Hyaluronan deficiency disrupts endothelial glycocalyx integrity and contributes to age-related arterial dysfunction

Abstract

Aging is accompanied by arterial dysfunction, defined by impaired endothelial function and increased large artery stiffness, yet the contribution of the endothelial glycocalyx (eGC) and its constituent, hyaluronan (HA), to this process remains incompletely understood. We tested the hypothesis that decreasing endothelial HA synthesis via a reduction in endothelial-specific HA synthase 2 (HAS2) compromises eGC integrity and contributes to arterial dysfunction with advancing age. Both endothelial-specific HAS2 reduction and aging impair eGC integrity, evidenced by impaired barrier function and reduced eGC thickness in vivo, and aging led to a reduction in circulating HA and an increase in syndecan-1. Similar to aging, reduced endothelial HA selectively attenuated flow-induced endothelium-dependent dilation (EDD) and eNOS activation, whereas agonist-induced EDD was preserved in young arteries despite the concomitant eGC dysfunction. In parallel, reduced HAS2 expression and advancing age were each associated with increased aortic stiffness and adverse arterial remodeling, including excessive collagen deposition and reduced elastin. Collectively, these findings identify endothelial HA as a key determinant of eGC integrity and endothelial function, linking impaired HA availability to the development of arterial dysfunction with advancing age.

Introduction

Aging is a primary risk factor for cardiovascular disease (CVD) and is accompanied by progressive arterial dysfunction. Two hallmarks of arterial aging are endothelial dysfunction and large artery stiffening, both of which contribute to impaired vascular homeostasis and heightened CVD risks [1]. Although endothelial dysfunction and large artery stiffness have been central targets for improving arterial health, the role of the endothelial glycocalyx (eGC) in these processes remains less clear.

The eGC is a carbohydrate-rich layer at the interface of flowing blood and the endothelium that supports barrier function and participates in shear-dependent signaling. Disruption of eGC integrity has been linked to endothelial dysfunction and large artery stiffening in aging and multiple disease states [2–5], yet the specific factors within the eGC that drive these changes are not yet fully characterized. Hyaluronan (HA) is a major eGC component and is particularly relevant because it is the principal non-sulfated glycosaminoglycan (GAG) within the eGC and is therefore more vulnerable to enzymatic and oxidative fragmentation than sulfated constituents [6–9]. Consistent with this vulnerability, eGC disruption is often accompanied by altered circulating HA and proteoglycan shedding markers [5], supporting the idea that HA homeostasis is closely tied to eGC integrity.

HA abundance in the eGC reflects a balance between synthesis and degradation [10]. HA is produced primarily by hyaluronan synthase 2 (HAS2), and aging is associated with reduced HAS2 expression and diminished eGC integrity [11]. In parallel, reactive oxygen species (ROS) and inflammatory mediators can accelerate HA fragmentation and eGC shedding [7, 12]. Together, these observations suggest that reduced HA availability may contribute directly to age-related deterioration of the eGC and downstream impairments in endothelial function and large artery stiffening.

To address this gap, we tested the hypothesis that reducing endothelial HA synthesis via endothelial-specific HAS2 reduction compromises eGC integrity and contributes to arterial dysfunction with advancing age. Using an endothelial-specific HAS2 reduction model in young and aged mice, we assessed eGC integrity in vivo and determined the functional consequences for both agonist- and flow-induced endothelium-dependent dilation (EDD). We further examined arterial stiffening and extracellular matrix remodeling as integrated outcomes of altered eGC and endothelial function.

Methods

Animal model

Inducible EC-specific HAS2 knockdown model was used to disrupt eGC properties in young (3–5 months) and old mice (21–25 months). Both male and female mice were included in the study. Because not all endpoints were measured in every animal, the distribution of males and females varied across individual experiments. Mice expressing EC-specific Cre-recombinase (Cdh5-Cre) were crossed with mice homozygous for floxed HAS2 (HAS2^f/f) and wild-type HAS2 (HAS2^+/+) to create HAS2^f/f Cdh5-Cre and HAS2^+/+ Cdh5-Cre mice for use as HAS2-deficient experimental animals (ecHAS2^−/−) and wild-type controls (ecHAS2^+/+), respectively (Fig. 1A-B). Young and old ecHAS2^−/− mice were generated by administering daily intraperitoneal injections of tamoxifen (4 mg/day) for 5 consecutive days, starting 2 weeks prior to the experiments (Fig. 1C). Young and old ecHAS2^+/+ mice were also administered the identical tamoxifen injection regimen to ensure that any observed effects were due to HAS2 reduction rather than tamoxifen exposure. Tamoxifen-treated ecHAS2^−/− and ecHAS2^+/+ mice were housed in standard mouse cages in an animal care facility at the VAMC-SLC on a 12:12 light/dark cycle. Mice were provided with a standard chow diet (Teklad Diet no. 8604, Envigo; protein: 24.3%, carbohydrate 40.9%, fat 4.7% by kcal) and water ad libitum. All animal studies followed the Guide for the Care and Use of Laboratory Animals and were approved by the University of Utah and Veteran’s Affairs Medical Center – Salt Lake City (VAMC-SLC).

Fig. 1.

Tamoxifen inducible endothelial cell-specific hyaluronan synthase 2 (HAS2) mice model. (A) Mice expressing endothelial cell (EC)-specific Cre-recombinase (Cdh5-Cre) were crossed with mice homozygous for floxed HAS2 (HAS2^f/f) and wild-type HAS2 (HAS2^+/+). (B) HAS2^f/f Cdh5-Cre and HAS2^+/+ Cdh5-Cre mice for use as HAS2-deficient experimental animals (ecHAS2^−/−) and wild-type controls (ecHAS2^+/+), respectively. (C) Tamoxifen (4 mg/day) was administered daily through intraperitoneal injection for 5 consecutive days, starting 2 weeks prior to the experiments. (D)Tamoxifen-induced HAS2 gene expression in young and old ecHAS2^+/+ and ecHAS2^−/− mice. Data are means ± SE. Individual data points represent biological replicates. N = 8–10 per group. *p ≤ 0.002 vs. young ecHAS2^+/+. †p = 0.013 vs. young ecHAS2^−/−

Cell culture model

HUVECs (passages 3–5) were cultured at 37 °C and 5% CO2 to promote eGC formation. To optimize eGC formation on the endothelial monolayer, HUVECs were seeded at a concentration of 5 × 10⁵ cells/mL on a pre-coated microfluidic single-channel slide (80176, Ibidi) and incubated until 70–80% confluency. Cells were then maintained under identical culture conditions for varying durations (48, 72, 120, and 168 h; Supplementary Fig. 1A). In addition, to optimize eGC degradation via enzymatic breakdown of HA, cells were treated with increasing concentrations of hyaluronidase (HYAL: 0, 2.5, 5, and 10 mg/mL; 389561, Sigma-Aldrich; Fig. 5E and Supplementary Fig. 1B) for 1 h. Based on these optimizations, an incubation duration of 120 h was sufficient for full eGC development, and HYAL at 10 mg/mL achieved robust eGC removal. These conditions were used for all subsequent experiments.

Fig. 5.

Flow-induced endothelium-dependent dilation (EDD) in young and old ecHAS2^+/+ and ecHAS2^−/− mice with and without hyaluronidase (HYAL) treatment. (A) The flow-induced EDD with incremental pressure gradient in carotid arteries. (B-D) The flow-induced EDD with incremental pressure gradient repeated with and without intraluminal treatment of HYAL. (E) The flow-induced in vitro model of endothelial glycocalyx formation and the effect of HYAL treatment in HUVECs. (F) Experiment was repeated to assess phosphorylated endothelial nitric oxide synthase expression with and without flow and HYAL treatment. Data are means ± SE. Individual data points represent biological replicates. N = 8–12 per group. *p ≤ 0.036 vs. young ecHAS2^+/+. †p < 0.0001 vs. young ecHAS2^−/−. ‡p ≤ 0.007 vs. control or static. §p = 0.003 vs. shear w/HYAL. Scale bars are 10 µm

To assess the role of HA within the eGC in the response to shear stress, experiments were performed using a sterile, closed-loop flow circuit in which culture medium was driven from a reservoir by a peristaltic pump, passed through a pulse dampener, and perfused through the microfluidic single-channel slide. The entire flow apparatus was maintained at 37 °C and 5% CO2. Cells were assigned to one of three conditions: (1) shear stress + HYAL, (2) shear stress + vehicle control, or (3) static control. For the shear stress + HYAL condition, cells were treated with HYAL and then exposed to laminar shear stress at 10 dyn/cm² for 48 h. For the shear stress + vehicle control condition, cells received PBS in place of HYAL and were exposed to the same shear stress regimen. Static controls were maintained in parallel without flow and without HYAL exposure. Cells were carefully washed with PBS and fixed with 4% paraformaldehyde for 15 min at room temperature. To visualize and quantify eGC, fixed cells were stained with wheat germ agglutinin (WGA-AF 647 conjugated, W32466, Invitrogen), which binds to N-acetyl-D-glucosamine and N-acetyl-D-neuraminic acid residues within the eGC, at 10 μg/mL for 20 min at 37 °C. In addition, cells were immunostained for phospho-eNOS (primary antibody: 1:200, PA5-104858, Invitrogen; secondary antibody: 1:500, AF488, A-11008, Invitrogen) and counterstained with 5 μg/mL Hoechst (H3570, Invitrogen). After two gentle washes with PBS, the microfluidic single-channel slides were filled with a water-based mounting medium (50001, Ibidi). Images were acquired using a Zeiss LSM 710 laser-scanning confocal microscope with 63 × and 40 × objectives on the day of the experiment. Image analysis was performed using ImageJ/FIJI (NIH, Bethesda, MD) and Imaris (Oxford Instruments, UK).

Glycocalyx related measurements

Mesenteric microcirculation was examined in vivo using intravital microscopy, as previously established [11]. Once anesthesia was induced with 2% isoflurane in 100% oxygen (2 L/min), mice were positioned supine on a temperature-controlled platform set to 37 °C. A small abdominal incision was made to expose the intestinal tract, after which a segment of intestine was gently externalized and placed in a reservoir filled with warm 0.9% saline solution. Externalized intestinal segments were continuously perfused with warm saline to maintain hydration and temperature throughout the imaging procedure. The quantification of microvascular structural and hemodynamic parameters, such as the perfused boundary region (PBR), eGC thickness, microvascular density, red blood cell (RBC) velocity, and flow, was performed using the automated GlycoCheck imaging and analysis platform (MicroVascular Health Solutions LLC, Alpine, UT), as previously discussed [11].

Briefly, the PBR was used as an index of eGC integrity and reflects the extent to which RBCs penetrate into the eGC layer. The GlycoCheck platform estimates this penetration by analyzing dynamic changes in RBC column width within individual microvessels and calculating the depth of RBC penetration into the luminal boundary region adjacent to the endothelial surface. Greater PBR values indicate deeper RBC penetration and are therefore indicative of greater eGC disruption. PBR was calculated using the equation shown below:

$$PBR=\frac{maximalperfuseddiameter-medianperfuseddiameter}{2}$$

The eGC thickness was determined by further manual inspection of the video datasets captured by the platform. Leukocyte passage transiently compresses the eGC, resulting in a temporary increase in RBC column width. The changes in the RBC column width were quantified by comparing measurements taken immediately before and after spontaneous leukocyte passage through individual capillaries. eGC thickness was calculated using the equation shown below:

$$eGCthickmess=\frac{\left(p,o,s,t,-,l,e,u,k,o,c,t,e,,p,e,r,f,u,s,e,d,,d,i,a,m,e,t,e,r\right)-\left(p,r,e,-,l,e,u,k,o,c,y,t,e,,p,e,r,f,u,s,e,d,,d,i,a,m,e,t,e,r\right)}{2}$$

The perfused microvascular density was defined as the total length of microvessels exhibiting blood flow within a specified tissue area with lumen diameters between 5 and 25 µm. Perfused microvascular density was calculated using the equation shown below:

$$Microvasculardensity=\frac{numberof4-25\mu mperfusedmicrovesselsegment\times 10\mu m}{tissuearearecorded}$$

The RBC velocity was quantified for individual vessel segments by applying cross-correlation analysis to longitudinal RBC intensity profiles measured between consecutive frames of recorded video sequences. To ensure reliable estimation of axial RBC displacement, only correlations with a coefficient of at least 0.80 were accepted. RBC velocity was calculated using the equation shown below:

$$RBCvelocity=\frac{RBCdisplacement}{timebetweenconsecutivevideoframes}$$

Microvascular flow was reported as a surface-normalized perfusion index expressed as the volume of RBC-containing blood traversing perfused microvessel segments per unit time and tissue area. Flow was calculated using the equation shown below:

$$Flow=\frac{volumeof4-25\mu mperfusedmicrovesselsegmentspertime}{tissuearearecorded}$$

Endothelial function

Agonist-induced EDD was assessed to investigate whether alterations in eGC integrity influence intrinsic nitric oxide (NO)-mediated vasodilatory signaling pathways. We performed dose-dependent responses in arteries to chemical agonists and antagonists, as previously discussed [13]. Mesenteric arterioles were dissected, cleared of perivascular connective tissue, and cannulated onto glass micropipettes in a pressure myography chamber (DMT Inc., Atlanta, GA). Arteries were warmed to 37 °C and pressurized to 68 cmH₂O with gradual increments. After equilibrium was reached for at least 60 min, arteries were pre-constricted with 2 μM phenylephrine (PE). Subsequently, EDD was measured in response to the cumulative addition of acetylcholine (ACh, 1 × 10⁻⁹ to 1 × 10⁻⁴ mol/L). To determine the contribution of NO to EDD, the ACh dose–response was repeated with the nitric oxide synthase inhibitor, L-NAME (0.1 mmol/L), for 30 min. To determine the contribution of superoxide to EDD, an additional artery was incubated with the superoxide dismutase mimetic, TEMPOL (1 mmol/L), for 60 min, and the ACh dose–response was repeated with and without L-NAME under the same conditions. To determine endothelium-independent dilation, changes in luminal diameter were measured in response to the cumulative addition of sodium nitroprusside (SNP, 1 × 10⁻¹⁰ to 1 × 10⁻⁴ mol/L). Arterial diameters were determined by MyoView software (DMT Inc., Atlanta, GA).

Flow-induced EDD was performed to examine whether modification of eGC mechano-sensing and mechano-transducing properties affects NO-mediated vasodilatory signaling pathways. Carotid arteries were dissected, cleared of perivascular tissue, and cannulated onto glass micropipette tips on both ends of the pressure myograph. Arteries were pre-constricted with 2 μM PE, and intraluminal flow was generated from a pressure gradient of 0 to 40 cmH₂O with 5 cmH₂O increments. To determine the contribution of HA to eGC mechano-sensing and mechano-transducing properties, 1 mg/mL HYAL was intraluminally infused and incubated for 1 h, and the experimental protocol was repeated. Because enzyme accessibility, eGC architecture, and exposure conditions differ substantially between cultured endothelial monolayers and intact pressurized arteries, HYAL concentrations were optimized independently for the in vitro and ex vivo preparations. Accordingly, 1 mg/mL intraluminal HYAL was used to assess the functional consequences of HA degradation on endothelial responses and was not intended to directly replicate the concentration used in cultured HUVECs. Arterial diameters were determined by MyoView software. All data are presented as percentage change in diameter at each pressure gradient relative to the pre-constricted diameter with PE.

Aortic stiffness

Aortic pulse wave velocity (PWV) was measured as an indicator of in vivo aortic stiffness, as previously described [14]. Briefly, mice were anesthetized with 2% isoflurane in 100% oxygen (2 L/min) and positioned supine on a temperature-controlled platform set to 37 °C. Pulse waveforms were placed simultaneously at the transverse aortic arch and abdominal aorta using 20 MHz Doppler probes (Indus Instruments, Webster, TX) and recorded with WinDAQ Pro + software (DataQ Instruments, Akron, OH, USA). The path length between probe positions was measured immediately after waveform collection using a scientific caliper. Pulse wave transit time between probes was analyzed in WinDAQ Waveform Browser (DataQ Instruments, Akron, OH) using the foot-to-foot method. Aortic PWV was calculated as the pulse waveform travel distance divided by the transit time.

Intrinsic mechanical wall stiffness

Aortic elastic modulus was quantified ex vivo as a measure of intrinsic mechanical wall stiffness, as previously established [15]. In short, the aorta was excised, placed in cold 0.9% saline solution, and gently cleared of perivascular connective tissue. Two thoracic aorta sections were prepared for testing, consisting of a 1 mm ring and a 2 mm ring positioned immediately distal to the initial segment. The 2 mm aortic ring was placed in a temperature-controlled bath at 37 °C filled with calcium-free PBS and mounted onto a wire myograph (DMT Inc., Atlanta, GA). Each aortic ring was pre-stretched by 1 mm from the initial position on the micropins for three intervals of 30 s. The ring diameter was then increased until a force of 1 mN was achieved. Aortic rings were incrementally stretched by 50 μm for 3 min per stage to reach equilibrium, continuing until structural failure. The 1 mm aortic ring was embedded in optimal cutting temperature solution and stored at −80 °C until tissue sectioning. Aortic wall thickness and diameter were acquired from this sample, and further details are described in the tissue histology section. A stress–strain curve was derived from the following equations:

$$Strain\left(\lambda \right)=\frac{▵d}{d_1}$$

where d is diameter and d₁ is initial diameter.

$$Stress\left(t\right)=\frac{\lambda +L}{2\times (H\times D)}$$

where L is one-dimensional load, H is wall thickness, and D is vessel length.

Tissue histology

OCT-embedded aortic rings were sliced into 8 μm sections, with 3 to 4 sections per slide averaged. Measurements of medial cross-sectional area, the lumen border, and the outer medial border were traced in ImageJ/FIJI (NIH, Bethesda, MD), and internal areas were measured. These areas were used to calculate medial cross-sectional area as the outer media border area minus the lumen area. Elastin content was quantified by Verhoeff’s Van Gieson stain as a percentage of the selected area, as described previously. An 8-bit grayscale was used for densitometric quantification of elastin content with ImageJ/FIJI (NIH, Bethesda, MD). Collagen content was quantified by trichrome stain as a percentage of the selected area, as described previously. Green channel images from an RGB stack were used for densitometric quantification of collagen content with ImageJ/FIJI (NIH, Bethesda, MD). Arterial nitrotyrosine (NT) content was quantified as described previously. Slides containing 3 to 4 8-μm aortic sections from each mouse were fixed for 15 min in 4% PFA, pH 7.4, at room temperature, followed by rehydration for 5 min in 1 × PBS. Samples were blocked in 1 × PBS containing 1% BSA and 0.1% Triton-X for 15 min, followed by incubation for 2 h with 1:50 primary antibody against NT (06–284, Millipore Sigma) in blocking solution. Samples were then washed 3 × 5 min in 1 × PBS, followed by 1 h incubation with 1:100 secondary AF647 antibody (A-21245, Invitrogen) in blocking solution, washed 3 × 5 min in 1 × PBS, and mounted with DAPI (17984–24, Electron Microscopy Sciences). Superoxide dismutase (SOD) content was quantified following the same protocol as above using 1:200 primary antibody against SOD (MA5-15520, Invitrogen) and 1:400 secondary AF647 antibody (A-21235, Invitrogen). Slides were imaged with an Olympus Fluoview FV1000 confocal microscope, and an equal number of 1-μm z-slices were imaged for each sample. Images were analyzed using ImageJ/FIJI (NIH, Bethesda, MD). NT and SOD content was quantified from thresholded images using the integrated density measure.

Electron paramagnetic resonance spectroscopy

To evaluate arterial superoxide levels, carotid arteries were cleared of perivascular tissue, and 2 mm sections were isolated. Arterial sections were washed once in cold PSS and then incubated for 1 h at 37 °C in modified Krebs-HEPES buffer containing 0.5 mmol/L CMH. Samples were then snap-frozen in liquid nitrogen and stored at −80 °C until analysis. Samples were analyzed on an MS300 X-band Electron Paramagnetic Resonance (EPR) Spectrometer (Magnettech, Novi, MI) using the following settings: microwave frequency 9.83 GHz, centerfield 3480 G, sweep 80 G, modulation amplitude 3.3 G, microwave power 40 mW, microwave attenuation 7, and receiver gain 30. A total of six sweeps were conducted, lasting 8.7 s per sweep. The running average of the six sweeps was collected, with the double integration (area under and over the baseline) of the triplet used to display the magnitude of the signal. The magnitude of this signal directly relates to the amount of superoxide trapped by the CMH. All values were normalized to WT superoxide levels.

Enzyme-linked immunosorbent assays

Serum concentrations of hyaluronan (K-1200, Echelon), syndecan-1 (ab273165, Abcam), and hyaluronidase activity (K-6000, Echelon) were measured according to the manufacturer’s protocol.

Quantitative polymerase chain reaction

Relative mRNA expression was measured in the aorta and carotid EC lysates by quantitative polymerase chain reaction (qPCR). The aorta was dissected, and perivascular tissues were removed. Both carotid arteries were dissected, and the inner lumen was flushed with QIAzol lysis reagent (79306, Qiagen) to obtain carotid EC lysate. RNA from aorta and carotid EC lysate was isolated using an RNeasy Mini Kit (74104, Qiagen) according to the manufacturer’s protocol. RNA isolated from aorta and carotid EC lysate was used to synthesize cDNA via the QuantiTect Reverse Transcription kit (205,311, Qiagen) according to the manufacturer’s protocol. qPCR was performed using SsoFast EvaGreen Supermixes (1725205, Bio-Rad) and a Bio-Rad CFX Real-Time system. The 2 ^− ΔΔCt method was used to quantify relative gene expression. Primer sequences were as follows: HAS2 (forward: CAGGAGCTGAACAAGATGCATTG, reverse: TCCGGTGTTCCAAAAAGGCA) and β-actin (forward: GGCTGTATTCCCCTCCATCG, reverse: CCAGTTGGTAACAATGCCATGT).

Blood pressure

Systolic and diastolic arterial blood pressure (BP) was measured non-invasively in conscious mice using a volume-pressure recording tail-cuff system (CODA System, Kent Scientific, CT), as previously described [16]. Heart rate was simultaneously recorded using the same apparatus. Prior to data collection, all mice underwent a 2–3 day acclimation period to the tail-cuff device. On the day of measurement, conscious mice were positioned in animal restrainers on a warmed platform maintained at 37 °C, and recordings were conducted in a quiet room held at 24 °C. Mice were allowed 15–20 min within the restrainers to acclimate and achieve a stable tail skin temperature of 30–35 °C. Each recording session began with 10 acclimation cycles to confirm signal stability, followed by 40 experimental cycles. Any measurements associated with excessive movement, behavioral artifacts, or insufficient tail blood volume/flow were excluded from analysis. The remaining valid readings were averaged to generate a single mean value per animal.

Statistics

Statistical analyses were performed with GraphPad Prism version 8.0.2. Cell culture data were analyzed using an independent t-test and one-way ANOVA with Tukey’s HSD post hoc tests. Aging and genotype main effects and interactions were analyzed using two-way ANOVA with Tukey’s HSD post hoc tests. Interaction effect sizes were calculated using partial η² and Post hoc power analyses were performed for selected primary endpoints using G*Power version 3.1 (Heinrich Heine University Düsseldorf, Düsseldorf, Germany). Data are presented as mean ± SEM. Significance was set at p < 0.05.

Results

EC-specific HAS2 gene expression

Carotid EC lysate HAS2 gene expression was significantly reduced in young ecHAS2^−/−, old ecHAS2^+/+, and old ecHAS2^−/− compared with young ecHAS2^+/+ mice after tamoxifen treatment (p ≤ 0.002, Fig. 1D). There was no additive effect in old ecHAS2^−/− compared with old ecHAS2^+/+ mice (p = 0.999).

eGC integrity

We first investigated whether HAS2 reduction, aging, or both affected eGC integrity. The PBR, an indicator of eGC barrier function, was assessed in vivo within the mesenteric microcirculation across microvessels ranging from 4 to 25 μm. PBR was significantly compromised after HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p = 0.001; Fig. 2A-B) and with aging (young vs old ecHAS2^+/+, p = 0.018). While HAS2 reduction and aging together impaired PBR (young ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.007) compared to young wildtype mice, there was no additional effect of HAS2 reduction in aged mice (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.966). The age × genotype interaction for PBR was significant (p = 0.012) and was associated with a partial η² of 0.17 and post hoc power of 0.78. Similarly, eGC thickness was diminished with both HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p < 0.0001; Fig. 2C) and aging (young vs old ecHAS2^+/+, p < 0.0001). While HAS2 reduction and aging together reduced eGC thickness compared to young wildtype mice (young ecHAS2^+/+ vs old ecHAS2^−/−, p < 0.0001), there was no further reduction in eGC thickness in aged mice after HAS2 reduction (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.893). The age × genotype interaction for eGC thickness was significant (p < 0.0001) and was associated with a partial η² of 0.74 and post hoc power of 1. Microvascular density was greater in old compared with young mice regardless of genotype across microvessels ranging from 4 to 25 μm (young vs old, p < 0.0001; Fig. 2D). RBC velocity was similar across genotypes and age (p ≥ 0.05; Fig. 2E). Microvascular flow was greater with HAS2 reduction and aging compared with young groups (old ecHAS2^−/− vs young ecHAS2^+/+ and young ecHAS2^−/−, p ≤ 0.039; Fig. 2F).

Fig. 2.

Endothelial glycocalyx (eGC) integrity in young and old ecHAS2^+/+ and ecHAS2^−/− mice. (A-C) The eGC barrier function was measured evaluating diameter‑dependent perfused boundary region (PBR), total PBR, and eGC thickness obtained from the mesenteric microcirculation. (D-F) The eGC-related hemodynamics were evaluated. (G–I) The circulating eGC-associated biomarkers, including serum hyaluronan, syndecan-1, and hyaluronidase activity. Data are means ± SE. Individual data points represent biological replicates. N = 9–18 per group. *p ≤ 0.034 vs. young ecHAS2^+/+

Serum concentrations of eGC-related biomarkers were investigated to confirm impairment of eGC integrity. Serum HA was significantly reduced with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p = 0.034; Fig. 2G) and aging (young vs old ecHAS2^+/+, p = 0.012). While HAS2 reduction and aging together diminished serum HA compared to young control mice (young ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.025), there was no additive effect of aging and HAS2 reduction (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.997). Syndecan-1, an eGC damage marker, increased in old compared with young mice regardless of genotype (young vs old, p ≤ 0.003; Fig. 2H). HYAL, a hyaluronan-degrading enzyme, did not differ across genotypes or age (p ≥ 0.05; Fig. 2I).

Agonist-induced EDD

ACh-mediated EDD was not different with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p ≥ 0.962; Fig. 3A-B) in young mice, but was suppressed with age (young vs old ecHAS2^+/+, p ≤ 0.018). HAS2 reduction in aged mice further reduced ACh-mediated EDD (old ecHAS2^+/+ vs old ecHAS2^−/−, p ≤ 0.021). The age × genotype interaction for ACh-mediated EDD was not significant (p = 0.119) and was associated with a partial η² of 0.06 and post hoc power of 0.37. In the presence of L-NAME, ACh-mediated EDD was significantly reduced to a similar extent in young groups (p ≤ 0.019; Fig. 3C). In old groups, L-NAME significantly reduced ACh-mediated EDD in ecHAS2^+/+ mice (p < 0.0001; Fig. 3D), whereas the reduction in ecHAS2^−/− mice did not reach statistical significance (p = 0.078). There was no difference in SNP-mediated dilation across all groups (p ≥ 0.05; Fig. 3E). In the presence of TEMPOL, the combined effect of HAS2 reduction and aging were prevented (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.272; Fig. 4A-D).

Fig. 3.

Agonist-induced endothelium-dependent dilation and –independent dilation in young and old ecHAS2^+/+ and ecHAS2^−/−mice. (A-B) The acetylcholine (ACh)–induced vasodilation dose–response and maximal vasodilation was measured in mesenteric arteries. (C-D) The ACh–induced vasodilation dose–response was repeated in the absence and presence of the nitric oxide synthase inhibitor, L-NAME.(E) The sodium nitroprusside, exogenous nitric oxide donor, -induced vasodilation dose–response. Data are means ± SE. Individual data points represent biological replicates. N = 8–12 per group. *p ≤ 0.04 vs. young ecHAS2^+/+. †p ≤ 0.004 vs. young ecHAS2^−/−. ‡p ≤ 0.021 vs. old ecHAS2^+/+

Fig. 4.

Effect of TEMPOL on agonist‑induced endothelium‑dependent dilation in young and old ecHAS2^+/+ and ecHAS2^−/− mice. (A-B) The acetylcholine (ACh)-induced vasodilation dose–response and maximal vasodilation measured in the presence of TEMPOL, the superoxide scavenger. (C-D) The ACh–induced vasodilation dose–response in the presence of TEMPOL was repeated in the absence and presence of the nitric oxide synthase inhibitor, L-NAME. Data are means ± SE. Individual data points represent biological replicates. N = 8–12 per group. *p ≤ 0.02 vs. young ecHAS2^+/+. †p ≤ 0.004 vs. young ecHAS2^−/−

Flow-induced EDD

Flow-induced vasodilation was diminished with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p < 0.0001; Fig. 5A) and aging (young vs old ecHAS2^+/+, p < 0.0001), but there were no additive effects of HAS2 reduction in the old mice (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.999). Subsequently, vasodilatory response to flow was suppressed after HYAL treatment in young and old wildtype and reduced HAS2 groups (young ecHAS2^+/+, young ecHAS2^−/−, old ecHAS2^+/+, old ecHAS2^−/− vs with HYAL counterparts, p ≤ 0.0001; Fig. 5B-C). However, the change in diameter with HYAL treatment, compared with and without HYAL, was significantly greatest in young ecHAS2^+/+ than in all other groups (p ≤ 0.036; Fig. 5D). The age × genotype interaction for flow-induced EDD was significant (p = 0.043) and was associated with a partial η² of 0.12 and post hoc power of 0.58.

We developed a flow-induced cell culture model using HUVECs. Prolonged incubation increased eGC formation (Supplementary Fig.  1A) and increasing doses of HYAL caused eGC degradation (Supplementary Fig. 1B). Based on these results, we adopted an incubation period of 120 h to develop sufficient eGC volume and a HYAL concentration of 10 mg/mL to achieve robust eGC removal for subsequent experiments. HUVECs incubated with HYAL had significantly diminished eGC volume compared with control cells without HYAL (p = 0.007; Fig. 5E). A peristaltic pump was used to expose HUVECs to physiologically relevant flow conditions with and without HYAL treatment. We found that laminar flow upregulated phosphorylated endothelial nitric oxide synthase (p-eNOS) expression (p = 0.0002; Fig. 5F), and that flow induced p-eNOS expression was reduced after HYAL treatment to a level comparable with static conditions (p = 0.118).

Aortic stiffness

Aortic stiffness, assessed by aortic PWV, was elevated with either HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p = 0.023; Fig. 6A) or aging (young vs old ecHAS2^+/+, p < 0.0001), but no further impairment was observed with the combination of HAS2 reduction and old age. (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.908). The age × genotype interaction for aortic PWV was not significant (p = 0.135) and was associated with a partial η² of 0.06 and post hoc power of 0.35. The change in intrinsic arterial properties was further assessed using the stress–strain curve (Fig. 6B). Aortic elastic modulus was increased with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p = 0.047; Fig. 6C) and aging (young vs old ecHAS2^+/+, p = 0.033), but no additive effect was of aging and HAS2 reduction was found (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.999). The age × genotype interaction for aortic elastic modulus was not significant (p = 0.073) and was associated with a partial η² of 0.11 and post hoc power of 0.48. Total collagen content was increased with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p < 0.0001; Fig. 6D) and aging (young vs old ecHAS2^+/+, p < 0.0001), but no additive effect was found (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.998). Meanwhile, total elastin content was markedly diminished with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p < 0.0001; Fig. 6E) and aging (young vs old ecHAS2^+/+, p < 0.0001) but was not further reduced by a reduction in HAS2 in the old mice (old ecHAS2^+/+ vs old ecHAS2^−/−, p = 0.974). Systolic and diastolic blood pressure did not change regardless of HAS2 reduction and aging (Supplementary Fig. 2A-B, p ≥ 0.897).

Fig. 6.

Mechanical and structural properties of aortas in young and old ecHAS2 ^+/+ and ecHAS2^−/− mice. (A) Pulse wave velocity (PWV), the measure of large artery stiffness, was performed in vivo from transverse aortic arch to abdominal aorta. (B-C) Intrinsic mechanical wall stiffness was measured ex vivo using aortic rings. (D) Aortic sections were stained for total collagen and elastin for quantification. Data are means ± SE. Individual data points represent biological replicates. N = 7–12 per group. *p ≤ 0.023 vs. young ecHAS2 ^+/+. †p ≤ 0.001 vs. young ecHAS2^−/−

Oxidative stress

Aortic NT, a biomarker of oxidative burden, was increased with aging (young vs old ecHAS2^+/+, p = 0.037; Fig. 7A) but not with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p ≥ 0.05). HAS2 reduction did not further increase NT content with aging (old ecHAS2^+/+ vs old ecHAS2^−/−, p ≥ 0.05). In agreement with these findings, superoxide measured by EPR spectroscopy in isolated carotid arteries was greater with aging (young vs old ecHAS2^+/+, p < 0.0001; Fig. 7B) but not with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p ≥ 0.05). No further increase in superoxide production with aging was detected after HAS2 reduction (old ecHAS2^+/+ vs old ecHAS2^−/−, p ≥ 0.05). In parallel, arterial SOD, an indicator of antioxidant capacity, was diminished with aging (young vs old ecHAS2^+/+, p = 0.034; Fig. 7C) but not with HAS2 reduction (young ecHAS2^+/+ vs young ecHAS2^−/−, p ≥ 0.05). No further reduction in SOD was observed with HAS2 reduction in aged mice (old ecHAS2^+/+ vs old ecHAS2^−/−, p ≥ 0.05).

Fig. 7.

Oxidative stress-related markers in arteries from young and old ecHAS2^+/+ and ecHAS2^−/− mice. (A) Quantification of arterial superoxide levels in an aortic ring using electron paramagnetic resonance spectroscopy. (B-C) Representative immunofluorescence images and quantification of superoxide dismutase and nitrotyrosine with nuclear counterstain. Data are means ± SE. Individual data points represent biological replicates. N = 6–10 per group. *p ≤ 0.049 vs. young ecHAS2^+/+. †p ≤ 0.006 vs. young ecHAS2^−/−. Scale bars are 100 µm

Discussion

Aging-related deterioration of the eGC has been implicated in arterial dysfunction, yet the specific contribution of HA loss to this process has remained poorly defined. In this study, we addressed this gap using an endothelial-specific HAS2 reduction model to directly examine how impaired HA synthesis influences eGC integrity and arterial function with advancing age. Our findings show that both reduced HA production and advanced age independently disrupt eGC structure, leading to impaired endothelial function and arterial stiffening. These results indicate that HA-dependent eGC loss is not merely a marker of aging but may contribute to arterial dysfunction. Collectively, these observations establish HA as a key determinant of eGC homeostasis and suggest that preserving HA within the eGC may be important for maintaining arterial health with aging.

Deterioration of the eGC is a well-recognized feature of multiple pathological states, including hypertension, diabetes, and chronic kidney disease, where elevated circulating syndecan-1 and HA reflect increased eGC shedding and loss of barrier integrity [3–5]. This degradation can occur through enzymatic pathways, including diverse protease activities [6, 17, 18], as well as through non-enzymatic mechanisms such as inflammation- and ROS-driven fragmentation of eGC constituents [7, 12]. Among these components, HA is uniquely susceptible to degradation because its non-sulfated structure offers less protection against enzymatic and oxidative stress than sulfated GAGs such as heparan sulfate and chondroitin sulfate [6–9]. Notably, this has been supported experimentally by findings that sulfated HA derivatives are resistant to enzymatic digestion whereas native HA is readily depolymerized [8, 9]. In our study, both reduced HA production and aging independently compromised eGC integrity, evidenced by greater RBC penetration into the eGC (i.e., higher PBR) and lower eGC thickness, consistent with a weakened endothelial barrier. These structural impairments occurred alongside lower serum HA and higher syndecan-1 in aging, a pattern more consistent with reduced HA synthesis rather than excessive HA shedding. At the same time, the elevation in syndecan-1 suggests that proteoglycan shedding remains increased, likely as a consequence of persistent, age-related eGC injury. Aged mice exhibited greater perfused microvascular density and flow. Because GlycoCheck quantifies perfused rather than anatomical microvascular density, these measurements may reflect changes in perfusion characteristics in addition to structural vascular alterations. Collectively, these findings support HA as an important contributor to eGC stability and suggest that reductions in HA associated with aging or reduced HAS2 expression drive the HA-dependent component of eGC degradation within the broader multicomponent structure of the eGC.

While the eGC is composed of multiple GAGs, proteoglycans, and associated plasma proteins, HA may be particularly relevant to arterial aging because of several unique properties. Unlike sulfated GAGs such as heparan sulfate and chondroitin sulfate, HA is especially susceptible to enzymatic and oxidative fragmentation, making it more vulnerable to age-related physiological insults [6–9]. In addition, HA exhibits molecular-weight-dependent biological activities, with high- and low-molecular-weight forms exerting distinct effects on vascular signaling and homeostasis [19]. Consequently, reductions in HA availability may have functional consequences that extend beyond simple loss of eGC mass. Nevertheless, because the present study did not directly evaluate other eGC constituents, we cannot exclude the possibility that reduced HA availability also induces secondary changes in broader glycocalyx architecture that contribute to the observed phenotype.

To determine whether structural changes in the eGC translated to altered endothelial function, we employed two complementary approaches: pharmacological stimulation of EDD via ACh, a muscarinic receptor agonist, and mechanical stimulation via fluid shear stress. These approaches provide a comprehensive assessment of endothelial function by separating agonist-induced NO signaling from eGC-dependent mechanosensing pathways that likewise activate NO-mediated vasodilation. Despite the multicomponent nature of the eGC, disrupting HA availability produced functional consequences, particularly for shear-dependent responses. In young arteries, agonist-induced EDD was preserved despite eGC impairment, indicating that receptor-driven NO signaling remained sufficient to preserve vasodilatory capacity when eGC-dependent mechanosensing was disrupted (Fig. 3A-B). This finding suggests that a compensatory reliance on agonist-driven pathways can maintain EDD when shear-dependent responses are limited; however, this compensatory mechanism appears to decline with aging. In contrast, flow-induced EDD was markedly reduced following either HAS2 reduction or selective HA degradation, indicating disruption of eGC-dependent mechanosensing pathways, a finding further supported by diminished shear-stimulated p-eNOS activation in endothelial cells. Earlier work demonstrated that enzymatic removal of HA from the glycocalyx markedly reduces flow-induced NO production while largely sparing agonist-induced NO signaling [20]. Our findings place this acute mechanistic concept into an age-relevant physiological context.

Age-related increases in oxidative stress likely contribute to the impaired EDD observed in old arteries. Elevated NT and reduced SOD expression reflect a detrimental shift in redox balance that diminishes NO bioavailability, thereby limiting both agonist- and flow-induced EDD. Consistent with this mechanism, TEMPOL partially restored ACh-mediated EDD in old arteries, indicating that superoxide plays a meaningful role in suppressing NO signaling with aging. In parallel, greater oxidative stress with age may increase the susceptibility of the eGC to structural weakening [7], thereby intensifying deficits in its mechanosensing capacity and contributing to further declines in endothelial function. In contrast to aging, HAS2 reduction alone did not induce oxidative stress or impair agonist-induced EDD, suggesting that these particular aspects of age-related arterial dysfunction involve mechanisms beyond reduced HA availability. Thus, reduced HA availability appears sufficient to impair eGC-dependent mechanosensing and flow-induced EDD, whereas additional aging-associated processes, including oxidative stress, may be required to produce the more generalized endothelial dysfunction observed in old arteries.

While the present study does not include a gain-of-function model, evidence from glycocalyx-targeted therapy in aged mice shows that HA supplementation can restore agonist-induced EDD, indicating that increasing HA availability within the eGC can improve NO signaling in an endothelium already affected by aging [2]. These findings suggest that the aged endothelium depends more heavily on HA to sustain NO signaling than the healthy endothelium. Although the underlying mechanisms are multifactorial, HA repletion likely enhances shear sensing [2], stabilizes the near-membrane redox environment by harboring SOD [21], and reduces local NO scavenging [22], thereby improving endothelial performance. Together, these findings support the concept that restoration of eGC integrity may represent a therapeutic strategy to improve arterial function with aging. Although HA restoration improves arterial function in aged animals, the extent to which these benefits reflect HA-specific effects versus broader restoration of eGC integrity remains unclear.

It is important to recognize that arterial dysfunction does not occur in isolation from impaired EDD but rather represents part of a continuum that includes structural changes in the arterial wall driving arterial stiffening. This interplay between functional and structural deterioration reflects a hallmark of arterial aging, where compromised eGC integrity not only blunts endothelial signaling but also promotes detrimental extracellular matrix remodeling. Our data indicate that both aging and reduced HAS2 expression independently contribute to arterial stiffening and adverse remodeling, characterized by elastin fragmentation and compensatory collagen accumulation. Although the magnitude of these histological changes warrants cautious interpretation, the directionality of the remodeling response was consistent across experimental groups. The observed increase in aortic PWV and elastic modulus underscores the functional consequences of these structural alterations. Likely, elevated oxidative stress and inflammatory signaling promote matrix crosslinking and vascular smooth muscle cell activation, which accelerates stiffening [23, 24].

The findings of this study provide important insight into the role of HA in eGC regulation and arterial function; however, several limitations should be acknowledged. This study specifically targeted HA synthesis by reducing endothelial HAS2 expression, but we did not evaluate differences in HA molecular weight, which represents an important functional property of the polymer, as high- and low-molecular-weight HA exert distinct effects on vascular signaling and extracellular matrix interactions [19]. Because our approach reduces overall HA availability without distinguishing between these molecular weight forms, we cannot determine whether the observed impairments stem from decreased total HA content, altered size distribution, or both. Future studies examining HA molecular-weight-specific effects will be important for clarifying how distinct HA sizes contribute to eGC regulation and arterial function. Second, the eGC also contains multiple sulfated and nonsulfated components that decline with aging, and the contribution of these constituents was not directly isolated here, limiting our ability to fully separate broader eGC remodeling from HA-specific effects. Nevertheless, our flow-induced EDD and cell culture experiments demonstrate that selective HA degradation alone was sufficient to blunt flow-induced EDD and reduce shear-stimulated p-eNOS activation, underscoring an important role for HA within the complex eGC structure. Third, this study did not include a restorative intervention targeting the eGC to determine whether the identified flow-induced EDD impairments are reversible in the same model. Although prior work from our group shows that HA supplementation in old mice can restore eGC structure and improve arterial function [2], whether the specific deficits observed in the HAS2-reduced model respond similarly requires further investigation. Fourth, although interaction effects were detected for several eGC-related and EDD outcomes, interaction effect sizes varied across primary outcomes (partial η² = 0.06–0.74), and post hoc power analyses indicated variable sensitivity across endpoints. In addition, endothelial HAS2 expression in old ecHAS2^+/+ mice were already markedly reduced and approached levels observed in young ecHAS2^−/− mice, raising the possibility of a floor effect in which further suppression of HAS2 expression is limited. Therefore, the absence of significant interactions for some measures should be interpreted with caution and should not be considered definitive evidence of mechanistic overlap between aging and HAS2 reduction. Finally, although both male and female mice were included, the study was not specifically powered to examine sex-dependent effects. The potential interactions between sex, age, and genotype cannot be excluded and warrant investigation in future studies.

In conclusion, this study identifies endothelial HA availability as a key regulator of eGC integrity and a determinant of arterial function with advancing age. The findings support the concept that maintaining HA-dependent eGC homeostasis is integral to preserving endothelial signaling capacity and limiting downstream arterial remodeling that contributes to stiffening. Collectively, these results position the eGC, and HA in particular, as a mechanistically grounded target for strategies aimed at maintaining arterial health with age.

Abbreviations

ACh

Acetylcholine

BP

Blood pressure

CVD

Cardiovascular disease

EDD

Endothelium-dependent dilation

eGC

Endothelial glycocalyx

eNOS

Endothelial nitric oxide synthase

EPR

Electron paramagnetic resonance

GAG

Glycosaminoglycan

p-eNOS

Phosphorylated-endothelial nitric oxide synthase

HA

Hyaluronan

HAS2

Hyaluronan synthase 2

HUVECs

Human umbilical vein endothelial cells

HYAL

Hyaluronidase

NO

Nitric oxide

NT

Nitrotyrosine

PBR

Perfused boundary region

PWV

Pulse wave velocity

qPCR

Quantitative polymerase chain reaction

RBC

Red blood cells

ROS

Reactive oxygen species

SOD

Superoxide dismutase

SNP

Sodium Nitroprusside

Author contributions

JL, LAL, and AJD designed the study. JL, JK, VRG, AA, and CB performed the experiments. JL analyzed the data. JL, LAL, and AJD interpreted the data. JL and AJD wrote the manuscript. All authors read and approved of the final version of the manuscript.

Funding

This work was supported by National Institutes of Health Awards R01 AG060395, R01 AG077751, R01 AG076748, T32 HL007576, T32 HL139451, Veteran's Affairs Merit Review Award I01 BX004492 and Nora Eccles Treadwell Foundation.

Declarations

Conflict of interest

The authors declare no competing interests.