Functional Gly297Ser Variant of the Physiological Dysglycemic Peptide Pancreastatin Is a Novel Risk Factor for Cardiometabolic Disorders

Pancreastatin (PST) is a proteolytically derived peptide from the acidic glycoprotein chromogranin A (CHGA) that is primarily expressed in endocrine, neuroendocrine, and neuronal tissues (1,2). Human PST exists in different molecular forms, including 29-mer PST (CHGA_273–301), 48-mer PST (CHGA_254–301), and 52-mer PST (CHGA_250–301), all containing the conserved functional C-terminal part (2); among these forms, the 52-mer peptide has been reported to be the major one in human plasma (3). Proteases like prohormone convertase 2 and carboxypeptidase H have been reported to be involved in the intracellular processing of PST (4,5). The carboxy-terminus of PST is amidated by peptide α-amidating mono-oxygenase (6).

PST exerts diverse biological effects, including inhibition of insulin secretion stimulated by physiological activators (e.g., glucose/glucagon) and pharmacological agents (e.g., sulphonylurea), inhibition of insulin-stimulated glucose uptake and translocation of GLUT4 in adipocytes, enhancement of hepatic gluconeogenesis, and suppression of insulin signaling (7). Mice lacking ChgA protein generated by systemic deletion of the Chga gene (8) showed reduced insulin resistance. Furthermore, the exogenous administration of human PST peptide in these Chga knockout mice caused increased insulin resistance and increased blood glucose levels along with enhanced synthesis of gluconeogenic genes (9,10). Consistently, infusion of PST through the brachial artery profoundly reduced human forearm glucose uptake and arteriovenous glucose gradient (11).

Resequencing of the CHGA locus by us and others (12,13) has led to the identification of several variants within the PST-encoding domain in various populations of the world (dbSNP database, https://www.ncbi.nlm.nih.gov/snp/rs9658664). Among these naturally occurring PST variants, the G297S variant (rs9658664) is located within the functionally active C-terminus of the peptide. Of note, the 297Ser variant peptide (PST-297S) has been shown to exhibit higher potency in inhibiting insulin-stimulated glucose uptake than the wild-type peptide (PST-WT) (11).

In our previous study (13), we analyzed the effect of this genetic variation on cardiovascular and metabolic disease states in a small Indian population (n = ˜400), in whom we found an association of the 297Ser allele with higher plasma glucose levels. In this study, we have extended our sample to a larger population (n = ˜4,300) comprising South Indian and North Indian subjects. This study revealed strong association of the p.Gly297Ser polymorphism with several cardiometabolic parameters (e.g., plasma glucose, glycosylated hemoglobin [HbA_1c], catecholamines, and blood pressure). We also used a combination of experimental (biochemical, cellular, and molecular assays, including insulin-stimulated glucose uptake, measurement of endogenous gluconeogenic gene transcripts, and receptor-ligand binding experiments) and computational (molecular modeling, docking, and molecular dynamics simulations) tools to unravel the mechanistic basis for the higher activity of the variant peptide (PST-297S) in comparison with PST-WT and to account for the higher disease risk in the carriers of the 297Ser allele. We provide evidence for higher-affinity binding of the variant peptide with insulin receptor (IR) and glucose-regulated protein 78 (GRP78) that may mediate the enhanced disease risk associated with the 297Ser allele.

We recruited 3,602 unrelated volunteers at three study centers in Chennai (South India) and 716 unrelated subjects at the Post Graduate Institute of Medical Education and Research in Chandigarh (North India). Details about the demographic and clinical parameters are given in Supplementary Tables 1 and 2.

All subjects gave written informed consent for the use of their blood samples for genetic and biochemical analyses. This study was approved by the Institute Ethics Committee at Indian Institute of Technology Madras, Chennai, India in accordance with Declaration of Helsinki (reference number: IITM IEC 2010027).

Exon-7 region of CHGA was resequenced in the first set of samples to detect the presence of single nucleotide polymorphisms (SNPs) in the PST domain (13), followed by genotyping for the p.Gly297Ser SNP (rs9658664) using TaqMan allelic discrimination method in the next set of samples.

Circulating levels of common biochemical parameters (e.g., insulin, glucose, HbA_1c, and cholesterols) in the subjects were measured using standard pathological laboratory methods, which have been described in detail in the Supplementary Material. Plasma catecholamine and PST levels were quantified using a radioimmunoassay kit (catalog number BA R-6500) and an ELISA kit (catalog number TM E-9000), respectively, from LDN Labor Diagnostika Nord GmbH & Co. (Nordhom, Germany).

HepG2 (human liver cell line), L6 (rat skeletal muscle cell line), and SH-SY5Y (human neuroblastoma cell line) cells were obtained from the National Centre for Cell Science (Pune, India). L6 cells were differentiated into myotubes by growing them in DMEM containing 10% FBS until they became confluent and then replacing the medium with differentiation medium (DMEM with 2% FBS) for 8 d (10). 3T3-L1 cells were grown and differentiated as described previously (13). PC12 cells were maintained and grown in DMEM containing 5% FBS and 10% horse serum (14). HEK-293 cells used for the binding assays were cultured in minimal essential media supplemented with 10% FBS and appropriate antibiotics. β-TC-6 cells were cultured in DMEM with high glucose supplemented with 15% heat-inactivated FBS and 1× penicillin/streptomycin.

PST-WT (PEGKGEQEHSQQKEEEEEMAVVPQGLFRG-amide) and PST-297S (PEGKGEQEHSQQKEEEEEMAVVPQSLFRG-amide; the variant serine residue is shown in boldface and underlined) peptides were synthesized and purified as described previously (13).

Norepinephrine secretion from SH-SY5Y and PC12 cells in the presence/absence of each PST peptide was estimated according to a previously described method (15). Glucose uptake by differentiated L6 and 3T3-L1 cells was assayed using a modified version of our previously reported protocol (13).

Total RNA samples from HepG2 cells treated with/without PST peptides (PST-WT or PST-297S) were isolated using our previously described method (16). Endogenous human (PCK-1) and glucose-6-phosphatase (G6PC1) and GRP78 transcript levels were estimated using real-time PCR with gene-specific primers (as detailed in the Supplementary Material). The relative gene expression levels for all of the genes were determined using the 2^−ΔΔCt method.

Molecular modeling of PST-WT and PST-297S 52-mer peptides was carried out using MODELER version 9v13 (17) with the previously derived 29-mer PST peptide structures (13) as templates. The homology-modeled structures were further refined using molecular dynamics simulations for 200 ns. The stability of the simulated structures was assessed by plotting root mean square deviations of Cα atoms with respect to the starting structure over the simulation period.

Next, we performed unbiased docking of the representative PST structures to GRP78 (Protein Data Bank [PDB] identification number 3LDO) monomer (18). We also carried out docking of PST peptides with IR (PDB identification number 2DTG) (19). We used a protein-docking program, ZDOCK (20), to predict the binding of PST peptides to GRP78 monomer and IR. Molecular interactions between PST peptides and GRP78 or IR were calculated using the PDBsum database (21). ZDOCK scores between GRP78/IR and PST peptides were calculated by using ZDOCK 3.0.2 scoring function considering IFACE statistical potential, shape complementarity, and electrostatic parameters (22).

The effect of PST-WT and PST-297S peptides on GRP78 ATPase activity was determined using a spectrophotometric Malachite Green Phosphate Assay Kit (MAK307; Sigma-Aldrich, St. Louis, MO) (23).

The inhibitory effects of PST-WT or PST-297S peptides on tunicamycin-stimulated GRP78 expression in HepG2 cells was determined using Western blot analysis (23).

GRP78- and IR-overexpression constructs were generated by cloning GRP78 cDNA (obtained from catalog no. 32701; Addgene, Watertown, MA) and IR cDNA (a kind gift from Dr. Frederick M. Stanley [24]) into pcDNA 3.1. Cells were transfected with either GRP78-FLAG or IR-FLAG overexpression constructs (8 μg/100-mm tissue culture dish) using TurboFect (Thermo Fisher Scientific, Waltham, MA) transfection reagent. At 48 h posttransfection, the cells were serum-starved for 6 h before being used for each of the individual experiments. Isolation of plasma membrane fraction from transfected HEK-293 cells or β-TC-6 insulinoma cells and Western immunoblotting were performed as previously described (25). The expression levels of the proteins were assessed using chemiluminescence postincubation with an anti-FLAG mouse monoclonal antibody (used at a dilution of 1:1,000; Sigma-Aldrich), followed by an anti-mouse secondary antibody (used at a dilution of 1:5,000).

For the competitive binding assay, 10 μg of plasma membrane (from HEK-293 or β-TC-6 cells) in binding buffer (75 mmol/L Tris-HCl, pH 7.5, 2 mmol/L EDTA, and 12.5 mmol/L MgCl₂) was used to assess the ability of PST-WT or PST-297S peptides to displace labeled ^[125]I-Tyr PST (catalog no. H-5604; Peninsula Laboratories International, San Carlos, CA) or ^[125]I-Tyr Insulin (catalog no. NEX420, Insulin, human ¹²⁵I; PerkinElmer, Waltham, MA). The peptides were added in increasing concentrations (100 pmol/L through 50 μmol/L for IR binding and 10 pmol/L through 1 μmol/L for GRP78 binding) along with 100 nCi of labeled PST or insulin. The binding reactions were incubated at 37°C for 1 h with constant shaking. Increasing concentrations (100 pmol/L through 50 μmol/L) of cold insulin were used as a positive control for displacement in studies pertaining to IR expression. Displacement was calculated as a percentage of receptors in the control samples that did not contain any competing peptides.

The experimental data for various cellular functions are representative of three or more separate experiments. All results are represented as mean ± SD from replicate cells or wells, as stated in the respective figure legends. To evaluate the effect of peptides on cellular activity or to compare between different conditions, one-way ANOVA followed by Tukey multiple-comparison post hoc test, Student t test, or linear regression analysis, as appropriate, was performed using Prism (GraphPad9 Software, San Diego, CA) or SPSS software (SPSS Inc., IBM, Chicago, IL). Analysis of genotypic frequencies for Hardy-Weinberg equilibrium was carried out as described (26). Odds ratios (ORs) were calculated using binary logistic regression analysis after age, sex, and BMI adjustments. Association of genotypes with various phenotypes was analyzed using SPSS software. The blood pressure data were adjusted for antihypertensive treatments as described previously (27), and subjects under medications (specifically hypoglycemic drugs) were not included in our analysis. A two-tailed P value of <0.05 was considered as statistically significant between groups. The power of the study was calculated using Quanto version 1.2.4 (Supplementary Table 3) (28).

The data sets and resources generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

In this study, we extended our previous analysis carried out in a small population of 410 individuals (13) to a larger population (∼4,300 subjects) comprising subjects from two Indian populations (South Indians at Chennai and North Indians at Chandigarh). The samples were analyzed by direct sequencing of the PST genomic region in the primary set of samples, followed by genotyping (Supplementary Fig. 1) of the p.Gly297Ser variant in a secondary set. The allele frequencies for this SNP in Chennai populations were: Gly297, 93.6% and Ser297, 6.4%; the frequencies in the Chandigarh population were: Gly297, 93.0% and Ser297, 7.0%. The genotype frequencies were in Hardy-Weinberg equilibrium for the entire population (χ² = 0.169; P = 0.681) as well as in each ethnic group (χ² = 0.096, P = 0.756 for South Indian ancestry; and χ² = 0.086, P = 0.7699 for North Indian ancestry). The genotype frequencies do not differ significantly (χ² = 0.622; P = 0.733) between the South Indian and North Indian populations.

This SNP has also been detected in other ethnic populations, including people of European, African, American, East Asian, and South Asian origin; the distribution of genotypes differed by ethnicity (χ² = 13,688; P < 0.0001) (Supplementary Table 4). Taken together, the PST p.Gly297Ser genetic variation occurs in a large section of people worldwide, with an especially high prevalence in the South Asian population.

The demographic, physiological, and biochemical parameters of the case subjects (comprising patients with type 2 diabetes [T2D]/hypertension [HTN]/impaired glucose tolerance [IGT]/coronary artery disease [CAD]) and control subjects are shown in Supplementary Tables 1 and 2. The case subjects had significantly higher BMI, systolic blood pressure, diastolic blood pressure (DBP), and mean arterial pressure than the control subjects; several biochemical parameters relevant to cardiometabolic diseases (e.g., plasma glucose, HbA_1c, and cholesterols) were also elevated in the case subjects.

Logistic regression analysis to check for the association of the Ser297 allele with these different disease conditions revealed the occurrence of the allele at higher frequencies in the disease groups than in the control subjects. As shown in Table 1, upon using the additive model (GG vs. GA), we found that the Ser297 allele increased the risk for T2D by ∼1.4-fold (P = 0.018); the disease risk remained significant after adjusting for age, sex, or BMI, individually or together. Upon using the dominant model (GG vs. GA+AA) as well, the risk for T2D was close to ∼1.3-fold (P = 0.022), which remained significant after adjusting for age, sex, or BMI, individually or together. Using both of the models, the risk for HTN was elevated in Ser297 carriers by ∼1.4-fold (P = 0.013 for the additive model [GG vs. GA] and P = 0.017 for the dominant model [GG vs. GA+AA]); the risks remained significant after adjusting for age or sex in both of the models. Similarly, this PST allele increased the risk for CAD by ∼1.5-fold (P = 0.008 for the additive model [GG vs. GA] and P = 0.017 for the dominant model [GG vs. GA+AA]); the risk remained significant after adjusting for age, sex, or BMI, individually under the additive model and for BMI under the dominant model. The Ser297 allele also enhanced the risk for metabolic syndrome (MS) (case subjects with one or more of the conditions: T2D, HTN, IGT, or CAD) by ∼1.4-fold (P = 0.002 for the additive model [GG vs. GA] and P = 0.003 for the dominant model [GG vs. GA+AA]); the risk remained significant after adjusting for age, sex, and BMI, individually or together, under both the additive and dominant models. No association was observed upon carrying out logistic regression using the additive model to compare wild-types with homozygous variants (GG vs. AA); this could be attributed to the low number of homozygous variants (n = 20) in the population.

PST p.Gly297Ser SNP was found to be associated with plasma glucose levels in the overall population. The Gly/Ser and Gly/Ser+Ser/Ser subjects displayed markedly higher random blood sugar (RBS) levels than those of the Gly/Gly individuals, by 11.83 mg/dL (P = 0.014) and 10.80 mg/dL (P = 0.022), respectively (Fig. 1A). Similarly, Gly/Ser and Gly/Ser+Ser/Ser subjects displayed significantly elevated postglucose blood sugar (PGBS) levels than those of their Gly/Gly counterparts by 13.68 mg/dL (P = 0.015) and 13.54 mg/dL (P = 0.014), respectively (Fig. 1B). Consistent with their elevated plasma glucose values (Fig. 1A and B), the Gly/Ser and Gly/Ser+Ser/Ser individuals displayed higher HbA_1c levels (6.00% vs. 5.80%; 42.11 vs. 39.94 mmol/mol; P = 0.004) and (6.00% vs. 5.80%; 42.13 vs. 39.94 mmol/mol; P = 0.003) than the Gly/Gly subjects (Fig. 1C and D). The P values for HbA_1c remained significant after adjustments for age, sex, and BMI (Gly/Gly vs. Gly/Ser, P = 0.005; Gly/Gly vs. Gly/Ser+Ser/Ser, P = 0.002). Stratification of the subjects based on their plasma RBS, PGBS, and HbA_1c levels showed that the minor allelic frequency (MAF) of the PST 297Ser allele, in general, gradually increased with increasing RBS, PGBS, and HbA_1c levels (Fig. 1E–H).

Although not statistically significant, the Gly/Ser (111.6 mg/dL) and Gly/Ser+Ser/Ser (112.5 mg/dL) individuals displayed higher fasting blood glucose levels, when compared with the Gly/Gly (107.8 mg/dL) individuals. Similarly, the log-transformed HOMA of insulin resistance index levels were also higher (but did not attain statistical significance) for the Gly/Ser (0.27) and Gly/Ser+Ser/Ser (0.26) individuals as compared with the Gly/Gly (0.23) individuals.

In view of the association of the PST 297Ser allele with higher plasma glucose levels (Fig. 1A and B), we examined whether PST-297S exhibits enhanced ability to inhibit insulin-stimulated glucose uptake as compared with PST-WT peptide in relevant cell types (adipocytes and myotubes). Insulin-augmented glucose uptake in differentiated 3T3-L1 adipocytes and differentiated L6 myotubes (by ∼135%, P < 0.01 and ∼150%, P < 0.001, respectively) (Fig. 2A). PST-WT showed only a modest or no inhibitory effect on insulin-stimulated glucose uptake in these cells; in contrast, PST-297S significantly diminished the insulin-stimulated glucose uptake in adipocytes (from ∼135% to ∼110%; P < 0.05) as well as in myotubes (from ∼150% to ∼128%; P < 0.001) (Fig. 2A).

Next, we tested whether PST-297S and PST-WT peptides alter the expression of gluconeogenic genes (e.g., PCK1 and G6PC1) in HepG2 liver cells. While PST-WT peptide increased the endogenous expression of PCK1 by ∼2.1-fold, the PST-297S peptide showed a ∼4.6-fold increment (P < 0.05) (Fig. 2B). Likewise, PST-WT increased the endogenous G6PC1 gene expression by ∼2.2-fold, whereas PST-297S caused a ∼5.6-fold increment (P < 0.05) (Fig. 2C).

In view of the strong association of PST with glucose homeostasis (Figs. 1 and 2), we checked whether the plasma PST levels differ between subjects with T2D and age-, sex-, and BMI-matched healthy (normal glucose tolerant) control subjects. Subjects with diabetes showed an ∼1.2-fold (P = 0.042) increase in their plasma PST levels compared with the control subjects (Supplementary Fig. 2).

As shown in Table 1, the Ser297 allele displayed association with HTN. Carriers of this allele also showed significantly higher levels of DBP than the Gly/Gly (WT) individuals (Gly/Gly [80.15 mmHg] vs. Gly/Ser [81.6 mmHg], P = 0.018; and Gly/Ser+Ser/Ser [81.6 mmHg], P = 0.017]). As plasma catecholamine levels regulate blood pressure, we measured the catecholamine levels in the plasma samples of Gly/Gly and Gly/Ser individuals. Gly/Ser subjects displayed elevated levels of plasma norepinephrine (by ∼l.6-fold; P = 0.0003) and epinephrine (by ∼1.5-fold; P = 0.015) compared with Gly/Gly subjects (Fig. 3A). The P values for both remained significant after adjustments for age, sex, and BMI (P = 0.0004 for norepinephrine and P = 0.036 for epinephrine). Consistent with these findings, the PST-297S variant peptide evoked enhanced [³H]norepinephrine secretion activity both in neuronal SH-SY5Y (∼117% over basal; P < 0.01) and neuroendocrine PC12 cells (∼123% over basal; P < 0.05) compared with the PST-WT peptide (∼109% over basal in case of both SH-SY5Y and PC12 cells) (Fig. 3B).

The identity of a functional receptor for PST remains elusive. Since PST peptides exhibit several anti–insulin-like activities in adipocytes and skeletal muscle cells (10), which is supported by our present study as well, we hypothesized that PST peptides might interact with the IR. In addition, recently, a few reports have also provided evidence for interaction of PST with GRP78 (23,29). Therefore, in the current study, we investigated whether PST-WT and PST-297S peptides display differential binding to IR and GRP78.

To begin with, predictions of interactions of PST peptides with IR and GRP78 were made in silico. First, we modeled the PST 52-mer peptides (PST-WT and PST-297S) and simulated them for 200 ns to obtain the best time-averaged, representative structures (Supplementary Fig. 3A and B); all of the residues were in the allowed regions of the Ramachandran plot (Supplementary Fig. 3C). The PST-WT structure consisted of a stable helix comprising the residues from Gln279 to Val293, whereas the PST-297S structure consisted of a stable helix comprising the residues from Glu278 to Val293 (Supplementary Fig. 3B). The percent persistence of conformations carrying α-helix during the simulation period was higher for PST-297S (∼80%) than for PST-WT (∼65%).

Next, we docked PST-WT and PST-297S peptides with IR. The peptides displayed remarkably different interactions with IR (Fig. 4A and B). Interestingly, PST binds to IR at the same region where insulin binds; the binding pocket in the IR is formed by two monomers such that PST peptides and insulin seem to share a common binding site (Fig. 4C). Briefly, 6 residues of PST-WT interact with 8 residues of IR monomer-1 through 1 hydrogen bond and 26 hydrophobic interactions. The hydrogen bond was formed between Glu255 (PST-WT) and Lys484 (IR) with a bond distance of 3.1 Å. Seven residues of PST-WT interact with 8 residues of IR monomer-2 via 102 hydrophobic interactions (Fig. 4D). In contrast, PST-297S interacts with IR monomer-1 via 5 hydrogen bonds and 372 hydrophobic interactions (between 25 residues of PST-297S and 23 residues of IR monomer-1); 4 residues of PST-297S interact with 4 residues of IR monomer-2 via 1 hydrogen bond and 98 hydrophobic interactions. The hydrogen bonds formed between PST-297S and IR were between: Glu286 and Ser526; Glu290 and Asn711; Leu298 and Tyr708; Glu274 and Gln328; and Glu269 and Arg114 (two hydrogen bonds) (Fig. 4E). The PST-297S and IR complex displayed lower binding free energy of −12.4 kcal/mol and dissociation constant of 8.06 × 10⁻¹⁰ mol/L, whereas these values for the PST-WT and IR complex were −12.02 kcal/mol and 15.3 × 10⁻¹⁰ mol/L (Supplementary Table 5).

Similarly, each PST peptide was docked onto the GRP78 monomer. Analysis of the docked complexes revealed that PST peptides exerted profoundly different interactions with GRP78 (Supplementary Fig. 4A and B). In brief, 9 residues of PST-WT interact with 13 residues of GRP78 via 1 hydrogen bond and 133 hydrophobic interactions. The hydrogen bond (distance = 2.51 Å) was established between Gly297 of PST-WT and Asn389 of GRP78 (Supplementary Fig. 4C). In contrast, 14 residues of PST-297S interact with 18 residues of GRP78 through 174 hydrophobic interactions and 3 hydrogen bonds; these hydrogen bonds were between Glu269 and Tyr396 (bond distance = 2.23 Å), Arg300 and Gly48 (bond distance = 2.87 Å), and Asp272 and Arg49 (bond distance = 2.77 Å) residues (Supplementary Fig. 4D). Consequently, the PST-297S and GRP78 complex displayed lower binding free energy (−9.18 kcal/mol) and dissociation constant (0.185 × 10⁻⁶ mol/L) than the PST-WT and GRP78 complex (binding free energy = −7.89 kcal/mol and dissociation constant = 1.64 × 10⁻⁶ mol/L) (Supplementary Table 5).

To validate the results of in silico docking of PST peptides to IR (Fig. 4), we performed competitive binding assays. An IR-expressing construct was generated and transfected into HEK-293 cells, and overexpression was confirmed using Western blot (Fig. 5A). Saturation binding assays identified the optimum concentration of the label to be 100 nCi of ^[125]I-Tyr insulin with reproducible binding upon use of 10 μg IR-expressing plasma membrane (Supplementary Fig. 5). Since insulin is known to bind to IR with high affinity, cold insulin was used as a positive control to confirm functional IR overexpression by assessing its ability to displace labeled ^[125]I-Tyr insulin under conditions of endogenous IR expression as well IR overexpression (Fig. 5B). Increasing concentrations of PST-WT and PST-297S were used to evaluate displacement of ^[125]I-Tyr insulin from the IR-overexpressing plasma membrane. Indeed, both of the peptides were capable of displacing ^[125]I-Tyr insulin (Fig. 5C), though PST-297S resulted in greater displacement, leading to lower area under the curve (AUC) (P = 0.0002) of the displacement curve compared with PST-WT (Fig. 5D). Assessment of the potencies of the two peptides with respect to cold insulin in causing displacement of ^[125]I-Tyr insulin showed that the effect of PST-297S was closer to insulin, as compared with PST-WT. When expressed as a ratio of insulin as 100%, PST-297S (∼72%) displayed significantly greater (P < 0.0001) binding than PST-WT (∼46%) (Fig. 5E). To further validate the binding of endogenous IR with PST peptides, we used β-TC-6 insulinoma cells. Similar to HEK-293 cells, we also observed both PST peptides displaced ^[125]I-Tyr insulin (Fig. 5F), and PST-297S showed greater displacement compared with PST-WT, as evidenced by lower AUC (P < 0.0001) of the displacement curve (Fig. 5G). The effect of PST-297S was closer to insulin as compared with PST-WT with respect to cold insulin in causing displacement of ^[125]I-Tyr insulin from plasma membrane of β-TC-6 insulinoma cells (P < 0.0001) (Fig. 5H). These in vitro binding studies support the in silico docking studies (Fig. 4) and show that PST peptides interact differentially with IR.

To ascertain if the two PST peptides differ in terms of their binding affinity toward GRP78 as well, we first compared the potencies of PST peptides in causing inhibition of GRP78 ATPase activity in vitro. This was achieved using a Malachite Green-phosphate assay that measures the liberated free phosphate release that corresponds to an increase in absorbance at 620–640 nm. Inhibition of GRP78 ATPase activity is believed to be important for the dysglycemic role of PST (23). The stronger inhibitory effect of PST-297S (P < 0.0001) on GRP78 ATPase activity, as compared with PST-WT (P = 0.0051) with respect to control (Fig. 6A), suggests that PST-297S exhibits more potent dysglycemic effects. In addition, PST-297S (P < 0.0001) was also capable of inhibiting tunicamycin-stimulated increase in GRP78 expression in HepG2 cells to a greater extent, as compared with PST-WT (P = 0.0016) (Fig. 6B and C).

To parse out the interactions of the PST peptides with GRP78, competitive binding assays were performed using a similar strategy to that used for the IR studies. Overexpression of GRP78 was achieved by generating a GRP78-expressing plasmid and transfecting it into HEK-293 cells (Fig. 6D). Following overexpression, the plasma membrane from the cells was isolated, and increasing concentrations of PST-WT and PST-297S were used to evaluate displacement of labeled ^[125]I-Tyr PST peptide. PST-297S peptide was capable of displacing labeled ^[125]I-Tyr PST to a greater extent, as compared with PST-WT (Fig. 6E), as evidenced by the lower AUC (P = 0.0021) for the displacement curve of PST-297S than that of PST-WT (Fig. 6F). This finding further suggests that PST-297S displays greater binding affinity toward GRP78 compared with PST-WT peptide, consistent with our in silico predictions (Supplementary Fig. 4).

PST is well known for its dysglycemic activities, such as inhibition of glucose-induced insulin secretion and insulin-stimulated glucose uptake in pancreatic islet β-cells/primary rat adipocytes (2,30). Exogenous PST administration caused insulin resistance in WT as well as CHGA-null (i.e., PST-deficient) mice fed with a normal chow diet; consistently, the PST-deficient mice exhibited protection against high-fat diet–induced insulin resistance (31). While there is no report on the association of CHGA with diabetes, elevated levels of circulating CHGA have been observed in several cardiovascular disease states, including HTN, myocardial infarction, and dilated/hypertrophic cardiomyopathy. There was always a notion, ever since the discovery of PST, that the circulating levels of the peptide might be elevated in subjects with diabetes (2,11). To investigate this, we measured the plasma PST levels in subjects with T2D and age/sex/BMI-matched healthy control subjects. Indeed, PST levels showed a significant increase in subjects with T2D as compared with the control subjects (Supplementary Fig. 2). Since the plasma immunoassay can detect unprocessed CHGA as well as processed PST, the elevated PST levels in subjects with T2D could be a reflection of either a higher amount of CHGA or a greater degree of CHGA processing to PST. Further analysis of PST levels in larger populations may establish elevated PST levels as an intermediate phenotype for metabolic disorders.

Previous studies discovered several genetic variations within the PST region in various human populations (12,13). Among these variants, p.Gly297Ser occurs within the functionally important C-terminus of the peptide and is the most frequent one, with an estimated prevalence in ∼300 million individuals worldwide. The current study revealed association of the PST 297Ser allele with elevated levels of biochemical and physiological parameters that are well-known risk factors for cardiometabolic diseases. Importantly, plasma glucose and HbA_1c (Fig. 1A–D) displayed significant associations with the PST 297Ser allele.

The plasma catecholamine levels were also significantly elevated in Gly/Ser subjects (Fig. 3A), an observation in line with more effective catecholamine secretion triggered by PST-297S peptide as compared with the PST-WT peptide in neuroblastoma and adrenal pheochromocytoma cells (Fig. 3B). Catecholamines have an adverse impact on various aspects of cardiac pathology, including atherosclerosis, a key risk factor for CAD (32,33). Indeed, therapeutic strategies to combat myocardial infarction include β-blockade, further underscoring the contribution of catecholamine action on myocardial damage. Of note, catecholamines are known to enhance both glycogenolysis and gluconeogenesis, leading to higher blood glucose levels and inhibition of insulin-mediated glycogenesis (34). In addition, elevated plasma catecholamine levels are linked to HTN (35). Therefore, it is conceivable that the PST 297Ser allele may enhance the risk for T2D/HTN/CAD/MS (Table 1) concomitant with its adverse effects on biochemical/physiological phenotypes. In addition, upon genotyping subjects with chronic kidney disease (CKD) from another cohort consisting of unrelated Indian subjects, we found that the PST 297Ser allele frequency in the population having CKD with T2D was ∼1.7-fold higher than that in the population having CKD without T2D (data not shown). Thus, the p.Gly297Ser variant also seems to enhance the disease risk in diabetic kidney disease.

To test whether the association of PST p.Gly297Ser variant with cardiometabolic traits in Indian populations is replicable in other populations, we examined several available genome-wide association studies (GWAS) data sets from different world populations (Supplementary Table 6; data obtained from https://t2d.hugeamp.org/). In general, the variant seemed to increase the disease risk (Supplementary Table 6). Among the cardiometabolic conditions tested, interestingly, the variant showed modest associations with T2D in two of the studies—the DIAGRAM 1000G GWAS and DIAMANTE T2D GWAS. In the DIAGRAM 1000G GWAS, which primarily focused on samples of European descent, the variant showed an association with T2D upon adjusting for BMI (n = 56,443; P = 0.024; OR = 1.234). In the DIAMANTE T2D GWAS carried out in a European population as well, the variant showed an association trend (P = 0.099; OR = 1.059) with T2D in a large sample size of n = 231,420. These directionally concordant findings in other populations of the world support our observation that p.Gly297Ser serves as a risk variant for cardiometabolic disorders. Of note, the findings of the BioMe AMP T2D GWAS suggested that subjects with diabetes with the p.Gly297Ser variant are more susceptible to develop several cardiovascular complications, such as peripheral vascular disease (n = 1,350; P = 0.038; OR = 2.588).

It is important to note that these GWAS have been carried out in European populations in whom the MAFs for this SNP are much lower than those in South Asian populations (Supplementary Table 4). Similar GWAS in South Asian populations may yield better associations of this SNP with cardiovascular phenotypes. In addition, many of the gene chip arrays (for example, Illumina Human610-Quad Bead chip and Illumina Human660-Quad Bead chip) used in GWAS do not probe for this SNP. Therefore, it is not surprising that many GWAS on T2D, particularly even in Indian subjects, have not identified rs9658664 as a risk variant for T2D.

Considering the observed differences in MAF for the p.Gly297Ser variant as well as phenotypic differences of cohorts in publicly available databases, we tested the SNP in the context of evolutionary aspect. This variant is under negative selection pressure, as evidenced by F_ST values <0.05 for all of the data sets (Supplementary Table 4).

Since we found the PST 297Ser variant to be associated with metabolic disease states (Table 1), we asked whether PST-297S peptide exerts differential activities as compared with PST-WT peptide. While PST-297S peptide significantly blunted insulin-stimulated glucose uptake in differentiated myotubes and adipocytes, the effect of PST-WT peptide was modest/insignificant (Fig. 2A). This observation is consistent with the higher potency and efficacy of the PST-297S peptide in inhibiting 2-deoxyglucose uptake in rat primary adipocytes (11). The enhanced effect of PST-297S might be due to differential interactions with the PST receptor (discussed below). The PST-297S peptide also profoundly augmented the expression of rate-limiting enzymes (e.g., PCK1 and G6PC1) involved in the gluconeogenesis pathway in HepG2 cells (Fig. 2B and C), an observation in line with the stronger activation of PCK1/G6PC1 promoters by the variant peptide in the same cells (13). Since liver is the major site for gluconeogenesis in mammals, such a finding in the liver cell line, HepG2, is of physiological relevance and suggests that the carriers of the PST 297Ser allele might have enhanced gluconeogenesis (36). The observation of elevated RBS and PGBS levels in the carriers of the PST 297Ser allele supports this premise (Fig. 1A and B).

Thus, the PST-297S peptide appears to be a gain of function variant showing higher activities for several cellular and molecular processes that we tested. In general, these PST peptides showed effects within the concentration range of 10–100 nmol/L (Figs. 2 and 3). The effect of PST-WT on forearm glucose uptake in human subjects was observed at ∼200 nmol/L (11). In addition, the glucose uptake inhibitory effect of PST-WT on adipocytes had an IC₅₀ of ∼600 pmol/L (11). Could such concentrations of PST be attained in vivo? The plasma PST levels in healthy human subjects vary in the range of 25–125 pg/mL (i.e., ∼5–25 pmol/L) (11). However, the concentration of the PST precursor protein CHGA in secretory vesicles has been reported to be ∼4 mmol/L, and the concentration of CHGA in the extracellular space in the vicinity of the exocytotic pore just after exocytosis has been estimated as ∼0.4 mmol/L (37). Thus, although the steady-state plasma concentration of these peptides may be in the picomolar range, their concentrations in the vicinity of secretory cells after its exocytotic release are likely to be much higher (perhaps in the nanomolar range). Hence, the effects of the PST peptides on various cellular processes are likely to be physiologically relevant.

Most of the effects of PST peptides are likely to be receptor-mediated since these effects are very prompt (within seconds to minutes). Although previous studies have suggested that PST activates a receptor signaling pathway that involves seven transmembrane-spanning receptors coupled to Gq-PLCβ-Ca²⁺-PKC signaling and nitric oxide–dependent pathways, the identity of the PST receptor/cellular interacting partners has remained elusive (38–42).

In addition, the mechanism of antagonistic action of PST on insulin signaling also remains unclear. It may be hypothesized that the circulating plasma PST may encounter many receptors expressed on the membrane. Since the exact receptor for PST is still unknown and PST peptides consistently show insulin counterregulatory effects, we asked whether PST peptides might interact with IR. Our in silico docking analysis revealed that both PST peptides and insulin bind to the IR at the same binding pocket (Fig. 4C), and PST-297S shows lower binding free energy and dissociation constant (i.e., more affinity) values toward IR than the PST-WT peptide (Supplementary Table 5). In corroboration, IR-peptide binding assays in vitro (in both HEK-293 cells and β-TC-6 insulinoma cells) provided evidence for stronger binding affinity for PST-297S (Fig. 5). Thus, enhanced anti-insulin effects of the PST-297S peptide might be due its higher affinity and stronger interactions with IR.

Recent studies suggested that PST binds with GRP78 (23,29). GRP78 is an endoplasmic reticulum chaperone that serves as an endoplasmic reticulum stress indicator (43) and is involved in various cellular processes (44), such as facilitating the folding and assembly of proteins (45) and regulating the unfolded protein response (46). GRP78 plays key roles in cardiac development, cardiomyocyte survival, and function (47). Moreover, GRP78 exerts cardioprotective effects in the face of ischemia/reperfusion injury (48). GRP78 overproduction in pancreatic β-cells protects against high-fat diet–induced glucose intolerance and insulin resistance (49). Given such an influence of GRP78, we asked whether the PST peptides could be differentially interacting with GRP78, resulting in the differential glucose-related phenotypes observed. Indeed, the docking of the PST peptides onto GRP78 revealed that PST-297S exerted higher affinity interactions as compared with PST-WT (Supplementary Fig. 4). In corroboration, binding assays demonstrated that PST-297S peptide displaced labeled ^[125]I-Tyr PST from GRP78-expressing plasma membranes to a greater extent as compared with PST-WT (Fig. 6).

Taken together, PST-297S peptide showed higher binding affinity to both IR and GRP78 than PST-WT peptide. These differential interactions provide a molecular basis for the enhanced activities of PST-297S over PST-WT for various cellular processes and may mediate the underlying deleterious phenotype for increased disease risk.

Very few individuals with the PST homozygous variant (i.e., Ser/Ser) (n = 20) were identified in our study population (n = 4,318). Moreover, these subjects did not have phenotypic data for all of the parameters. Therefore, a direct comparison of their parameters with the WT (i.e., Gly/Gly) subjects was not meaningful. In view of this limitation, we combined the Ser/Ser homozygotes with the Gly/Ser heterozygotes for most of the analyses. It would be interesting to test genotype–phenotype associations between PST Gly/Gly and PST Ser/Ser subjects in larger study populations in future.

We discovered a common genetic variation (p.Gly297Ser) within the physiological dysglycemic peptide PST, a proteolytic fragment of the prohormone CHGA that is expressed in secretory vesicles across endocrine/neuroendocrine/neuronal cell types. The 297Ser allele was associated with elevated levels of several cardiometabolic traits, including plasma glucose, HbA_1c, DBP, and catecholamines in human subjects. In corroboration, human carriers of the 297Ser allele were associated with an increased risk for cardiometabolic diseases in human populations. Our multidisciplinary studies provided molecular basis for enhanced effects of the variant peptide (as compared with the WT peptide) on several relevant cellular/molecular functions that might lead to an increased disease risk in the carriers of 297Ser allele. A schematic of our findings is presented in Fig. 7. These results have implications for interindividual variations in glucose/catecholamine homeostasis and ultimately for pathogenesis of cardiometabolic disease states.

Acknowledgments. The authors thank Dr. Frederick M. Stanley, NYU Langone Medical Center, for providing the IR cDNA and Dr. Manikandan Narayanan, Department of Computer Science and Engineering, IIT Madras, for the help in genetic selection analysis. The authors also thank the high-performance computational facility at Indian Institute of Technology (IIT) Madras. V.R.C., M.Ki., and L.S. thank IIT Madras, Council of Scientific and Industrial Research, and the Department of Science and Technology, Government of India, respectively, for the research fellowships. P.K.R.A. and D.V. received research fellowships from the University Grants Commission, Government of India.

Funding. This work was supported by a grant from the Science and Engineering Research Board, Department of Science and Technology, Government of India (SR/SO/HS-084/2013A) to N.R.M., V.R., and V.M. This work was also supported by a grant from the Ministry of Human Resource Development, Government of India (SPARC/2018-2019/P652/SL) to N.R.M. and S.V.N.P. This work was also supported in part by a grant from the Department of Biotechnology, Government of India (BT/PR4820/MED/12/622/2013) to A.K.M.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. N.R.M. conceived and designed the study. P.K.R.A., M.Ki., S.D.M., V.R.C., R.G., D.V., S.R., L.S., B.S.S., D.R.I., S.M., and S.Sh. carried out the experimental and computational analyses. M.S.R., M.Kh., A.K.M., J.R.G., S.Se., A.S.M., V.M., V.R., S.V.N.P., and N.R.M. provided equipment, reagents, or clinical samples and reviewed and edited the manuscript. P.K.R.A., M.Ki., V.R.C., and N.R.M. wrote the manuscript. N.R.M. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.