Expression markers of human skeletal muscle associated with disorders of glucose metabolism in the basal and postprandial state

INTRODUCTION

Obesity is one of the main contributors to the development of socially significant chronic metabolic diseases, such as metabolic syndrome and type 2 diabetes mellitus (T2D). In Russia, the number of patients with T2D is growing and today is about 3% of the population [1]. Skeletal muscles play a key role in the carbohydrate metabolism in the body, since even at rest their contribution to insulin-dependent glucose consumption exceeds 80% [2]. This means that dysregulation of insulin-dependent glucose uptake by skeletal muscle impairs carbohydrate metabolism and can lead to obesity and T2D.

The insulin resistance of skeletal muscle is mainly associated with impaired insulin-dependent signaling regulating translocation of glucose transporter GLUT4 and postprandial uptake of glucose from the bloodstream. However, it turned out that in individuals with obesity and T2D, skeletal muscles in a basal state (fasting) have pronounced changes in the phosphorylation of many signaling molecules that are not part of the canonical insulin signaling cascade, as well as a number of transcription factors [3], which is in good agreement with the change in the transcriptional profile of the muscles of patients [4–6]. These changes can be caused by a change in the blood insulin and other hormones and metabolites in the basal state, as well as the sensitivity of skeletal muscles to them. Indeed, a number of studies have found significant correlations of gene expression in fasting skeletal muscle with blood insulin levels and some markers of metabolic disorders. For example, a study of the transcriptome of skeletal muscles in individuals with different body mass index revealed 180 significant correlations of gene expressions with the insulin resistance index (Homeostatic Model Assessment for Insulin Resistance, HOMA IR), including, among other things, some regulators of lipid metabolism and the AKT-mTOR signaling pathway that regulates protein synthesis and proteolysis, gene expression, etc. [7]. In addition, a meta-analysis of transcriptomic data showed that the expression of the insulin receptor gene (INSR) and its substrate (IRS2) negatively correlate with the level of fasting insulin in skeletal muscle of individuals with different levels of insulin sensitivity [8].

On the other hand, a change in basal expression of skeletal muscle genes can be regulated by a change in the content of insulin and other substances caused by regular meals. Indeed, several studies have shown that an increase in blood insulin during clamp testing (up to 100 mU/L or ~0.6 nM) after 3–4 hours alters the expression of several hundred genes in skeletal muscle of healthy individuals [9,10]. In addition, significant associations of gene expression in muscle tissue and adipose tissue with carbohydrate metabolism during insulin stimulation (clamp test) have already been found [11]. However, it must be noted that in healthy individuals and in obese individuals with or without T2D, there might be a significant difference in the increase and absolute level of insulin during everyday meals from those obtained under the clamp test conditions and between themselves. Therefore, a great interest exists in studies examining changes in glucose metabolism and gene expression in skeletal muscle in reaction to meals, normalized by body weight [12–14]

PURPOSE OF THE STUDY

The purpose of the study is to identify expression markers (genes) of human skeletal muscle associated with impaired glucose metabolism in the basal state and after meals, normalized by body weight.

MATERIALS AND METHODS

Study design

A multicenter interventional prospective comparative study.

Study location and time

The organization of the study was described in our previous publications [15][16]. Studies with patients were carried out at the Endocrinology Research Centre from April 2022 to May 2023. Studies with healthy volunteers were carried out at the Institute of Biomedical Problems, Russian Academy of Sciences, during the period from March 2022 to September 2023.

Populations

The characteristics of the volunteers are shown in Table 1. The study included 8 healthy volunteers («H») (inclusion criteria: BMI<25; absence of diagnosed T2D), 8 obese patients («Ob») (BMI>30; absence of diagnosed T2D) and 8 obese patients with T2D (BMI>30; diagnosed T2D). T2D patients were on glucose-lowering therapy; this therapy was interrupted for 1–2 days (sodium glucose cotransporter-2 inhibitors, dipeptidyl peptidase-4 inhibitors, sulfonylurea derivatives, biguanide [metformin]) and/or for 7 days (glucagon-like peptide-1 receptor agonist [semaglutide]) prior to the mixed meal tolerance test.

Description of intervention

All volunteers completed a short form survey SF-12 [17] to subjectively assess their physical abilities. Venous blood samples from v. cephalica were taken in the morning on an empty stomach (09:00) and 30 min, 60 min, 90 min, 120 min and 180 min after ingestion of mixed food product Resource 2.0 (Nestle Health Science, France, 3 ml or 6 kcal/kg b.w.; protein : fat : carbohydrate ratio 1:2.3:2.7 kcal/kg b.w.) (Figure 1A). Needle biopsy was taken from the middle of m. vastus lateralis under local anesthesia (2 mL of 2% lidocaine), before and one hour after meals (the second biopsy was taken 10 cm proximal to the first) as described elsewhere [15][16].

Laboratory test methods

Glycated hemoglobin (HbA_1c) in whole blood was determined by high performance liquid chromatography using D10 analyzer (BioRad, USA), serum glucose was determined using Architect c8000 automatic analyzer (Abbott Diagnostics, USA); immunoreactive insulin and C-peptide were determined in serum using Cobas 6000 electrochemiluminescence analyzer (Roche, Switzerland) as described elsewhere [13][14].

The whole-genome RNA sequencing and bioinformatic data processing were performed as described elsewhere [16]. In brief, total RNA was isolated using spin columns (RNeasy Mini Kit, Qiagen, Germany) and its concentration was measured (Qubit 4 fluorometer and Qubit RNA HS Assay Kit (Thermo Scientific, USA). After RNA integrity assessment (RIN>7; capillary electrophoresis on TapeStation 4150 and High Sensitivity RNA ScreenTape Kit, Agilent, Germany), chain-specific libraries were prepared with the NEBNext Ultra II Directional RNA Library Preparation Kit (New England BioLabs, USA) and then sequenced (75 nucleotides from one end, covering about 55 million reads per sample) on NextSeq 550 analyzer (Illumina, USA).

Bioinformatics and statistical analysis

RNA reads were mapped to the human genome (primary assembly GRCh38.p13) using the Hisat2 v.2.2.1 software. The number of uniquely mapped reads for the exons of each gene was calculated using the Rsubread package (R software environment). Genes that changed expression were determined using the DESeq2 method with correction for Benjamin-Hochberg multiple comparison (p_adj<0.05; R environment), expression level was determined using TPM value (Kallisto software, v0.46.2).

After removal of low-expression genes (TPM<1), correlations between protein-coding gene expression and biochemical parameters were determined using Spearman rank correlation coefficient; only strong relationships were used for further analysis: |ρ|>0.7 and p<0.001 (Figures 1B, 3A).

For biological interpretation and finding correlations between basal gene expression and biochemical parameters only those genes were taken whose expression differed by more than 1.5-fold between healthy volunteers and patients (DESeq2 test, p_adj<0.05) (Figures 1B, 3B).

For biological interpretation and finding correlations between changes in gene expression after meals and biochemical parameters only those genes were used whose expression changes in response to meals significantly differed by more than 1.5-fold between healthy volunteers and patients (Mann-Whitney test, p<0.05).

Kruskal-Wallis univariate analysis of variance corrected for multiple comparison (Dunn test) at significance level of 0.05 was used to assess differences between samples for biochemical parameters.

Protein classes were determined using the KEGG BRITE database.

Ethical review

The study was carried out in accordance with the Helsinki Declaration and approved by the Biomedical Ethics Committee, Institute of Biomedical Problems, Russian Academy of Sciences (Minutes No. 613 of March 29, 2022) and the Local Ethics Committee of the Endocrinology Research Centre (Minutes No. 4 of February 14, 2022). Written informed consent was obtained from all volunteers.

RESULTS

Characteristics of groups and mixed meal tolerance test

Obese patients (Ob) and obese patients with diabetes mellitus (T2D) differed from healthy patients (H) by elevated basal insulin and C-peptide levels, obese patients with type 2 diabetes mellitus had elevated glucose and glycated hemoglobin levels (Table 1). As we described earlier [15], the dynamics of biochemical parameters after meal differed between groups. In Ob and T2D groups, there was an increased postprandial level and an increase in the area under the curve (iAUC) of C-peptide and insulin (Figures 2A and B). In contrast to the control group, glucose levels in patients were increased up to the second hour after eating (as expected), whereas iAUC was increased only in the T2D group (Figure 2).

Correlations of basal gene expression with parameters of glucose metabolism

In the basal state, 526, 174 and 397 genes were found which changed expression compared to healthy controls in the Ob group, T2D group and pooled group of patients, respectively. However, only 38 of these showed significant and strong correlations (|ρ|> 0.7) with basal insulin, C-peptide, fasting glucose, or glycated hemoglobin. In order to identify potential biologically significant correlations, we selected 11 genes, the expression of which differed between the healthy control and any group of patients more than 1.5-fold (Figure 3A, Table 2, supplementary figure). When analyzing the relationship of basal expression with a change in glucose level in response to meal, significant correlations were found for four genes: positive for NFIL3 and VGLL2, and negative for CADM1 and GPCPD1) (Figure 3A, Table 2, supplementary figure).

Correlations of postprandial gene expression changes with measures of glucose metabolism

Significant correlations between healthy volunteers and patients were found for 37 genes; among them, only two genes showed a significant (by >1.5) change in the amplitude of postprandial expression – MEOX1 (Mesenchyme Homeobox 1) and TXNDC12 (Thioredoxin Domain Containing 12) (Figure 3B, supplementary figure).

DISCUSSION

In our study, we examined associations between biochemical parameters of the blood characterizing glucose metabolism abnormalities and gene expression in skeletal muscle in the basal state and after meals in healthy subjects, obese patients without and with T2D. It is important to note that the change in these indicators was studied in nearly physiological conditions, namely after the intake of mixed food normalized by body weight. The study identified 17 genes, the expression of which was associated with metabolic disorders and the level of insulin, C-peptide or glucose (on an empty stomach or in response to a meal) (supplementary figure).

Fasting insulin and C-peptide expression markers

Insulin is an anabolic hormone that regulates many processes in skeletal muscle. In insulin resistance, chronically elevated insulin levels and/or excessive postprandial increments can lead to alterations in insulin-dependent cell signaling and gene expression in skeletal muscle. We identified several expression markers in skeletal muscle that correlate closely with basal levels or postprandial changes in insulin levels. Among them, genes were found for which a role in the regulation of glucose metabolism and the development of metabolic disorders was already demonstrated.

For example, by analogy with our study, which showed that the basal expression of the adipomyokin gene FSTL1 positively correlates with fasting levels of C-peptide and insulin (Figure 4A), the authors of [18] found that the level of FSTL1 in the blood is positively associated with obesity and insulin resistance. On L6 myoblasts and primary dog myoblasts, it was shown that FSTL1, regulates AMPK-dependent glucose uptake and expression of GLUT4 [19]. Moreover, infusion of FSTL1 to dogs with the induced heart failure positively influenced metabolism of fat acids and glucose oxidation [20]. These data and our results suggest that insulin-induced increases in FSTL1 expression in patients may be one of the compensatory mechanisms aimed at prevention of disorders in glucose uptake in muscles.

The negative correlations of the expression of the SMOC1 gene with the level of insulin and C-peptide found in our study are consistent with the data on the positive correlation of its protein level in plasma with insulin sensitivity in humans [21]. In mice, intraperitoneal administration of SMOC1 or increased expression of its gene in the liver improved glucose metabolism and insulin sensitivity by increasing the expression of gluconeogenesis regulatory genes [21]. In humans, the relationship between blood SMOC1 levels and carbohydrate metabolism disorders is still unclear, however, our study points to the potential role of changes in this gene expression in skeletal muscle in the development of metabolic disorders.

The UBTD1 and MFAP5 genes had a positive correlation of basal expression with fasting C-peptide levels. This is consistent with other studies where they are described as expression markers positively correlating with fasting glucose level, glucose consumption rate (UBTD1), or degree of obesity or fasting insulin level (MFAP5) [11][22].

Expression markers associated with fasting glucose level

Hyperglycemia is a key cause of impaired functioning of various tissues, organs and systems, including skeletal muscles; in particular, its relationship with the development of diabetic myopathy was found [23]. However, the molecular mechanisms of the negative effects of hyperglycemia on skeletal muscle are still poorly understood. We found several expression markers correlated with postprandial glucose levels. For example, basal GPCPD1 expression was negatively correlated with glucose change in response to the mixed meal tolerance test (Figure 4B). The effect of GPCPD1 on the development of insulin resistance in humans has not been studied, but it has been shown in mice that inactivation of this gene in muscles causes glycerophosphocholine accumulation, hyperglycemia and insulin resistance, which is consistent with data on an increase in glycerophosphocholine in skeletal muscle of elderly people and patients with T2D [24].

The role of other genes whose expression was found to be associated with postprandial glucose alteration, has not been studied in relation to the regulation of glucose metabolism in skeletal muscles. However, in primary pancreatic beta cell culture, CADM1 gene expression has been shown to be associated with insulin secretion and glucose level [25], and in primary hepatocyte culture, the NFIL3 gene regulates glucose production and gluconeogenesis [26].

Postprandial gene expression changes associated with insulin and C-peptide levels

Postprandial changes in the expression of only two genes (transcription factor MEOX1 and endoplasmic reticulum protein TXNC12) correlated with different parameters of glucose metabolism; the role of these genes in regulating glucose metabolism is still unknown. It is worth noting that we included in our analysis only genes with marked differences (>1.5-fold) in postprandial response between patients and healthy individuals. Nevertheless, it is interesting to note the negative correlation between the change in the expression of the beta-2 adrenoceptor gene ADRB2 (1.3-fold difference in expression change between healthy and patients) and the changes in insulin and C-peptide (Figure 4). ADRB2 has been repeatedly associated with insulin resistance and has several polymorphisms reliably associated with obesity [27][28], and its expression positively correlates with fat loss [29].

STUDY LIMITATIONS

The results of our study were obtained on a small sample; therefore, subsequent studies are required to verify the markers we found. In addition, the presence of comorbidities, as well as the use of glucose-lowering drugs by patients (despite their withdrawal the day before the study) could affect the variability of the data obtained. In our study, only two patients took exclusively glucose-lowering therapy, the rest additionally took other drugs, mainly for control of arterial hypertension and hypercholesterolemia.

CONCLUSION

In our study, 17 expression markers (genes) associated with indicators characterizing disorders of glucose metabolism in the basal state and after meals were identified. The validity of our prediction is supported by the fact that for 7 of them, a relationship with metabolic disorders was already demonstrated, and FSTL1 and SMOC1 genes were already considered as potential therapeutic targets for blood glucose control. Other expression markers identified in our study are promising candidates for future targeted studies aimed at investigating the mechanisms of insulin resistance development and may be considered as potential therapeutic targets. In addition, it should be noted that we examined early transcriptome response to mixed food intake; investigating this response at later stages (and looking for associations with impaired glucose metabolism) appears to be a promising area for further research.

SUPPLEMENTARY MATERIALS

Supplementary figure is available at https://zenodo.org/records/10998825

FURTHER INFORMATION

Funding. The study was carried out with the financial support of the Russian Science Foundation as part of Project No. 21-75-10146.

Conflict of interest The authors declare no clear and potential conflicts of interest related to the content of this paper.

Contributions of authors. P.A. Makhnovsky – scientific leadership of the study, bioinformatic and statistical analysis of data and their biological interpretation, writing and editing the manuscript; N.S. Kurochkina – preparation of RNA samples, RNA sequencing, making significant edits to the manuscript; T.F. Vepkhvadze – conducting interventions and their organization, preparation of RNA samples, making significant edits to the manuscript; A.O. Tomilova – work with patients, conducting interventions and their organization, making significant edits to the manuscript; E.M. Lednev – conducting interventions and their organization, statistical analysis of data, making significant edits to the manuscript; M.V. Shestakova – scientific leadership of the study, organization and concept of the study, biological interpretation of the results, making significant edits to the manuscript; D.V. Popov – scientific leadership of the study, organization and concept of the study, biological interpretation of the results, writing the manuscript.

All authors approved the final version of the manuscript before publication, agreed to be responsible for all aspects of the manuscript, ensuring proper investigation and resolution of issues related to the accuracy or fidelity of any part of the manuscript.

Acknowledgments. The authors express their gratitude to I.A. Sclyanik, Cand. Med. Sci. (Diabetes Institute of the Endocrinology Research Centre) for help in working with patients.