Hypertensive stretch regulates endothelial cell inflammation and apoptosis through ceramide metabolism

Abstract

Hypertension and cardiovascular disease (CVD) are associated with elevated plasma levels of ceramides (Cer), a type of membrane sphingolipids (SPLs). Increased cellular Cer levels are known to cause vascular endothelial cell (EC) dysfunction. However, Cer metabolism changes due to EC exposure to high-magnitude “hypertensive” cyclic stretch (HCS) and their role in EC dysfunction are poorly defined. Cultured human ECs exposed to HCS (15% elongation at 1 Hz, 24 h) exhibited increased oxidative stress, upregulation of genes for cytokines and leukocyte adhesion molecules, increased inflammatory response to a low concentration of tumor necrosis factor-α, and increased apoptosis compared to ECs exposed to “normotensive” CS (NCS; 5% elongation at 1 Hz, 24 h). Lipidomics analysis of EC pellets at 4 or 24 h of CS detected no difference in Cer levels and significantly higher sphingomyelin (SM) levels at 24 h of HCS compared to NCS. Cer immunostaining showed significantly higher Cer levels at both the perinuclear and peripheral subcellular regions in HCS compared to the corresponding regions in NCS. HCS significantly increased the concentrations of certain long-chain Cer in the extracellular media compared to NCS. Pharmacological inhibition of key enzymes in Cer biosynthesis, i.e., de novo synthesis and SM hydrolysis pathways, significantly inhibited HCS-induced EC inflammation and apoptosis, suggesting that Cer generated from SM, via the sphingomyelinase family of enzymes, and accumulated at specific subcellular compartments may be responsible for the HCS-induced EC dysfunction. In summary, Cer act as mechanotransducers that connect hypertensive stretch to EC inflammation and apoptosis, and promote CVD.

Graphical Abstract

NEW & NOTEWORTHY

Human vascular endothelial cell (EC) exposure to high-magnitude (“hypertensive”) cyclic stretch (HCS), via alterations in sphingolipid (SPL), specifically ceramide (Cer), metabolism, enhanced their sensitivity to proinflammatory stimuli and compromised their viability, compared to physiologic-magnitude (“normotensive”) cyclic stretch (NCS). Pharmacological inhibitors of the de novo synthesis and sphingomyelin (SM) hydrolysis pathways protected ECs from the harmful effects of HCS, suggesting that these Cer-generating pathways may be appropriate targets for preventing vascular endothelial injury in early hypertension.

INTRODUCTION

Hypertension, the most important contributor to the development of cardiovascular disease (CVD) and heart failure (HF), has been associated with vascular endothelial cell (EC) dysfunction and vascular damage [1–3]. In an effort to mimic the in vivo early stages of hypertension prior to vessel remodeling, wall stiffening, and reduced distensibility, others have exposed cultured human ECs from different vascular beds to prolonged cyclic stretch (CS) of high magnitude (15–20% elongation at 1 Hz). These studies showed that “hypertensive” CS (HCS) causes an increase in EC reactive oxygen species (ROS) production, cytoskeletal remodeling, activation of tyrosine kinase signaling pathways and transcription factors, expression of cytokines and EC adhesion molecules for leukocytes, and cell apoptosis, compared to normotensive CS (NCS; 5% elongation at 1 Hz) [4–8]. Furthermore, increasing the intraluminal pressure in isolated arterioles from human adipose tissue was found to shift the mediator of flow-induced vasodilation (FIV) from nitric oxide (NO) to hydrogen peroxide (H₂O₂), suggesting that the hypertension-induced ROS production reduces NO bioavailability, a critical event in endothelial dysfunction and CVD development [9, 10].

Sphingolipids (SPLs) are a class of bioactive lipids that regulate many cellular functions, including survival, migration, proliferation, differentiation, and immune responses [11–13]. The key bioactive SPLs known to play an important role in EC function are ceramides (Cer) and sphingosine-1-phosphate (Sph-1P or S1P) [14–18]. In both humans and rats, hypertension has been associated with increased plasma Cer levels compared to normotensive controls [19]. In addition, hypertensive rats were shown to have increased Cer levels in arterial tissues, suggesting that Cer may contribute to vascular injury in hypertension [19]. The latest clinical trials have utilized the plasma concentrations of certain long-chain Cer species or the total Cer:S1P ratio as biomarkers for predicting CVD risk and adverse outcomes, besides the traditional biomarkers of increased plasma low density lipoprotein-cholesterol (LDL-C) and triglycerides [20–22]. Specifically for ECs, when cultured ECs were incubated with palmitate, a substrate for Cer biosynthesis, the increased intracellular Cer levels were found to trigger EC inflammation via NADPH oxidase-mediated ROS production, phosphatase 2A activation, and decreased endothelial NO synthase (eNOS) activity and NO bioavailability, supporting the causal role of Cer in EC dysfunction [23, 24]. Similarly, cultured EC incubation with an exogenous lipophilic synthetic Cer analog was shown to increase biosynthesis of endogenous long-chain Cer species leading to EC apoptosis [25]. Importantly, endogenous Cer accumulation in arterioles from coronary artery disease patients was found to mediate endothelial dysfunction, as demonstrated by the shift in the FIV mediator from NO to H₂O₂ [26].

SPLs are distributed to different subcellular membranes, where they undergo further metabolism, and, hence, the subsequent signaling is majorly determined by their compartmentalization [27, 28]. Cer are generated through three main pathways: a) The de novo synthesis pathway on the cytosolic side of the smooth endoplasmic reticulum (ER) membrane synthesizes Cer from serine and palmitoyl-CoA via a series of enzymatic reactions catalyzed initially by serine palmitoyltransferase (SPT) and further downstream by dihydroceramide desaturase (DES), prior to its transfer to the Golgi apparatus and subsequent participation into pathways to produce sphingomyelin (SM) and other complex SPLs, b) the SM hydrolysis pathway that hydrolyzes SM to Cer via either neutral sphingomyelinase (nSMase) in the inner leaflet of the plasma membrane or acid sphingomyelinase (aSMase) in the endolysosomal compartment, and c) the SPL salvage pathway primarily in lysosomes, where complex SPLs, including SM, are degraded into Sph that can then regenerate Cer in the ER [11, 29, 30]. Although it is generally accepted that, at least certain, Cer species are responsible for endothelial dysfunction in hypertension and other cardiometabolic disorders [19, 24, 26, 31], specific information on Cer metabolism changes in ECs exposed to HCS and the relative contribution of different Cer synthesis pathways in HCS-induced EC dysfunction is currently lacking. We hypothesized that human EC exposure to HCS vs. NCS may have differential effects on cellular and subcellular Cer metabolism, and that pharmacological interventions that inhibit key enzymes in Cer synthesis pathways may prevent the dysfunctional phenotype of HCS-exposed ECs.

Our study showed that 24 h of HCS increases human EC oxidative stress and inflammatory gene and protein expression compared to 24 h of NCS. Anti-Cer immunostaining showed significantly higher Cer accumulation at both a perinuclear region, where ER density is highest, and a peripheral region of HCS-exposed ECs compared to NCS-exposed ECs, despite the fact that whole-cell lipidomics analysis did not detect a significant difference in Cer levels between the two CS profiles. Furthermore, there was a significant increase in whole-cell SM levels and in the levels of several long-chain Cer released into the extracellular space following HCS compared to NCS, suggesting that differential and spatially-defined metabolic reprogramming takes place in ECs depending on the magnitude of CS. Pharmacological inhibitors of key enzymes in the de novo Cer synthesis and the SM hydrolysis pathways, i.e., SPT, DES, nSMase, and aSMase, were found to significantly inhibit the HCS-induced increases in EC inflammation and apoptosis compared to NCS. Our findings collectively suggest that certain Cer pools, specifically those at the plasma membrane and endolysosomes that are generated via the SMase family of enzymes, may be responsible for the increased Cer release into the extracellular space and the endothelial dysfunction in early hypertension. Identifying candidate molecular targets in Cer biosynthesis may pave the way for new pharmacological interventions to delay or inhibit CVD development in hypertensive patients.

MATERIALS AND METHODS

EC Culture and Exposure to CS

Pooled primary human umbilical vein ECs (HUVECs; Lonza, C2519A) were cultured in endothelial growth media-2 (EGM-2; Lonza, CC-3162) at 37 °C and humidified 5% CO₂/95% air. ECs at passages 3–7 were seeded at 6–8×10⁵ cells/well onto collagen I-coated flexible-bottomed BioFlex^® 6-well culture plates (Flexcell International, BF-3001C) (Fig. 1, A and B). At 24 h after seeding, 2 ml of fresh EGM-2 were added to confluent EC monolayers in each well. ECs were exposed to CS using the FX-6000T^™ Flexcell Tension system (Flexcell International) equipped with a 25-mm BioFlex loading station designed to provide equiaxial strain across the membrane surface. ECs were exposed to CS at either 5% elongation (NCS) or 15% elongation (HCS) and duration of either 4 or 24 h (Fig. 1C). Equiaxial strain (% elongation) was determined by videos of motion of dots made by a marker on the membrane, and confirmed by the vendor. ECs maintained under static conditions were seeded onto identical plates as those exposed to CS, and kept in the incubator. In some experiments, ECs were preincubated for 1 h prior and throughout CS or static exposure with the following specific pharmacological inhibitors: the SPT inhibitor myriocin (1 μM; Sigma-Aldrich, M1177), the DES inhibitor GT-11 (10 μM; Avanti Research, 857395), the nSMase, in particular nSMase2, inhibitor GW4869 (1 μM; Cayman Chemical, 13127), or the aSMase inhibitor ARC39 (C10 bisphosphonate; 5 μM; Avanti Research, 860480). The inhibitor concentrations were selected from a range of concentrations previously shown to significantly inhibit enzyme activity and signaling in ECs or epithelial cells exposed to cytokines and other proapoptotic stimuli, and, in the case of GW4869, to significantly inhibit the nSMase2-mediated release of small extracellular vesicles (sEVs; <100 nm in diameter, also called exosomes) from activated HUVECs [32–36]. In addition, a range of concentrations for each compound were tested for their ability to inhibit the upregulation of EC receptors, vascular cell adhesion molecule-1 (VCAM-1) and intercellular adhesion molecule-1 (ICAM-1), in HUVECs stimulated by tumor necrosis factor-α (TNF-α, 0.5 ng/ml for 4h; ThermoFisher Scientific, A42551), as described under EC Protein Expression Analysis by Western Blotting. The concentration that showed maximal inhibition of TNF-α-induced inflammation without causing any cytotoxicity was chosen for testing.

Fig. 1.

Schematic of Flexcell Tension system and CS profiles. A and B: The FX-6000T^™ Flexcell Tension system applies uniform equiaxial CS to cultured EC monolayers on collagen I-coated flexible membranes at the bottom of BioFlex 6-well plates by using a cylindrical loading post beneath each membrane and cycles of vacuum and positive pressure. C: Elongation (%) of the membrane, and the overlaying EC monolayers, vs. time profiles representing static (0% elongation), NCS (5% elongation at 1 Hz), and HCS (15% elongation at 1 Hz) conditions. Only 10 s are shown, but EC exposure to static, NCS, and HCS was continuous for up to 24 h. Created in BioRender. Manzar, Z (2026).

EC Protein Expression Analysis by Western Blotting

HUVECs at the end of CS or static period were collected from the BioFlex^® 6-well culture plates via trypsinization and centrifugation (500g, 5 min, 4 °C). Cell pellets were washed twice with cold phosphate-buffered saline (PBS) and lysed by adding RIPA buffer (1:10 dilution; Abcam, ab156034) and Halt^™ protease and phosphatase inhibitor cocktail, EDTA-free (1:100 dilution; ThermoFisher Scientific, 78441) and incubated for 15 min on ice. Lysates were homogenized via sonication and centrifuged at 14,000g for 10 min. The supernatant was collected, and protein levels were quantified using the Pierce^™ BCA Protein Assay kit (ThermoFisher Scientific, 23227) according to manufacturer’s instructions. Equal amounts of protein for each sample were added to Abby protein analysis plates (12–230 kDa separation module; ProteinSimple, SM-W001). Total protein for each well was quantified using the Replex^™ (ProteinSimple, RP-001) and total protein detection module (ProteinSimple, DM-TP01). All experiments were conducted under reducing conditions. Quantification of chemiluminescence area under the curve and total protein normalization was done with Compass for Simple Western v7.1. Primary antibodies used were: 4-hydroxynonenal (4-HNE) (1:30; R&D Systems/Biotechne, MAB3249; RRID:AB_664165), VCAM-1 (1:100; R&D Systems/Biotechne, MAB8091; RRID:AB_3659023), and ICAM-1 (1:20; Proteintech, 10831–1-AP; RRID:AB_2264494). At the end of CS or static period and prior to collection for VCAM-1 and ICAM-1 Western blotting, ECs were incubated, or not, statically with TNF-α (0.5 ng/ml) for an additional 4 h.

EC Protein Carbonyl Colorimetric Assay

HUVECs at the end of CS or static period were collected from the BioFlex^® 6-well culture plates via trypsinization and centrifugation (1500g, 10 min, 4 °C). Cell pellets were lysed by adding RIPA buffer (1:10 dilution) and Halt^™ protease and phosphatase inhibitor cocktail, EDTA-free (1:100 dilution) and incubated for 15 min on ice. Lysates were homogenized via sonication on ice and centrifuged at 10,000g for 15 min at 4 °C. Protein carbonyl content in the supernatant was quantified spectrophotometrically at 370 nm following the manufacturer’s protocol for the Protein Carbonyl Colorimetric Assay Kit (Caymen Chemical, 10005020). Protein from the remaining supernatant was quantified using the Pierce^™ BCA Protein Assay kit to express carbonyl content as nmoles/mg of protein.

EC Gene Expression Analysis by RT-qPCR

Total RNA was extracted from HUVECs at the end of CS or static period using the RNeasy mini kit (Qiagen, 74104). Reverse transcription (RT) was performed using the iScript cDNA Synthesis kit (Bio-Rad, 1708890). Quantitative real-time PCR (qPCR) was performed in duplicate using the iTaq Universal SYBR Green Supermix (Bio-Rad, 1725120) and the CFX Connect Real-Time PCR Detection System (Bio-Rad, 1855201). The PCR program consisted of an initial step at 95 °C for 30 s, 40 cycles of denaturation at 95 °C for 3 s, annealing/extension at 60 °C for 20 s, followed by melting at a gradient from 65 to 95 °C. Gene expression was normalized to housekeeping genes GAPDH and ACTB. Normalized fold expression was calculated and plotted using the ΔΔCt method. All primer sequences were verified for specificity using Primer-BLAST (RRID:SCR_003095). Expression levels of only some genes, for which primers were available, were assessed using Droplet digital PCR (ddPCR) as previously [37]. Briefly, ddPCR reactions were prepared in 96-well plates using cDNA, ddPCR Supermix, gene-specific primers, and fluorescent probes following the manufacturer’s guidelines for the QX200 Droplet Digital PCR system (Bio-Rad; RRID:SCR_019707). Reaction mixtures were partitioned into water-in-oil droplets using the Bio-Rad Automated Droplet Generator and transferred to a new 96-well plate, which was subsequently sealed with heat-applied foil. Amplification of target transcripts and the reference gene hypoxanthine phosphoribosyltransferase 1 (HPRT1) was performed under the following thermal cycling conditions: 95 °C for 10 min for enzyme activation, 40 cycles of 94 °C for 30 s and 56 °C for 60 s for denaturation and annealing/extension, respectively, followed by a 98 °C hold for 10 min to terminate the reaction. Droplets were analyzed using the QX200 Droplet Reader, and data were processed with QuantaSoft software (Bio-Rad; RRID:SCR_025696) after applying fluorescence thresholds derived from negative controls. P values were calculated by comparing the relative expression of Cer metabolic enzyme genes compared to HPRT1 in all biological replicates. Primer sequences are listed in Table 1.

Table 1.

Primer sequences used for gene expression analysis.

Gene	Primers (5’ to 3’)		Method
VCAM-1	Forward	CAGTAAGGCAGGCTGTAAAAGA	qPCR
	Reverse	TGGAGCTGGTAGACCCTCG
ICAM-1	Forward	ATGCCCAGACATCTGTGTCC	qPCR
	Reverse	GGGGTCTCTATGCCCAACAA
MCP-1	Forward	CAGCCAGATGCAATCAATGCC	qPCR
	Reverse	TGGAATCCTGAACCCACTTCT
IL-6	Forward	ACTCACCTCTTCAGAACGAATTG	qPCR
	Reverse	CCATCTTTGGAAGGTTCAGGTTG
IL-8	Forward	ACTGAGAGTGATTGAGAGTGGAC	qPCR
	Reverse	AACCCTCTGCACCCAGTTTTC
ACTB	Forward	CCAACCGCGAGAAGATGA	qPCR
	Reverse	CCAGAGGCGTACAGGGATAG
GAPDH	Forward	CCACAGTCCATGCCATCAC	qPCR
	Reverse	CCACCACCCTGTTGCTGTA
SPTLC2	Forward	GAAATCCAACCACAACGACAC	ddPCR
	Reverse	AATGAAGACTCTCCAGTAGTGC
DEGS2	Forward	AATCCCAGACACCTTCAC	ddPCR
	Reverse	ACTGGATCACCTTCAATGTGG
CERS1	Forward	GCCTTCCACAACCTCCTG	ddPCR
	Reverse	AACTGGGTAACAAGCAGAGTC
CERS2	Forward	CACTGCGTTCATCTTCTACCA	ddPCR
	Reverse	GCTCTATCCTGCCTTCTTTGG
CERS5	Forward	CCGATTATCTCCCAACTCTCAA	ddPCR
	Reverse	GCCAATTATGCCAAGTATCAGC
CERS6	Forward	TGACTCCGTAGGTAAATACATAAAGG	ddPCR
	Reverse	CAATCAGGAGAAGCCAAGCA
SMPD1	Forward	GAGAGAGATGAGGCGGAGA	ddPCR
	Reverse	CTGGCTCTATGAAGCGATGG
SMPD2	Forward	GGATCTTCAACCTCAACTGCT	ddPCR
	Reverse	GAAGTCCTGCTCACTCCA
SGMS1	Forward	GTACAGATAGTCCCCACACAT	ddPCR
	Reverse	GCATTTCAACTGTTCTCCGAAG
SGMS2	Forward	TCCTACGAACACTTATGCAAGAC	qPCR
	Reverse	CCGGGTACTTTTTGGTGCCT
ASAH1	Forward	GCCAGTTATGACCCAGGTATC	ddPCR
	Reverse	TGTACTTCAATAGTAGCAGAAGACA
SPHK1	Forward	GTATGAATGCCCCTACTTGG	qPCR
	Reverse	AACACACCTTTCCCATCCT
SPHK2	Forward	AGCGTGGTAGCCACTTCAG	qPCR
	Reverse	GAGCAGTGTACCGATGCCA
HPRT1	Forward	GTATTCATTATAGTCAAGGGCATATCC	ddPCR
	Reverse	AGATGGTCAAGGTCGCAAG

EC Lipid Extraction

Following CS or static exposure, ECs were collected from the BioFlex^® 6-well culture plates via trypsinization and centrifugation. The media was aspirated, and each cell pellet was gently resuspended in 1 ml of cold PBS. A 62.5 μl aliquot was removed and lysed by adding RIPA buffer and Halt^™ protease and phosphatase inhibitor cocktail, EDTA-free, and quantified using the Pierce^™ BCA Protein Assay kit. The remaining suspension was centrifuged (500g, 5 min, 4 °C), PBS was aspirated, and the dry cell pellet was stored at −80 °C until lipid extraction was performed using the previously published protocol [38]. Briefly, cell pellets were resuspended in 1 ml of cold methanol and transferred to 1.5 ml microcentrifuge tubes. Each sample was sonicated on an ice block (40% power, 10 s on-10 s off for 30 s) followed by centrifugation (16,900g, 15 min, 4 °C). After centrifugation, 900 μl of the methanol extract was transferred to a labeled 1-Dram glass vial, and another 1 ml of cold methanol was added to the microcentrifuge tube. The solution was subjected to another round of sonication and centrifugation, followed by transfer of 900 μl of extract to the designated 1-Dram vial. Extracts were dried down under a nitrogen evaporator, resuspended in methanol spiked with internal standards C17-Cer, C17-GluCer, and C17-SM, and normalized based on their total protein content.

EC Pellet Lipid Analysis by LC-QTOF-MS

Targeted lipid analysis was conducted using liquid chromatography coupled to quadrupole time-of-flight mass spectrometry (LC-QTOF-MS) on an Agilent 1260 high-performance liquid chromatography system interfaced with an Agilent 6530 Jet Stream electrospray ionization QTOF mass spectrometer. Mass spectral data were acquired over an m/z range of 50–1700. Chromatographic separation was achieved using previous protocols [38]. Three biological replicates per condition, NCS and HCS, were analyzed in both positive and negative electrospray ionization modes. Quantitative analysis was carried out using Agilent MassHunter Quantitative Analysis software (RRID:SCR_015040), into which the raw LC-QTOF-MS data were imported. The m/z ratios were extracted, and peak areas corresponding to each lipid species were manually integrated. The relative abundance was calculated by normalizing each lipid species to the raw abundance of the internal standard. The internal standard corrected abundances were then normalized to the internal standard-corrected abundance of the respective NCS control. Data was plotted as fold change relative to the NCS control. Total lipids were quantified by adding the internal standard corrected abundance for each lipid species per experimental condition and normalizing to the total lipids of the NCS control.

EC Conditioned Media Lipid Analysis by HPLC-MS/MS

Following exposure to NCS or HCS, both ECs and media were collected from the BioFlex^® 6-well culture plates. ECs were collected via trypsinization, centrifugation (500g, 5 min, 4 °C), and resuspension in 1 ml of cold PBS. After two centrifugations with PBS to clean the pellet, cells were lysed by adding RIPA buffer and Halt^™ protease and phosphatase inhibitor cocktail, EDTA-free, and total protein was quantified using the Pierce^™ BCA Protein Assay kit. The conditioned media was centrifuged (500g, 5 min, 4 °C) to remove any floaters (negligible, since the monolayers were intact at the end of CS), and stored at −80 °C until processed. Concentrations of Cer, S1P, and some additional SPLs in the media were measured at the Medical University of South Carolina (MUSC) Lipidomics Shared Resource using HPLC-MS/MS, as previously described [39–41], normalized to EC total protein, and expressed in picomoles/mg protein. Total Cer concentration was calculated by adding the concentrations of all Cer species for each experimental condition.

EC Survival Analysis by Flow Cytometry

EC apoptosis and necrosis were measured by staining with Invitrogen^™ FITC-conjugated Annexin-V (ThermoFisher Scientific, A13199) and Propidium Iodide Ready Flow^™ Reagent (ThermoFisher Scientific, R37169), respectively. Briefly, ECs exposed to static, NCS, or HCS for 24 h were collected via trypsinization and centrifugation. The pellet was resuspended in 1 ml of 1X Annexin binding buffer (140 mM NaCl, 2.5 mM CaCl₂, 10 mM HEPES) at a concentration of 1–5×10⁵ cells/ml and left to incubate for 45 min. Cells were stained with 1.75 μl of FITC-conjugated Annexin V (determined by dye titration experiments) on ice for 15 min protected from light. Cells were washed twice with 1 ml of 1X Annexin binding buffer to remove excess of Annexin V-FITC by centrifugation (500g, 5 min, 4 °C). The pellet was resuspended in 200 μl of 1X Annexin binding buffer, stained with 5 μl of propidium iodide on ice for 5–15 min protected from light, and the stained cells were analyzed at a flow cytometer (BD Fortessa).

EC Immunofluorescence Against Cer

ECs exposed to NCS or HCS were washed with PBS and fixed with 4% formaldehyde (Sigma Aldrich, NC1810005) in PBS for 10 min at 37 °C. The fixed cells were washed three times with PBS. The flexible bottoms of BioFlex^® 6-wells were cut, and segments of membrane were placed in 8-well μ-slides (ibidi, 50-305-795) to reduce antibody volumes required. Cells were permeabilized with 150 μl of 0.1% Triton X-100 (ThermoFisher, AAA16046AE) in PBS and incubated for 10 min at room temperature. The permeabilization solution was removed and 150 μl of 3% BSA blocking solution (Blocker^™ BSA; ThermoFisher, 37525) was added to each well and incubated for 30 min at 37 °C. The blocking solution was removed and 1:200 anti-C16–24 Cer antibody, previously shown to specifically bind C16:0, C18:0, C20:0, and C24:0/C24:1 Cer [42], diluted in 0.1% BSA blocking solution was added and incubated overnight at 4 °C. Each well was washed three times with PBS and incubated with secondary antibody (goat anti-rabbit IgG Alexa Fluor^™ Plus 488; ThermoFisher Scientific, A32731; RRID:AB_2633280) in 0.1% BSA blocking solution. The secondary antibody solution was removed and 150 μl of 1:2,000 Hoechst dye (ThermoFisher, H3570; RRID:AB_3675235) in PBS was incubated for 10 min. The solution was removed, and membranes were washed with PBS. Membranes were inverted and placed on coverslips (cells facing downward). Fluorescence images were captured at 63x in a DMi8 inverted epifluorescence microscope (Leica) using LAS X software followed by Thunder 3D imaging (RRID:SCR_016556), as described previously [43]. Images were processed using CellProfiler cell imaging analysis software and three objects were identified: The anti-C16–24 stain (CeramideStain), the perinuclear region, and the peripheral region. A secondary mask corresponding to the CeramideStain object was made using the nucleus and the anti-C16–24 fluorescence signal around the nucleus, and was used to define (most of) the area of each cell. The perinuclear region was chosen as an area 5 μm wide surrounding the nucleus, where most of the ER, the Golgi apparatus, and perinuclear mitochondria are located [44, 45], and the mean fluorescence intensity (MFI) of the anti-C16–24 fluorescence signal in the perinuclear region was measured. The peripheral region, where peripheral mitochondria, endolysosomes, and the plasma membrane are located [44, 45], was defined by removing the perinuclear region from the CeramideStain object and the MFI was calculated over the remaining cell area. Outlines of the perinuclear region and CeramideStain object were overlaid on top of the original images. MFI for each of the three objects was measured and plotted for each cell for a total of 120 cells in images collected from 3 independent runs at 24 h of NCS and HCS.

Statistical Analysis

Results were expressed as mean±SEM. Student’s unpaired t-test was used to compare the means of two groups. Statistically significant differences among the means of three or more groups were determined using one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparisons post hoc test. Statistical symbols were as follows: *, P≤0.05; **, P≤0.01; ***, P≤0.001; and ****, P≤0.0001. Statistical analyses were performed using GraphPad Prism 9 software (RRID:SCR_002798) on a PC.

RESULTS

HCS Induces Oxidative Stress and Primes ECs for Inflammation

ECs exposed to static, HCS, or NCS for 24 h were analyzed for oxidative stress and inflammatory gene and protein expression. Anti-HNE Western blotting showed HCS-exposed ECs to have significantly higher levels of 4-HNE protein adducts, a product of lipid peroxidation and a biomarker for oxidative stress, compared to NCS-exposed ECs (Fig. 2A and S1). 4-HNE expression in static ECs tended to be higher, but not significantly different, from that in NCS-exposed ECs (Fig. 2A and S1). A protein carbonyl assay showed HCS-exposed ECs to have significantly more protein carbonylation, an irreversible form of protein oxidation caused by oxidative stress, compared to both NCS-exposed and static ECs (Fig. 2B). The HCS-induced oxidative environment was accompanied by a significantly increased expression of inflammatory genes encoding EC adhesion molecules for leukocytes, VCAM-1 and ICAM-1, and the cytokine monocyte chemoattractant protein-1 (MCP-1) compared to the corresponding gene expression in NCS (Fig. 2C). These findings suggest that EC exposure to the high-magnitude tensile forces of prolonged HCS is sufficient for activating oxidative and inflammatory signaling. However, gene expression levels for interleukin-6 (IL-6) and IL-8 in HCS were not any different from those in NCS, and for some genes, their expression levels in static ECs were comparable or even higher than those in NCS or HCS (Fig. 2C).

Fig. 2.

HCS caused increased EC oxidative stress and the expression of several inflammatory genes compared to NCS, and also primed ECs for inflammatory protein expression in response to subsequent cytokine stimulation. A: Western blot analysis of 4-HNE adducts at histidine residues in lysates of ECs exposed to 24 h of static, NCS, or HCS (protein loaded per lane at 0.2 mg/ml). Quantification of 4-HNE staining showed significantly higher lipid peroxidation, an index of oxidative stress, in HCS-exposed ECs compared to NCS-exposed ECs. B: Protein carbonyl content was measured using a protein carbonyl colorimetric assay based on the reaction between 2,4-dinitrophenylhydrazine and protein carbonyls to produce a colored protein-hydrazone product. Absorbance measurements showed significantly higher protein carbonylation, an index of oxidative stress, in HCS-exposed ECs compared to NCS-exposed and static ECs. C: RT-qPCR analysis of gene expression for adhesion molecules and cytokines showed HCS to significantly upregulate VCAM-1, ICAM-1, and MCP-1 expression compared to NCS (24 h). D and E: Preconditioning with HCS amplified, whereas preconditioning with NCS inhibited, the TNF-α (0.5 ng/ml, 4 h post-treatment)-induced increases in VCAM-1 (protein loaded per lane at 0.2 mg/ml) (D) and ICAM-1 (protein loaded per lane at 0.15 mg/ml) (E) protein expression. Data are presented as mean±SEM from n=3 independent experiments. * P≤0.05; ** P≤0.01; *** P≤0.001; **** P≤0.0001.

When EC lysates were analyzed for VCAM-1 and ICAM-1 protein expression, it was found that 24 h HCS did not elicit inflammatory cell adhesion molecule protein changes (Fig. 2, D and E). However, when ECs were exposed to static, NCS, or HCS followed by static post-treatment with a low concentration of TNF-α (0.5 ng/ml, 4 h), HCS-preconditioned ECs showed significantly higher TNF-α-induced VCAM-1 and ICAM-1 protein expression compared to that in either static or NCS-preconditioned ECs (Fig. 2, D and E). Besides an observed amplification of the EC inflammatory response to TNF-α by HCS preconditioning, NCS preconditioning was found to exert a protective effect in inflammatory protein expression compared to either static or HCS preconditioning (Fig. 2, D and E). The concentration of TNF-α used for post-CS treatment was determined based on submaximal expression of VCAM-1 and ICAM-1 in static ECs treated with increasing concentrations of TNF-α (0.05–5 ng/ml) for 4 h (Fig. S2). These findings suggest that HCS primes/sensitizes ECs to upregulate pro-inflammatory signaling in response to subsequent chemical stimuli, whereas NCS protects ECs from developing the dysfunctional phenotype.

HCS Induces Time-Dependent SPL Metabolism Changes Relative to NCS

Briefly, SPL metabolism unfolds around the synthesis and catabolism of Cer species (Fig. 3A), which are fatty acid derivatives of sphingoid bases, saturated or mono-unsaturated with carbon chains ranging from 14 to 26 atoms [11, 14, 18]. Cer is de novo synthesized on the ER membrane via condensation of serine and palmitoyl-CoA mediated by SPT (Fig. 3A). The product 3-ketodihydrosphingosine is reduced to dihydrosphingosine (dhSph), the substrate of Cer synthases (CerS) forming dihydroceramide (dhCer), which is then converted to Cer by DES. Cer is transferred to the Golgi, via CERT or vesicular trafficking, and gets converted to either SM by SM synthase (SMS), ceramide-1-phosphate by CERK, or glycosylceramide (and complex glycosphingolipids) by GCS. SM is transferred to the inner leaflet of the plasma membrane and to the endolysosomal compartment, where it is hydrolyzed to Cer by nSMase and aSMase, respectively, via the SM hydrolysis pathway. SM and other complex SPLs in late endosomes and lysosomes break down into Sph that, upon transport to the ER, regenerates Cer via the SPL salvage pathway. Cer is converted into Sph by CDase, that is metabolized into S1P by SphK. S1P is primarily released into the extracellular space, but it can also be hydrolyzed to Sph or degraded into hexadecenal and ethanolamine phosphate (Fig. 3A).

Fig. 3.

HCS caused a significant increase in EC SM, but not Cer, levels compared to NCS at 24 h (based on lipidomics analysis of EC pellets). A: Overview of SPL/Cer metabolism, described in detail in the text. Created in BioRender. Manzar, Z (2026). B and C: LC-MS analysis of Cer in EC pellets at 4 (B) and 24 h (C) of HCS, plotted as fold change relative to NCS. D: Total Cer levels at the end of 4 and 24 h CS did not differ between HCS- and NCS-exposed ECs at either time point, although several long-chain Cer trended toward significance. E-G: LC-MS analysis of SM in EC pellets at 4 (E) and 24 h (F) of HCS, plotted as fold change relative to NCS. At 24 h, but not at 4 h, HCS-exposed ECs exhibited significantly higher levels of every long-chain SM (F) and total SM (G) compared to NCS-exposed ECs. H and I: Total Cer (H) and total SM (I) levels at 24 h of NCS and HCS (fold change relative to NCS) showed no significant changes when followed, or not, by TNF-α post-treatment (0.5 ng/ml, 4 h). Data are presented as mean±SEM from n=7 independent experiments (4 h) and n=3 independent experiments (24 h). * P≤0.05; ** P≤0.01; *** P≤0.001.

To examine the effect of CS on SPL metabolism, whole-cell pellets from ECs exposed to NCS or HCS for 4 or 24 h were subjected to targeted lipidomics analysis. Cer levels remained relatively unchanged with only a few Cer species reaching (C16:1-Cer) or approaching (C14:0-Cer, C24:0-Cer, C24:1-Cer, C26:1-Cer) significant difference compared to NCS at 4 and 24 h (Fig. 3, B and C, and Table S1). As a result, there was no significant difference in total Cer between HCS (plotted as fold change relative to NCS) and NCS at either time point (Fig. 3D). Differences in SM remained modest at 4 h, with only a significant decrease in C18:0-SM (Fig. 3E, and Table S2). However, major SPL metabolic remodeling occurred at 24 h, with HCS significantly increasing the levels of every long-chain SM species (Fig. 3F, and Table S2), resulting in significantly higher total SM levels in HCS compared to NCS at 24 h (Fig. 3G). To assess whether the post-treatment with TNF-α (0.5 ng/ml, 4 h), employed only when studying VACM-1 and ICAM-1 protein expression, was causing additional changes to SPL metabolism, ECs were exposed to NCS or HCS followed, or not, by static incubation with TNF-α, and pellets were processed for lipidomics analysis. The low-dose TNF-α treatment had no additional effect on individual Cer and SM abundances in HCS-exposed ECs compared to NCS-exposed ones (Fig. S3, and Table S3). As a result, total Cer levels were the same in HCS-exposed ECs with or without TNF-α post-treatment and total SM levels maintained their 1.5- to 2-fold increase in HCS independently of TNF-α post-treatment (Fig. 3, H and I), suggesting that CS alone is responsible for any alterations in SPL levels.

To better understand the regulation of SPL metabolism by CS, mRNA levels for genes that encode key enzymes in SPL metabolic pathways (Fig. 3A) were measured at 12 h of CS. HCS caused significant decreases in expression of genes that encode for enzymes in the de novo synthesis pathway (SPTLC2, CERS1, CERS5), but it also caused significantly lower expression of ASAH1 that encodes acid ceramidase (aCDase), the CDase isoform in lysosomes that hydrolyzes Cer to Sph [46], compared to NCS (Fig. 4). Interestingly, HCS-exposed ECs had significantly higher expression of SGMS2, the SMS isoform on the plasma membrane that converts Cer to SM [46], compared to NCS (Fig. 4), which may explain the increased SM levels in HCS (Fig. 3F). Collectively, these data demonstrate that HCS triggers reprogramming of EC SPL metabolism that favors SM abundance compared to NCS.

Fig. 4.

HCS and NCS caused differential expression of genes that encode for key enzymes in SPL metabolism at 12 h. HCS-exposed ECs had significantly higher mRNA levels of the gene that expresses SMS2, and significantly lower mRNA levels of the genes that express SPT, CERS1, CERS5, and CDase, compared to NCS-exposed ECs. Data are presented as mean±SEM from n=3 independent experiments. * P≤0.05; ** P≤0.01.

HCS Distributes Cer to the Cell Periphery and Increases its Extracellular Release

Immunofluorescence with an anti-C16–24 Cer antibody was performed to determine the subcellular distribution of Cer in ECs exposed to either 4 or 24 h of CS; ECs exposed to 24 h of CS were either in the absence or presence of the specific SPT inhibitor myriocin (1 μM). Myriocin has been shown to lower Cer synthesis, and, hence, decrease the concentrations of long-chain Cer and total Cer, at concentrations varying from 100 nM to 10 μM in different cell types, including ECs and epithelial cells, activated by various stimuli [47–50]. At either 4 or 24 h, HCS-exposed ECs showed higher Cer staining compared to NCS-exposed ECs (Fig. 5, A and B). Myriocin reduced Cer staining at 24 h of HCS and NCS; the myriocin effect was more apparent in HCS (Fig. 5C). Digital processing of Cer immunofluorescence images (40 cells at 4 h and 120 cells at 24 h from 3 independent experiments) revealed that the MFI per cell was significantly higher in HCS compared to NCS at either 4 or 24 h, and myriocin significantly reduced the MFI in both NCS and HCS at 24 h (Fig. 5D). At 24 h of NCS, Cer immunostaining was almost exclusively at the perinuclear region (a 5 μm wide extension around the nucleus), whereas it was distributed more to the peripheral region (the rest of the cell area) at 24 h of HCS (Fig. 5E). Myriocin reduced both the perinuclear and peripheral Cer staining of HCS-exposed ECs to levels comparable to NCS-exposed ECs (Fig. 5E). Quantification of Cer immunofluorescence at the perinuclear and peripheral regions revealed that HCS significantly increased Cer levels at both regions compared to the corresponding regions in NCS, and myriocin significantly decreased Cer accumulation at both subcellular locations in HCS (Fig. 5F). Although to a lesser extent, myriocin also decreased Cer levels at both regions in NCS; the effect was significant at the peripheral region (Fig. 5F). These findings suggest enhanced Cer de novo synthesis and trafficking to the cell periphery under HCS compared to NCS.

Fig. 5.

HCS significantly increased cellular Cer levels at 4 and 24 h compared to NCS; Cer immunostaining levels were significantly increased at both the perinuclear region and cell periphery at 24 h compared to NCS (based on Cer immunofluorescence); HCS effects were inhibited by myriocin. A-C: Fluorescence microscopy images of Cer immunostaining of NCS- and HCS-exposed ECs at 4 h (A), 24 h (B) and 24 h in the presence of myriocin (C). The white rectangles in NCS and HCS images (B and C) indicate areas selected for magnification. D: Quantification of MFI (at least 40 ECs from n=3 independent experiments) showed significantly higher Cer levels in HCS compared to NCS at 4 and 24 h; myriocin significantly reduced Cer levels in NCS- and HCS-exposed ECs at 24 h. E: Selected areas from images in B and C are shown under higher magnification including the masks used to define a perinuclear region (extending 5 μm around the nucleus) in each cell. F: Quantification of Cer MFI (at least 120 ECs from n=3 independent experiments) showed significantly higher Cer levels at both the perinuclear and peripheral regions compared to NCS and compared to ECs exposed to HCS in the presence of myriocin. Data are presented as violin distributions with individual data points and internal lines denoting the median and upper and lower quartiles. ** P≤0.01; *** P≤0.001; **** P≤0.0001.

When Cer concentrations were measured in EC conditioned media, HCS-exposed ECs released significantly more long-chain Cer species C18:0-Cer, C22:0-Cer, and C24:1-Cer into the media, and there was an increasing trend in total released Cer, compared to NCS-exposed ECs at 24 h (Fig. 6A and B). DhSph and S1P concentrations were also significantly higher in the media from HCS-exposed ECs compared to NCS-exposed ECs (Fig. 6C). Collectively, our findings suggest that HCS promotes Cer accumulation at the cell periphery and higher secretion of certain long-chain Cer into the extracellular space compared to NCS.

Fig. 6.

HCS caused a significant increase in the release of certain long-chain Cer, and an increased trend in released total Cer, in conditioned media compared to NCS at 24 h. A: Conditioned media from HCS-exposed ECs had significantly higher concentrations of several long-chain Cer compared to conditioned media from NCS-exposed ECs. B: Total Cer levels in media from HCS-exposed ECs tended to be higher compared to those in media from NCS-exposed ECs. C: HCS-exposed ECs released significantly greater amounts of dhSph and Sph-1P in the extracellular media compared to NCS-exposed ECs. Data are presented as mean±SEM from n=3 independent experiments. * P≤0.05; ** P≤0.01.

Cer Synthesis Pathways Mediate HCS-Induced EC Inflammation and Apoptosis

To assess the role of Cer in the expression of EC surface inflammatory markers VCAM-1 and ICAM-1 under HCS, ECs were treated with specific pharmacological inhibitors of key enzymes in Cer generating pathways during 1 h of static preincubation, 24 h of either NCS or HCS, and 4 h of static TNF-α (0.5 ng/ml) stimulation. Each of the inhibitors at the concentrations tested, i.e., the SPT inhibitor myriocin (1 μM), the DES inhibitor GT-11 (10 μM), the nSMase2 inhibitor GW4869 (1 μM), and the aSMase inhibitor ARC39 (5 μM), has been shown by others to inhibit the corresponding enzyme activity and intracellular signaling in activated ECs or epithelial cells [32–36]. In addition, we chose the concentration of each inhibitor based on maximal inhibition of the TNF-α (0.5 ng/ml, 4 h)-induced upregulation of VCAM-1 and/or ICAM-1 expression in static ECs without any cytotoxicity (Fig. S6, VCAM-1 blots for all inhibitors, and ICAM-1 blot for ARC39). More specifically, myriocin (1 μM) and GT-11 (10 μM) were employed to block the de novo synthesis pathway (Fig. 7A). GT-11 was found to significantly inhibit the TNF-α-induced increases in VCAM-1 and ICAM-1 expression in HCS-preconditioned ECs, bringing their expression down to similar levels with those in NCS-preconditioned ECs, whereas myriocin had no significant effect (Fig. 7, B and C, and Fig. S4). Cer generation via the SM hydrolysis pathway was blocked using GW4869 (1 μM) and ARC39 (5 μM) (Fig. 7D). GW4869 was found to significantly inhibit the TNF-α-induced increases in VCAM-1 and ICAM-1 expression in HCS-preconditioned ECs, bringing expression down to similar levels with those in NCS-preconditioned ECs, whereas ARC39 significantly inhibited only ICAM-1 expression (Fig. 7, E and F, and Fig. S5).

Fig. 7.

HCS, but not NCS, exposure primed ECs for enhanced TNF-α-induced inflammation through Cer generated via the de novo synthesis and SM hydrolysis pathways. A: De novo Cer synthesis was inhibited using the SPT inhibitor myriocin and the DES inhibitor GT-11. Created in BioRender. Manzar, Z (2026). B and C: Western blots showed that GT-11 (10 μM) abolished the TNF-α-induced upregulation of VCAM-1 (B) and ICAM-1 (C) expression in HCS-preconditioned ECs, whereas myriocin (1 μM) had no significant effect. D: Cer generation via the SM hydrolysis pathway was inhibited using the nSMase2 inhibitor GW4869 and the aSMase inhibitor ARC39. Created in BioRender. Manzar, Z (2026). E: Western blots showed that GW4869 (1 μM) abolished the TNF-α-induced increase in VCAM-1 expression in HCS-preconditioned ECs. F: Western blots showed that either GW4869 (1 μM) or ARC39 (5 μM) abolished the TNF-α-induced ICAM-1 upregulation in HCS-preconditioned ECs. Data are presented as mean±SEM from n=3 independent experiments. * P≤0.05; ** P≤0.01; *** P≤0.001.

Flow cytometric analysis of Annexin-V and propidium iodide labeling of ECs at the end of 24 h of CS showed HCS to induce significantly higher EC apoptosis compared to either static or NCS, whereas NCS significantly lowered apoptosis compared to static (Fig. 8, A and B). Using each of the specific pharmacological inhibitors to block the de novo synthesis and SM hydrolysis pathways significantly decreased the HCS-induced apoptosis, reducing it to similar levels with those in NCS-exposed ECs (Fig. 8, C and D). Despite its inability to inhibit inflammation (Fig. 7, B and C), myriocin significantly protected from HCS-induced apoptosis, as did GT-11, GW4869, and ARC39. The findings from the inflammatory marker expression and cell viability measurements collectively identify Cer as key effector molecules that link HCS to EC dysfunction, specifically to inflammation and apoptosis.

Fig. 8.

Cer generation is responsible for the increased apoptosis in HCS-exposed ECs compared to NCS-exposed ECs. A and B: EC apoptosis and necrosis were assessed using an Annexin V/propidium iodide flow cytometry-based assay following a 24 h exposure to static, NCS, or HCS. Characteristic flow cytometry data are shown in (A). Q1: Necrotic cells, Q2: late apoptotic and early necrotic cells, Q3: early apoptotic cells, Q4: viable cells. Total apoptotic cells were quantified by adding the percentages of ECs in Q2 and Q3 (B). C and D: Each of the Cer synthesis inhibitors, myriocin (1 μM), GT-11 (10 μM), GW4869 (1 μM), and ARC39 (5 μM), significantly inhibited the HCS-induced apoptosis, detected using flow cytometry (characteristic data shown in C and quantified in D). Data are presented as mean±SEM from n=3 independent experiments. * P≤0.05; ** P≤0.01; *** P≤0.001.

DISCUSSION

Using an in vitro model to mimic the mechanical environment of the arterial wall in early hypertension, the present study identified Cer as key mediators of EC dysfunction due to HCS. EC exposure to HCS triggered oxidative stress, inflammatory signaling, and apoptosis. These HCS-induced EC functional changes were accompanied by whole-cell and subcellular changes in SPL metabolism, including increased whole-cell SM levels, pronounced Cer accumulation at the cell periphery, and enhanced release of several long-chain Cer into the extracellular space. Importantly, pharmacological inhibition of key enzymes in the two major Cer generating pathways, the de novo synthesis and SM hydrolysis, inhibited the HCS-induced EC inflammation and apoptosis, suggesting that Cer are, at least in part, responsible for the dysfunctional EC phenotype observed under pathological mechanical stretch.

Our findings agree with prior work showing HCS to promote oxidative stress and inflammatory activation in ECs. Specifically, HCS has been shown to increase Ca²⁺ influx via stretch-activated channels, produce ROS from sources such as NADPH oxidase, xanthine oxidase, and the mitochondria, and lead to upregulation and transactivation of vascular endothelial growth factor receptor 2 (VEGFR2) followed by RhoA-mediated disassembly of adherens junctions/increased monolayer permeability and activation of nuclear factor (NF)-κB signaling [5–8, 51–53]. Increased HCS-induced ROS production in fibroblasts was previously shown to mediate the generation of 4-HNE [54], in agreement with our findings in HCS-exposed ECs (Fig. 2A). Increased protein carbonylation has also been shown in the aorta, heart, and kidneys of spontaneously hypertensive rats compared to their normotensive controls [55], consistent with our in vitro findings in HCS-exposed ECs compared to NCS-exposed ECs (Fig. 2B). While some found increased ICAM-1 protein expression in HCS-exposed ECs, they also showed that prior exposure to HCS exacerbated the inflammatory response to post-treatment with either TNF-α or lipopolysaccharide (LPS) for 4–6 h [6, 7]. Similarly, we showed that HCS preconditioning primes ECs to significantly upregulate the expression of NF-κB signaling products VCAM-1 and ICAM-1 compared to NCS preconditioning (Fig. 2, D and E). Exposure to physiological CS (NCS) was previously shown, via activation of small GTPase Rap1, to maintain the integrity of cell junctions, downregulate the pro-inflammatory RhoA/NF-κB signaling, and prevent ICAM-1 upregulation following post-treatment with LPS [7, 8]. Our data support the protective role of NCS based on the demonstrated inhibition of the TNF-α (0.5 ng/ml, 4 h post-treatment)-induced VCAM-1 and ICAM-1 expression in NCS-preconditioned ECs (Fig. 2, D and E). The differential effects of priming with HCS vs. NCS suggest that mechanical forces in the vasculature dictate the endothelial phenotype and sensitivity in response to subsequent chemical stimuli. Due to CS-induced changes in transcription (Fig. 2C) and post-transcription/translation (Fig. 2, D and E), ECs appear to exhibit a mechanical force-regulated “memory” that determines their future function and survival.

To the best of our knowledge, nobody has studied the effects of CS on EC SPL/Cer metabolism. In a typical cell, the majority of membrane lipids are phospholipids (structural lipids); SPLs make up ~10–15% of total cellular lipids. Most of SPLs are SM and only 1–3% are Cer [29]. Cer are highly concentrated in the smooth ER (site of de novo synthesis), especially in mitochondria-associated membranes (MAMs), moderately in the Golgi, mitochondria, and endolysosomes, and low in the plasma membrane [56]. Changes in Cer levels at these subcellular locations are very important; e.g., increased Cer in the ER means increased de novo synthesis and can lead to ER stress, increased Cer in the mitochondria has been associated with mitochondrial dysfunction and activation of apoptosis, and increased Cer at the plasma membrane is known to induce membrane curvature/blebbing and promote the release of sEVs into the extracellular media, which, via paracrine Cer transfer, can cause dysfunction/injury to recipient cells [57–60]. Increased total Cer levels have been found to occur in plasma of adults with essential hypertension compared to normotensive adults, as well as in arterial tissue and plasma of spontaneously hypertensive vs. normotensive rats, suggesting that abnormal levels of certain Cer species and total Cer in arterial tissues and plasma may play a causative role in CVD [17, 19, 61]. In the case of HCS-exposed ECs, the demonstrated oxidative stress, via activation of the redox-sensitive Cer-generating enzymes CerS, aSMase, and nSMase, was expected to increase Cer production, and increased Cer levels via upregulation of expression of pro-inflammatory cytokines (and EC adhesion molecules for cytokines) were expected to further enhance the production of ROS and Cer [62–65]. Contrary to that, lipidomics analysis of EC pellets revealed limited changes in levels of individual Cer species and no significant increase in total Cer in HCS compared to NCS at either 4 or 24 h (Fig. 3, B–D).

Instead, our lipidomics analysis showed time-dependent changes in SM abundance leading to a significant increase in both individual and total SM levels at 24 h of HCS compared to NCS (Fig. 3, E–G). No clear association has been established between plasma SM levels and incidence of CVD in humans [66, 67]. SM is both a structural lipid that helps stabilize organized membrane environments and a signaling substrate that is produced in two ways: Cer is transported to the Golgi, where SMS1 transfers a phosphocholine headgroup from phosphatidylcholine onto Cer to produce SM, which is then transported to the plasma membrane. SMS2 acts at the plasma membrane by utilizing Cer and phosphatidylcholine to synthesize new SM. SM is continuously degraded by nSMase and aSMase, and Cer is generated at the plasma membrane and endolysosomes, respectively [11, 27, 30]. It is known that 90% of the cell SM is found on the outer leaflet of the plasma membrane, and, following cell injury, translocates to the inner leaflet via the action of a Ca²⁺-sensitive scramblase, where it drives the nSMase2-mediated Cer production [66, 68, 69]. In a healthy cell, the reaction rate and equilibrium state that converts Cer to SM via SMS is strongly favored over the reverse reaction (SMase-mediated SM hydrolysis to Cer) keeping Cer levels in check. However, under cell stress and apoptosis (EC apoptosis was found to occur at 24h of HCS, Fig. 8), the rate of the reverse reaction increases significantly; the reverse reaction can convert up to ~50% of plasma membrane SM into Cer within less than 1 h [70]. Cer produced via SMases, especially nSMase on the plasma membrane, then induces membrane blebbing and gets released into the extracellular media in the form of Cer-enriched sEVs [57–59]. This might explain both the maintenance of total cellular Cer at low levels in HCS, like those in NCS (Fig. 3, B–D), and the significant increase in concentrations of several long-chain Cer (and the increasing trend in total Cer) in the media of ECs exposed to HCS compared to NCS (Fig. 6, A and B). In summary, we speculate that, in HCS-exposed ECs, the plasma membrane SM quickly converts into Cer and this Cer gets released in the form of sEVs into the extracellular media, where it was detected. Oxidative stress is known to induce secretion of Cer-rich sEVs [59], and Cer release into the media may be the reason why most of the Cer immunolabeling was detected intracellularly, and not at the plasma membrane (Fig. 5, A–B and E). However, changes in both Cer and SM levels may have occurred earlier than the 4 h time point of lipidomics analysis of EC pellets (Fig. 3, B and E). It is possible that, early on during HCS, Cer was increasing and SM was decreasing, until Cer and SM reached their steady-state levels at similar to and higher than their corresponding NCS levels.

Since, in order to study CS effects on VCAM-1 and ICAM-1, we added a post-treatment with TNF-α (0.5 ng/ml, 4 h) and TNF-α (20 ng/ml) is known to activate the nSMase and aSMase enzymes/increase Cer production in ECs and epithelial cells [71, 72], we repeated the lipidomics analysis following 24 h of CS and 4 h of post-treatment with TNF-α. Post-treatment did not cause any additional changes in the levels of Cer or SM species (Fig. 3, H and I; Fig. S3). Hence, in our study, any association between Cer and the HCS-induced VCAM-1 and ICAM-1 expression is due to the HCS preconditioning, and not due to post-treatment with a low concentration of TNF-α.

Gene expression analysis for key enzymes in SPL metabolism provided further insight into the effects of 24 h of HCS vs. NCS (Fig 4). Under NCS, ECs exhibited significantly higher expression of genes that encode the de novo synthesis enzymes SPTL, CERS1, and CERS5, along with increased expression of ASAH1, which encodes the aCDase isoform that hydrolyzes Cer to Sph in lysosomes [46]. Lower aCDase expression in HCS-exposed ECs may lead to lysosomal Cer accumulation compared to NCS-exposed ECs. HCS-exposed ECs had increased expression of SGMS2 that encodes the SMS2 isoform responsible for generating SM from Cer at the plasma membrane [46], in agreement with the significant increase in SM under HCS compared to NCS. Expression of the SMase-encoding genes SMPD1 (encodes aSMase) and SMPD2 (encodes nSMase1) did not differ significantly between CS conditions, and SMPD3 (encodes nSMase2) expression was very low (not shown). Overall, the transcriptional changes confirmed the differential modulation of gene expression for key enzymes in SPL metabolism by HCS vs. NCS.

Overall, the lack of detection of changes in Cer using whole-cell (bulk) lipidomics may be due to the fact that changes occur at specific subcellular locations, such as the plasma membrane and lysosome or mitochondrial membranes, and these ultimately determine cell function [27, 28]. If, for example, the lipid concentration inside one organelle is increased or decreased while other subcellular locations keep their levels unchanged, that change can go undetected by bulk lipidomics that reports an average concentration across the whole cell [73, 74]. The Canals group recently showed that the plasma membrane Cer in epithelial cells amounts to 0.1% of total cellular Cer, and, thus, even several-fold changes would not be detectable by mass spectrometry of the total Cer pool [75]. In contrast to the findings from lipidomics analysis, Cer immunofluorescence at 4 and 24 h of CS showed HCS to cause significant changes in total cellular Cer immunofluorescence signals compared to NCS, as well as a time-dependent Cer redistribution to specific subcellular regions (Fig. 5). Specifically, at either 4 or 24 h of HCS, Cer MFI was significantly higher relative to that in NCS (Fig. 5, A–B and D). At 4 h of either HCS or NCS, Cer was localized at the perinuclear region, but, at 24 h of HCS, and not NCS, it was redistributed to the cell periphery (Fig. 5, A–B and E). This agrees with previous studies where treatment of epithelial cells with H₂O₂ caused nSMase2 trafficking from the Golgi/the perinuclear region to the plasma membrane and increased the Cer production at that location [59, 76]. To quantify the Cer distribution at subcellular locations, we digitally marked a perinuclear region that extended 5 μm around the nucleus and, thus, included most of the ER, and measured Cer MFI at both the perinuclear and peripheral sections of HCS- and NCS-exposed ECs (Fig. 5, E and F). Changes in Cer immunofluorescence levels at the perinuclear region can be attributed mainly to de novo synthesis in the ER and secondarily to trafficking through the Golgi and perinuclear mitochondria. Changes in Cer immunofluorescence levels at the periphery can be attributed to Cer trafficking to mitochondria, endolysosomes and the plasma membrane, as well as to Cer generation from SM at lysosomes and the plasma membrane. A significant increase in Cer immunofluorescence levels in both the perinuclear and peripheral regions of HCS-exposed ECs compared to the corresponding regions of NCS-exposed ECs suggested enhanced Cer de novo synthesis and trafficking under HCS (Fig. 5F).

The inclusion of myriocin in the media during HCS provided additional evidence that the HCS-induced increase and redistribution of Cer were dependent, at least in part, on de novo synthesis (Fig. 5, C and D). Inhibition of SPT with myriocin significantly reduced Cer immunostaining in both NCS- and HCS-exposed ECs, demonstrating the importance of de novo synthesis in Cer generation (Fig. 5, C and D). Notably, myriocin significantly decreased Cer MFI at both the perinuclear and peripheral regions of HCS-exposed ECs, indicating that the inhibitor reduced the overall Cer available for accumulation at different locations throughout the cell (Fig. 5, E and F). The localized increase in Cer in the cell periphery, as detected by immunolabeling, without any significant overall change of total Cer levels, as detected by mass spectrometry, has been documented previously and reflects the low proportion of plasma membrane Cer relative to total Cer [59, 75]. In agreement with our in vitro findings, an increase in either flow rate/pressure in perfused rat lung vasculature or in flow rate over cultured ECs was found to induce an acute (within min) increase in nSMase activity in plasma membrane fractions and trigger localized Cer production [77, 78]. Similarly, when HCS was applied on engineered cardiac constructs consisting of cardiomyocytes and fibroblasts, it was shown to increase caveolae-based nSMase activity and the generated Cer depressed cardiac conduction velocity [79]. Based on all the above, our observed redistribution and increased accumulation of Cer at the cell periphery under HCS, but not NCS, is probably due to an ROS-mediated increase in SMase trafficking and activity, with nSMase2 being the leading enzyme candidate.

Targeted lipid analysis of conditioned media showed significant increases in long-chain Cer species C18:0-Cer, C22:0-Cer, and C24:1-Cer, total Cer, dhSph and S1P in the media from HCS-exposed ECs compared to NCS-exposed ECs (Fig. 6, A and C). While circulating levels of C16:0-Cer, C18:0-Cer, and C24:1-Cer, as well as their ratio to C24:0-Cer, are considered prognostic markers of future CVD [20, 80], and two out of those three markers were found to be significantly higher in HCS relative to NCS (Fig. 6A), the significantly higher S1P levels in HCS relative to NCS (Fig. 6C) are EC protective [81, 82]. It is known that sEVs form from endolysosomal compartments and are released into the extracellular space via inward budding of the plasma membrane, a process regulated by the nSMase2-Cer axis [83]. Oxidative stress in epithelial cells was shown to trigger the release of sEVs that were enriched with Cer, and the release was inhibited by nSMase and aSMase inhibitors [59]. In addition, a study detected Cer in large EVs (lEVs; 100 nm-1 μm in diameter, also called microparticles or microvesicles) released from ECs under oxidative stress due to hyperglycemic conditions [57]. Overall, the detected Cer accumulation at the periphery of HCS-exposed ECs together with the significantly higher concentration of several long-chain Cer in HCS media compared to NCS media suggest that the Cer in HCS media exists primarily in the membranes of released sEVs, and possibly also lEVs.

The role of the two major Cer-generating pathways, de novo synthesis and SM hydrolysis, were interrogated by pharmacologically inhibiting SPT with myriocin, DES with GT-11, nSMase with GW4869, and aSMase with ARC39 (Fig. 7, A and D). Both GT-11 and GW4869 significantly decreased the TNF-α-induced VCAM-1 and ICAM-1 expression in HCS-preconditioned ECs to levels comparable to NCS-preconditioned ECs, while they had no significant effect on the VCAM-1 and ICAM-1 expression in NCS-preconditioned ECs (Fig. 7, B–C and E–F). Myriocin was unable to decrease the TNF-α-induced VCAM-1 and ICAM-1 expression in HCS-preconditioned ECs. GT-11 at 10 μM may be more effective in lowering total Cer compared to myriocin at 1 μM, and/or the SPL products from the de novo synthesis pathway between SPT and DES may regulate inflammation. Indeed, dhCer, an intermediate expected to accumulate during DES inhibition, has been shown to activate autophagy and help maintain cell homeostasis [84], and this may, in part, explain the higher protection afforded by GT-11 compared to myriocin. The potential effects of myriocin on the levels of soluble ICAM-1 and VCAM-1, and/or other inflammatory markers, i.e., MCP-1, IL-1β, IL-6, and IL-8, in the conditioned media were not examined. Interestingly, ARC39 significantly inhibited only the ICAM-1 upregulation in HCS-preconditioned ECs. The differential effects of GW4869 and ARC39 on the TNF-α-induced cell receptor expression in HCS-preconditioned ECs may be due to distinct roles of plasma membrane vs. lysosomal Cer pools in the transcriptional control of gene expression, and subsequent protein expression. Overall, the fact that HCS-induced inflammation was inhibited by these pharmacological inhibitors suggests that Cer synthesized at the ER via de novo synthesis and at the plasma membrane and endolysosomes via SM hydrolysis (catalyzed by nSMase and aSMase, respectively) is responsible for the inflammation in HCS-exposed ECs. Hence, Cer at the plasma membrane and endolysosomes act as mechanotransducers that contribute to endothelial dysfunction in hypertension.

Our flow cytometry analysis of EC apoptosis/death at the end of 24 h of CS showed HCS to cause significantly higher apoptosis compared to NCS and static, while NCS caused significantly lower apoptosis compared to static (Fig. 8, A and B). This agrees with studies by others who showed that HCS activated mTOR, cleaved caspase-3, and stimulated EC apoptosis, whereas NCS activated Akt, phosphorylated/inactivated BAD, and inhibited EC apoptosis [85, 86]. There is also in vivo evidence that, in early hypertension, ECs are proinflammatory and proapoptotic [87]. The inhibitors of SPL metabolism enzymes, previously tested for their effects on inflammation, were also tested for their effects on apoptosis and each of them was found to significantly inhibit the HCS-induced increase in apoptosis (Fig. 8, C and D), suggesting that Cer production from the de novo synthesis and SM hydrolysis pathways is responsible for the EC apoptosis under HCS. A mechanism by which Cer leads to EC death is based on the fact that nSMase2-induced Cer production can inactivate the gate of the EC mechanosensitive PIEZO1 channel, causing sustained Ca²⁺ influx, increased mitochondrial ROS, opening of the mitochondrial permeability transition pore (mPTP), and cell apoptosis [88]. One interesting finding was that myriocin inhibited the HCS-induced increase in apoptosis, although it was unable to inhibit the increase in VCAM-1 and ICAM-1 expression. Cer are known to accumulate in the mitochondrial outer membrane and form channels allowing for cytochrome c release and activation of the mitochondrial pathway of apoptosis. By lowering Cer de novo synthesis and mitochondrial Cer levels, myriocin was previously shown to preserve mitochondrial integrity and cell survival [89, 90]. Last, our static controls have had significantly higher apoptosis compared to NCS (Fig. 8, A and B). Others have reported similar levels of apoptosis in static control HUVECs [91, 92]. Others have also shown a significant decrease in apoptosis, and increased resistance to death in response to injurious stimuli following prolonged EC exposure to either NCS or atheroprotective shear stress, suggesting that ECs that have been preconditioned at physiologic mechanical environments may be a more appropriate control than static ECs [4, 93, 94].

The present study has several limitations: One limitation is the use of HUVECs, instead of human ECs from arterial sites. Although our in vitro CS model captures the tensile component of mechanical forces, it does not encompass all the forces ECs experience on the vessel wall [1]. The model also does not reproduce the complex in vivo chemical environment shaped by products released, not only from ECs, but also from blood cells and smooth muscle cells. Lipidomics analysis was performed on cell pellets, which limits resolution of compartment-specific Cer pools. Future studies incorporating subcellular analyses for the plasma membrane, ER, and mitochondria would help refine these insights. While pharmacological inhibitors are commonly used in SPL research, genetic approaches are needed for better molecular specificity. For example, endothelial-specific deletion of SPTLC2 in mice showed a basal level of EC Cer to be protective, because it was necessary for resting eNOS phosphorylation and blood pressure homeostasis [47]. Despite those limitations, the present study demonstrated that EC exposure to pathological CS induces spatially-defined Cer generation and Cer release to the extracellular space that trigger inflammation and apoptosis. Targeting Cer-generating pathways, especially those mediated by the SMase enzymes, may represent a promising strategy to mitigate endothelial dysfunction and delay/prevent CVD development in hypertensive patients.

Supplemental Materials

Supplemental Figs. S1-S6: https://doi.org/10.6084/m9.figshare.32190897

Supplemental Tables S1-S3: https://doi.org/10.6084/m9.figshare.32249100

Data Availability

Data associated with Figs. 2, 3, 6, and 7 are available in the online supplementary materials. Further data will be made available upon reasonable request.