Circulating adiponectin versus apoM relationship with cardiometabolic status

The clinical and biological characteristics of cohort A are presented in Supplementary Table S1. Cohort A comprised overweight men (BMI, 26.7 kg/m², [SD, 3.5]), of who 39.6% were patients with CAD (cases).

Compared with the controls, the CAD patients had less physical activity and were diagnosed or treated more often for hypertension, dyslipidemia, and a larger proportion was suffering from T2D (25.4% versus 6.9% in the controls). The body fat percentage and waist circumference were also higher in cases than in controls. Among the biological markers, the total cholesterol and LDL-C were lower in cases, probably reflecting the effects of lipid-lowering drugs in patients. However, individuals with CAD displayed higher levels of TG, hs-CRP, and γ-GT. HDL markers (HDL-C and apoA-I) were lower. Finally, the cases exhibited higher HOMA-IR and lower adiponectin levels than the control participants, while the apoM levels were similar between the two groups.

Blood levels of apoM and adiponectin were then considered according to other disease conditions, including T2D, dyslipidemia, and hypertension (Supplementary Table S2). Adiponectin and apoM levels were lower in participants diagnosed or treated for T2D. Unlike adiponectin, apoM levels were lower in participants with hypertension, while they were higher in subjects with dyslipidemia than in the individuals without these conditions. These observations were weakened when only the treatment conditions were considered (Supplementary Table S3), except for adiponectin, for which the levels were lower in participants treated for dyslipidemia and T2D.

Circulating apoM is a negative determinant of insulin resistance

Correlations between circulating apoM and biological and clinical parameters relative to cardiovascular and T2D risk factors were investigated in the entire study population (cohort A). These correlations were compared with equivalent correlation for adiponectin and HOMA-IR (Table 1).

Table 1 Spearman correlation coefficients of adiponectin, apoM and HOMA-IR with anthropometric parameters, biological markers and environmental factors in cohort A.

The blood level of apoM did not show any significant correlation with that of adiponectin. However, similar to adiponectin, apoM exhibited a negative association with hs-CRP (r = −0.28) and HOMA-IR (r = −0.29), while it was positively associated with HDL markers (r = 0.32, for HDL-C and apoA-I). Unlike adiponectin, apoM did not correlate with TG, but it was positively associated with total cholesterol (r = 0.36), LDL markers (LDL-C and apoB100, r = 0.27; and r = 0.20, respectively), and apoC-III (r = 0.19). No association was observed between apoM and cardiac functions (systolic blood pressure, heart rate), or environmental factors (smoking, physical activity), except for a positive correlation between apoM and alcohol consumption (r = 0.19). Of note, apoM was negatively associated with fat mass (r = −0.26) and age (r = −0.25).

In addition to the expected negative correlations with apoM and adiponectin, HOMA-IR displayed negative correlations with HDL markers (apoA-I and HDL-C) and physical activity, and positive correlations with adiposity markers (BMI, fat mass, and waist circumference), TG, apoC-III, hs-CRP, and γ-GT (Table 1).

To further identify the determinants of IR among these variables associated with HOMA-IR, we conducted a multiple regression analysis with HOMA-IR as the dependent variable. The model included apoM and adiponectin as potent explanatory variables, along with BMI, waist circumference, physical activity, TG, HDL-C, hs-CRP, γ-GT, apoC-III, apoB100, T2D, dyslipidemia, hypertension, and case-control status. This model explained 39% of the variability in HOMA-IR and revealed that diabetes status and four variables were determinants of HOMA-IR variability, which included apoM but not adiponectin (Table 2). Among the identified determinants, waist circumference and γ-GT showed the strongest positive associations with HOMA-IR, explaining 13.6% and 9.6% of the variability, respectively. In contrast, apoM displayed an inverse association, accounting for 5.6% of the variability in HOMA-IR. Notably, no significant interaction between apoM and case-control status was observed (P = 0.58). Furthermore, even after removing apoM from the set of variables at entry, adiponectin still did not contribute to the variability of HOMA-IR (data not shown).

Table 2 Multiple linear regression analysis on HOMA-IR in cohort A.Increased circulating apoM is closely related to insulin resistance relief after sleeve gastrectomy

To further study the association between apoM and IR, we examined the changes in circulating apoM after SG in patients with obesity and analyzed their correlation with the improvement in HOMA-IR. Anthropometric and biochemical characteristics of these patients at baseline (i.e., before gastric sleeve surgery) and one year after surgery are presented in Supplementary Table S4. One year after surgery, all individuals showed a decrease in adiposity markers (BMI, fat mass, and waist circumference) and displayed an improved lipid profile, including reduced atherogenic lipids (TG and LDL-C) and increased HDL markers (HDL-C and apoA-I). Additionally, there was a significant decrease in HOMA-IR (from 4.26, [SD, 1.49] to 1.39, [SD, 0.69]) and an increase in adiponectin levels (from 5.24 µg/L, [SD, 1.61] to 10.47 µg/L [SD, 4.07]). Circulating levels of apoM were not significantly changed (Supplementary Table S4); however, we found a strong inverse relationship between the changes in HOMA-IR and changes in circulating apoM (r = −0.71), while no significant association was observed with changes in adiponectin (Fig. 1). Of note, there was a trend of a positive correlation between changes in circulating apoM and changes in adiponectin (r = 0.57, P = 0.08, not shown).

Fig. 1: Changes in circulating apoM and adiponectin relationship with improvement of insulin sensitivity after bariatric surgery.

Sera from 11 individuals with obesity (cohort B) were tested for apoM (panel A) and adiponectin (panel B) levels before and one year after sleeve gastrectomy. Changes in adipokine concentrations were correlated to the changes in the Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) index. Data are expressed as percentage change from baseline. Linear correlation coefficients and P-values are displayed in each graph. The dotted lines represent 95% confidence intervals.

Positive association between ADIPOQ and APOM mRNA level in AT

We compared the gene expression of adiponectin and apoM in human subcutaneous AT from individuals with overweight or obesity (cohort C, Fig. 2A and Supplementary Table S5). ADIPOQ gene expression in AT was higher than APOM gene expression and both were positively correlated (r = 0.44, Fig. 2A). Additionally, HOMA-IR negatively correlated with AT expression of ADIPOQ (r = −0.15 Fig. 2B), and APOM gene expression in AT (r = −0.12, Fig. 2C).

Fig. 2: ADIPOQ and APOM gene expression in adipose tissue from overweight and obese individuals.

ADIPOQ and APOM mRNA levels were measured in subcutaneous adipose tissue of 267 men and women with overweight or obesity (cohort C). Linear regression analysis of ADIPOQ and APOM gene expression (panel A). Linear regression of APOM (panel B) and ADIPOQ (panel C) gene expression to HOMA-IR. Linear correlation coefficient and P-value are presented. Dotted lines represent 95% confidence interval.

Inflammatory factors downregulate ADIPOQ and APOM gene expression in adipocytes

Due to the close association of IR with AT inflammation, we aimed to mimic the impact of inflammatory conditions on ADIPOQ and APOM gene expression in adipocytes. We treated hMADS adipocytes with conditioned media from pro-inflammatory M1-like, or anti-inflammatory M2-like polarized ThP-1 macrophages. The former media led to down-regulation of both ADIPOQ and APOM gene expression (Fig. 3A). A strong correlation between ADIPOQ and APOM gene expression was observed in these adipocytes (r = 0.91, Fig. 3B). Similarly, a treatment with CRP also downregulated the expression of ADIPOQ and APOM genes in hMADS cells (Fig. 3C), while the positive correlation between ADIPOQ and APOM mRNA levels persisted under this condition (r = 0.82, Fig. 3D).

Fig. 3: Inflammatory factors downregulate ADIPOQ and APOM gene expression in adipocytes.

Effect of conditioned media from non-polarized, M2-like polarized or M1-like polarized ThP-1 cells on ADIPOQ gene expression in hMADS adipocytes (panel A). Association between ADIPOQ and APOM gene expression in hMADS adipocytes treated for 48 h with conditioned media from non-polarized (gray dots), M2-like polarized (black dots) or M1-like polarized (white dots) ThP-1 cells (panel B). ADIPOQ and APOM gene expression in hMADS adipocytes after C-reactive protein treatment for 48 h (panel C). Association between ADIPOQ and APOM gene expression in hMADS adipocytes treated for 48 h with C-reactive protein (panel D). Results are presented as mean ± SD of 3 independent experiments (panels A and C). Linear correlation coefficients and P-values are displayed in each graph. The dotted lines represent 95% confidence intervals (panels B and D). Data were analyzed by Kruskal–Wallis’ test (panels A and C). *P < 0.05, **P < 0.01.

Adiponectin promotes APOM gene expression

Adipocytes express specific adiponectin receptors, whereas apoM lacks known specific receptor. To further investigate the relationship between the two adipokines, we examined the effect of adiponectin on APOM gene expression. Treating hMADS adipocytes (Fig. 4A) or HepG2 hepatocytes (Fig. 4B) with increasing concentrations of adiponectin resulted in a dose dependent increase in APOM expression, with a mean 1.6-fold [SD, 0.9]) or 1.5-fold [SD, 0.3] rise at 1 µM adiponectin for adipocytes and hepatocytes, respectively.

Fig. 4: Adiponectin upregulates APOM gene expression in adipocytes and hepatocytes.

APOM mRNA level upon 48 h of globular adiponectin treatment in hMADS adipocytes (panel A) and in HepG2 hepatocytes (panel B). Results are expressed as percentage of control and are presented as mean ± SD of 3 independent experiments or 1 experiment, respectively. Data were analyzed by Kruskal–Wallis’ test. *P < 0.05, **P < 0.01.