A total of 366 subjects were included in this study, including 180 MetS patients with an average age of 26 (25–30) years and 159 healthy controls with an average age of 25 (24–28) years and 27 normal individuals [age 25 (24-29)] for intervention study. These participants were randomly selected from the 2017-2020 cross-sectional survey of obesity among young women in Chongqing, as a section of the Chongqing physique and health survey project. MetS was diagnosed based on Chinese Diabetes Society criteria (CDS guideline 2017) (16). MetS was diagnosed as follows: (1) abdominal obesity (waist circumference, WC ≥ 90 for male or ≥ 85 cm for female); (2) blood pressure (BP) ≥ 130/85 mmHg or receiving anti-hypertensive medication; (3) triglycerides (TG) ≥ 1.7 mmol/L. (4) high-density lipoprotein- cholesterol (HDL-C < 1.04 mmol/L). (5) fasting blood glucose (FBG) ≥ 6.1 mmol/L or 2-h blood glucose after the glucose challenge (2h-BG) ≥ 7.8mmol/L or known history of T2DM. Patients with the following diseases were excluded, such as heart failure, liver cirrhosis, liver and kidney failure, long-term use of steroids, cancer, infection, etc. Individuals were defined as normal glucose tolerance (NGT), impaired glucose tolerance (IGT), or T2DM according to WHO 1998 diagnostic criteria (17). In this study, MetS patients were diagnosed for the first time without any drug and lifestyle intervention. The control group was healthy subjects from physical examination, who had normal blood glucose, no family history of T2DM and hypertension, no clinical evidence of any disease, and did not take any drugs. The experimental design was shown in Figure 1 . This study was approved by the Human Research Ethics Committee of Chongqing Medical University and has been registered on atchictr.org (ChiCTR1800019776).

Anthropometric measurements, including height, weight and blood pressure (BP), were performed at 8:00 am, and body mass index (BMI) was calculated. The waist circumference (WC) was determined as the minimum value between the iliac crest and the lateral costal margin. After 12 hr of overnight fasting, the blood sample was taken for biochemical measurements, including fasting blood glucose (FBG), blood glucose during oral glucose tolerance test (OGTT), glycosylated hemoglobin, insulin, triglyceride (TG), total cholesterol (TC), free fatty acid (FFA), HDL-C and LDL-C as described previously (17). The percentage of fat in vivo (FAT%) was examined using bioelectrical impedance (BIA-101; RJL Systems, Shenzhen, China). As previously reported, the homeostasis model assessment index (HOMA-IR) was calculated to assess IR (18).

OGTT was performed for study individuals at 8:00 am. after fasting for 12 h. Venous blood samples were drawn for GPHB5, adiponectin, glucose, and insulin (19).

Euglycemic-hyperinsulinemic clamp (EHC) Test was performed on all subjects as previously described (20). Briefly, after overnight fasting, the test started at 8:00 am, and catheters were inserted into the left and right elbow anterior veins for collecting blood and infusing glucose and insulin. During the EHC, regular human insulin (1 mU/kg/min) was infused for two hours, and 20% glucose was infused at a variable rate to maintain blood glucose at fasting levels. The EHCs lasted for 120 min, and blood glucose was measured every 10 min. Glucose disposal rate (GDR) was defined as the glucose infusion rate (GIR) during the stable phase of EHC, which was related to the M-value. Venous blood was collected at 0, 80, 100, 110, and 120 min during the clamping and stored at - 80° C for insulin and GPHB5 determination.

A lipid-infusion experiment and the EHC were performed in 25 normal individuals (8 males and 17 females; age 25.84 ± 3.66 yr; BMI 21.17 ± 2.02 kg/m²) to establish a lipid-induced IR. Briefly, after 10-h of fasting, a polyethylene catheter was inserted into the anterior cubital vein at 8:00 am for the infusion of the glucose and insulin. Another catheter was inserted into the contralateral forearm vein for blood collection. Lipid and heparin (0.4 U/kg/min) were infused at a rate of 1.5 ml/min for 4 hours (21).

20 healthy individuals participated in the cold-exposure experiment, including 9 males and 11 females (age 25.95 ± 3.24 yr; BMI 21.77 ± 1.86 kg/m²). After 12-h fasting, individuals with light clothing were placed on a bed with a water-circulated cooling blanket (ThermoBlanket, P&C-AII, Hengbang Technology) at 8:00 am. The water temperature was regulated by electromyography. After 30 min of exposure to 27°C, a blood sample was drawn, and the water temperature was cooled to 18°C, and then reduced by 2°C every 3 min until 12°C. This temperature was maintained for 5 min, and then blood was drawn for GPHB5 measurement.

20 young individuals 8 males and 12 females (age 26.16 ± 3.51 yr; BMI 21.50 ± 2.08 kg/m²) participated in the physical exercise test. After 12 hours of fasting, at 8:00 am, a treadmill exercise was performed on these individuals, with 60% of maximal oxygen consumption for 45 min. Blood samples were collected for the measurement of GPHB5 concentration at four-time points: baseline, after exercise, and rest for 60 and 120 min after exercise.

Fifty-two women with MetS (age 28.2 ± 4.3 yr) participated in the Glucagon-like peptide-1 receptor agonists (GLP-1RA, Lira) intervention study. All participants were given informed written consent about the side effects of Liraglutide (lira) at the onset of this study. All women did not use any drugs three months before enrollment. Women with a history of acute pancreatitis and thyroid tumors were excluded from this intervention study. Treatment was started after the basic evaluation, and the dose of the lira was increased from 0.6 mg to 1.8 mg/d sc once a day for 24 weeks. GTT, ITT, EHC, insulin, biochemical indexes, and GPHB5 levels were measured at the 12^th and 24^th weeks of treatment. All women underwent anthropometric measurements before and after GLP-1RA treatment. Blood samples were collected at 8:00 am on day 0, week 12, and week 24 of GLP-1RA treatment for GPHB5 and biochemical parameters measurement. Fasting blood samples were collected at 8:00 am on day 0, week 12, and week 24 of GLP-1RA treatment for GPHB5, insulin, and blood glucose measurement.

Circulating GPHB5 was measured by an ELISA kit (Kete Biotechnology Co., Ltd, Jiangshu, China) following the manufacturer’s protocol. The sensitivity of this kit was 10 pg/ml, and intra- and inter-assay variations (CV) were both < 10%. The kit is highly sensitive and specific for the determination of human GPHB5 and is not interfered with by other cytokines. Serum adiponectin levels were also measured by an ELISA kit from Aviscerabio science Inc., USA (sk00010-01), with intra- and inter-assay CV of 4% - 8% and 8% - 12%, respectively, as previously reported (19).

GPHB5 gene was mapped into the Search Tool or the Retrieval of Interacting Genes (STRING) database (v11.0) with a 0.4 confidence score to explore the protein-protein interaction (PPI) relationship between the hub gene and interacting genes. The adding nodes function of STRING was used to depict the third generation of relationship for the hub genes, which is the indirect relationship of the PPI network (22). Gene Ontology (GO) and Kyoto Encylopedia of Genes and Genomes (KEGG) analyses were performed using the clusterProfiler package. Gene classification was performed according to the biological processes (BP), cellular components (CC), and molecular functions (MF) from GO analysis. q-value < 0.05 indicates statistical significance.

GPHB5 mRNA expression was performed by RT-PCR assay, as previously described (20). Gene expressions were analyzed using the comparative threshold cycle (Ct) method in relation to the levels of the ß-actin. The sequences for GPHB5 were: F: 5’-CCAGACAG GTGACAGTGAAGC-3’ and R: 5’- ACATCGGACAGCCATAGGG-3’.

Data were expressed as mean ± SD or median with an interquartile range. All statistical analyses were performed by using the SPSS software package (version 2.0, SPSS Inc.). Two-tailed p-value < 0.05 was considered statistically significant. A Kolmogorov-Smirnov test was used to analyze whether each variable was normally distributed. Non-normal distribution data were logarithmically transformed before analysis. Linear regression analysis was used to evaluate the correlation between circulating GPHB5 and other indicators. Multiple regression analysis was performed to evaluate the statistically significant variables of circulating GPHB5. The quartile-quartile (Q-Q) plot was depicted using the R statistical package to show the expected distribution of p values in logistic regression analysis relative to the null hypothesis (http://www.r-project.Org). To provide a tool for identifying MetS patients using circulating GPHB5, we calculated the receiver operating characteristic (ROC) curve as an independent predictor and obtained a cut-off point value. Cochran-Armitage trend and Row Mean Scores tests were performed to analyze the tendency of serum GPHB5 concentrations associated with MetS. ROC curve analyses were performed to determine the value of using circulating GPHB5 to predict MetS. In the OGTT, the area under the glucose (AUCg) and insulin (AUCi) curve was determined according to the trapezoidal rule. The equation: n = ( Z α / 2 σ ε μ ) 2 (σ, standard; µ, mean; Z_α/2 = 1.96, α = 0.05, ε = 10%) was used for estimating sample size.² In statistical analyses, p < 0.05 was considered significant.

We initially conducted a bioinformatics investigation employing Internet data to understand the relationship between GPHB5 and metabolism. PPI network showed 11 genes interacting with GPHB5, including NPS, TSHB, TSHR, FSHR, POMC, LHCGR, GPHA2, AVP, GNRHR, and KISS1R, some of which are involved in metabolism and energy balance, such as POMC and KISS1R (23, 24). Further, we found ten more interacting proteins through additional functions ( Figure 2A ). KEGG analysis was performed with a q-value < 0.05. Twelve pathways were significantly enriched in KEGG analysis, including autoimmune thyroid disease, vasopressin- regulated water reabsorption, cortisol synthesis and secretion, GnRH secretion, aldosterone synthesis and secretion, regulation of lipolysis in adipocytes, thyroid hormone synthesis, melanogenesis, ovarian steroidogenesis, GnRH signaling, neuroactive ligand-receptor interaction, and cAMP signaling pathways, etc. Some are related to glucose and lipid metabolism and energy balance ( Figure 2B ). Using the human genome as a background variable (q-value < 0.05), the GO analysis annotated 56 BP and 16 MF pathways. Some pathways were related to feeding behavior and energy metabolism, such as OPRM1and melanocortin receptor family, etc. ( Figure 2C ). The results from bioinformatics analysis preliminarily suggest that GPHB5 may be associated with glucose and lipid metabolism and energy equilibrium.

To identify the source of circulating GPHB5, we analyzed the expression of GPHB5 mRNA in tissues and organs in mice. We found higher expression in the heart, brain, liver, and skeletal muscle ( Figure S1 ). Therefore, we believe that the main sources of GPHB5 may be the heart, brain, and liver.

The general clinical characteristics of MetS women and control individuals are shown in Table 1 . Compared with the control group, MetS women had higher BMI, WC, blood pressure (BP), TG, TC, LDL-C, FBG, two-hour blood glucose after glucose overload (2-h BG), fasting insulin (FIns), 2-h serum insulin after glucose overload (2 h-Ins), HOMA-_IR, AUCi, AUCg, HbA1c, visceral adiposity index (VAI) and body adiposity index (BAI). However, HDL-C and M-value were lower in MetS patients. In addition, thyroid hormones (T3 and T4) and thyroid-stimulating hormone (TSH) were comparable in MetS patients and healthy controls ( Table S1 ). As expected, serum adiponectin levels were significantly lower in MetS patients than in normal controls ( Table 1 and Figure 3A ).

To investigate the distribution of circulating GPHB5 concentration, serum GPHB5 levels in 159 average women were measured by ELISA kit. The results showed that serum GPHB5 concentrations ranged from 1.35 to 5.12µg/L for 95% in the normal controls ( Figure 3B ). Q-Q plot showed that the error distribution was a standard normal distribution ( Figure 3C ). Interestingly, in contrast to low serum adiponectin levels, circulating GPHB5 concentrations were significantly elevated in MetS patients ( Table 1 and Figure 3D ). Furthermore, obese/overweight (OB) individuals had lower adiponectin levels and higher GPHB5 levels than lean individuals ( Figures 3E, F ). These results indicated that GPHB5 might be related to metabolic disorders and obesity.

We performed linear and multivariate regression analyses to understand the relationship between GPHB5 and other indicators. Linear regression analysis revealed that circulating GPHB5 was significantly positively correlated with BMI, WHR, BP, FBG, 2 h-BG, FIns, 2 h-Ins, LDL-C, FFA, HbA1c, HOMA-IR, AUCi, AUCg, and BAI, while negatively correlated with HDL-C, M-value, and Adiponectin ( Table 2 ). The dependent variable GPHB5 and other parameters (independent variables) were used for regression analysis to establish a multiple regression model. The results showed that HbA1c, AUCg, and BAI were independent influencing factors of serum GPHB5 ( Table S2 ). The regression equation of gphb5 was: Y_GPHB5 =-19.21 + 0.496× HbA1c + 0.96 × AUCg + 0.718 × BAI (R = 0.571, R2 = 0.326).

We further stratified the serum GPHB5 concentration according to the MetS components in MetS patients. Serum GPHB5 concentration is increased with the increased components of MetS ( Figure 3G ). Additionally, Cochran–Armitage trend and row mean score tests revealed that increased GPHB5 levels had a linear trend related to MetS ( Table S3 ). To calculate the probability of MetS, serum GPHB5 levels were divided into four quartiles (1, < 2.23 µg/L; 2, 2.23-3.13 µg/L; 3, 3.13-3.92µg/L; 4, > 3.92 µ g/L). Logistic regression analysis revealed that odds ratios for developing MetS in the quartiles 3 and 4 were higher than the quartile 1 and 2 [95% confidence interval (CI) 3.08; 3.92µg/L for quartile 3 and 95% CI 3.69; 5.53µg/L for quartile 4 vs. quartile 1, both p < 0.01, Figure 3H ].

To investigate whether GPHB5 can predict MetS, a ROC curve analysis was performed. The AUC of ROC was 0.65, the sensitivity was 79%, and the specificity was 60.6%, respectively ( Figure 3I ). The cut-off value for serum GPHB5 to predict MetS was 3.27µg/L.

To further understand the relationship between blood glucose, insulin, and GPHB5, we conducted OGTT and EHC experiments in MetS and healthy individuals. The results showed that serum GPHB5 levels did not change significantly during the OGTT in both normal controls and MetS individuals (from 3.43 ± 1.15 to 3.55 ± 1.35 µg/L at 30 min, and then to 3.51 ± 1.18 µg/L at 60 min, and finally to 3.48 ± 1.32 μg/L at 120 min), while AUC_GPHB5 in MetS patients was significantly higher than that of controls ( Figure 4A ). During the OGTT, circulating GPHB5 levels did not change with blood glucose and insulin levels in normal individuals ( Figures 4B, C ). In addition, changes in circulating GPHB5 levels were similar to adiponectin in response to the glucose challenge ( Figure 4D ).

To observe the effect of elevated insulin levels on GPHB5, we engaged in an EHC to increase insulin levels under normal blood glucose conditions ( Figure 4E upper). Blood glucose was maintained at 4-6 mmol/L during the clamping, while insulin level increased from 12.4 ± 1.2 to 67.3 ± 5.4 mU/L. At the steady-states of the clamp, the M-values in MetS patients were significantly lower than those in the control group, suggesting that the whole-body insulin sensitivity in MetS individuals is reduced (5.69 ± 2.81 vs. 8.87 ± 2.46 mg/kg/min, p <0.01; Figure 4E bottom). Although serum insulin concentrations significantly increased during the clamping, circulating GPHB5 levels remained unchanged in MetS patients and controls compared with baseline (3.52 ± 1.29 vs. 3.43 ± 1.15 μg/L; Figure 4F ). However, AUC_GPHB5 was significantly higher in MetS individuals (5.61 ± 1.68 vs. 4.15 ± 1.36 μg·h/L; Figure 4F ). These data indicate that GPHB5 may not be regulated by insulin in vivo.

To understand whether acute-increased FFA levels affect GPHB5 release, we performed a 4-h lipid-infusion combined with 2-h EHC in healthy individuals ( Figure 4G upper). After 2-h lipid infusion, FFA concentration increased significantly from 0.46 ± 0.1 to 1.46 ± 0.2 mmol/L. As shown in Figure 4G (bottom), during the stable-state of the EHC, lipid-infusion significantly reduced M-values, suggesting that the acute increase of FFA resulted in an IR in vivo. However, 2-h lipid perfusion did not increase circulating GPHB5 concentration. At the beginning of EHC, elevated insulin levels led to a slight increase in GPHB5 concentration, but did not reach statistical significance (from 2.86 ± 1.21 μg/L at 80 min to 3.2 ± 1.26 μg/L at 100 min and then to 3.16 ± 1.34 μg/L at 120 min; Figure 4H ). Compared with EHC alone, lipid-infusion did not change GPHB5 levels ( Figure 4I ). These results further suggest that GPHB5 may not be regulated by insulin and FFA.

To understand the effects of thermogenesis and energy consumption on GPHB5 in vivo, we conducted cold-exposure ( Figure 5A upper) and physical activity ( Figure 5B upper) experiments in healthy subjects. During the cold-exposure investigation, we found that there were no significant changes in serum GPHB5 levels when the body temperature decreased from 27 °C to 12 °C (3.16 ± 0.81 vs. 2.94 ± 0.95 μg/L; Figure 5A bottom). Furthermore, there was no significant change in serum GPHB5 level before and after physical activity (3.01 ± 0.98 vs. 3.06 ± 0.91μg/L; Figure 5B bottom). These data show that short-term thermogenic effects and energy consumption do not regulate the secretion and release of GPHB5 in vivo.

To investigate the impact of improved insulin sensitivity on GPHB5 levels, we used Lira therapy for 24 weeks in 70 MetS patients. After GLP-1RA intervention for 6 months, BMI, WHR, Fat %, BP, TG, TC, LDL-C, FFA, FBG, 2-h BG, FIns, 2-h Ins, HOMA-_IR, BAI, and VAI in MetS patients decreased significantly compared with those before treatment, while HDL-C was increased in these patients ( Table 3 ). In addition, glucose tolerance was significantly improved after treatment ( Figures 6A, B ). As expected, the M-values of the clamp after treatment in MetS women increased significantly compared with that before treatment ( Table 3 and Figure 6C ). These data indicate an improved insulin sensitivity in vivo. Importantly, after GLP-1RA treatment, serum adiponectin levels in MetS women increased significantly ( Table 3 and Figure 6D ), while serum GPHB5 levels decreased significantly ( Table 3 and Figure 6E ).