecystectomy itself. The samples of VAT (about 1 g) were collected into PBS-containing sterile tubes and were immediately transported on wet ice (4°C) for preliminary processing from the operating room to the laboratory within the same building. The samples were weighed and 250 mg of fat samples were homogenized with Lysis/Binding Buffer (700 μl) and frozen at −80°C. Total RNA was isolated using the mirVana™ miRNA Isolation Kit (Invitrogen™, Carlsbad, CA, USA), as suggested by the manufacturer. The purity and concentration of isolated RNA were measured with a Pico100 Spectrophotometer (Picodrop). Only samples of RNA reaching ~ 2 (A260/A280) were chosen for expression profiling.

RT-QPCR: Expression profiling was preceded by experimental and RefFinder-based selection of 2 pairs of the most stable reference genes, using either TaqMan Array Human Endogenous Control Plates (Applied Biosystems, Foster City, CA, USA) or evaluating expression levels of numerous miRNAs and well-known reference genes (U6, RNU48, RNU44), as described earlier [13]. Our analysis indicated that PPIA and POLR2A are the most suitable references for mRNA expression profiling, while miR-93-5p and miR-17-5p were chosen for normalization of miRNA expression data. Expression levels of the selected genes and miRNAs were performed with TaqMan assays depicted in Tables 2 and 3, as described in subsections RNA isolation and RT-qPCR and miRNAs selection, miRNA isolation, and RT-qPCR of Material and Methods concerning HPA-v cells. In addition to miRNA examined in HPA-vs, we evaluated the expression of miR-26a-5p and miR-26b-5p ex vivo, which are the molecules studied previously in HPA-vs [36].

IN SILICO ANALYSIS: Functional analysis of upregulated miRNAs in VAT of T2DM female subjects was evaluated with miEAA 2.0 (WikiPathways, miRPathDB) [34] and miRSystem [43]. In over-representation analysis (ORA) involving the WikiPathways database, the result pathways were refined to those with strong experimental evidence and significantly annotated to a minimum of 4 out of 5 tested miRNAs. It was performed with P value adjustment using FDR (Benjamini-Hochberg) adjustment, and significance level P<0.05. Data are presented as a wordcloud of the top 100 categories by P value.

The miRNA enrichment analysis involving miRSystem was refined to pathways which were significantly annotated to at least 4 miRNAs. Moreover, each hit included only validated targets, the O/E ratio (observed to estimated ratio) was greater than or equal to 2, and the number of total genes in a pathway was greater than or equal to 25 and less than or equal to 500. Functional annotation included KEGG, Biocarta, Pathway Interaction Database (human only), Reactome (human only).

STATISTICAL ANALYSIS: Collected data were checked for equity of variance and normality of data distribution via Levene and Shapiro-Wilk tests, respectively. To evaluate differences among the 3 studied groups in terms of age, biochemical, and anthropometric parameters, Kruskal-Wallis one-way ANOVA with Dunn’s post hoc test was used (Supplementary Table 1). The mRNA and miRNA expression data were analyzed among the 3 studied groups based on −ΔCt values with either parametric or nonparametric one-way ANOVA (Tukey’s/Dunn’s post hoc tests) depending on the results of the Levene and Shapiro-Wilk tests. Two-way ANOVA was used to evaluate the association of expression changes of miRNAs and mRNAs of the selected genes with sex. Univariate correlation among levels of individual mRNAs, miRNAs, sex, age, and anthropometric and biochemical parameters were checked via Spearman’s rank correlation, while correlation heatmaps were generated with MultiExperimentViewer software 4.9.0 (The Perl Foundation) [44]. Analysis of covariance was performed for expression data derived from female VATs only while taking into account either age, BMI, WHR, TC, or HDL-C. Statistical analyses were performed with STATISTICA 10.0 software (StatSoft, Tulsa, OK, USA), while GraphPad Prism 6.0 software (GraphPad Software, Inc.) was used to generate graphs of expression data. P<0.05 was considered significant.

Results

In Vitro Studies

THE IMPACT OF HIGH GLUCOSE ON EXPRESSION LEVELS OF GENES AND MICRORNAS EXAMINED DURING HUMAN VISCERAL ADG (HPA-V CELLS): Firstly, we investigated expression changes of the examined genes during the progression of normoglycemic visceral ADG (N, NN, NNN), which reflected a physiological process and served as a control to measure the impact of HG. mRNA expression of ATM, TIGAR, and NFKB1 showed a similar pattern during normoglycemic ADG, being decreased in Ads cultured in NG since the differentiation stage (Figure 3). This may indicate their similar role and molecular interconnection during visceral ADG. On the other hand, the transcription of INSR, a well-known marker of ADG, was increased since the differentiation stage; whereas the SOD2 mRNA level was upregulated only in mature Ads (NNN vs NN, NNN vs N). In contrast, HG evoked a decrease of TIGAR mRNA only after the differentiation stage (HH vs H). It appears that expression of ATM, NFKB1, SOD2, and INSR is stable from preadipocytes to mature Ads during the course of ADG conducted under chronic HG (H, HH, HHH), which stays in dramatic contrast to their expression pattern observed during the physiological process (N, NN, NNN).

While comparing ADG in HG and NG, none of the examined genes was transcriptionally affected by HG at the stage of preadipocytes (H vs N). At the second stage of ADG, either chronic or intermittent HG, resulted in higher TIGAR and NFKB1, as well as lower INSR mRNA expression, showing signs of hyperglycemic memory in differentiated Ads. Mature Ads exposed to chronic HG showed: downregulated mRNA levels of SOD2 and INSR; upregulated mRNA levels of ATM and TIGAR with a tendency for increase of NFKB1 (P=0.07). Each variant exposed to intermittent HG had similarly elevated mRNA levels of ATM and TIGAR in mature Ads, with analogously lower transcription of INSR (NNH, NHN, HNN vs NNN). The downregulation of INSR in each variant of HG-treated mature Ads (NNH, NHN, HNN, HHH) provided further evidence of a decreased response of Ads to insulin, and thus, glucose utilization, as well as indicated a mere hyperglycemic memory. SOD2 mRNA was increased in Ads exposed to HG only during the maturation stage (NNH vs NNN), highlighting the diverse direction of its changes depending on the duration and stage of exposure to HG (SOD2 downregulation, HHH vs NNN). Altogether, the expression of all of the examined genes, except for SOD2, might be a subject of hyperglycemic memory in differentiated and/or mature visceral Ads.

Regarding miRNAs, we observed a decreased level of solely miR-93-3p after the differentiation stage (NN vs N; Figure 4). By contrast, 4 of the studied miRNAs manifested changes in transcriptional levels during ADG under chronic HG. Specifically, we observed upregulation of miR-145-5p and a downregulation of miR-222-3p (HHH vs H, HH vs H), along with decreased let-7g-5p and miR-16-5p exclusively after adipocyte maturation (HHH vs HH). These results indicate that HG affects a set of miRNAs, altering their physiological expression pattern during visceral ADG.

While comparing ADG in HG and NG, we found a decrease of only miR-93-3p in HG-treated preadipocytes (H vs N). Additionally, chronic, but not intermittent, HG exposure resulted in an elevation of let-7g-5p and miR-16-5p at the stage of differentiated Ads (HH vs NN). Considering mature Ads, we noticed a decrease of miR-10b-5p under chronic HG.

The impact of intermittent HG treatment was noticeable exclusively in mature Ads and resulted in reduced miRNA levels. For instance, we detected a HG-induced decrease of transcription of miR-145-5p and miR-21-5p, which was in line with the observed upregulation of SOD2. Intermittent exposure to HG elicited also a reduction of levels of miR-10b-5p (NHN vs NNN) and miR-132-3p (HNN vs NNN), implying rather a stage-specific manner of their targets’ transcriptional modulation.

Simultaneously, we observed that mRNA expression of SOD2 and INSR changed oppositely to their potentially modulating miRNAs, eg, INSR and let-7g-5p (HH vs NN, chronic HG), SOD2 and miR-145-5p or miR-21-5p (HHH vs NNN). Thus, we speculate that the observed changes in gene expression could occur with the contribution of miRNAs.

Altogether, the results suggest that the expression of examined genes and some of their epigenetic regulators (miRNAs) is affected by the glycemic condition during ADG, especially in differentiated and mature Ads. The obtained results encouraged us to perform an ex vivo data evaluation in VAT of subjects with NGM, IFG, and T2DM. These 3 groups of subjects were designed in a way that precluded significant differences in age, and to some extent mitigated the impact of central and overall adiposity (WHR, BMI).

EX VIVO STUDIES – DESCRIPTION OF THE STUDY SUBJECTS:

We enrolled 52 subjects, including 34 females and 18 males, into the molecular examination study and assigned them to 3 groups: NGM (normal glucose metabolism), IFG (impaired fasting glucose), and T2DM (diabetic). Both female and male subjects were assigned to the study groups based on FPG and HbA1c levels with avoidance of significant differences in age, body mass, and BMI (Supplementary Table 1). However, the impact of adiposity was not overly eliminated, as all subjects (normal weight, overweight, obese) were assigned to the study groups accordingly to BMI criteria.

Women of all study groups were characterized by a similar distribution of adipose tissue (visceral) according to WHR (median WHR ranging from 0.90 to 0.93). We noticed that female T2DM patients had higher values of FI and HOMA-IR compared to NGM female subjects, as well as increased FI and HbA1c levels in relation to IFG female probands. This provides further confirmation that the subjects were properly classified and assigned to the study groups. We also found an increase of TG and TG/HDL in T2DM female subjects compared to NGM ones. Decreased levels of TC and LDL-C in T2DM females could possibly be caused by the recommended statin administration to treat dyslipidemia in T2DM and to prevent cardiovascular complications [45]. Surprisingly, FI and HOMA-IR were indifferent between IFG and NGM females. Differences in FI may arise from an early stage of development of the glucose metabolism disturbances in IFG, where IR has not yet induced hyperinsulinemia, which was consistent with the observation of Dulkadiroğlu et al [46].

Taking into account the description of the studied subjects: i) the studied subgroups of female subjects were well-matched in terms of age, WHR, body mass, and waist circumference; (ii) the low number of male study subjects; and (iii) the impact of adipose tissue distribution, which was not eliminated in male study subjects; we decided to focus only on analysis of gene and miRNA expression changes in female VAT samples. The results of analyses of expression of the examined genes and miRNAs and expression correlations in VAT of male study subjects are shown in Supplementary data (Supplementary Table 2 and Supplementary Figures 1, 2).

EXPRESSION CHANGES OF THE EXAMINED GENES AND MIRNAS IN VAT: In general, we observed a substantial level of overlap of gene expression levels between the female subgroups. However, we still detected increased levels of NFKB1 and TIGAR in VAT of T2DM women, as compared to either IFG or NGM group (Figure 5). Expression of several studied miRNAs (miR-10b-5p, miR-132-3p, miR-20a-5p, miR-21-5p, miR-26a-5p) was higher in T2DM as compared to the NGM group (Figure 6). The changes in expression of miR-10b-5p and miR-21-5p seemed to reflect the upregulated mRNA level of their modulatory gene, NFKB1, which was not confirmed by the analysis of correlation between NFKB1 and miRNAs. This could be associated with a (slight) tendency of miR-10b-5p expression to similarly increase in both IFG and T2DM females as compared to NGM ones (IFG vs NGM, P=0.074). While comparing IFG and NGM females, we also observed that an increase of miR-145-5p nearly reached statistical significance (P=0.0558, post hoc IFG vs NGM=0.050). These results, in line with our previous research, could indicate that the expression of distinct miRNAs varies depending on the severity of hyperglycemia (IFG vs T2DM). Finally, our observations may suggest a NF-κB-dependent modulation of transcription of miR-10b-5p and miR-21-5p in VAT of T2DM subjects.

CORRELATION ANALYSIS AMONG EXPRESSION OF THE EXAMINED GENES AND MIRNAS IN VAT: Next, we compared miRNAs and mRNAs expression detected in all of the female VAT samples via correlation analysis (Figure 7). We found 6 significant correlations among mRNAs and miRNAs, which could imply their regulatory relationship. Expression of exclusively one of the upregulated miRNAs, miR-21-5p, showed a moderate positive correlation with the level of SOD2 (r=0.42), its well-known target gene, and with TIGAR. We also noticed an emerging potent and positive correlation between the expression of SOD2 and miR-145-5p (r=0.67), both being unchanged in VAT of T2DM females. The only negative correlation was found between levels of TIGAR and miR-16-5p; however, similarly to the above-mentioned correlation with miR-21-5p, it has not been mechanistically validated so far. Finally, let-7g-5p and miR-374a-5p manifested slight and positive correlations with their direct targets, INSR and ATM, respectively.

We also found that TIGAR, NFKB1, ATM, and INSR were almost always positively correlated with each other, except for the NFKB1-ATM pair, while uncorrelated with SOD2 levels. This provides further implication of interrelated roles of these 4 genes in VAT, simultaneously pointing out the different role of SOD2 (the antioxidative one).

Considering the upregulated miRNAs, we detected several moderate-to-strong positive correlations between the levels of: miR-26a-5p and miR-10b-5p; miR-132-3p and miR-214-3p; along with a pronounced correlation between miR-132-3p and miR-21-5p. This suggests that they might be either commonly regulated by an upstream factor or they might regulate similar biological phenomena in VAT.

CORRELATION ANALYSIS OF EXPRESSION OF THE EXAMINED GENES AND MIRNAS WITH AGE AND CLINICAL PARAMETERS IN VAT: Firstly, levels of miR-99b-5p and NFKB1 demonstrated either a positive or negative slight correlative relationship with age, respectively (Figure 8). As we expected, none of the tested molecules was correlated with body mass and BMI. However, levels of miR-99b-5p and TIGAR showed a slight positive correlation with WHR, while miR-16-5p was negatively correlated with this parameter. Taking into account the other detected correlations, such as the ones related to glucose metabolism and IR (eg, FPG, FI, TG/HDL, HOMA-IR), the above-mentioned molecules might be especially associated with VAT accumulation and resultant metabolic repercussions. Additionally, the levels of nearly all of the upregulated genes or miRNAs in VAT of T2DM females, except for NFKB1, had at least a moderate positive correlation with FPG levels. The expression of NFKB1, TIGAR, miR-132-3p, miR-21-5p, and miR-26a-5p was positively correlated with at least 3 out of 4 examined indicators of glucose metabolism and IR (FPG, FI, HOMA-IR, HbA1c). By contrast, miR-10b-5p and miR-20a-5p were correlated with FPG and/or HbA1c levels, but were uncorrelated with parameters of IR. This could suggest their slightly different role in molecular signaling of VAT, as compared to 3 upregulated miRNAs (miR-21-5p, miR-132-3p, and miR-26a-5p) found to be broadly connected with IR. In particular, miR-21-5p was moderately and positively correlated with all of the measured indicators of glucose metabolism and IR. The latter observation is especially interesting as miR-21-5p can be modulated by NF-κB, or target NFKB1 and SOD2, which are genes involved in inflammation and antioxidative defense.

Considering the parameters of lipid metabolism, we found a moderate positive correlation between TG levels and miR-21-5p or miR-26a-5p and a weak correlation between TG and TIGAR levels. However, the rest of the correlations were negative and/or regarded not significantly changed molecules, as follows: miR-16-5p, miR-26b-5p, miR-374a-5p, and SOD2. It appears that the expression of miR-374a-5p and SOD2 might be more specifically associated with lipid metabolism, which could explain their stable expression levels among the studied groups. On the other hand, the rest of the molecules (miR-16-5p, miR-26a/b-5p, miR-21-5p, TIGAR) seem to be connected with both glucose and lipid metabolism.

Expression of let-7g-5p, miR-145-5p, miR-214-3p, miR-222-3p, miR-93-3p, and ATM was not changed and was uncorrelated with any examined parameter. However, we cannot exclude the possibility that the role of some of these miRNAs in VAT could be associated rather with ADG, or other phenomena strictly related to obesity, such as DNA damage response. We were unable to assess such changes in our study due to the predominance of mature Ads over differentiating ones in VAT, and a similar level of visceral adiposity among studied groups.

ANALYSIS OF COVARIANCE:

As we observed several prominent correlation coefficients, we decided to perform multivariate analyses taking into account several parameters (age, BMI, WHR, TC, and HDL-C). Other parameters were either correlated with FPG and HbA1c (being strictly associated with assignment to the study group, eg, TG) or redundant, such as body mass or waist or hip circumference, making BMI and WHR representative values. In most cases, the difference in gene expression was still significant. Specifically, NFKB1, miR-20a-5p, miR-21-5p, and miR-26b-5p were still significantly upregulated while adjusting for any of the aforementioned parameters. TIGAR was significantly changed after adjustment for age, BMI, TC, or HDL-C. miR-132-3p was still upregulated while adjusting for age, WHR, BMI, or HDL-C. Finally, miR-10b-5p was significantly upregulated only when adjusted for age and HDL-C. Thus, the expression of only miR-10b-5p was partially associated with indicators of both central and overall adiposity. Surprisingly, miR-145-5p was significantly upregulated in VAT of IFG subjects when adjusting for BMI.

IN SILICO ANALYSIS OF THE SET OF 5 UPREGULATED MIRNAS IN VAT OF FEMALE T2DM SUBJECTS: To recognize the potential biological phenomena regulated by 5 miRNAs upregulated in VAT of diabetic women (miR-10b-5p, miR-132-3p, miR-20a-5p, miR-21-5p, miR-26a-5p), we performed a functional analysis. The silico tool, miEAA, indicated the association of examined miRNAs with the phenomena typically associated with HG-afflicted VAT (ADG, cell cycle, and its control, miRNA Regulation of DNA Damage Response, DNA damage response only ATM dependent) (Table 4). These data are consistent with several top results generated by the miRSystem, also showing association with p53 pathway and downstream insulin signaling of pathways (PI3K/Akt) (Supplementary Table 3). Altogether, these results may suggest a contribution of 5 upregulated miRNAs to the signaling associated with DNA damage along with the regulation of ADG.

EXPRESSION OF THE EXAMINED GENES AND MIRNAS IN VAT – TWO-WAY ANOVA ANALYSIS:

In addition to analyses conducted for samples collected from female subjects, we performed a preliminary analysis regarding the impact of sex. Two-way ANOVA with Tukey’s post hoc test demonstrated that the miR-21-5p level was significantly upregulated when comparing female T2DM to NGM subjects (P=0.014). Additionally, we found an upregulation of TIGAR expression in female T2DM when compared to male IFG subjects (P=0.045). These preliminary results possibly suggest that sex may affect the expression of some of examined mRNAs and miRNAs in VAT. However, mainly due to low number of male subjects, these results should be interpreted with great caution and research needs to be repeated on groups with increased and equal numbers of male and female subjects.

Discussion

In the current study we demonstrated that the expression pattern of inflammation-, oxidative stress- and glucose metabolism-related genes is greatly changed during physiological visceral ADG. To the best of our knowledge, we may be one of the first teams to evaluate the expression pattern of NFKB1 and TIGAR during white ADG. Moreover, similarly decreased mRNA expression observed in expression patterns of NFKB1, ATM, and TIGAR may indicate their molecular interconnection during visceral ADG. Our data also are in line with reports showing increased expression of INSR and SOD2 during white visceral ADG [47–49], and suggest the contribution of all examined genes to the development of functional visceral Ads.

We mainly aimed to determine the impact of hyperglycemic conditions on the expression of the studied genes along with their targeting miRNAs. ADG of the studied HPA-v cells was functionally validated using Raman spectroscopy and lipid droplet staining in our previous report [38]. In contrast to normoglycemic ADG, HG stabilized the expression of all of the examined genes, except for TIGAR. This implies that HG may alleviate expression changes of the examined genes, which could underlay phenotypic alterations of HPA-vs, including their hypertrophy [38]. Similarly, Ads that were HG-treated during preadipocyte proliferation had stable expression of ATM, NFKB1, and INSR during ADG, which was significantly different from NG conditions, but only in differentiated and/or mature Ads. Interestingly, the expression of TIGAR was not completely stabilized by this early hyperglycemic hit, but it was the same as the expression under chronic HG treatment. Strikingly, expression changes of ATM, INSR, and TIGAR upon intermittent HG were highly similar to the impact of chronic HG in mature Ads, indicating the effect of hyperglycemic memory. This is in line with our previous reports and the one provided by Andersen et al, who demonstrated that preadipocytes have a memory of a donor (obese/obese with T2DM) in cell culture, based on the remodeling of DNA methylation and expression changes of genes associated with PPARγ, for example [36,40,50]. In contrast, another type of modulation was noticed for SOD2, and its increase in NNH cells could serve as a response for short-term and acute stressors, in contrast to its downregulation in HHH variant, and being in line with previous reports [51,52]. Additionally, when compared to NG, the chronic HG-induced modulations of NFKB1, INSR, and TIGAR, for example, were also consistent with the ones observed in retinal ganglion cells, cardiomyocytes, and insulin-resistant adipocytes [53–55].

Taken together, the detected changes in gene expression suggest that chronic HG can elicit: (i) reactive oxygen species (ROS) accumulation; (ii) ATM expression increase along with its potential activation; (iii) antioxidative defense via blunting glycolysis and increasing glucose influx into the phosphopentose pathway (PPP) via the ATM/p53/TIGAR pathway to produce NADPH needed for reduction of glutathione; and (iv) activation of NF-κB and initial induction of SOD2 expression. This sequence of events possibly led to the exhaustion of antioxidative systems, including downregulation of SOD2, being accompanied by reduction of INSR, and, possibly, exacerbation of glucose metabolism dysregulation.

Thus, on the basis of the analysis of gene expression in Ads, one could speculate about: (i) a coregulatory relationship among ATM, INSR, and TIGAR; (ii) metabolic memory induced by HG for genes involved in glucose metabolism and DNA damage response; (iii) reduced response of Ads to insulin, and thus, glucose utilization [47], and (iv) induction of well-known HG-triggered oxidative stress [56].

Among the miRNAs analyzed in Ads, many of the observed expression changes (induced by either chronic or intermittent HG) were convergent with their putative targets. Moreover, the expression changes of ATM and TIGAR (also in either chronic and intermittent HG) agree with the previously reported changes of miR-26a-5p and miR-26b-5p, respectively [36]. Similar to the gene expression analysis, we observed that miR-10b-5p manifested signs of epigenetic memory under intermittent HG. Thus, HG could change the expression of some miRNAs, even during a single ADG stage, to negatively regulate the examined genes, and therefore lead to HG-induced detrimental outcomes. Moreover, to the best of our knowledge, we are the first to present a significant reduction of miR-93-3p in HG-exposed preadipocytes, but unrelated to the expression of the predicted targets assessed in our research. Upregulation of let-7g-5p and miR-16-5p in chronic HG-exposed differentiated Ads, underlining similarity of their expression profiles, has not yet been suggested by any cluster analysis. Expression of miRNAs in Ads needs to be explored in ex vivo studies.

Our in vitro experiment suggested the existence of metabolic memory, which is achieved, at least in part, through epigenetic mechanisms. This encouraged us to evaluate our results in VAT of NGM, IFG, and T2DM subjects. To accomplish our aim, we determined whether the level of dysglycemia observed in IFG is sufficient to induce the expression changes observed in diabetic subjects. Preliminary analysis of subjects’ anthropometric and biochemical parameters revealed that the results respecting male subjects could be affected by the not overly eliminated impact of adipose tissue distribution. Additionally, considering the low number of male subjects and the fine balance of age, WHR, body mass, waist circumference in the female subgroups, we decided to focus mainly on female subjects in our ex vivo experiment. The observed upregulations of NFKB1 and TIGAR in VAT of T2DM females were in line with the results of exposure to chronic HG in vitro – upregulation of these genes in differentiated adipocytes and upregulated TIGAR in mature adipocytes. Nevertheless, VAT samples of IFG and T2DM females showed distinct expression patterns of NFKB1 and TIGAR along with 5 miRNAs (miR-10b-5p, miR-132-3p, miR-26a-5p, miR-20a-5p, and miR-21-5p). Moreover, our results may indicate that molecular abnormalities of the examined molecules are more prominent in VAT of diabetic female patients rather than in VAT of IFG ones. Consistently, the correlation and the in silico analyses also further supported the connection of the upregulated miRNAs with glucose metabolism, IR, inflammation, and ADG in T2DM-affected VAT [57]. VAT of T2DM patients is generally characterized by the increased level of p53 and its activation, mainly due to DNA damage response signaling via phosphorylated ATM. Activated p53 directly induces TIGAR to rescue repercussions of oxidative stress via blunting glycolysis and producing NADPH [37]. There is a little information about the expression of TIGAR in diabetic environment, including the research on HG-treated cardiomyocytes manifesting increased levels of acetylated p53 and expression of TIGAR, in line with our research [53]. Moreover, taking into account that activation of p53 is concomitant with an upregulation of NF-κB [37], our results seem to be consistent with the elevated mRNA level of NFKB1 observed by Camarena et al in epicardial AT of diabetic patients [58]. A recent study on GDM reported upregulated placental NFKB1 of GDM women associated with insulin resistance, which was in line with our correlation analysis [59]. Considering the abundance of upregulated pro-inflammatory NF-κB in various tissues in dysregulation of diabetic glucose metabolism, our study supports its contribution to systemic low-grade inflammation and glucose metabolism dysregulation [8]. In contrast to our results, the upregulation of SOD2 was observed in T2DM subjects, specifically in VAT-derived mitochondria [60]. Lack of changes of ATM and INSR in IFG and T2DM subjects could be associated with similar age, BMI, and WHR levels among female study participants. Indeed, Pereira et al observed no changes in expression of INSR in SAT of T2DM subjects, whereas decreased levels in VAT of obese subjects were associated with BMI [61,62]. This may also explain the lack of changes in expression of at least NFKB1 in IFG subjects as compared to NGM ones. We also cannot exclude a possibility of changes in the phosphorylation status of ATM in VAT of IFG and T2DM subjects, which could confirm the activation of the DNA damage response pathway and a cascade of events associated with oxidative stress and IR. In our in vitro model, the changes observed in the variants treated with intermittent HG were not reflected in IFG subjects. Although we similarly did not observe an upregulation of NFKB1 in VAT of IFG subjects, our in vitro results still did not provide answers regarding the unchanged expression of TIGAR.

Contrary to our expectations, we did not observe any miRNA to be changed in IFG subjects. However, we observed a trend of increased miR-145-5p in IFG subjects as compared to NGM ones, which was partially in line with a similar reduction of miR-145-5p level in all single-HG hit variants, along with the unchanged levels for chronic HG one. A few studies also recorded reduced miR-145-5p levels after a short exposure of cardiomyocytes or vascular smooth muscle cells (VSMCs) to HG [63–65]. Hence, fluctuations of HG could evoke cardiovascular T2DM complications with a possible contribution of miRNAs [66]. Especially with respect to oxidative stress, as SOD2 is a presumed target of miR-145-5p, these results seem to be consistent with the study by Sun et al showing even greater ROS production in Ads treated with intermittent HG as compared to stable HG [67]. On the other hand, expression of miR-10b-5p appeared to be similarly upregulated in T2DM and IFG females, which was inconsistent with its decrease induced by chronic and intermittent HG in vitro. This discrepancy in comparison to in vitro miRNA expression (eg, miR-20a-5p or miR-21-5p) was most plausibly connected with the presence of other types of cells, including the inflammation-inducing ones (eg, macrophages) in VAT. In our previous research, the second member of the conserved miR-10 family, miR-10a-5p, was the most upregulated miRNA in VAT of IFG and T2DM females and thus indicated to be subjected to hyperglycemic memory [13]. Moreover, in accordance with our results on VAT samples, Zhou et al observed increased expression of miR-20a-5p was increased in VAT of ob/ob mice, which is considered as a murine model of T2DM [68]. Interestingly, Nunez Lopez et al reported in 2022 that pioglitazone treatment of T2DM patients reduced levels of adipose-derived circulating miR-20a-5p, which was associated with improved insulin sensitivity [69]. To the best of our knowledge, research on miRNAs in diabetic patients has mainly focused on miRNAs circulating in blood. A common adipose tissue-related miRNA, miR-21-5p, has increased expression in blood of T2DM subjects [70,71] and it recently gained attention as a potential biomarker of improved dysglycemia [72] or T2DM remission after sleeve gastrectomy [73]. miR-26a-5p was also upregulated in plasma of T2DM patients with irritable bowel syndrome [74]. We speculate that due to the upregulation of miRNAs associated with inflammation and IR, VAT may contribute to systemic dysregulation of glucose metabolism of T2DM patient, with a possible role of VAT-derived exosomes [21,75,76]. In contrast, several studies reported downregulation of some of the detected changes in expression of miR-26a-5p and miR-10b-5p [77–80], but most of them investigated the impact of obesity, which was to some extent mitigated in our study. Moreover, Assmann et al found an increase of circulating miR-132-3p was accompanied by decrease of miR-132-3p expression detected in omental AT of T2DM subjects, but the latter was associated with macrophage infiltration into this fat depot [81].

Surprisingly, we detected predominantly positive correlations among the miRNAs and their target genes in females’ VAT (especially between SOD2 and miR-145-5p), which differed from our in vitro observations. Therefore, the signaling of these molecules might be interconnected, but their expression changes could not be detected in VAT of hyperglycemic subjects due to a low number of differentiating cells and predominance of mature Ads and macrophages in VAT. Nevertheless, Diaz et al performed global analyses and showed that miRNAs and their mRNA targets remained either in positive or negative correlative relationships in examined material. This led to the conclusion that there is a complex bidirectional molecular network involving a feed-forward regulation dependent on transcription factors or regulations among miRNAs and miRNAs, or mRNAs and mRNAs [82]. This aspect makes it especially difficult for observational studies to confirm the negative impact of miRNAs on their targets. However, we did notice a moderate negative relationship between TIGAR and miR-16-5p, which has not been reported previously and could serve as a topic of a new investigation.

Additionally, expression changes of a majority of the examined molecules were found in female, but not male, VAT samples, which could suggest an impact of sex on gene expression. A sex-underlain distinction of miRNAs expression was earlier implied in animal studies on white AT [83,84], and recently, by our team [13]. Nevertheless, such a hypothesis should be confirmed by further studies involving equal numbers of males and females, and our present results should be interpreted with great caution due to the relatively small number of males in the studied groups.

Considering the strong and weak points of our study, we evaluated expression levels of numerous molecules in rarely examined material (HPA-v cells and VAT). In vitro studies provided previously unrecorded data regarding expression changes of several molecules and miRNAs during physiological visceral ADG. Moreover, we demonstrated that HG triggers changes in the expression profile of inflammation- and oxidative-stress-related molecules during visceral ADG, implying that the impact of HG on the expression of some molecules may be preserved during the process of Ads formation. VAT possesses a reservoir of preadipocytes, which are cells ready to differentiate and replace dysfunctional Ads on biological request. Taking into account the low number of studies conducted on VAT taken from IFG and T2DM subjects, our matching of female study subjects is unique, even though we failed to entirely eliminate the impact of overall and abdominal adiposity. Additionally, our preliminary results may possibly suggest a sex-underlain regulation of the expression of some molecules in VAT, although the number of males was lower, and they predominantly lacked visceral obesity as compared to female subjects [13]. We are aware that the presented observations comparing the expression of miRNAs and genes should be taken as suggestions which need to be validated in further research using, for example, gain/loss of function experiments. Further studies are also needed to assess the biological relevance of the reported changes in gene and miRNA expression observed in in vitro and ex vivo experiments. Nevertheless, taking into account the complexity of epigenetic RNA interference (which could be evoked by multiple miRNAs), the presented results may help to decipher the miRNA-mRNA regulatory network responsible for oxidative stress and insulin resistance in visceral adipose tissue (VAT) of T2DM patients.

Conclusions

Our results obtained in vitro imply that HG impacts the expression of the examined genes not only in mature Ads, but also during ADG. Moreover, the expression changes detected in HPA-vs showed signs of hyperglycemic memory. Therefore, HG may alter the expression of miRNAs and genes broadly associated with inflammation during ADG, thus potentially facilitating further hyperplastic, hypertrophic, inflamed, and vulnerable to ROS-induced cellular damage VAT. Accordingly, our ex vivo study showed that VAT of T2DM, but not IFG, females is characterized by upregulation of NFKB1 and TIGAR along with miR-10b-5p, miR-132-3p, miR-20a-5p, miR-21-5p, and miR-26a-5p. The detected changes suggest altered regulation of inflammation, oxidative stress, and glycolysis in VAT of subjects who experience prolonged and markedly raised blood glucose levels.