Exploring differential miRNA expression profiles in muscular and visceral adipose tissue of patients with severe obesity

Abstract

Background

This study aims to investigate the differential miRNA expression profile between the visceral white adipose tissue and the skeletal muscle of people with obesity undergoing bariatric surgery.

Methods

Skeletal muscle and visceral adipose tissue samples of 10 controls and 38 people with obesity (50% also with type 2 diabetes) undergoing bariatric surgery were collected. miRNA expression profiles were analyzed using Next-Generation Sequencing and subsequently validated using RT-PCR.

Results

Approximately 69% of miRNAs showed similar expression in both tissues, however, 55 miRNAs were preferentially expressed in visceral adipose tissue and 53 in skeletal muscle. miR-122b-5p was uniquely identified in skeletal muscle, while miR-1-3p and miR-206 were upregulated in skeletal muscle. Conversely, miR-224-5p and miR-335-3p exhibited upregulation in visceral adipose tissue. Notably, distinctions related to the presence of type 2 diabetes were observed solely in the expression of miR-1-3p and miR-206 in visceral adipose tissue.

Conclusions

This is the first study unveiling distinct miRNA expression profiles in paired samples of visceral adipose tissue and skeletal muscle in humans. The identification of obesity-specific miRNAs in these tissues opens up promising avenues for research into potential biomarkers for obesity diagnosis and treatment.

Introduction

According to the World Health Organization, overweight and obesity are defined as abnormal or excessive fat accumulation that may impair health, and they are normally measured by the body mass index (BMI) [1], being a BMI higher than 40 kg/m² classified as severe obesity [2]. For patients with a high level of obesity, bariatric surgery remains the best option for losing the excessive weight and improving obesity-related comorbidities such as metabolic diseases, hypertension, or dyslipidemia [3].

Obesity is directly related with skeletal muscle and adipose tissue dystrophy, in fact, it is widely established that in people with obesity, white adipose tissue adipocytes become hypertrophic and hypoxic. Also, the tissue changes to a more pro-inflammatory phenotype, richer in M1 macrophages, neutrophils, B-cells, NKT-cells, etc. [4, 5]. In addition, obesity causes insulin resistance in white adipose tissue, liver, and skeletal muscle, which, together with impaired insulin secretion by pancreatic beta cells, has been postulated to be the major cause of the development of type 2 diabetes mellitus in people with obesity [6]. However, although the relationship between obesity and the development of type 2 diabetes is widely accepted, it is still unclear why some people with obesity develop diabetes and others do not.

MicroRNAs (miRNAs) are small non-coding RNA molecules that have emerged as key regulators of metabolic homeostasis [7]. They are present in different tissues related to metabolic processes, including skeletal muscle, which is highly affected during a hyperinsulinemic clamp [8], and adipose tissue, which not only serves as an energy reservoir, but also plays an important role in glucose homeostasis [9]. Furthermore, miRNAs can be secreted from cells and transported through circulation to other tissues, thus becoming important biomarkers of disease [10].

Regarding the important role of visceral adipose tissue and skeletal muscle in the development and progression of various metabolic diseases, this study aims to investigate the differential miRNA expression profile between the visceral white adipose tissue and the skeletal muscle of people with severe obesity undergoing bariatric surgery. We believe that studying the miRNA profile of these two tissues opens a door to explain the metabolic duality of obesity, which in some cases favors the development of other pathologies such as diabetes, while in others acts as a protective factor.

Methods

Study population and sample acquisition

For this study, the participation of all patients undergoing bariatric surgery at the Central University Hospital of Asturias (HUCA) was requested. A total of 38 volunteers with obesity aged between 29 and 62 years were included in the study. In addition, 10 volunteers without obesity or type 2 diabetes who underwent abdominal surgery for other reasons (mainly eventrations), and excluding any tumor pathology, were included in the study, and classified as controls. Volunteers were then subdivided into a discovery cohort (N = 6) and a validation cohort (N = 48), which also include patients from the discovery cohort. All subjects were recruited at the General Surgery Service and the Endocrinology and Nutrition Service of the HUCA. Informed consent was obtained from all volunteers and the study protocol was approved by the HUCA ethical committee (Project identification code: CEImPA:2020.419; acceptation date: 21st October 2020) which is consistent with the principles of the Declaration of Helsinki.

Fresh visceral adipose tissue and skeletal muscle were obtained simultaneously during the surgery procedure under sterile conditions and immediately transported to the laboratory. Specifically, visceral adipose tissue is obtained from the greater omentum, while muscle biopsies are collected from the transversus abdominis or external oblique, in the left subcostal region in the midclavicular line. A laminar flux cabin was used for tissue manipulation to avoid contamination, and blood vessels were cleaned out by using tweezers and scissors. Subsequently, samples were frozen at −80 °C immediately. Additionally, a routine blood biochemical analysis was performed a few days before surgery.

Library preparation and small-RNA next-generation sequencing (NGS) in the discovery cohort

Muscular and visceral adipose tissue samples were analyzed by NGS for small-RNA total expression. Small-RNA isolation, library preparation, quality control and next-generation sequencing procedures were accomplished by Arraystar INC (Rockville, Maryland, USA). Briefly, total RNA of each sample was extracted from 50–100 mg of tissue by using TRIzol (Invitrogen) reagent. Then, total RNA was used to prepare the miRNA sequencing library. RNA extraction was performed with NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, Ipswich, Massachusetts, USA), Ribo-Zero Magnetic Gold Kit (Human/Mouse/Rat) (Epicentre, an Illumina Company, Madison, Wisconsin, USA) and NEB Multiplex Small RNA Library Prep Set for Illumina according to the manufacturer’s instructions, which include the following steps: (1) 3’-adapter ligation; (2) 5’-adapter ligation; (3) cDNA synthesis; (4) PCR amplification; (5) size selection of 135–155 base pairs (bp) PCR amplified fragments (corresponding to 15–35 nucleotides (nt) small RNAs). Libraries were quantified with Agilent 2100 Bioanalyzer, the DNA fragments in well mixed libraries were denatured with 0.1 M NaOH to generate single-stranded DNA molecules, captured on Illumina flow cells, amplified in situ and finally sequenced for 51 cycles on Illumina NextSeq 500 (Illumina, CA, USA) according to the manufacturer’s instructions.

Bioinformatic analysis of miRNA data

Raw sequencing data quality was evaluated using FastQC software (0.11.9) [11]. Cutadapt (3.4) was used to trim full and truncated adapter sequences and set a minimal sequence length of 17 nt [12]. Alignment on hg38 human reference genome was performed using Bowtie1 (1.0.0) and annotation and read counting of predicted miRNAs from miRbase v22.1 was done applying mirDeep2 (2.0.1.2) package [13, 14]. The small RNA that are obtained through miRDeep2 are mainly the mature sequences, which are eventually produced by more than one precursor throughout the genome. Therefore, all mature miRNAs with the same names were taken from different precursors and averaged for each (rounded up).

To identify differentially expressed miRNAs, read counts were analyzed using the edgeR (v.3.36.0) package from Bioconductor in R environment (v.4.1.3) [15]. miRNA data normalization was performed using the trimmed mean of M values (TMM) method and calculated the effective library size. To minimize the error due to sample size, the p value was adjusted using Benjamini–Hochberg tests [16]. miRNAs were considered detectable if they had expression levels higher than 100 counts per million (CPM) in more than half of the samples in the discovery cohort (E-MTAB-13008). In addition, possible batch, or confounder effects, including potential population bias, were handled by surrogate variable analysis (SVA) using the package sva (v3.32.1) and adding all surrogate variables found as covariates to the model [17]. Subsequently, differential expression analysis used the quasi-likelihood negative binomial generalized log-linear model (GLM) functions, provided by the edgeR package. Statistical significance for differentially expressed miRNA was defined as p values < 0.05, and absolute log₂ fold changes ≥ ±1.5. In addition, only differentially expressed miRNAs with logCPM greater than 5 were considered for subsequent validation.

GO and KEGG pathway analysis of the differentially expressed miRNAs

The target gene network of the differentially expressed miRNAs was analyzed using miRnet website (https://www.mirnet.ca) through the mitTarBase v8.0 database [18]. The GO analysis of target genes was then carried out based on three terms: biological processes (BP), molecular functions (MF) and cellular components (CC). Additionally, the related biological pathways were analyzed by Kyoto Encyclopedia of Genes and Genomes (KEGG) [19].

RNA isolation and quantification in the validation cohort

For RNA extraction, tissues were firstly crushed using a mortar and pestle in liquid nitrogen to prevent sample heating and then, RNA was obtained with the SPLIT RNA Extraction kit (Lexogen) according to manufacturer’s instructions. Isolated total RNA was reverse transcribed into cDNA using the TaqMan advanced miRNA cDNA synthesis kit (Life Technologies, California, USA). miRNA expression analysis was carried out by quantitative PCR using TaqMan^® Gene Expression assays (Applied Biosystems; Table S1) and the Applied Biosystems Prism 7900HT Sequence Detection System (Applied Biosystems) as previously reported [20]. U6 was used as an endogenous control as previously reported [21]. Gene expression data are expressed as target miRNA expression relative to the corresponding housekeeping mean gene expression (ΔCT = CT miRNA – CT value of the housekeeping gene). The relative expression of each miRNA was reported as 2^−ΔCT.

Statistical analysis

For RT-PCR validation, statistical analysis was performed with JASP software (0.14.1). The Shapiro–Wilk test was conducted to assess sample normality and then, non-parametric Paired Willcoxon test or Kruskal–Wallis test followed by Dunn Post Hoc test were performed and Bonferroni correction stated. Spearman test was used to evaluate correlations between variables. A p value ≤ 0.05, corrected by FDR was considered significant.

Results

Demographic and biochemical profile of participants

Anthropometric and clinical profile of patients included in both the discovery and the validation cohort are present in Table 1. Regarding the validation cohort, significant differences were observed in age, BMI, Glucose, HbA1c, LDL cholesterol and triglycerides among the three groups. Additionally, BMI was significantly increased in both groups of people with obesity versus the control group and both glucose and HbA1c between the group of people with diabetes and the two non-diabetic groups. Finally, LDL cholesterol levels were found significantly reduced in the group of people with obesity and diabetes compared to controls, while triglycerides are significantly higher in the group of patients with both obesity and diabetes compared with the group of people with obesity but not diabetes.

Table 1.

Demographic and biochemical profile.

	Discovery	Validation				p value (Bonferroni correction)
		All	noOB_noT2D (I)	OB_noT2D (II)	OB_noT2D (III)	I vs. II	I vs. III	II vs. III
N (Male/Female)	6 (0/6)	48 (22/26)	10 (7/3)	19 (6/13)	19 (9/10)	–	–	–
Age (years)	54.5 [36.0–61.0]	49.5 [29.0–62.0]*	50.5 [39.0–61.0]	48.0 [29.0–59.0]	53.0 [33.0–62.0]	0.229	1.000	0.061
BMI (kg/m2)	42.4 [34.6–49.6]	42.7 [21.5–55.8]***	25.6 [21.5–27.8]	45.8 [39.4–55.8]	43.1 [33.5–55.8]	<0.001	<0.001	1.000
Glucose (mmol/L)	8.55 [4.61–10.55]	5.99 [4.4–16.0]***	5.16 [4.7–6.7]	5.77 [4.4–10.1]	8.6 [5.7–16.0]	0.445	<0.001	0.013
Total Ch (mmol/L)	3.9 [1.6–5.2]	4.3 [1.3–6.8]	5.0 [4.1–6.6]	4.1 [2.2–6.8]	4.2 [1.6–6.6]	0.295	0.141	1.000
HDL Ch (mmol/L)	0.8 [0.39–1.16]	0.97 [0.39–1.73]	1.14 [0.88–1.45]	1.01 [0.70–1.58]	0.91 [0.39–1.73]	1.000	0.531	1.000
LDL Ch (mmol/L)	2.20 [0.75–3.16]	2.41 [0.62–4.45]*	3.36 [3.05–4.09]	2.44 [0.62–3.85]	2.25 [0.75–4.45]	0.096	0.018	1.000
TG (mmol/L)	1.33 [0.99–1.72]	1.22 [0.91–8.64]*	1.11 [0.78–2.78]	1.05 [0.91–1.70]	1.54 [0.99–8.64]	0.780	1.000	0.043
HbAc1 (%)	6.8 [5.0–8.9]	5.60 [4.50–9.40]***	5.10 [4.90–5.10]	5.35 [4.50–5.90]	7.00 [5.40–9.40]	0.960	0.003	<0.001

Data are given as median [min–max] for continuous variables and number for categorical variables.

OB obesity, T2D type 2 diabetes, BMI body mass index, Ch cholesterol, HDL high density lipoproteins, LDL low density lipoproteins, TG triglycerides, HbA1c glycated hemoglobin A1c.

Kruskal–Wallis and post hoc test differences among the three groups; ***p < 0.001; *p < 0.05.

Visceral adipose tissue and skeletal muscle differential microRNA profile in the discovery cohort

A total of 345 miRNAs were identified by NGS technique in both the visceral adipose tissue and the skeletal muscle of women with obesity undergoing bariatric surgery. The criteria for differential expression between tissues was established as log₂ fold change ≥1.5 or ≤−1.5 and p value < 0.05. With these criteria, a total of 108 miRNAs (31% of the total identified miRNAs) showed a significant differential expression profile between tissues; 55 miRNAs were upregulated in the visceral adipose tissue (Table S2), and 53 miRNAs were upregulated in the skeletal muscle (Table S3 and Fig. S1A, B).

Spearman correlation was performed to explore any potential relationships between miRNA expression in both tissues, and different biochemical variables including the lipidic profile (HDL Cholesterol, LDL Cholesterol, Total Cholesterol and Triglycerides) and the glucose profile (HbA1c and blood glucose) (Table S4). We could observe greater number of miRNAs related with the lipidic profile in the adipose tissue compared with the skeletal muscle. On the other hand, although more miRNAs from the skeletal muscle were influenced by blood glucose, the same number of miRNAs in both tissues were affected by the HbA1c percentage. Interestingly, among all the significant correlations, only the relationship between hsa-miR-374a-3p with HDL and hsa-miR-331-5p with total cholesterol were found to be common in both tissues, with the latter correlation exhibiting opposite directions. At this point, and due to the small sample size, consisting of only six patients, we decided to correct our significance cut point to 0.0001 based on the number of miRNAs included in the analysis (0.05/345 miRNAs). Then, significancy was only conserved between visceral adipose tissue hsa-miR-941 expression and LDL cholesterol.

Bioinformatic tools were then used to find the predicted target genes for the differentially expressed miRNA. Among the 108 differentially expressed miRNAs, only 103 were successfully mapped, leading to the identification of a total of 10,144 target genes with which the GO and KEGG pathway analysis were investigated and classified by the number of target genes, and the enrichment score calculated as −log(p value). For GO:BP, the term with the highest enrichment score was “negative regulation of transcription from RNA polymerase II promoter” (GO:0000122; p = 3.96 × 10⁻⁸; 378 target genes) (Fig. S2A); For GO:MF, the term with the highest enrichment score was “adenyl ribonucleotide binding” (GO:0032559; p = 1.07 × 10⁻¹¹; 242 target genes) (Fig. S2B); For GO:CC, the term with the highest enrichment score was “Nucleoplasm part” (GO:0044451; p = 5.28 × 10⁻¹¹; 574 target genes) (Fig. S2C). Subsequently, based on the KEGG pathway analysis for all predicted target genes was finally performed. Excluding cancer-related terms, 66 pathways were found to be affected by these miRNAs (data not shown), being the top ten affected pathways represented in Fig. S2D, and highlighting the “Wnt signaling pathway” (hsa04910; p = 6.10 × 10⁻⁹; 113 target genes), the “insulin signaling pathway” (hsa04310; p = 3.33 × 10⁻⁶; 102 target genes) and “TGF-beta signaling pathway” (hsa04350; p = 8.92 × 10⁻⁶; 66 target genes).

miRNA depot dependent expression in the validation cohort

Among all the differentially expressed miRNAs in the discovery cohort, five miRNAs were selected for their validation in an extensive cohort: hsa-miR-206, hsa-miR-122b-5p and hsa-miR-1-3p which were found upregulated in the skeletal muscle, and hsa-miR-224-5p and hsa-miR-335-3p which were found upregulated in the visceral adipose tissue. For this validation, miRNA expression was quantified by real time PCR with Taqman probes.

Regarding the cohort of people with obesity, among all the selected miRNAs, hsa-miR-122b-5p could not be detected in the visceral adipose tissue, and hsa-miR-206 and hsa-miR-1-3p were found to be significantly upregulated in the skeletal muscle (p < 0.001, Fig. 1A–C). By contrast, hsa-miR-224-5p and hsa-miR-335-3p were upregulated in the visceral adipose tissue (p < 0.001, Fig. 1D, E). These changes were also observed in the control group (Fig. S3).

Fig. 1. Box plot images of the six selected miRNAs from the discovery cohort in the validation cohort.

A hsa-miR-1-3p; B hsa-miR-206; C hsa-miR-122b-5p; D hsa-miR-335-3p; E hsa-miR-224-5p. Mann–Whitney test was applied for tissue comparison. ***p < 0.001.

Spearman correlation was performed to observe if there was any relation between the expression of the six selected miRNAs in both tissues and the different biochemical variables analyzed (Fig. 2 and Table S5). Regarding skeletal muscle, hsa-miR-1-3p was positively correlated with hsa-miR-122b-5p and hsa-miR-206, hsa-miR-206 was positively correlated with miR-122b-5p and hsa-miR-224-4p with hsa-miR-335-3p. In the same way, in the visceral adipose tissue, we observed a positive correlation between hsa-miR-1-3p and hsa-miR-206, hsa-miR-224-5p and hsa-miR-335-3p, hsa-miR-206 with both hsa-miR-224-5p and hsa-miR-335-3p and finally hsa-miR-224-5p and hsa-miR-335-3p were also positively correlated.

Fig. 2. Spearman matrix correlation plot between the selected miRNAs in both the skeletal muscle and the visceral adipose tissue and biochemical variables.

BMI body mass index, HbAc1 glycated hemoglobin A1c, Ch cholesterol, HDL high density lipoproteins, LDL low density lipoproteins, TG triglycerides.

We also compared miRNA expression between tissues, observing a positive correlation of visceral adipose tissue hsa-miR-224-5p expression with skeletal muscle hsa-miR-122b-5p. However, significance disappears when we correct the p value by the number of miRNAs analyzed, being reduced to 0.01 (0.05/5 miRNAs).

In addition to the aforementioned correlations, we have also observed a negative correlation between visceral adipose tissue hsa-miR-335-3p expression with HDL cholesterol levels.

Effect of type 2 diabetes in the skeletal muscle and the visceral adipose tissue miRNA profile

To analyze the influence of type 2 diabetes in the expression of those selected miRNA, people with obesity from the validation cohort were subdivided into two different groups: people with obesity and type 2 diabetes (OB_T2D), and people with obesity without type 2 diabetes (OB_noT2D). Also, a group of people without obesity and diabetes (noOB_noT2D) was included in this analysis (Fig. 3A–I). Regarding skeletal muscle, no significant changes were observed. By contrast, when analyzing the visceral adipose tissue, hsa-miR-1-3p, hsa-miR-224-5p and hsa-miR-206 were modified (p = 0.027, p = 0.032 and p = 0.062 respectively). Additionally, both hsa-miR-1-3p (Fig. 3F) and hsa-miR-206 (Fig. 3G) were found to be significantly upregulated in the OB_noT2D group compared with the OB_T2D group (p = 0.023 and p = 0.056 respectively).

Fig. 3. Box plots of differentially expressed miRNAs in the validation cohort, regarding on the presence of type 2 diabetes.

A–E Skeletal muscle F–I visceral adipose tissue [red: OB_T2D—people with obesity and type 2 diabetes; green: OB_noT2D—people with obesity and without type 2 diabetes; blue: noOB_noT2D—people without obesity and without type 2 diabetes]. Wilcoxon test with Bonferroni correction was applied for tissue comparison ***p < 0.001; **p < 0.01; *p < 0.05; ^{^}p < 0.06.

Discussion

Obesity was considered as a risk factor for other pathologies for decades, rather than being recognized as a pathology itself. However, since the last years several medical societies, while controversial, have stated the need to consider obesity as a pathology [22]. Obesity is defined as an excessive fat accumulation around the body and measured by the body mass index. However, it is increasingly acknowledged that BMI is not enough for obesity classification, and other anthropometric and molecular factors must be studied [23, 24]. In fact, obesity is a multifactorial disease that may arise due to an imbalance of multiple molecular mechanisms, and it does not always have the same detrimental effect on health [25, 26]. In the same way, although obesity was always associated with the visceral adipose tissue, other tissues and organs can also be affected, turning obesity in an important risk factor for multiple pathologies and becoming one of the main health problems worldwide [27].

Given the important role that both visceral adipose tissue and skeletal muscle have in metabolic dysfunction [28], and given the ability of miRNAs to be secreted into the bloodstream by different tissues and transported throughout the body by microvesicles and exosomes, our goal was to investigate the expression profile of miRNAs in these two tissues, not only to determine the differences between the two, but also to be able to analyze the different expression profile of miRNAs in those people with obesity who develop type 2 diabetes and those who maintain normal blood glucose levels despite obesity. For this, we simultaneously examined paired visceral adipose tissue and skeletal muscle miRNA profile from six people with obesity. As expected, both tissues showed a specific miRNA expression profile, however, among the 345 miRNAs identified, only 108 miRNAs (31% of miRNAs) showed significant differences among groups, which means that 69% of the identified miRNAs are similarly expressed in both tissues.

The differential genomic and miRNA profile of different human white adipose tissue depots has been widely studied [29, 30], however, differences between visceral adipose tissue and skeletal muscle have only been studied in different animal models as pigs [31] and yak [32], and the differential miRNA profile between visceral adipose tissue and skeletal muscle in human obesity is poorly explored.

Among all the differentially expressed miRNAs, five were then selected for their validation in a larger cohort, two of them upregulated in the visceral adipose tissue compared with the skeletal muscle, and three downregulated (thus, overexpressed in the skeletal muscle), being all the observed changes validated in a larger cohort. hsa-miR-1-3p, which was found upregulated in the skeletal muscle compared with the visceral adipose tissue (170 times overexpressed), although no specific of the muscle, is highly expressed on it, mostly in heart muscle, and therefore, it is associated with the development of cardiovascular pathologies [33]. In fact, in a previous study we have associated the circulating level of hsa-miR-1-3p with the development of vascular complications in type 1 diabetes [34]. Also, hsa-miR-206, previously described as key modulator of skeletal muscle development [35], was overexpressed (more than 2000 times) in the skeletal muscle of patients with obesity. This significantly increased expression of hsa-miR-1-3p and hsa-miR-206, allows us to corroborate the different origin of both tissues. Additionally, hsa-miR-122b-5p which was defined as liver specific miRNA [36], was found in skeletal muscle but not in visceral adipose tissue. While not expressed in the adipose tissue, changes in circulating hsa-miR-122b-5p have previously been related with obesity, weight loss and metabolic syndrome [37, 38].

Regarding adipose tissue, no miRNA has been reported in the literature to be specific of this tissue. Nonetheless, its role in adipogenesis, lipolysis, and even insulin resistance has been described [7, 39]. hsa-miR-224-5p and hsa-miR-335-3p were previously related with adipocyte differentiation [40] and adipogenesis [41] respectively. Also, hsa-miR-335-3p was associated with obesity-related β-cell dysfunction [42–44].

It is important to recognize that different miRNAs could potentially target numerous mRNAs, and that at the same time, a single mRNA could contain multiple binding sites for miRNAs, therefore, not only the study of each single miRNA expression, but also the study of the interaction and correlation between different miRNAs is essential for the correct understanding of different metabolic processes. Among the miRNAs selected for this study, those most strongly correlated with each other were similarly correlated in both tissues, underscoring miRNAs’ importance as biomarkers, as they travel through the bloodstream to the entire organism. Additionally, by using a bioinformatic approach, we could observe a direct implication of the differentially expressed miRNAs with different metabolic related pathways, highlighting the insulin signaling pathway. In fact, the interplay of both obesity and type 2 diabetes is widely stablished. However, although closely related diseases, do not always go hand in hand, being not only β-cells, but also skeletal muscle and adipose tissue are pivotal players in the development of glucose intolerance in type 2 diabetes mellitus [45]. In fact, it has been previously stated the direct relation between visceral adipose tissue and skeletal muscle and their potential role in metabolic dysfunction in obesity [46, 47]. This is where miRNAs appear to be useful for preclinical diagnosis and may help us identify those people with obesity, who are at greater risk of developing diabetes in the future. Different studies have stated the importance of miRNAs in the prognosis of obesity after bariatric surgery [48–50], as well as in the development of type 2 diabetes [51], however, not many are also investigating the joint role of both diseases in miRNA expression profile. In addition, most of these works are carried out at the level of circulating expression of miRNAs, and not directly in the target tissues [52–54]. Accordingly, we decided to analyze the miRNA expression regarding on the presence of obesity and type 2 diabetes, for which only the validation cohort could be analyzed, due to the small number of individuals in the discovery cohort. Thus, the five selected miRNAs, were re-analyzed but grouping patients not only by obesity, but also by the presence of type 2 diabetes mellitus. Only differences in the expression of both hsa-miR-1-3p and hsa-miR-206 were observed, being both upregulated in the group of people with obesity but without diabetes compared with the group of people with both conditions. Interestingly, hsa-miR-1-3p was previously described to be involved on insulin resistance and prediabetes both in mice and humans [55, 56]. Additionally, in pilot study by Wojciechowska et al., serum levels of hsa-miR-1-3p, along with 3 other miRNAs, might be predictive biomarkers of type 2 diabetes remission after bariatric surgery [54], highlighting the role of miRNAs in the prognosis of people with obesity undergoing bariatric surgery, not only achieving the desired weight loss, but also resolving associated comorbidities.

In this study, we describe the miRNA expression profile in both visceral adipose tissue and skeletal muscle of the same patients with severe obesity, with the aim of analyzing the differences and similarities between these two tissues.

On the other hand, different obesity phenotypes have been defined in recent years, leading to the term “metabolically healthy obesity” to define those people with obesity with non-pathological metabolic, insulin, lipid and inflammatory profiles. Thus, we believe that miRNAs may play a fundamental role in the identification and classification of patients with different metabolic profiles, although more studies are needed in this area.

Limitations of the study

One of the main limitations of this study is the small sample size of the discovery cohort, therefore, although this serves as a pilot study and more extensive research should be conducted, we suggest that further study of these miRNAs and their corresponding target genes could improve our understanding of the metabolic complexities of obesity.

In addition, further exploration could include the establishment of a new discovery cohort explicitly designed to study miRNA profiles in people with obesity and with type 2 diabetes. This initiative would provide a more complete picture of the factors that influence the onset of type 2 diabetes in people with severe obesity.

Another important limitation we found is the lack of a real control group, because although patients who underwent minor surgeries (mainly eventrations) were selected, any surgery can cause inflammation and affect the final result of the study. In addition, the strict criteria for this control group results in a smaller cohort size, but the large metabolic differences in this group make the sample size valid.

Author contributions

Design: Carmen Lambert, Paula Morales-Sánchez, María Moreno Gijón, José Manuel Fernandez-Real and Elías Delgado. Conduction/Data collection: Carmen Lambert, Paula Morales-Sánchez, Ana Victoria García, Elsa Villa-Fernández, Jèssica Latorre, Estrella Olga Turienzo Santos, Lorena Suárez-Gutierrez, Raquel Rodríguez Uría and Sandra Sanz Navarro. Analysis: Carmen Lambert, Paula Morales-Sánchez, Miguel García-Villarino, Jessica Ares-Blanco, Pedro Pujante, Lourdes María Sanz Álvarez and Edelmiro Menéndez-Torre. Writing manuscript: Carmen Lambert, Paula Morales-Sánchez, María Moreno Gijón, José Manuel Fernandez-Real and Elías Delgado. All authors have read and agreed to the published version of the manuscript.

Funding

This study has been funded by Instituto de Salud Carlos III (ISCIII) through the project PI19/01162 to ED and co-funded by the European Union. A non-conditional grant from Menarini laboratory group was also received. CL is recipient from a Sara Borrell grant from ISCIII (CD23/00037). PMS is recipient from a pre-doctoral grant from the Spanish Association Against Cancer (AECC) (PRDAS18003FERN). We thank Fundación Caja Rural and Sociedad Asturiana de Diabetes, Endocrinología, Nutrición y Obesidad for their continuous support.

Data availability

The NGS data generated in this study have been deposited in the ArrayExpress database under accession code E-MTAB-13008.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41366-024-01683-4.