Role of adropin in arterial stiffening associated with obesity and type 2 diabetes

Abstract

Adropin is a peptide largely secreted by the liver and known to regulate energy homeostasis; however, it also exerts cardiovascular effects. Herein, we tested the hypothesis that low circulating levels of adropin in obesity and type 2 diabetes (T2D) contribute to arterial stiffening. In support of this hypothesis, we report that obesity and T2D are associated with reduced levels of adropin (in liver and plasma) and increased arterial stiffness in mice and humans. Establishing causation, we show that mesenteric arteries from adropin knockout mice are also stiffer, relative to arteries from wild-type counterparts, thus recapitulating the stiffening phenotype observed in T2D db/db mice. Given the above, we performed a set of follow-up experiments, in which we found that 1) exposure of endothelial cells or isolated mesenteric arteries from db/db mice to adropin reduces filamentous actin (F-actin) stress fibers and stiffness, 2) adropin-induced reduction of F-actin and stiffness in endothelial cells and db/db mesenteric arteries is abrogated by inhibition of nitric oxide (NO) synthase, and 3) stimulation of smooth muscle cells or db/db mesenteric arteries with a NO mimetic reduces stiffness. Lastly, we demonstrated that in vivo treatment of db/db mice with adropin for 4 wk reduces stiffness in mesenteric arteries. Collectively, these findings indicate that adropin can regulate arterial stiffness, likely via endothelium-derived NO, and thus support the notion that “hypoadropinemia” should be considered as a putative target for the prevention and treatment of arterial stiffening in obesity and T2D.

NEW & NOTEWORTHY Arterial stiffening, a characteristic feature of obesity and type 2 diabetes (T2D), contributes to the development and progression of cardiovascular diseases. Herein we establish that adropin is decreased in obese and T2D models and furthermore provide evidence that reduced adropin may directly contribute to arterial stiffening. Collectively, findings from this work support the notion that “hypoadropinemia” should be considered as a putative target for the prevention and treatment of arterial stiffening in obesity and T2D.

INTRODUCTION

Stiffening of the vasculature, a feature of obesity and type 2 diabetes (T2D), is a causal factor and independent prognosticator of cardiovascular morbidity and mortality (1–3). Despite the indisputable recognition that arterial stiffening contributes to the pathogenesis of cardiovascular disease, the mechanisms underlying arterial stiffening remain largely unknown.

Adropin is a highly conserved 76-amino acid peptide encoded by the energy homeostasis-associated (i.e., Enho) gene and is heavily expressed in the liver, where it was first identified ∼15 years ago (4). Adropin plays a role in maintaining energy balance and regulating lipid and glucose metabolism (4–6). Beyond its well-established metabolic effects, increasing evidence indicates that adropin also exerts cardiovascular effects. For example, adropin has been shown to regulate cardiac energy substrate flexibility (7–10), promote angiogenesis (11), decrease endothelial inflammation (11, 12), activate endothelial nitric oxide (NO) synthase (eNOS) (11), and cause endothelium-dependent vasodilation (13). All this suggests that adropin signaling may be beneficial to the cardiovascular system.

Data are also available indicating that circulating levels of adropin are depressed in various chronic diseases associated with arterial stiffening and cardiovascular disease, including obesity and T2D (14–27) (reviewed in Refs. 28–30). Herein, we first confirmed that severity of obesity and T2D inversely associate with hepatic mRNA expression of adropin in a cohort of male and female patients that underwent liver tissue sampling during bariatric surgery. In a separate cohort, we also corroborated that low circulating levels of adropin in individuals with T2D are associated with increased arterial stiffness. The phenomenon that circulating levels of adropin are reduced in disease states characterized by arterial stiffening and are inversely correlated with indices of arterial stiffness prompted the hypothesis that diminished levels of adropin may contribute to the development of arterial stiffening in obesity and T2D. To begin to address this hypothesis, we tested whether adropin deficiency in mice causes arterial stiffening, thus phenocopying obesity and T2D. Furthermore, we posited that the converse is also true; that is, adropin exposure destiffens the vasculature in obese T2D mice. Mechanistically, we examined the role of NO signaling in mediating the vascular destiffening effects of adropin.

METHODS

Ethics and Approvals

All human study procedures conformed to the Declaration of Helsinki and were approved by the University of Missouri Institutional Review Board (IRB, No. 2008181, No. 2012106, No. 2028142, and No. 2008258). Written informed consent was obtained from all subjects before any procedures. All animal study procedures received prior approval by the University of Missouri Institutional Animal Care and Use Committee and were conducted in accordance with the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals.

Cohort of Patients with Bariatric Surgery

Forty-five female and male patients with clinical obesity [36 female (F)/9 male (M), 46.3 ± 2.0 yr of age, body mass index (BMI) = 48.3 ± 1.1 kg/m², HbA1c = 6.5 ± 0.2%] who underwent elective bariatric surgery with liver tissue sampling at the University of Missouri Hospital, as part of a larger study (31), and from whom we had liver mRNA available were included in this retrospective analysis. Specifically, subjects were segregated into fixed clusters according to their BMI and HbA1c values (≤40 or >40 kg/m² and <6.5 or ≥6.5%, respectively). A BMI above 40 kg/m² is considered severe obesity and an HbA1c of 6.5% or higher is indicative of diabetes. Mean values of hepatic adropin mRNA expression are reported on the z-axis of a three-dimensional histogram.

Cohort of Healthy Individuals and Individuals with T2D

Forty-two individuals with a clinical diagnosis of T2D and confirmed HbA1c ≥ 6.5% (21 F/21 M, 56.1 ± 1.6 yr of age, BMI = 35.5 ± 0.9 kg/m², HbA1c = 8.1 ± 0.2%), along with 33 aged-matched healthy controls (20 F/13 M, 51.8 ± 1.9 yr of age, BMI = 23.3 ± 0.4 kg/m², HbA1c = 5.2 ± 0.1%), and from whom we either had fasting plasma available, data on carotid-to-femoral pulse wave velocity (PWV), or both, were included in this retrospective analysis. Subjects were free of overt cardiovascular, hepatic, and autoimmune diseases; cancers; tobacco use; excessive alcohol consumption (>14 drinks/wk for men, and >7 drinks/wk for women); pregnancy; and uncontrolled hypertension (≥180 mmHg systolic, or ≥110 mmHg diastolic). It should be noted that a small subset of these participants was included in a previous publication examining an unrelated question, i.e., the effects of chronic heating on metabolic and vascular outcomes (32). Participants were admitted to the laboratory after an overnight fast for a blood draw, anthropometric measures, and assessment of arterial stiffness. Participants rested supine for 15 min before the assessment of arterial stiffness via carotid-to-femoral PWV using the cuff-based SphygmoCor XCEL system (AtCor Medical, Itasca, IL), according to current recommendations (3, 33) and as previously described (34). The SphygmoCor XCEL device enables the simultaneous acquisition of carotid and femoral pulse waves, via tonometer and leg cuff, respectively. Transit time between carotid and femoral pressure waves was calculated with the foot-to-foot method. Wave foots were identified using intersecting tangent algorithms. PWV was calculated as distance traveled by the pulse wave divided by pulse transit time and reported as meters per second.

Mouse Models

Leptin receptor deficient homozygous (db/db) and heterozygous (db/+) male mice were purchased from The Jackson Laboratory (Bar Harbor, MA). Adropin knockout male mice and wild-type littermates were obtained from Saint Louis University. This mouse model has been previously described in detail by Kumar et al. (5). All mice were group housed (n = 2–5) with ad libitum access to chow [3.35 kcal/g food, Laboratory Rodent Diet 5001, Laboratory Diet (35)] and water. Mice were housed in an environmentally controlled animal facility maintained at 23°C on a 12-h:12-h light/dark cycle from 0700 to 1900 h.

In Vivo Adropin Administration

Db/db mice (9 wk old) were assigned to two groups, using block randomization to balance initial body weight, and infused with either vehicle (phosphate-buffered saline, PBS, n = 7) or adropin (63 mg/kg/h, n = 12, NovoPro, Cat. No. 314322, Shanghai, China) for 4 wk using an osmotic minipump (Alzet Model 2001; Durect, Cupertino, CA). Pumps were implanted subcutaneously into the dorsum and replaced after 2 wk under isoflurane anesthesia, as previously described (36).

Ex Vivo Treatments in Isolated Mesenteric Arteries

Small mesenteric arteries were selected as the model of choice for this project because 1) they are considered resistance arteries and thus contribute to the regulation of blood pressure, 2) multiple arteries of the same size can be isolated from a single animal and exposed to different experimental conditions, thus allowing for paired analyses, and 3) it is a nonsex-specific vascular bed extensively used across animal models (37). To determine the impact of adropin on stiffness in diabetes, mesenteric arteries were excised from db/db mice, cleaned of perivascular adipose and connective tissue, and flushed of blood. Two mesenteric arteries per animal were then incubated with VascuLife EnGS (Lifeline Cell Technology, Frederick, MD) culture media (2% FBS) under standard culture conditions (37°C, 5% CO₂) and treated with vehicle (ddH₂O) versus adropin (10 ng/mL, Phoenix Pharmaceuticals, Burlingame, CA) for 24 h. To assess the role of NO in mediating adropin effects, the experiment above was repeated with cotreatment of N^G-nitro-l-arginine methyl ester (l-NAME, 300 μM) in mesenteric arteries isolated from separate db/db mice. l-NAME was added 30 min before vehicle or adropin. To examine the direct effect NO signaling on arterial stiffness, mesenteric arteries from db/db mice were cannulated in a pressure myography chamber (Living Systems Instruments, St. Albans City, VT). After vessels were corroborated as having no leaks, arteries were pressurized at 70 mmHg and incubated in physiological salt solution (PSS), containing (in mM) 145.0 NaCl, 4.7 KCl, 2.0 CaCl₂, 1.2 MgSO₄, 1.0 NaH₂PO₄, 5.0 dextrose, 3.0 3-(N-morpholino)propanesulfonic acid (MOPS) buffer, 2.0 pyruvate, and 0.02 EDTA at pH of 7.4 at 37°C containing either vehicle or sodium nitroprusside (SNP, 10 mM, Sigma-Aldrich, BCCB2459) for 4 h. After incubation, treated arteries underwent mechanical testing under passive conditions to assess arterial stiffness, as described below. As a point of clarification, isolated arteries treated with adropin for 24 h were kept in the incubator under standard culture conditions to ensure viability. Given the shorter half-life of SNP (38), arteries were only treated with SNP for 4 h and kept viable in the pressure myography chamber set at the bench. Maintaining the vessels pressurized provides greater cytoskeletal dynamics.

Ex Vivo Assessment of Arterial Stiffness in Isolated Mesenteric Arteries

Mesenteric arteries were assessed for stiffness under passive conditions. Briefly, cleaned mesenteric arteries were cannulated and pressurized to 40 mmHg for 40 min in PSS at 37°C to acclimate before mechanical testing. After the 40-min acclimation period, the bath was replaced with Ca²⁺-free PSS and arteries were exposed to increasing intraluminal pressure from 5 to 120 mmHg with a stepwise increase in intraluminal pressure maintained for 2 min to achieve diameter and wall thickness plateaus. Diameter and wall thickness curves were used to calculate the circumferential strain and stress, and from them the incremental modulus of elasticity (E_inc), as previously reported (39). After mechanical testing, arteries were fixed with 4% paraformaldehyde (PFA) at 70 mmHg for 30 min for further wall composition experiments.

Confocal Microscopy Imaging of Isolated Mesenteric Arteries

The amount of nuclear material and filamentous-actin (F-actin) were measured in fixed db/db mesenteric resistance arteries after mechanical testing. Briefly, fixed mesenteric arteries were cannulated, permeabilized, and incubated in Syto 63 (250 nM, Thermo Fisher Scientific, Cat. No. S11345) and Alexa Fluor 546 phalloidin (200 nM, Thermo Fisher Scientific, Cat. No. A22283) for 1 h at room temperature in PBS and subsequently washed with PBS. Images of nuclei and F-actin were obtained using a Leica SP5 confocal-multiphoton microscope with a ×63 water immersion and 1.2 numerical aperture objective (Leica Microsystems, Inc., Morrisville, NC). Imaris software (Bitplane, Inc., Concord, MA) was used to render three-dimensional reconstructions. An unbiased Matlab script was used to quantify the volume of the molecules of interest (40). Volumetric data were normalized to the volume of nuclei.

Whole Blood and Plasma Analysis

Assessment of HbA1c in humans was performed at the University of Missouri Diabetes Diagnostic Laboratory or at Quest Diagnostics, Columbia MO. Mouse HbA1c and plasma concentrations of glucose, insulin, total cholesterol, and nonesterified fatty acids (NEFAs) were assessed at the University of California Davis Mouse Metabolic Phenotyping Center. Human and mouse plasma adropin concentrations were determined via a commercially available enzyme immunoassay per the manufacturer’s instructions (Phoenix Pharmaceuticals, Burlingame, CA).

Cell Culture Experiments

Human aortic endothelial cells (Lonza: Cat. No. CC-2535, Morristown, NJ) were obtained and cultured under standard culture conditions in VascuLife EnGS culture media (2% FBS) and treated with vehicle (ddH₂O) versus adropin (10 ng/mL) for 24 h in the presence or absence of l-NAME (300 μM). l-NAME was added 30 min before vehicle or adropin. Endothelial cells used for immunohistochemical imaging were seeded on 15-well-ibidi plates (Ibidi, Cat No. 81506, Fitchburg, WI). Cells used for Western blot and atomic force microscopy (AFM) analyses were seeded on 60-mm cell culture dishes. Cells used for nitrite analysis were seeded on 6-well cell culture dishes.

Human coronary artery smooth muscle cells (Thermo Fisher Scientific, Cat. No. C0175C, Waltham, MA) were obtained and cultured under standard culture conditions in Medium 231 (Thermo Fisher Scientific, Cat. No. M231500), with 5% FBS; with smooth muscle cell supplements (Thermo Fisher Scientific, Cat. No. S-007–25) and treated with vehicle (ddH₂O) versus SNP (10 mM) for 1 h. To further examine the effects of NO signaling on actin polymerization, the experiment above was repeated with cotreatment of LIM kinase inhibitor 3 (10 mM, LIMKi, Calbiochem, Cat. No. 435930, San Diego, CA) or jasplakinolide (100 nM, Thermo Fisher Scientific). LIMKi and jasplakinolide were used to inhibit and force actin polymerization, respectively. LIMKi and jasplakinolide were added 30 min before vehicle or SNP. Cells used for immunohistochemical imaging were seeded on 15-well-ibidi plates. Cells used for AFM were seeded on either 40-mm or 60-mm cell culture dishes.

Measurement of Nitrite in Cell Culture Supernatant

Nitrite, a byproduct of NO, was assessed using the gold-standard method of ozone-based reductive chemiluminescence (CLD88, Eco Physics, Ann Arbor, MI) according to the manufacturer recommendations and as previously described (41–43). Briefly, endothelial cell supernatant samples (50 µL) were injected in duplicate into a purge vessel containing an acidified iodide solution serving as a nitrite reducing agent (50 mg potassium iodide dissolved in 1-mL filtered ddH₂0 with 4-mL glacial acetic acid), which was then purged with pure nitrogen in line with the CLD88 gas-phase NO analyzer. The chemiluminescence signal was captured using eDAQ Chart v5.5.27 and $\text{N}{\text{O}}_2^{-}$ was quantified using the Flow Injection Analysis (FIA) software extension (ADInstruments, Australia). The FIA software calculated area under the curve for each sample peak, which was then converted to a concentration using a calibrated standard curve of known sodium nitrite (NaNO₂) and normalized to protein content using the bicinchoninic acid protein assay.

Vascular Cell Cortical Stiffness

Cultured endothelial and smooth muscle cells were assessed for cortical stiffness via AFM. Briefly, individual cells were exposed to nanoindentation at room temperature using a silicon nitride cantilever (Bruker, Cat. No. MLCT, Billerica, MA) on either an MFP-3D Atomic Force Microscope (Asylum Research, Inc., Goleta, CA) mounted on an Olympus IX81 microscope (Olympus Ins., Tokyo, Japan) or a NanoWizard IV Atomic Force Microscope (Bruker, JPK) mounted on a Leica DMi8 automated microscope (Leica Microsystems, Inc., Morrisville, NC). At least 40 force curves were obtained from each cell at a location of approximated one-third the distance from the edge of the cell to the nucleus, with seven cells assessed per dish. Elastic moduli were calculated from the force curves using a custom-made Phython script that, in an unbiased manner, identifies and removes curves with excessive noise and the remaining curves are fitted to the Hertz model, as previously described (40).

Immunohistochemical Imaging of Endothelial and Smooth Muscle Cells

Immunohistochemical images of endothelial and smooth muscle cells were obtained from cells plated on 15-well-ibidi plates. After incubation with the respective treatments, cells were fixed in 4% PFA at room temperature for 30 min and permeabilized with 0.5% Triton X-100 for 15 min. Cells were incubated for 1 h in 1.5 nM 4′,6-diamidino-2-phenylindole (DAPI) and 200 nM Alexa Fluor 563 phalloidin (Thermo Fisher Scientific, Cat. No. A12380) to stain for nuclei and F-actin, respectively. Images were acquired with either a Leica SPE confocal microscope (endothelial cells, Leica Microsystems, Inc.) or Leica THUNDER Imager 3 D Cell Culture microscope (smooth muscle cells, Leica Microsystems, Inc.). Fluorescence intensity was quantified using Imaris software (Bitplane, Inc., Concord, MA) and normalized by the total number of cells. Values are expressed as fold difference.

Adropin mRNA Expression via RT-PCR

Samples from mice and humans were processed to assess expression of Enho, the gene encoding adropin. Tissues were homogenized in TRIzol reagent (TissueLyser LT, Qiagen, Valencia, CA) and total RNA was isolated using Qiagen’s RNeasy lipid tissue protocol and assayed using a Nanodrop spectrophotometer (Thermo Fisher Scientific) to assess purity and concentration. First-strand cDNA was synthesized from total RNA by using the High-Capacity cDNA Reverse Transcription kit (Agilent Technologies, Salt Lake City, UT). mRNA expression was analyzed by RT-PCR using either the CFX96 real time PCR system (C1000 thermocycler, Bio-Rad Laboratories, Hercules, CA) or Quant Studio 5 applied biosystems (Thermo Fisher Scientific). The Enho sequences were as follows: human: forward: 5′-ATTGAGGCAGCTCCACTGTC-3′, reverse: 5′-CTGGAGTTGGGACTGGATTC-3′; mouse: forward: 5′-CTCAACTCAGGCCCAGGA-3′, reverse: 5’GCTGTCCTGTCCACACAC-3′. The glyceraldehyde-3-phosphate dehydrogenase (GAPDH) sequences were as follows: human: forward: 5′-CACCAGGGCTGCTTTTAACTCTGGTA-3′, reverse: 5′-CCTTGACGGTGCCATGGAATTGC-3′; mouse: forward: 5′-GGAGAAACCTGCCAAGTATGA-3′, reverse: 5′- TCCTCAGTGTAGCCCAAGA-3′. All data were normalized to the corresponding GAPDH mRNA and analyzed using the 2⁻Δ^ΔCt method, where C_t is threshold cycle.

Western Blot Analysis

Specific protein content was assessed in endothelial cell lysates prepared in RIPA buffer (Invitrogen, Cat. No. 1861278, Waltham, MA) and phosphatase inhibitors (Invitrogen, Cat. No. 1862495). Proteins within samples were separated in Criterion Tris-Glycine eXtended-PAGE precast gels (Bio-Rad) and transferred onto polyvinylidene difluoride membranes. Specific proteins were probed using the following antibodies: anti-phospho-eNOS (Ser1177; 1:500, Cat. No. 9570, Cell Signaling Technology, Danvers, MA) and anti-eNOS (1:500, Cat. No. 32027, Cell Signaling). The individual protein band intensities were quantified via densitometry via Bio-Rad ChemiDoc XRS+ System (Bio-Rad). Phosphorylated eNOS was normalized to total eNOS. Values are expressed as fold difference.

Proteomics

Proteomics analysis of adropin-treated endothelial cells was performed at Charles W. Gehrke Proteomics Center, a research core facility at the University of Missouri. Briefly, following a 24-h incubation with either vehicle or adropin (10 ng/mL), endothelial samples were homogenized in 100-µL RIPA buffer and sonicated. Samples were then heated to 65°C for 20 min and finally centrifuged for 10 min at 16,000 g. The supernatant was transferred to a new tube and protein was precipitated overnight at −20°C with four volumes of cold acetone added to each sample. Each protein sample was centrifuged and washed with 80% acetone twice, and protein pellets were then resuspended with 6 M urea, 2 M thiourea, and 100 mM ammonium bicarbonate. Protein was quantified using Pierce 660-nm protein assay method following manufacturer’s instructions. Details on spectral library built-up, DIA data, statistical analysis, and full list of proteins that showed to be significantly changed with adropin are described in the Supplemental material (all Supplemental material is available at http://doi.org/10.6084/m9.figshare.21030286).

Statistical Analysis

Statistical analyses were performed using GraphPad Prism version 9 (GraphPad Prism Software, La Jolla, CA). The Shapiro–Wilk test was used to determine if data were normally distributed. The ROUT test was used to determine if data had outliers based on a false discovery rate, with Q set to 5%. The Mann–Whitney U nonparametric test was used to compare data not normally distributed. Two-tailed, unpaired Student’s t test was used to compare independent samples, whereas the two-tailed, paired Student’s t test was used to compare paired samples. The E_inc slopes were calculated for each artery and was compared with its respective control via Student’s unpaired and paired t test, according to experimental design. The E_inc data are represented as a simple linear regression to represent the group mean slopes. A two-way repeated-measure analysis of variance (ANOVA) was performed to assess group by time effects of glucose tolerance. In addition, Pearson’s correlation or partial correlation analyses were performed to examine the relationship between adropin, PWV, and other cardiovascular risk factors when appropriate (SPSS Statistics software version 26, IBM Corporation, Armonk, NY). Data are represented as means ± SE (44). The results were considered significant when P < 0.05. Whenever possible, investigators were blinded to the treatment conditions. However, it should be acknowledged that for several of the protocols, the investigator administering the treatments was the same one performing the measurements and thus blinding could not be ensured. Data analysis was performed by a single investigator and repeated by a second independent investigator for confirmation.

RESULTS

Obesity and T2D Are Associated with Reduced Adropin Levels and Increased Arterial Stiffness in Humans

As displayed in Fig. 1A, levels of BMI and HbA1c inversely associate with hepatic adropin mRNA expression in a cohort of male and female patients that underwent liver tissue sampling during bariatric surgery. That is, those individuals with a BMI > 40 kg/m² and HbA1c ≥ 6.5% had lower adropin mRNA expression in the liver relative to those with a BMI ≤ 40 kg/m² and HbA1c < 6.5% (P < 0.05). In Fig. 1B, in a separate cohort, we show that individuals with T2D exhibited lower plasma concentrations of adropin, relative to healthy subjects (P < 0.05) and that this was associated with increased arterial stiffness as assessed by carotid-to-femoral PWV (P < 0.05). Sex did not influence the differences between healthy subjects and T2D in plasma adropin or PWV (i.e., sex by group interactions were not significant; P > 0.05).

Figure 1.

Decreased adropin is associated with increased arterial stiffness. A: human hepatic adropin mRNA expression is inversely associated with glycosylated hemoglobin (HbA1c) and body mass index (BMI) in subjects that underwent a liver biopsy during bariatric surgery (n = 45, F = 36, M = 9, n = 3–24/cluster). B: plasma adropin concentrations of healthy subjects (n = 33, F = 20, M = 13) and subjects with type 2 diabetes (T2D; n = 42, F = 21, M = 21). Carotid-to-femoral pulse wave velocity in healthy subjects (n = 33, F = 20, M = 13) and subjects with T2D (n = 36, F = 19, M = 17). Pearson correlation between plasma adropin concentration and pulse wave velocity in healthy subjects (n = 33, F = 20, M = 13) and subjects with T2D (n = 36, F = 19, M = 17). C: hepatic adropin mRNA expression in db/db and db/+ male mice (n = 9/genotype). Plasma adropin concentrations in db/db and db/+ male mice (n = 9–10/genotype). Incremental modulus of elasticity (E_inc) of mesenteric arteries of db/db and db/+ male mice (n = 9–10/genotype). D: adropin mRNA expression in various tissues harvested from adropin knockout and wild-type littermate male mice (n = 5–14/genotype). Final body weight of adropin knockout and wild-type littermate male mice (n = 9–14/genotype). Blood glucose concentration during a glucose tolerance test in adropin knockout and wild-type littermate male mice (n = 10–14/genotype). E_inc of mesenteric arteries from adropin knockout and wild-type littermate male mice (n = 9–14/genotype). Student’s unpaired t tests were performed in A–D. E_inc data (C and D) are represented using simple linear regression. Mann-Whitney test was performed for the comparison of adropin mRNA expression in the brain (D). A two-way, repeated-measures ANOVA was performed to assess glucose tolerance (D). *P < 0.05 compared with control.

There was a significant correlation between adropin and PWV (R = −0.26, P = 0.033, Fig. 1B). Adjusting for BMI or HbA1c eliminated the significance of the correlation between adropin and PWV (BMI-adjusted: R = −0.131, P = 0.285; HbA1c-adjusted: R = −0.139, P = 0.315); whereas adjusting for age or MAP did not (age-adjusted: R = −0.26, P = 0.032; MAP-adjusted: R = −0.297, P = 0.016). There was no significant correlation between adropin and MAP (R = −0.06, P = 0.583). Age and PWV (R = 0.458, P < 0.01) as well as BMI and PWV (R = 0.675, P < 0.01) were significantly correlated.

Mesenteric Arteries from Adropin Knockout Mice Are Stiffer, Thus Recapitulating the Stiffening Phenotype Observed in db/db Mice

As shown in Fig. 1C, db/db mice exhibited lower mRNA expression of adropin in the liver and lower adropin concentrations in plasma relative to db/+ littermates (P < 0.05). db/db mice also displayed increased stiffness in isolated mesenteric arteries as assessed by pressure myography (P < 0.05). In Fig. 1D, as confirmation of the adropin knockout mouse model, we show that knockout mice demonstrated reduced levels of adropin mRNA across tissues, relative to wild-type littermates (P < 0.05). Plasma insulin concentrations were reduced in adropin knockout mice (P < 0.05), but no significant differences in body weight, glucose tolerance, or other metabolic parameters were noted between knockout and wild-type mice (P > 0.05, Table 1, Fig. 1D). Despite the overall similar metabolic phenotype between genotypes, mesenteric arteries from adropin knockout mice were stiffer compared with arteries from wild-type counterparts (P < 0.05).

Table 1.

Circulating metabolic parameters of wild-type and adropin-knockout littermates

	Wild Type	Adropin Knockout
n	9	13
Glucose, mg/dL	246.5 ± 18.4	243.0 ± 8.7
Insulin, pg/mL	1,117.2 ± 264.7	659.7 ± 121.7*
Free glycerol, mg/dL	25.1 ± 1.3	22.7 ± 1.5
Triglycerides, mg/dL	63.5 ± 7.1	64.9 ± 6.7
Free-fatty acids, mEq/L	0.4 ± 0.1	0.4 ± 0.0
Total cholesterol, mg/dL	90.2 ± 5.0	84.8 ± 2.4
HbA1c, %	4.8 ± 0.1	4.8 ± 0.3

Values are means ± SE; n, number of animals. *P < 0.05 compared with wild type.

Exposure of Endothelial Cells or Isolated Mesenteric Arteries from db/db Male Mice to Adropin Reduces F-Actin Stress Fibers and Stiffness: Role of NO

In Fig. 2A, we report that adropin stimulation increased phosphorylation of eNOS in endothelial cells as well as nitrite concentration, a by-product of NO, in the cell culture supernatant (P < 0.05). Figure 2B shows that prolonged exposure of endothelial cells to adropin reduced F-actin stress fibers and cellular stiffness, as assessed by AFM (P < 0.05). Notably, these effects of adropin were abrogated when cells were cotreated with l-NAME, a NOS inhibitor (P > 0.05). Similarly, as illustrated in Fig. 2C, prolonged exposure of isolated mesenteric arteries from db/db mice to adropin reduced F-actin stress fibers and stiffness (P < 0.05); effects that were also abrogated by l-NAME (P > 0.05).

Figure 2.

Adropin reduces F-actin stress fibers and stiffness in endothelial cells (EC) and isolated mesenteric arteries of db/db male mice: role of NO. A: phosphorylation of eNOS Ser1177 relative to total eNOS in human aortic EC treated with vehicle vs. adropin (10 ng/mL) for 30 min (n = 8–9/condition); representative Western blot images are also included. Only the top band was responsive to adropin and selected for quantification and analysis. Nitrite concentrations in the supernatant of human aortic EC treated with vehicle vs. adropin (10 ng/mL) for 24 h (n = 11/condition). B: volume of F-actin stress fibers in human aortic EC treated with vehicle vs. adropin (10 ng/mL) for 24 h (n = 20–24/condition); representative confocal microscope images of F-actin (yellow) and nuclei (blue) are also included (scale bar = 30 mm). Cortical stiffness of human aortic EC treated with vehicle vs. adropin (10 ng/mL) for 24 h (n = 18/condition). Below panels are repeat experiments in the presence of N^G-nitro-l-arginine methyl ester (l-NAME; added 30 min before adropin; n = 17–18/condition). C: volume of F-actin content and incremental modulus of elasticity (E_inc) of isolated mesenteric arteries from db/db male mice treated with vehicle vs. adropin (10 ng/mL) for 24 h (n = 8/condition); representative images of confocal microscope images of F-actin (yellow) and nuclei (blue) are also included (scale = 30 mm). Below panels are repeat experiments in the presence of l-NAME (added 30 min before adropin; n = 7–8/group). Student’s unpaired (A and B) and paired t tests (C) were performed, according to experimental design. E_inc data (C) are represented using simple linear regression. *P < 0.05 compared with control.

Stimulation of Smooth Muscle Cells or Isolated Mesenteric Arteries from db/db Male Mice with SNP, an NO Mimetic, Reduces F-Actin Stress Fibers and Stiffness

As shown in Fig. 3A, stimulation of smooth muscle cells with SNP decreased F-actin stress fibers (P < 0.05). This effect of SNP was abrogated by cotreatment of cells with LIMK inhibitor or jasplakinolide (P > 0.05). Congruently, as displayed in Fig. 3B, stimulation of smooth muscle cells or db/db mesenteric arteries with SNP reduced stiffness as assessed by AFM and pressure myography, respectively (P < 0.05).

Figure 3.

Stimulation of vascular smooth muscle cells (VSMCs) with sodium nitroprusside (SNP), an NO mimetic, reduces F-actin stress fibers and stiffness. A: volume of F-actin stress fibers in human coronary artery VSMCs treated with vehicle vs. SNP (10 µM) for 1 h (n = 28–32/condition); representative fluorescent images of F-actin (yellow) and nuclei (blue) are also included (scale bar = 30 mm). Below panels are repeat experiments in the presence of LIMKi or jasplakinolide (added 30 min before SNP; n = 14–16/condition). B: cortical stiffness of human coronary artery VSMCs treated with vehicle vs. SNP (10 µM) for 1 h (n = 6–7/condition). Incremental modulus of elasticity (E_inc) of mesenteric arteries from db/db male mice treated with vehicle vs. SNP (10 µM) for 4 h (n = 14–16/condition). Student’s unpaired (A and B) and paired (B) t tests were performed according to experimental design. E_inc data are represented using simple linear regression. *P < 0.05 compared with control.

In Vivo Treatment of db/db Mice with Adropin Reduced Stiffness in Mesenteric Arteries

As illustrated in Fig. 4, treatment of db/db mice with adropin for 4 wk via osmotic minipumps increased plasma concentrations of adropin, compared with mice treated with vehicle. Adropin treatment did not cause significant changes in body weight or other circulating indices of metabolic function (Table 2; P > 0.05). Despite the lack of adropin effect on metabolic function, mesenteric arteries from adropin-treated mice were less stiff compared with arteries from vehicle-treated mice (P < 0.05).

Figure 4.

In vivo treatment of db/db male mice with adropin for 4 wk reduces mesenteric arterial stiffness. Plasma adropin concentration in db/db male mice implanted with osmotic minipumps containing either vehicle or adropin (63 mg/kg/h, n = 5–11/group). Final body weight of db/db male mice after 4 wk of vehicle or adropin administration (n = 7–12/group). Incremental modulus of elasticity (E_inc) of mesenteric arteries after 4 wk of vehicle or adropin administration (n = 6–12/group). Student’s unpaired and paired t tests were performed, as appropriate. E_inc data are represented using simple linear regression. *P < 0.05 compared with control.

Table 2.

Circulating metabolic parameters of db/db mice after a 4-wk vehicle vs. adropin administration

	Vehicle	Adropin
n	7	12
Glucose, mg/dL	798.1 ± 48.3	843.2 ± 32.0
Insulin, pg/mL	8,736.2 ± 1,277.8	8,038.9 ± 735.8
Free glycerol, mg/dL	47.6 ± 6.1	38.8 ± 2.4
Triglycerides, mg/dL	164.4 ± 34.7	151.3 ± 15.6
Free-fatty acids, mEq/L	0.4 ± 0.0	0.5 ± 0.0
Total cholesterol, mg/dL	129.7 ± 6.9	124.1 ± 5.9
HbA1c, %	7.6 ± 0.4	7.8 ± 0.2

Values are means ± SE; n, number of animals.

Molecular Signature of Endothelial Cells Treated with Adropin

The proteomic analysis identified 421 differentially expressed proteins, of which 404 increased whereas 17 decreased, in endothelial cells treated with adropin for 24 h. Unbiased ingenuity pathway analysis revealed that organization of the cytoskeleton was one of the top activated networks associated with disease and biological function (Fig. 5, Supplemental Tables S1, S2, and S3, P = 7.92E-5; Z-score, 2.708). The top upstream translational regulator, also linked with the cytoskeletal organization, was switch/sucrose nonfermentable-related (SWI/SNF), matrix-associated, actin-dependent regulator of chromatin subfamily A, member 4 (SMARCA4, Fig. 5, P = 8.18E-3; Z-score, 3.341). In addition, the unbiased ingenuity pathway analysis revealed an activation of the cell survival network in concert with a downregulation of apoptotic network (Supplemental Fig. S1 and Supplemental Tables S4, S5, S6, and S7, P = 0.00197 and P = 7.28E-4; and Z-scores, 4.456 and −3.551, respectively).

Figure 5.

Proteomic analysis of endothelial cells treated with vehicle vs. adropin. Upregulation of pathways linked to organization of the cytoskeleton in human aortic endothelial cells treated with vehicle vs. adropin (10 ng/mL) for 24 h (n = 10/condition). Data expressed as average log2 ratio relative to vehicle. Student’s unpaired t test was performed for all protein comparisons (see Supplemental material for further description of the statistical analysis and individual proteins within figure).

DISCUSSION

Increased arterial stiffness contributes to the pathogenesis of cardiovascular disease and represents an independent predictor of adverse cardiovascular events, morbidity, and mortality (1–3, 45–48). A better understanding of factors regulating arterial stiffness in obesity and T2D is critical for the identification of novel therapeutic targets to reduce cardiovascular disease burden. In this work, we further establish that adropin is decreased in obesity and T2D in humans and preclinical animal models, as well as provide evidence that reduced adropin may directly contribute to arterial stiffening.

Specifically, in liver samples from male and female patients that underwent bariatric surgery, we show that levels of BMI and HbA1c inversely associate with hepatic mRNA expression of adropin, corroborating previous findings (15). In a separate cohort, we show that individuals with T2D also exhibit lower plasma concentrations of adropin, relative to healthy subjects, and that this is associated with increased arterial stiffness. Subsequently, we provide several lines of evidence that collectively support the role of adropin in regulating arterial stiffness, as elaborated below.

First, we report that increased stiffness in mesenteric arteries of db/db mice, a genetic mouse model of obesity and T2D, is accompanied with reduced mRNA expression of adropin in the liver and lower adropin concentrations in plasma. Notably, we show that mesenteric arteries from adropin knockout mice are also stiffer, relative to arteries from wild-type counterparts, thus recapitulating the stiffening phenotype we and others (49) observed in db/db mice. Of note, the stiffening effect caused by adropin ablation appears to be independent of obesity and metabolic dysfunction because metabolic derangements in this mouse model are primarily or only manifested when animals are subjected to another insult (e.g., an obesogenic diet) (5, 50).

We next provided evidence that overnight exposure of isolated mesenteric arteries from db/db mice to adropin has a destiffening effect. To examine the role of NO in mediating this beneficial effect of adropin, we first documented that adropin does indeed activate eNOS and increase NO production in endothelial cells, a finding that is consistent with prior work by others (11). Importantly, we show that the destiffening effect of adropin in db/db mesenteric arteries is abolished when arteries are cotreated with the NOS inhibitor l-NAME, suggesting that the stiffness-reducing effect of adropin is at least in part mediated by NO.

Although collagen deposition and cross-linking are important determinants of arterial stiffness, another important driver of stiffness at the cellular level is cytoskeletal actin polymerization (40, 45, 46, 51). Here we show that the destiffening effect of adropin in db/db mesenteric arteries is accompanied by reduced content of F-actin stress fibers and that the actin depolymerization effects of adropin are also abolished by NOS inhibition. Similarly, endothelial cells treated with adropin become less stiff, as assessed by AFM, and this destiffening effect is paralleled with reduced F-actin stress fibers, as well as abrogated by NOS inhibition. These findings indicate that the actin depolymerization and stiffness-reducing effects of adropin, in both isolated arteries and endothelial cells, are NO dependent. Notably, we are not asserting that reduced endothelial rigidity with adropin physically contributes to the lowering of whole artery stiffness. Rather, based on previous evidence that reduced endothelial cell stiffness promotes eNOS activation (52), a more conceivable scenario is that adropin-induced destiffening of endothelial cells potentiates NO production and that, subsequently, NO signaling in smooth muscle cells promotes actin depolymerization and reduced stiffening. Consistent with this proposition, we show that stimulation of smooth muscle cells with SNP, an NO mimetic, reduces F-actin stress fibers and cellular stiffness. As proof of concept, we also confirmed that prolonged treatment of db/db mesenteric arteries with SNP results in a destiffening effect. Because LIMK inactivates cofilin to favor formation of actin networks and stress fibers (39, 40) and earlier work indicates that NO signaling can act upstream of LIMK and reduce its activity (e.g., by inhibiting tissue transglutaminase activity and the RhoA-ROCK pathway) (39, 53–55), here we reasoned that inhibition of LIMK would prevent SNP-induced actin depolymerization. Indeed, this is what we observed. Reciprocally, we show that treatment of smooth muscle cells with jasplakinolide, an approach to experimentally force actin polymerization and stabilization, also prevents SNP-induced actin depolymerization. Collectively, these findings are consistent with a scenario in which adropin-induced endothelial-derived NO drives smooth muscle cell actin depolymerization and consequently reduce cellular stiffness (Fig. 6). In this regard, there is precedence in the literature demonstrating that changes in smooth muscle cell stiffness can substantially contribute to changes in whole artery stiffness (56, 57).

Figure 6.

Schematic illustrating the interpretation of the results. Adropin-induced nitric oxide (NO) production in endothelial cells (EC) promotes vascular smooth muscle cell (VSMC) actin depolymerization and reduced stiffness. Destiffening of VSMCs likely contributes to reduced whole artery stiffness. Not depicted here, adropin also causes actin depolymerization and reduced stiffness in EC. Although endothelial rigidity does not physically contribute to whole artery stiffness, it is possible that decreased EC stiffness augments endothelial nitric oxide synthase (eNOS) activation and, consequently, increased NO signaling in VSMCs.

To further examine the role of adropin as a potential therapeutic target for vascular destiffening in obesity and T2D, we chronically treated db/db mice with adropin using surgically implanted subcutaneous osmotic minipumps. In concert with the hypothesis, we found that in vivo treatment of db/db mice with adropin reduced stiffness in mesenteric arteries. No significant changes in body weight or circulating metabolic parameters were observed after the 4 wk of adropin treatment, suggesting that the destiffening effects of adropin in obesity and T2D are likely independent of changes in metabolic function.

Finally, as an approach to provide a better understanding of signaling pathways and molecules modulated by adropin and thus stimulate future research, we performed a proteomic analysis in adropin-treated versus untreated endothelial cells. We found that a total of 421 proteins were differentially expressed in response to adropin treatment. An unbiased ingenuity pathway analysis revealed that a large fraction of differentially expressed proteins were associated with organization of the cytoskeleton. In response to the adropin treatment, the most upregulated transcriptional regulator linked to the organization of the cytoskeleton was SMARCA4 [gene encoding for SWI/SNF complex containing brahma-related gene 1 (BRG1), a transcriptional activator associated with angiogenesis] (58, 59). The observation that adropin induces an endothelial molecular signature indicative of cytoskeletal organization is in alignment with the abovementioned finding that adropin reduced cytoskeletal actin polymerization and cortical stiffness in endothelial cells. Furthermore, in corroboration of previous reports that adropin decreases apoptosis in vascular tissues (11, 12), ingenuity pathway analysis also revealed that treatment of endothelial cells with adropin stimulated the activation of cell survival pathways and repression of apoptotic pathways.

Several aspects of this investigation warrant further consideration. First, the reason for including carotid-to-femoral PWV in humans was to add greater translational relevance to the project; however, it should be acknowledged that carotid-to-femoral PWV reflects large elastic artery stiffness, whereas stiffness in our mouse models was examined in small arteries. Even though the mechanisms driving stiffening are likely not the same across the arterial tree, smooth muscle cytoskeletal remodeling is a determinant of arterial wall stiffening in both large and small vessels (57, 60–62). Also, although most data supporting the clinical relevance of aortic stiffening come from human studies in which measures of carotid-to-femoral PWV were obtained (63–65), data are also available indicating that stiffening of more peripheral and small arteries leads to poor perfusion and impaired blood pressure control (40, 66–71). Another disconnect between the human and mouse arterial stiffness outcomes of the current study is the in vivo versus ex vivo nature of the measures. Measures of PWV are impacted by physiological factors such as sympathetic activity and blood pressure (72, 73), whereas in our ex vivo preparation, isolated arteries are kept in Ca²⁺-free conditions thus removing any active tone and allowing for the assessment of stiffness of the wall material. For mechanistic studies, this could be considered a strength of the ex vivo measurement. Second, identification of adropin receptors in the vasculature remain largely elusive. Although elucidation of the receptor responsible for adropin signaling in the endothelium was outside the scope of this work, a prior study showed that in vascular endothelial growth factor receptor 2 (VEGFR2)-silenced endothelial cells, adropin-induced activation of eNOS was abrogated, suggesting that VEGFR2 may be an upstream mediator of adropin-induced eNOS activation (11). Third, although other studies have examined and reported the effects of adropin on endothelium-dependent dilation (13, 20), and have served as the impetus for the current work focused on arterial stiffness, the lack of endothelial function outcomes in the present report can be considered a limitation. Fourth, findings from the ex vivo treatment of arteries or cells with adropin for 24 h suggest that adropin-induced destiffening effects can occur rather quickly. The acute nature of these adropin effects is consistent with the idea that cytoskeletal remodeling can occur rapidly (i.e., within minutes/hours) (60, 74–76); nevertheless, time course studies are needed to establish the kinetics of these changes. Fifth, follow-up telemetry studies are needed to determine if the observed changes in arterial stiffness in our mouse model of adropin deficiency and/or adropin treatment are associated with, or independent of, changes in blood pressure. Lastly, although data from men and women were used for establishing the inverse association between magnitude of BMI/HbA1c and adropin expression in the liver, as well as the inverse relationship plasma adropin and arterial stiffness, it should be acknowledged that only male mice were included in the present studies. Accordingly, the extent to which findings from our preclinical models can be translated to females requires further investigation.

In aggregate, herein we extend prior work on cardiovascular effects of adropin by providing the first evidence that loss of adropin alone causes an increase in arterial stiffness that recapitulates the effects of obesity and T2D. Conversely, we show that adropin exposure reduces obesity and T2D-associated arterial stiffening, likely through a pathway involving NO. The significance of these observations is notable in that arterial stiffening is a strong and independent predictor of life-threatening cardiovascular events. Accordingly, this work supports that consideration should be given to “hypoadropinemia” as a potential target for the prevention and treatment of arterial stiffening in obesity and T2D.

SUPPLEMENTAL DATA

Supplemental Fig. S1 and Supplemental Tables S1–S7: http://doi.org/10.6084/m9.figshare.21030286.

GRANTS

This work was supported, in part, by National Institutes of Health Grants R01 HL137769 (to J.P.), R21 DK116081 (to C.M-A.), R01 HL088105 (to L.A.M-L.), and R01 DK113701 (to E.J.P., R.S.R., and J.A.I.) and funds from the Cardiometabolic Disease Research Foundation (to J.P.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

T.J.J., L.A.M-L., and J.P. conceived and designed research; T.J.J., F.I.R-P., F.J.C-A., R.N.S., R.J.P-M., E.E.B-C., N.J.M., N.S., H.N.M.R., S.F., M.M-Q., Y.L-F., and S.B. performed experiments; T.J.J., F.I.R-P., F.J.C-A., R.N.S., R.J.P-M., E.E.B-C., N.J.M., N.S., and H.N.M.R. analyzed data; T.J.J., F.I.R-P., A.A.B., H.S.S., J.A.I., E.J.P., R.S.R., C.M-A., L.A.M-L., and J.P. interpreted results of experiments; T.J.J., F.I.R-P., and J.P. prepared figures; T.J.J. and J.P. drafted manuscript; T.J.J., F.I.R-P., R.N.S., N.J.M., A.A.B., H.S.S., J.A.I., E.J.P., R.S.R., C.M-A., L.A.M-L., and J.P. edited and revised manuscript; T.J.J., F.I.R-P., F.J.C-A., R.N.S., R.J.P-M., E.E.B-C., N.J.M., N.S., H.N.M.R., S.F., M.M-Q., Y.L-F., A.A.B., S.B., H.S.S., J.A.I., E.J.P., R.S.R., C.M-A., L.A.M-L., and J.P. approved final version of manuscript.