The metabolic footprint of compromised insulin sensitivity under fasting and hyperinsulinemic-euglycemic clamp conditions in an Arab population

The metabolic footprint of compromised insulin sensitivity under fasting and hyperinsulinemic-euglycemic clamp conditions in an Arab population

Characteristics of study subjectsThis study included 47 [age 30.4 ± 5.1 (21–43)] healthy non-obese males of Arab descent (BMI 24.8 ± 2.46, range 16.9–28), normal HbA1C levels (HbA1C [%]: 5.1 ± 0.25, range 4.4–5.6), fasting glucose < 5.6 mmol/l, and blood glucose response to an oral glucose tolerance test (OGTT, 75 g) at 2 h below 7.8 mmol/l. We assessed the subjects’ IS using HIEC5. The experimental design is presented in Fig. 1A.Figure 1Subject characteristic. (A) Experimental design. 47 clinically healthy lean (body mass index (BMI) < 28) male subjects of Arab descent were enrolled for the study. The subjects underwent hyperinsulinemic-euglycemic clamp to determine their levels of insulin sensitivity (IS). The samples were collected before the clamp (T1) and at 120 min of the clamp (T2). Subjects were divided into the low (L), moderate (M) and high (H) responders to insulin based on their levels of IS depicted by the M-value. The samples collected at T1 and T2 were probed for the metabolic composition. Association analysis between level of metabolites and three levels of IS (L, M, H) was conducted at T1 and T2. (B) Levels of IS across the subject were determined at hyperinsulinemic-euglycemic clam and were depicted by M-values; the subjects were grouped into the low (M-value ≤ 5.6), moderate (M-value > 6) and high (M-value ≥ 12) responders to inulin, (C) association between BMI and IS. (D) Association between glycated hemoglobin (HbA1C) and IS levels. At T1 subjects levels of IS are depicted in gray scale from light grey indicating low IS into dark grey indicating high IS. At T2 subjects levels of IS are depicted in green scale from light green indicating low IS into dark grey indicating high IS.The HIEC revealed a wide range of the levels of IS as defined by the M-value [mg/kg/min] (M-value 9.69 ± 4.18, range 2–19.5) across the tested individuals (Fig. 1B). The distribution of the subjects’ M-values is presented in Supplementary Fig. 1. Following previous studies by Tam et al.4 individuals were grouped based on their level of IS into low (M-value ≤ 5.6), moderate (5.6 < M-value ≤ 12) and high (M-value > 12) responders to insulin. The study subjects were selected to have a BMI (below 28) and normal HbA1C (HbA1C % < 6). No significant association was detected between IS and these variables (Fig. 1C,D).Metabolic differences between the fasting state and hyperinsulinemic-euglycemic clamp reflect a metabolic switch in energy utilizationWe conducted metabolome-wide profiling of serum samples collected at the fasting state (T1) and under euglycemic clamp conditions (T2) using two mass-spectrometry-based platforms, including the broad non-targeted metabolic profiling HD4 platform19 and the Lipdyzer complex lipid platform (CLP)20. Together, we identified 1896 molecules, 905 on the HD4 platform and 990 on the CLP platform, including 1697 named metabolites and 199 unnamed molecules (Supplementary Table 1).We conducted a principle component analysis (PCA) including all detected metabolites to assess the impact of HIEC on overall body metabolism. We observed a clear separation between metabolic profiles between the fasting state (T1) and HIEC conditions (T2) (Fig. 2A).Figure 2Impact of HIEC on body metabolism. (A) Principal component analysis (PCA) analysis reveals metabolic differences at the fasting state (T1) and under HIEC (T2). (B–I) Box plots present alternation patterns between fasting and HIEC conditions, and reflect body metabolism at fasting and fed state, respectively. Grey indicate the fasting state and green indicate HIEC. The p-value was defined by the statistical contrast between T1 and T2 and is depicted by “*”. TAG56:1(FA16:0): TAG—Triacylglycerol; 56—total number of carbons on all three fatty acid (FA) chains; (:1) one of the FA chains is unsaturated; and (FA16:0)—one FA chain consist of 16 carbons (16) and (:0)—is fully saturated.The biochemical differences between postprandial and fasting state are reflecting switch in the energy utilization from glycolytic pathway to fatty acid oxidation21. Under fasting, the glucose level is maintained by hepatic gluconeogenesis, and the organism energetic needs are sustain by fatty acids (FA), released from triacylglycerols, catabolized in the process of beta-oxidation, ketone bodies and amino acids released from protein degradation21. We asked whether metabolic differences between fasting and HIEC would reassemble the metabolic switch, observed between fasting and postprandial state. Using a mixed linear model (see “Material and methods”), we identified 551 metabolites that showed significant alterations between fasting and HIEC at a false discovery rate (FDR)22 of 5% (Supplementary Table 2). The vast majority of the associated metabolites (514 out of 551) were elevated under fasting conditions in comparison to the HIEC state, including predominantly fatty acids, acylcarnitines, bile acids as well as molecules involved in branched chain amino acids (BCAA), aromatic amino acids (AAA), methionine and urea cycle metabolism. In contrast, molecules involved in carbohydrate metabolism and acyl cholines were elevated under HIEC conditions. Glucose was maintained at the same level at the fasting state and HIEC (Fig. 2B). Increase in the pyruvate level was observed under HIEC (Fig. 2C). Levels of TAG’s (TAG56:1 Fig. 2D, full list in Supplementary Table 2), fatty acids [Palmitate (16:0) Fig. 2E, full list in Supplementary Table 2], glycerol (Fig. 2F) and 3-hydroxybutyrate (Fig. 2G) decreased under HIEC conditions. The HIEC triggered a decrease in the level of amino acids including BCAA (Leucine Fig. 2H) and AAA (tryptophan Fig. 2I) (full list with altered amino acids in Supplementary Table 2). Interestingly, we even identified molecules from the clinical intervention: Subjects were given lidocaine as a local anesthetic prior to the clamp, and we found lidocaine and N-ethylglycinexylidide, which is a breakdown product of lidocaine, in most blood samples at T2 and very low levels at T1 (Supplementary Fig. 2).These results indicate that metabolic differences between the fasting state and HIEC reflect the metabolic switch in the pathways of energy utilization from beta-oxidation of fatty acids under fasting to glucose utilization under glucose and insulin presence. This experimental setting provides unique and well defined conditions for further investigation of the impact of IS on body metabolism under conditions where fatty acids or glucose are utilized as primary energy source, as we show next.Elevated levels of triacylglycerols and diacylglycerols at baseline are the hallmark of compromised insulin sensitivityWe asked whether different levels of IS are impacting body metabolism in a fasting state, under conditions where fatty acids are utilized to fulfill the body’s energetic demand. The differences between the estimated metabolite levels in the three IS groups (low, moderate, and high) at the fasting-state (T1) were assessed (see “Material and methods”). We identified 336 metabolites showing FDR significant associations with the levels of IS (low, moderate, and high) at the fasting-state (T1) (Supplementary Table 3). The identified metabolites were predominantly lipids, such as TAGs (268 molecules with different lipid side chain compositions), diacylglycerols (29 molecules), and phosphatidylcholines (eight molecules), which were all elevated in subjects with low IS. The top ten molecules associated with IS are presented in Fig. 3; six of those lipid molecules contain an eicosapentaenoate (20:5n3) fatty acid chain. Only six out of 336 identified metabolites were not directly involved in lipid metabolism. Those metabolites were two nucleotides (uridine and 3-ureidopropionate), two amino acids involved in glutamate metabolism (alpha-ketoglutaramate and pyroglutamine) as well as two xenobiotics (betonicine and 2-hydroxyacetaminophen sulfate). The levels of urine and 3-ureidopropionate were elevated in subjects with low IS (Fig. 3).Figure 3Triacylglycerols and diacylglycerols differentiate subjects with low IS level in the fasting state. Box blots presents top ten metabolites showing significant association with the levels of IS. The p-value was defined by the statistical contrast between estimated metabolite levels and IS [low (L) and high (H)] at subjects’ baseline (T1), depicted by “*”. Gray and green color gradient indicate metabolite levels after overnight fasting and hyperinsulinemic-euglycemic clamp, respectively. Light color tone indicate subjects with low IS and dark color tone indicate subjects with high IS.Given that majority of metabolites associated with the levels of IS were triacylglycerols and diacylglycerols, we further tested, whether lipids monitored using clinical chemistry assays, such as total cholesterol (TC), total triacylglycerols (TG), high-density lipoproteins (HDL), and (LDL) low-density lipoproteins, differ across the subjects with low, moderate and high levels of IS. We observed significant differences in TC (p-value = 2.67 × 10–3), TG (p-value = 1. 78 × 10–2), and LDL (p-value = 2.06 × 10–2) but not HDL across the subjects with various IS levels (Supplementary Table 4). Next, we investigated the correlations between all metabolites and clinical chemistry readouts (TC, TG, HDL and LDL). The correlation analysis between metabolites and the levels of TC, TG, HDL, and LDL is presented in Supplementary Table 5. We found strong correlations (r > 0.7) between 24 metabolites and TC, between 66 metabolites and TG, as well as between 13 metabolites and LDL. The levels of TC correlated predominantly with cholesterol esters (CE), phosphatidylcholines (PC) and sphingomyelins (SM) consisting of long and very long chain fatty acids. The levels of LDL correlated with CE and SM. The levels of TG correlated with various TAGs and DAGs. We found that 71% of metabolites that correlated (r > 0.7) with the levels of TG overlapped with the metabolites significantly associated with the levels of IS. In contrast, none of the metabolites correlating with either TC or LDL levels overlapped with metabolites that associated with the levels of IS (Supplementary Fig. 3). Only a weak correlations (r < 0.5) were observed with HDL. These results indicate metabolic differences between subjects with low and high IS at the fasting state, which manifested in accumulation of triacylglycerols and diacylglycerols, and suggest a metabotype, which is specific to compromised IS, and is predominantly driven by dysregulated TG homeostasis.Metabolic dysregulation associated with early onset of compromised insulin sensitivity in response to insulinWe further tested whether a low IS would affect the metabolic responses to HIEC, under which glucose is utilized as a primary source of energy. First, we investigated the impact of HIEC on the metabolites already dysregulated at the fasting state. We found that under HIEC conditions, di- and tri-acylglycerols remain elevated in subjects with low IS, which suggest that those alterations are independent of the primary energy source.Next, we identified 39 metabolites showing FDR significant association with the levels of IS (low, moderate, and high), defined by the statistical contrast between estimated metabolite levels in the three groups under HIEC (Table 1). The associated metabolites are from various metabolic pathways, including branched chain and aromatic amino acid metabolism, carbohydrate, and lipid catabolism. All identified amino acids, including the BCAAs isoleucine and leucine, the aromatic amino acids (tryptophan, phenylalanine, and tyrosine), as well as methionine and histidine were elevated in subjects with low IS. The elevated level of fatty acids [e.g. docosapentaenoate (22:5n3), eicosapentaenoate (20:5n3), dihomo-linolenate (20:3n3), myristoleate (14:1n5), and myristate (14:0)] in individuals with a low IS were also observed. Levels of almost all carbohydrates were lower in subjects with a low IS, with the exception of mannose, which is interesting as mannose was also inversely associated with a genetic higher diabetes risk at the GCKR locus23.Table 1 List of metabolites showing FDR significant association with IS level under hyperinsulinemic-euglycemic clamp.These results reveal metabolic dysregulations associated with low IS under HIEC, and indicate metabolic responses, which are impaired under the insulin action and glucose availability, further suggesting its potential involvement in progression toward IR and T2D.Impaired metabolic response to insulin in clinically healthy subjects replicates in T2D patientsLastly we tested whether the metabolic dysregulations observed under HIEC in clinically healthy subjects with low IS might be implicated in the future progression toward IR and T2D. We conducted replication in an independent cohort of 17 subjects, which included healthy controls (seven subjects) and T2D (ten subjects). The HIEC was performed as we previously reported24 and the M-values [mg/kg/min] were (M-value: 4.78 ± 2.81, range 0.75–10.8). The subjects were grouped into insulin sensitive (M-value > 4.7) and insulin resistance (M-value ≤ 4.7) based on the conservative definition of Bergman et al.6 The metabolic profiling was conducted using a non-targeted metabolomics platform after overnight fasting (T1) and 120 min. at HIEC (T2). We identified 22 metabolites showing FDR significant association with IS in the insulin-sensitive and insulin-resistant groups under HIEC. Among the metabolites showing FDR significant association with IS in clinically healthy subjects of Arab descent, we replicated the same trend at the nominally significant p-value for almost all amino acids apart of histidine (Table 1). The association of mannose, isoleucine, leucine, tyrosine, and uridine with IS in clinically healthy subjects was replicated at the FDR significance level in the T2D patients of this independent cohort (Fig. 4). In the replication cohort, we observed an FDR significant association of mannose (p-value = 2.45 × 10–3) and tyrosine (p-value = 3.01 × 10–3) with IS at the baseline (T1). Among the lipids, polyunsaturated fatty acids (PUFA) were not replicated.Figure 4Different metabolic responses to insulin under hyperinsulinemic-euglycemic clamp in clinically healthy subjects replicate in independent cohort consist of T2D (A–E). Box blots presents metabolites showing FDR significantly different responses to HIEC, associated with the levels of IS. The p-value was defined by the statistical contrast under hyperinsulinemic-euglycemic clamp (T2), between estimated metabolite levels and IS [low (L) and high (H)] in clinically healthy or between estimated metabolite levels and insulin-resistance/insulin-sensitive group in replication cohort of T2D. Gray and green color gradient indicate metabolite levels after overnight fasting and hyperinsulinemic-euglycemic clamp, respectively. Light color tone indicate subjects with low IS and dark color tone indicate subjects with high IS. The replication cohort is depicted by blue (fasting-state) and orange (at HIEC) color. Light blue—insulin resistant (R), dark blue—insulin-sensitive (S), light orange—insulin resistant (R), dark orange—insulin sensitive (S).These results further suggest that impaired responses of mannose, branch chain amino acid, tyrosine and uridine to insulin, observed in clinically healthy subjects with low IS, could be considered as key metabolic dysregulations, which together with accumulated lipid triggers progression toward IR and T2D phenotype.

Via Source link