PGC1α Regulates the Endothelial Response to Fluid Shear Stress via Telomerase Reverse Transcriptase Control of Heme Oxygenase-1

Abstract

Objective:

Fluid shear stress (FSS) is known to mediate multiple phenotypic changes in the endothelium. Laminar FSS (undisturbed flow) is known to promote endothelial alignment to flow which is key to stabilizing the endothelium and rendering it resistant to atherosclerosis and thrombosis. The molecular pathways responsible for endothelial responses to FSS are only partially understood. In this study we determine the role of PGC1α-TERT-HMOX1 during shear stress in vitro and in vivo.

Approach and Results:

Here we have identified peroxisome proliferator gamma coactivator-1α (PGC1α) as a flow-responsive gene required for endothelial flow alignment in vitro and in vivo. Compared to oscillatory FSS (disturbed flow) or static conditions, laminar FSS (undisturbed flow) showed increased PGC1α expression and its transcriptional co-activation. PGC1α was required for laminar FSS-induced expression of telomerase reverse transcriptase (TERT) in vitro and in vivo via its association with estrogen-related receptor alpha (ERRα) and Kruppel-like factor 4 (KLF4) on the TERT promoter. We found that TERT inhibition attenuated endothelial flow alignment, elongation, and nuclear polarization in response to laminar FSS in vitro and in vivo. Among the flow-responsive genes sensitive to TERT status, heme oxygenase-1 (HMOX1) was required for endothelial alignment to laminar FSS.

Conclusions:

These data suggest an important role for a PGC1α-TERT-HMOX1 axis in the endothelial stabilization response to laminar FSS.

Graphical Abstract

Introduction

The endothelium exerts considerable control over vascular homeostasis with important roles governing vascular tone, inflammation, and metabolism ^{1, 2}. Normal endothelial function is characterized by a quiescent cell phenotype that is non-proliferative, non-migratory, and exhibits a cell surface that prevents thrombosis, inflammation, and lipid deposition, thereby resisting atherosclerosis and vascular disease ^{3, 4}. A key stabilizing stimulus for endothelial quiescence is laminar fluid shear stress (also called undisturbed flow) passing over the cell surface, a common feature of straight vascular segments with little to no curvature ⁵. Endothelial cells reorient and change their shape in order to align their long axis to physiological fluid shear stress (FSS) ⁶. In contrast, curved and branched arteries experience multidirectional and chaotic FSS (also called disturbed flow). Areas with disturbed flow promote a less stable, activated, and dysfunctional endothelial phenotype that is more susceptible to inflammation and atherosclerosis ^{5, 7–11}. Thus, endothelial response to fluid shear stress is a critical element of vascular homeostasis.

Endothelial cell responses to fluid shear stress involve both mechanotransduction of the flow signal and coordinated regulation of signaling pathways and gene expression that dictate the phenotypic consequences. A number of mechanosensors and mechanotransducers have been implicated in flow sensing, including ion channels, PECAM-1, G protein-coupled receptors, junctional proteins, VEGF receptors, and even primary cilia ^{5, 7, 12, 13}. Several pathways have been implicated in the phenotypic response to physiologic vs. pathologic flow. For example, disturbed flow enhances NF-κB, Yap/Taz, β-catenin and the Smad2/3 pathways that cooperate to promote remodeling that includes activation of inflammatory mediators and, perhaps, induction of endothelial to mesenchymal transition ^14–16. In contrast, laminar flow leads to activation of the KLF2/4, Notch, and Alk1-Smad1/5 pathways that induce genes required for vascular stability ^17–19. In particular, KLF2 and KLF4 have been implicated in promoting vasodilation and the inhibition of both inflammation and thrombosis ²⁰. Among the genes regulated by KLF2 and KLF4, NOS3 codes for the endothelial isoform of nitric oxide synthase (eNOS) ²⁰. This enzyme contributes importantly to the vascular homeostatic environment by promoting vasodilation and limiting atherothrombosis. We recently reported that the transcriptional coactivator, peroxisome proliferator gamma coactivator-1α (PGC1α), was an important determinant of eNOS expression in vitro and in vivo ²¹. Here we sought to determine if PGC1α and its downstream targets play a role in endothelial responses to fluid shear stress.

Material and Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Animals

C57BL/6J, Tie2-Cre (#008863) and VE-Cadherin-Cre (#006137) strains of mice were obtained from the Jackson Laboratory. PGC1α transgenic mice ²¹, conditional PGC1α ²², TERT global knockout ²³ and conditional TERT ²⁴ mice were described previously. For the creation of endothelial knockout lines, conditional PGC-1α allele (LoxP sites flanking exons 3–5, a kind gift from Dr. Bruce Spiegelman, Harvard University) ²² or TERT allele (LoxP sites flanking exons 1–2) mice were crossed with either VE-Cadherin-Cre or Tie2-Cre mouse lines on the C57BL/6J background. These endothelial-knockout PGC-1α (PGC1α-ECKO) or TERT-ECKO mice were compared to Cre control mice. Endothelial specific PGC1α transgenic mice (PGC1α-ECTG) were produced with human PGC-1α expression under the control of the mouse vascular endothelial cadherin promoter (VE-Cad) as described ²¹.

All mouse experiments were performed according to the relevant ethical regulations. Mice were housed in facilities either accredited by the American Association for Laboratory Animal Care or approved by the Ethics Committee of the University Hospital Mainz and Landesuntersuchungsamt Koblenz (23 177–07/G 12–1-080 and 23 177–07/G 17–1-066). Animal protocols were approved by the Institutional Animal Care and Use Committee of the Brigham and Women’s Hospital, Boston.

Voluntary Exercise in Mice

Exercise experiments were performed with 4-week old animals that were kept in individual cages for six weeks equipped with a running wheel and a mileage counter. Exercise training was performed voluntarily. The running distance of Tie2Cre control and PGC1α-ECKO mice did not differ significantly.

Cell Culture

Human aortic endothelial cells (HAECs) (#PCS-100–011) and human umbilical vein endothelial cells (HUVECs) (#PCS100010) were purchased from ATCC and cultured in endothelial cell growth basal medium-2 containing bullet kit growth factor supplements (EBM-2, Lonza), 5% fetal bovine serum, 100 units/mL penicillin, 100 μg/mL streptomycin, and 2 mM L-glutamine (Invitrogen). Cultured human ECs between passages 2 and 6 were utilized for experiments.

Adult mice were sacrificed for lung endothelical cell (MLEC) isolation according to approved guidelines as mentioned above. Pooled mouse lungs were immediately harvested and minced. Tissues were then transferred to a fresh conical tube containing digestion solution, 50 mg/mL Type 1 filtered collagenase (Worthington) in DMEM media, and incubateed at 37 °C for 1 hour. The enzyme digested tissue was filtered through a 70-μm and 40-μm cell strainers. The cell suspension was pelleted by centrifugation at 300xg for 10 min at room temperature. The isolated cells were further used for sorting and direct mRNA isolation or for cell culture experiments ²⁵.

For direct use of the cells in RTqPCR measurements, two separation steps were performed to isolate endothelial cells using CD31 MACS (CD31 Micro Beads mouse, 130–097-418, Miltenyi Biotech) and ICAM Dynabeads (Rat anti-mouse CD102/ICAM2, 553326, BD Bioscience) according to the manufacturer’s protocols. For cell culture experiments, pellets were resuspended in MLEC growth media containing a 1:1 mixture of DMEM and Ham’s F-12 (Gibco), 20% fetal bovine serum, 50 mg endothelial mitogen (Cell Applications), 50 mg/mL heparin (Sigma), 100 units/ml penicillin, 100 μg/ml streptomycin and plated on 0.2% gelatin-coated cell flasks. Media was changed daily. At 80% confluency, cells were collected in a conical tube with 25uL Dynabeads® sheep anti-Rat IgG (Invitrogen) coated with ICAM-2 (BD Biosciences Pharmingen) for the first selection. The cell-Dynabead suspension was set to rotate at 4 °C for 40 minutes and harvested on a magnetic rack. The beads were resuspended in MLEC growth medium and plated into a new 0.2% gelatin-coated cell flask. When cells reached 80% confluency, the above steps were repeated for the second selection. MLEC used for experiments were between passages 2–4. Preliminary experiments did not reveal any effect of sex on in vitro cell culture experiments consistent with prior experience ²⁵. Therefore, cell culture work here generally involved pooled cells isolated from 2–3 mice regardless of sex except in Figure 1 MLECs were from male mice and data in Figures 3 and 6 were derived from female MLECs.

Figure 1. PGC1α regulates endothelial cell function during fluid shear stress.

A-B) PGC1α mRNA expression (A) or protein expression (B) from human aortic endothelial cells (HAECs) were measured by RT-qPCR or western blots after cells were subjected to static, laminar or oscillatory fluid shear stress (FSS) for 48 hours (n = 5). C) Bright-field image of mouse lung endothelial cells (MLECs) isolated from either wild-type (WT) or PGC1α-ECKO mice after exposure to laminar FSS. D) MLECs from WT and PGC1α-ECTG mice were exposed to laminar FSS, and RT-qPCR was performed for the indicated genes (n = 5). E) HAECs were treated with scrambled or PGC1α siRNA, and RT-qPCR was performed for the indicated genes after exposure to laminar FSS for 48 hours (n = 5). F-G) Aortae were isolated from either WT (F) or PGC1α-ECTG (G) mice, mRNA isolated, and RT-qPCR performed in the indicated regions reflecting disturbed or laminar flow (n = 5). (H) Sample sites of disturbed (oscillatory) vs. laminar fluid shear stress in mouse aorta. (I) En face staining with β-Catenin and DAPI in WT and PGC1α-ECKO aorta. Scale bar, 20 μm. (J) Composite data of length/width ratio of the endothelium in the thoracic region of mouse aortae (n = 18–24). All experiments were repeated 3–5 times. Statistically significant differences were measured by Student’s t-test or one-way ANOVA with post-hoc comparison as appropriate with control group. The data are mean ± SEM.

Figure 3. PGC1α regulates TERT expression.

A-B) HAECs were either treated with scrambled or ERRα siRNA (A) or scrambled or KLF4 siRNA (B), and RT-qPCR was performed for different shear stress-related genes after exposure of cells to laminar FSS for 48 hours (n = 5). C) HAECs were lysed and immunoprecipitation (IP) was performed with control IgG or ERRα antibody and immunoblotting was done with antibodies against PGC1α and KLF4. Lysates were examined by probing with GAPDH antibody. D-E) MLECs were isolated from control and PGC1α-ECTG mice, and either RT-qPCR (D) or immunoblot analysis (E) was performed with the probes and antibodies as indicated. F) Lysates prepared from HAECs treated with control shRNA or shRNA against PGC1α (48 hrs) were examined by immunoblot analysis using antibodies for TERT, PGC1α and GAPDH. G) MLECs were isolated from WT and PGC1α-ECKO mice and mRNA expression was measured by RT-qPCR for TERT gene before and after exercise (n = 6). H) Aortae were isolated, and mRNA expression was measured for TERT by RT-qPCR in the arch and thoracic region of WT mice (n = 5). I-J) ChIP-qPCR analysis of PGC1α recruitment to the TERT promoter region was performed in HAECs (n = 5). All the experiments were repeated 3–6 times. Statistically significant differences between groups are indicated. Statistically significant differences were measured by Student’s t-test or one-way ANOVA with post-hoc comparison as appropriate with control group. P values vs. control group (black) + exercise group (red) by Student’s t-test. The data are mean ± SEM.

Figure 6. PGC1α-TERT regulates HMOX1 expression.

A) MLECs were isolated from control and PGC1α-ECTG mice and immunoblot analysis was performed with the antibodies as indicated. B) MLECs were isolated from WT and PGC1α-ECKO mice aortae, and mRNA expression was measured by RT-qPCR for the HMOX1 gene before and after exercise (n = 6). C) HUVECs were exposed to laminar or oscillatory FSS with or without TERT inhibitor and immunoblot analysis was performed with the antibodies as indicated. D) HAECs were either treated with scrambled or HMOX1 siRNA and RT-qPCR was performed for different genes related to endothelial function after exposure to laminar FSS for 48 hours (n = 6). E) Bright-field image of HAECs after exposure to laminar FSS in the presence of either control (CuPP 0.25 μM) or two different HMOX1 inhibitors (ZnPP 0.25 μM and OB 24 hydrochloride 0.25 μM). F) Cell and mitochondrial morphology (MitoRed fluorescence) were imaged in the presence of laminar flow with and without HMOX1 inhibition (Scale bar, 5 μm). G) Schematic diagram of LSS-induced PGC1α-TERT-HMOX1 pathway. n = 6 in each group. All the experiments were repeated 3 – 5 times. Statistically significant differences between groups are indicated. Statistically significant differences were measured by Student’s t-test or one-way ANOVA with post-hoc comparison as appropriate with control group. P values vs. control group (black) + exercise group (red) by Student’s t-test). The data are mean ± SEM.

Adenoviral Constructs

Adenoviral vectors were used for both overexpression (24 hrs) and knockdown (48 hrs) of PGC1α. The PGC1α expressing vector was a kind gift from Dr. Bruce Spiegelman, Harvard University and ERRα vector was a kind gift from Dr. Anastasia Kralli, Scripps Research Institute ²⁶. Control viruses (siCtl, LacZ, and GFP) from Vector BioLabs (Malvern, PA) were used for comparison. Cells were typically infected at a multiplicity of infection (MOI) of 10 to 50 with a control adenovirus at the same MOI.

Transfections

Transfection assays were performed using 100nM small interfering human RNA oligonucleotides (ON-TARGET plus SMART pool; Horizon Discovery Dharmacon, Lafayette, CO) for control (D-001810–10), TERT (L-003547–00), KLF4 (L-005089–00), ERRα (L-003403–00), and HMOX1 (L-006372–00) in DharmaFECT 3 reagent (Horizon Discovery Dharmacon,T-2003) for 6 hours in OptiMEM (Invitrogen, Waltham, MA)²⁷. Media was then changed to EBM-2 growth medium. For flow experiments, after 48 hours of siRNA treatment, cells were exposed to either laminar flow 12 dynes/cm² or oscillatory flow of 4 dynes/cm² (Ibidi system) for 48 hours.

Total RNA Preparation and Quantitative Polymerase Chain Reaction

Cell and tissues were lysed with TRIzol reagent (Life Science Technologies), and total RNA was extracted using the RNeasy Plus Micro Kit (Qiagen). Total RNA was reverse transcribed with oligo (dT) primers for cDNA synthesis using an iScript cDNA synthesis kit (Bio-Rad). The expression of mRNA was examined by quantitative PCR analysis using a QuantStudio™ 6 Flex Real-Time PCR System (Applied Biosystems). TaqMan^© assays were used to quantitate

ERRα (Hs00607062_gH), HEY1 (Hs05047713_s1, Mm00468865_m1), HMOX1 (Hs01110250_m1, Mm00516005_m1), ICAM1 (Mm00516023_m1Hs00164932_m1), KLF2 (Hs00360439_g1, Mm00500486_g1), KLF4 (Hs00358836_m1, Mm00516104_m1), PGC1α (Hs01016719_m1, Hs01016724_m1, Mm01208835_m1) TERT (Mm01352136_m1), VCAM1 (Mm01320970_m1, Hs01003372_m1), MMP2 (Mm00439498_m1), SCA1(ATXN1) (Mm00485928_m1), SM22 α (TAGLN) (Mm00441660_m1), TBP (Mm00446973_m1), HPRT (Hs02800695_m1, Mm00446968_m1), GAPDH (Hs99999905_m1, 4352339E-0904021), B2M (Hs99999907_m1, Mm00437762_m1) mRNA (Applied Biosystems). The 2^-ΔΔCT method was used for relative quantification of gene expression ^{28, 29}. Expression of HPRT, GAPDH, TBP and B2M were used to normalize each sample.

Immunoblot Analysis

Cell extracts were prepared using Triton lysis buffer (TLB buffer) [20 mM Tris (pH 7.4), 1% Triton X-100, 10% glycerol, 137 mM NaCl, 2 mM EDTA, 25 mM β-glycerophosphate] with proteinase inhibitors (Sigma #11873580001) and phosphatase inhibitors (Sigma #4906837001) ³⁰. Protein extracts (50 μg) in β-mercaptoethanol containing SDS sample buffer were separated in 4% to 12% gradient SDS-polyacrylamide gels (Bio-Rad #456–8094), transferred to nitrocellulose membranes (Bio-Rad #170–4271, Hercules, CA) and incubated with primary antibody at 1:1000 dilution. Immunocomplexes were visualized with horseradish peroxidase-conjugated secondary antibodies and detected with a Clarity western ECL substrate (Bio-Rad #170–5061, Hercules, CA). Images were acquired on a chemiluminescent imager (Bio-Rad Chem-Doc Imaging System).

Antibodies and Reagents

Primary antibodies for immunoblots were obtained from Abcam (HMOX1 #52947, TERT # ab32020, Cambridge, MA), Cell Signaling (ERRα #13826, Danvers, MA); R&D (KLF4 #AF3640); Novus Bio (PGC1α #NBP-04676, Centennial, CO); BD Pharmingen (CD31 #550274, San Jose, CA) and Sigma (β-catenin, #C2206, St. Louis, MO). Control antibodies were obtained from Proteintech (GAPDH #HRP-60004 and β-Actin #HRP-60008, Rosemont, IL). The TERT inhibitor, BIBR1532 was purchased from Selleck Chemicals (Houston, TX). The HMOX1 inhibitor, zinc (II) protoporphyrin IX (ZnPP) and control, copper protoporphyrin IX (CuPP) were purchased from Sigma. The HMOX1 inhibitor 1-[[2-[2-(4-Bromophenyl)ethyl]-1,3-dioxolan-2-yl]methyl]-1H- imidazole hydrochloride (OB 24 hydrochloride) was purchased from Tocris Bioscience (Minneapolis, MN).

Immunoprecipitation

Cell extracts were prepared using Triton lysis buffer and incubated (16 hrs., 4°C) with either 3 μg non-immune control rabbit IgG (Cell Signaling #2729) or with ERRα antibody (Cell Signaling #13826) to 500 uL of cell lysate. Immunocomplexes were isolated using Protein G Sepharose beads (Santa Cruz #SC2002, Santa Cruz, CA) and washed 4–5 times with lysis buffer. Bead pellets were resuspended and boiled in β-mercaptoethanol containing Laemmli sample buffer, separated in 4% to 12% gradient SDS-polyacrylamide gels (Bio-Rad #456–8094), transferred to nitrocellulose membranes (Bio-Rad #170–4271, Hercules, CA) and incubated with rabbit anti PGC1α antibody with 1:1000 dilution. Immunocomplexes were visualized with horseradish peroxidase-conjugated goat anti-rabbit IgG (Cell Signaling, #7074) and detected with a Clarity western ECL substrate (Bio-Rad #170–5061).

Chromatin Immunoprecipitation (ChIP)

Cultured HAECs were exposed to 1% oxygen for 30 min at 37°C and processed using a ChIP-IT® kit (Active Motif #53008, Carlsbad, CA) according to the manufacturer’s instructions. Briefly, the cells were fixed in 1% formalin and homogenized in lysis buffer. Lysed cells were sheared with sonication ten times each with a pulse of 20 seconds and 30-seconds of rest on ice between shearing steps. Sheared samples were incubated with Protein G Beads and either non-immune IgG (Sigma #NI01), Polymerase II (Santa Cruz #sc56767), or PGC1α (Novus Bio #NBP-04676, Centennial, CO) antibodies on an end-to-end rotator overnight at 4°C. The beads binding target chromatin were washed on a magnetic bar with washing buffer. Elucidated chromatin targets were amplified with quantitavive PCR using a primer pair (5’-CAGAAGTTTCTCGCCCCCTT-3’ and 5’- GAGGCCAACATCTGGTCAC-3’) specific for the TERT promoter.

Whole Mount Aorta En Face Immunofluorescence Staining

Prior to excision, aortae were perfused serially via the left ventricle of the heart with: i) 0.5 mM EDTA in PBS; ii) 4% paraformaldehyde, 7.5% sucrose, 0.5 mM EDTA, in PBS; and iii) 0.5 mM EDTA in PBS, respectively. Dissected aortae were fixed in 4% paraformaldehyde for 30 minutes and then permeablized with 0.1% Triton X-100 in PBS at room temperature, followed by an overnight incubation (4°C) with a rabbit anti-β-Catenin antibody (Sigma #C2206) in 1:1000 diluted blocking solution (Dako-Agilent, K800621, Carpinteria, CA). Secondary anti-rabbit IgG antibody conjugated with Alexa Plus 555 (Thermo #14–387-071, Waltham, MA) at 1:1500 dilution was used for visualization of endothelial borders. Nuclei were counter stained with 1μg/mL Hoechst 33342 (Cell Signaling #4082). Vessels were mounted on a coverslip with ProLong anti-fade mounting medium (Thermo #P36962, Waltham, MA). Images were acquired with confocal microscope (Carl Zeiss) and ZEN 2012 software (Carl Zeiss). Since our cell culture work revealed no impact of sex, composite data from en face staining was typically sampled 4–8 times in aortae isolated from 4–5 mice of either sex. Representative staining in Figure 1 was derived from a male mouse and that in Figure 5 representes a female mouse.

Figure 5. TERT regulates mitochondrial structure and function.

A) Mitochondrial morphology and mass (MitoTracker fluorescence) were measured in the presence of laminar flow with and without TERT inhibition (iTERT- BIBR1532) and compared with oscillatory flow (Scale bar, 5 μm). B) Quantification of mitochondrial mass (n = 5) and morphology during FSS (n = 22–25). C) ROS production in HAECs either treated with control or TERT inhibitor (n = 9–13). D) En face staining with β-Catenin and Hoechst 33342 in WT and TERT knockout aorta (Scale bar, 10 μm). E) Composite data of length/width ratio of the endothelium in WT and TERT knockout aortae. n = as indicated in each group. All experiments were repeated 3 – 6 times. Statistically significant differences were measured by Student’s t-test or one-way ANOVA with post-hoc comparison as appropriate with control group. The data are mean ± SEM.

Endothelial Shear Stress Exposure In Vitro.

In vitro endothelial cell fluid flow experiments were conducted using either the Ibidi system or a parallel plate flow chamber system as previously described ³¹. The Ibidi system used confluent endothelial cells seeded in 0.2% gelatin coated μ-Slide I Luer chambers (Ibidi #80176). Fluid flow conditions involved either unidirectional steady flow (12 dyn/cm²), or bidirectional oscillatory flow (±4 dyn/cm²) using the Ibidi Pump System (Ibidi #10902) for 48-hour treatment. The second system used a parallel plate flow chamber with a peristaltic pump that generates fully antegrade pulsatile flow (maximum, minimum and mean wall shear stress equal 6.7, 2.7, and 4.8 dyn/cm², respectively) or net antegrade flow with a flow reversal component (maximum, minimum and mean wall shear stress equal 1.6, −1.1, and 0.3 dyn/cm², respectively). The steady flow and the fully antegrade pulsatile flow waveforms are referred to as undisturbed flow (UF) since the flow is unidirectional. The bidirectional oscillatory flow and the net antegrade flow with a flow reversal component waveforms are referred to as disturbed flow (DF) because of the multidirectional nature of the waveforms. The fluidic units were maintained in 37 °C incubators with 5% CO2.

MitoTracker Staining

Mitochondria in live cells were stained with either MitoTracker obtained from ThermoFisher (MitoTracker Red, M7512, Carlsbad, CA) or CytoPainter MitoRed from Abcam (#ab176832, Cambridge, MA) according to the manufacturer’s instructions. In brief, the cells were incubated with 1:1000 diluted MitoTracker/MitoRed in a growth medium for 15 minutes at 37°C incubator and followed by 1μg/mL Hoechst 33342 (Cell Signaling #4082) for 5 min at room temperature. Following the incubation, cells were washed in the growth medium twice and images were acquired.

ROS Measurement

Amplex Red

As an index of ROS generation, we used the Amplex Ultra Red reagent, 10-acetyl-3,7-dihydroxyphenoxazine (Molecular Probes; A36006), which reacts with hydrogen peroxide (1:1 stoichiometry) in the presence of horseradish peroxidase (HRP) to form resorufin. Endothelial cells were cultured to confluency in 12-well plates and incubated with Kreb’s HEPES buffer (118mM NaCl, 22mM HEPES, 4.6mM KCl, 2.1 mM MgSO₄, 0.15mM Na₂HPO₄, 0.41mM KH₂PO₄, 5mM NaHCO₃, 5.6mM Glucose, 1.5mM CaCl₂) for 30 min. The Amplex Ultra Red and HRP were then added and fluorescence (excitation 544 nm; emission 590 nm) was determined as a function of time (2h) in 96-well black plates (Corning) at 37°C using a fluorescent plate reader (Spectramax, Molecular Devices). Vascular hydrogen peroxide formation in aortic tissue was measured by an HPLC-based Amplex Red assay as described earlier using aortic ring segments (3 mm in length) ³².

MitoSOX™ Red and DHE

Total cellular ROS and mitochondrial ROS levels were examined using DHE and MitoSOX™ Red, respectively, according to the manufacturer’s instructions. Cells were briefly washed and loaded with DHE (10 μM) or MitoSOX™ Red (5 μM) at 37°C for 20 min in the dark and washed three times with warm buffer. Pre-loaded cells were incubated at room temperature, 37°C, 40°C or 42°C for 1, 2, or 3 h, respectively. To analyze topographic ROS production in aortic tissue, aortic cryo-sections were used as described previously ^{33, 34}.

Immunofluorescent Staining

Cells were grown on #1 thick cover glass (EMS #72290–09, Hatfield, PA) or in flow chambers. Cells were fixed with 2% PFA or methanol for 10 min followed by permeabilization with 0.1% Triton X-100 (Sigma Aldrich #X100) in PBS. Fixed cells were blocked in blocking solution (Dako-Agilent, K800621) and primary antibodies were incubated at 1:1000 dilution overnight at 4 °C. The primary antibodies used were GOLPH4 #ab28049, HMOX1 #ab52947, TERT #ab32020 (Abcam, Cambridge, MA). Immunocomplexes were visualized with anti-rabbit secondary antibody conjugated with Alexa Plus 555 (Thermo #14–387-071) or Alexa Plus 488 (Thermo A11034) at 1:1500 dilution. Slides were mounted with an anti-fade mounting medium (Vector Laboratories, Burlingame, CA) containing DAPI and imaged at the Microscopy Resources On the North Quad (MicRoN) facility at Harvard Medical School, Boston.

Image Analysis

The ImageJ processing software (National Institute of Mental Health, Bethesda, MD) was used to import fluorescent images and separate individual color channels. All area and intensity values were measured from the green channel. Under digital magnification, the width and height of each hyperfluorescent vessel were manually outlined and the encompassed measurement in pixels converted to μm using the scale bar. The average width/height ratio of experimental vessels was compared to the control vessels. For cell polarization, the locations of the nucleus and Golgi were identified, and the angle between the two points relative to the direction of flow was quantified. Three fields were used for quantification from each condition of the experiments. The wind rose plots were compiled using Origin software (OriginLab 2019, Northampton, MA).

Isometric Measurements of Aortic Function.

Thoracic aortic rings (2 mm in length) were mounted on 200 μM pins in a 6-mL vessel myograph (Danish Myo Technology) containing physiological salt solution (PSS) consisting of 130mM NaCl, 4.7 mM KCl, 1.18mM KHPO₄, 1.17 mM MgSO₄, 1.6 mM CaCl₂, 14.9 mM NaHCO₃, 5.5 mM dextrose, and 0.03 mM CaNa₂/EDTA. Vessels were stretched to 1g basal tension at 37°C and aerated with 95% O₂ and 5% CO₂. Vessels were equilibrated in PSS for 1h, followed by two consecutive contractions with PSS containing 60 mM potassium and 1 μM phenylephrine, then with KPSS alone. Rings were then washed, allowed to return to basal tension, and subjected to concentration-response curves to increasing concentrations of phenylephrine, acetylcholine, and nitroglycerin. Percentage of relaxation was calculated in reference to the degree of phenylephrine pre-contraction.

Telomerase Length

Telomere length was analyzed using a quantitative polymerase chain reaction (RT-qPCR)-based method previously described ^{35, 36}. The relative telomere length was calculated as the ratio of telomere repeats to a single-copy gene (SCG) (T/S ratio). The acidic ribosomal phosphoprotein PO (36B4) gene was used as the SCG. All qPCRs were performed in duplicates. The primers used for the telomere and the SCG amplification were as follows: telomere forward: 5′ GGT TTT TGA GGG TGA GGG TGA GGG TGA GGG TGA GGG T, telomere reverse: 5′ TCC CGA CTA TCC CTA TCC CTA TCC CTA TCC CTA TCC CTA; SCG forward: 5′ CAG CAA GTG GGA AGG TGT AAT CC and SCG reverse: 5′ CCC ATT CTA TCA TCA ACG GGT ACA A.

Mouse Lung Endothelial Cell Transfection

PGC1α-ECKO MLECs were transfected using either pcDNA3 (Invitrogen, V790–20) or pBABEhygroHigheGFPhTERT (Addgene, 28169) plasmids with DharmaFECT 3 reagent for 6 hours in OptiMEM. MLECs were then changed to growth media containing a 1:1 mixture of DMEM and Ham’s F-12 (Gibco), 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin. After 48 hours of plasmid transfection, cells were exposed to laminar flow 12 dynes/cm² (Ibidi system) for 48 hours. Wild type MLECs were used as a control.

Statistical Analysis

All data are expressed as mean ± SEM, and the numbers of independent experiments are indicated. Our analysis strategy involved first testing equal variance and normality of data to determine whether parametric or nonparametric tests were to be used. This strategy lead us to statistical comparisons between 2 groups by use of the parametric Student t-test and multiple groups using one- or two-way ANOVA with a post-hoc Tukey–Kramer multiple comparison as indicated in legends. A p-value is provided up to 3 decimal digits in figures. All statistics were done using StatView version 5.0 (SAS Institute, Cary, NC) or GraphPad Prism version 9 (GraphPad Software, La Jolla, CA).

Results

PGC1α is needed for endothelial cell response to fluid shear stress (FSS)

Endothelial cell alignment in the direction of flow involves nuclear polarization and elongation in response to physiological levels of fluid shear stress ⁶. We examined the response of human aortic endothelial cells (HAECs) to oscillatory vs laminar FSS (disturbed vs undisturbed flow). As expected, oscillatory FSS yielded random HAECs orientation, whereas laminar FSS produced HAECs alignment in the direction of flow (Supplemental Figure. 1A and 1B). Laminar FSS also led to increased expression of key flow-responsive genes, including KLF2 ¹⁸, KLF4 ^{19, 37}, HMOX1 ³⁸, and HEY1 ³⁹ in both HAECs and human umbilical vein endothelial cells (HUVECs) (Supplemental Figure. 1C–D). Since PGC1α has implications for endothelial function ^{21, 40}, we examined its expression as a function of FSS. We found that PGC1α mRNA and protein levels were significantly higher with laminar shear stress (LSS) compared to static control in the human endothelium (Figure. 1A–B). Similarly, PGC1α mRNA was upregulated with laminar compared to oscillatory flow in two different human endothelial cell types (Figure. 1B and Supplemental Figure. 2). These data suggest that PGC1α may play an important role during physiological fluid shear stress.

To determine the direct roles of PGC1α, we utilized loss- and gain-of-function strategies. We isolated mouse lung endothelial cells (MLECs) from WT mice or endothelial cell specific PGC1α knockout (PGC1α-ECKO) mice. Wild-type MLECs subjected to laminar FSS for 48 hours exhibited flow alignment, whereas PGC1α−ECKO MLECs did not (Figure. 1C). Next, a gain-of-function strategy was used with MLECs from mice with endothelial specific PGC1α overexpressing transgenic mice (PGC1α-ECTG), that exhibit ~60% increase in endothelial PGC1α compared to WT MLECs ²¹. We found that the static PGC1α-ECTG MLECs exhibited upregulation of the flow-responsive genes KLF4 ^{8, 19} and HMOX1 ³⁸ compared to wild-type cells (Figure. 1D). Moreover, PGC1α-ECTG endothelium demonstrated suppressed ICAM1 and VCAM1 mRNA expression (Figure. 1D) compared to wild type endothelium. Conversely, HAECs with PGC1α suppression by siRNA displayed blunted upregulation of HMOX1 and KLF2 ¹⁸ in response to laminar FSS compared to the scrambled siRNA control (Figure. 1E).

To determine the impact of PGC1α in vivo, we harvested intact aortic segments from areas of disturbed (inner arch) and laminar (descending thoracic) flow as in Figure. 1H. We observed relatively greater PGC1α mRNA expression in the laminar flow segments compared to areas of disturbed flow (Figure. 1F–G). Similarly, mRNA expression of HMOX1, KLF2, and KLF4 were upregulated in the laminar vs. disturbed flow regions of the aorta in WT and PGC1α-ECTG mice (Figure. 1F–G). Western blot analysis of aortic segments from areas of disturbed and laminar flow indicated that PGC1α and HMOX1 protein expression in both WT and PGC1α-ECTG mice were significanly upregulated in laminar flow areas of the aorta compared to areas of disturbed flow (Supplemental Figure. 3A–3B). We also isolated endothelium from disturbed and laminar flow aortic segments in WT mice and found that endothelial PGC1α and HMOX1 gene expression were significantly upregulated in laminar vs. distubed flow segments (Supplemental Figure. 3C).

We used en face staining of the aorta with the endothelial junction marker, β-catenin, to assess morphological endothelial responses to fluid shear stress. Endothelial cells adapt to laminar FSS by exhibiting planar cell polarity in the direction of flow ^{41, 42}. The endothelial length to width ratio is a key index of flow alignment and planar cell polarity⁴³. Qualitative assessment of en face staining demonstrated a highly elongated polygon endothelial cell shape and orientation to the vessel’s axis in the descending aorta of WT mice (Figure. 1I–J). Conversely, we observed a reduced endothelial length to width ratio in PGC1α-ECKO descending aorta compared to WT littermate controls (Figure. 1I–J). Collectively, these data suggest that PGC1α is a key element of the endothelial morphological and genetic response to fluid shear stress.

Endothelial PGC1α is required for the endothelial response to exercise

Prolonged exercise is known to increase vascular FSS into the high physiological range and is associated with the upregulation of key FSS-responsive genes that include NOS3, KLF2/4, and HMOX1 ^44–46. In WT mice subjected to chronic exercise, PGC1α mRNA expression in the aorta (Fig. 2A) and lung endothelium (Fig. 2C) was significantly greater than sedentary animals. We also observed that HMOX1, as well as telomerase reverse transcriptase (TERT), a gene sensitive to PGC1α status in vascular smooth muscle ⁴⁷, were upregulated in WT mice with chronic exercise, but not in PGC1α-ECKO mice (Figure. 2A). Since PGC1α contributes to reactive oxygen species (ROS) detoxification ⁴⁸, we examined ambient vascular ROS as a function of exercise. Superoxide, mitochondrial ROS, and H₂O₂ were estimated by dihydroethidium (DHE), MitoSOX, and Amplex Red, respectively. We found that chronic exercise reduced all three indices of ROS in WT mice; however, this effect was lost in PGC1α-ECKO animals (Figure. 2B). Endothelial function determined as aortic relaxation to acetylcholine (Ach) improved with chronic exercise (Figure. 2D) in control mice, but this effect was lost in PGC1α-ECKO animals. Endothelial independent relaxation (using nitroglycerin, NTG) was unchanged (Figure. 2E), suggesting that the beneficial effect of exercise requires the presence of PGC1α specifically in vascular endothelium and not smooth muscle (Figure. 2D–E). These data imply that PGC1α is an important component of endothelial function during exercise and is required for full exercise-induced HMOX1 and TERT upregulation.

Figure 2. PGC1α is required for endothelial functional responses to exercise.

A) Aortae were isolated from WT or PGC1α-ECKO mice and gene expression determined in total aortic tissue by RT-qPCR for the indicated genes before and after exercise; n= 7/group (PGC1α mRNA expression), n=18–32/group (TERT) and n=12–21/group (HMOX1). B) Reactive oxygen species (ROS) production was measured as indicated in WT and PGC1α-ECKO mice before and after exercise; n=8–12/group (MitoSOX), n=16–17/group (DHE) and n=10–20/group (AmplexRed). C) MLECs were isolated from WT mice with or without exercise, and PGC1α mRNA expression determined by RT-qPCR before and after exercise. D) Endothelial function measured as aortic isometric force in response to acetylcholine (Ach; n=10–12). E) Nitroglycerin-mediated smooth muscle cell function by treatment and genotype (NTG; n= 8–12;). Statistically significant differences were measured by Student’s t-test or either one-way or two-way ANOVA with post-hoc comparison as appropriate with control group. The data are mean ± SEM. P values vs. control mice (Black); vs. control + exercise mice (Red) and vs. PGC1α-ECKO mice (Blue).

eNOS controls PGC1α, TERT and HMOX1 expression during FSS

Laminar flow can induce endothelial nitric oxide synthase (eNOS; NOS3) activation and nitric oxide production in the endothelium ^49–51. This shear stress activates eNOS, which can lead to endothelial protection by triggering a large number of molecular protective intracellular pathways ^49–51. In addition, nitric oxide is well known to activate TERT expression and activity, ⁵² as well as PGC1α expression ⁵³. Therefore, we examined the roles of eNOS in PGC1α, TERT and HMOX1 expression during FSS using eNOS shRNA in HAECs during FSS. Our data showed that eNOS is required for FSS mediated PGC1α, TERT and HMOX1 expression in human endothelium (Supplemental Figure. 4).

PGC1α impacts FSS-dependent HMOX1 expression via ERRα and KLF4

PGC1α impacts gene expression by the co-activation of transcription factors ⁵⁴. The transcription factors ERRα and KLF4 play a critical role in regulating endothelial function ^{21, 55}. In cardiac myocytes, ERRα and KLF4 interact with each other and can form a complex with the PGC1α protein ⁵⁶. Therefore, we examined the role of ERRα and KLF4 in PGC1α-mediated endothelial responses to laminar FSS. First, using siRNA directed against ERRα in HAECs, we observed that ERRα controls endothelial upregulation of HMOX1 mRNA and protein expression in response to laminar FSS (Figure. 3A and Supplemental Figure. 5A). Similarly, KLF4 siRNA inhibited HMOX1 mRNA and protein expression in the setting of laminar FSS (Figure. 3B and Supplemental Figure. 5B). In contrast, KLF4 and ERRα had no reciprocal effect on the other’s expression with laminar FSS (Figure. 3A–B). As HMOX1 expression depends on both ERRα and KLF4, we tested their interaction in endothelial cells. ERRα immunoprecipitation demonstrated that it exists in a complex with PGC1α and KLF4 in human endothelial cells (Figure. 3C). These data suggest that ERRα with KLF4 play an important role in PGC1α regulated HMOX1 regulation.

PGC1α dictates telomerase reverse transcriptase expression in the endothelium

One identified PGC1α-dependent gene is telomerase reverse transcriptase (TERT) that, along with the telomerase RNA component (TERC), forms the telomerase complex that has been implicated in vascular aging ⁴⁷. In cultured endothelial cells, oxidative stress stimulates nuclear export of TERT to the mitochondria as a protective mechanism ⁵⁷, and inhibition of telomerase impairs flow-mediated nitric oxide bioactivity in human arterioles ⁵⁸. However, the role of TERT in endothelial responses to FSS is incompletely understood. Therefore, we used PGC1α gain- and loss-of-function to examine its implications for TERT in FSS responses. We found that PGC1α-ECTG MLECs exhibited upregulation of TERT mRNA and protein (Figure. 3D–E), whereas PGC1α shRNA impaired endothelial TERT protein expression (Figure. 3F). Also, we examined TERT expression as a function of FSS. Similar to the aortic PGC1α (Figure. 2A), exercise-induced FSS upregulated TERT mRNA in both aorta and MLECs isolated from WT mice, but not in PGC1α-ECKO mice (Figure. 2A, 3G). Furthermore, we found that similar to PGC1α, TERT expression is higher in the laminar flow (thoracic) region of the aorta than in the disturbed flow (inner arch) region (Figure. 3H).

To determine how PGC1α controls the expression of TERT mRNA, we performed a chromatin immunoprecipitation (ChIP) assay for the human TERT promoter (GenBank# AH007699), which has two potential binding sites for KLF4 (Figure. 3I; position 6612bp, 8649bp) and one for ERRα (Figure. 3I; 8271bp). ChIP of the human TERT promoter with PGC1α antibody indicated that PGC1α binds to the TERT promoter region (Figure. 3I–J). Collectively, these data demonstrate that PGC1α is required for TERT expression.

TERT is required for endothelial alignment, elongation, and polarization

We next examined the impact of TERT on endothelial cell responses to FSS using both pharmacologic and molecular approaches. Endothelial cells treated with the TERT inhibitor, BIBR1532 (iTERT) ⁵⁹, exhibited impaired orientation and alignment in the direction of laminar FSS compared to vehicle-treated cells (Figure. 4A). Pharmacologic TERT inhibition had no impact on PGC1α expression (Figure. 4B). Next, we treated HAECs with siRNA directed against TERT (siTERT) and observed ~80% reduction in TERT expression levels (Supplemental Figure. 6). Suppression of TERT expression also inhibited HAECs visual alignment to laminar FSS compared to scrambled siRNA control (Figure. 4C). This lack of alignment corresponded to a reduced length to width ratio in response to laminar FSS in the presence of either pharmacologic or molecular TERT inhibition (Figure. 4C–D). Another endothelial response to laminar FSS is nuclear polarization with the Golgi directed against the direction of flow ⁶⁰. We examined this phenomenon using the Golgi marker, GOLPH4, and found that TERT inhibition prevented nuclear polarization towards laminar flow (Figure. 4E–F). Collectively, these data show that TERT plays an important role in endothelial cell alignment and polarization in response to laminar FSS.

Figure 4. TERT is required for endothelial alignment to the flow.

A) Bright-field image of HAECs after exposure to laminar FSS in the presence of either control or TERT inhibitor. B) mRNA was isolated and PGC1α expression for cells treated with the vehicle control or TERT inhibitor was measured by RT-qPCR (n = 6). C) HAECs morphology was measured by staining for CD31/DAPI as a function of pharmacologic (iTERT) or genetic (siTERT) inhibition of TERT (Scale bar, 25 μm). D) Length to width ratio of CD31 stain in HAECs was measured after TERT inhibition in the presence of laminar FSS (n = 15–19). E) HAEC nuclear polarization towards the direction of laminar flow was measured with GOLPH4 (Golgi) and DAPI (nuclei) staining with and without TERT inhibitor (Scale bar, 10 μm). F) Compass plots of Golgi/nuclear angle as a function of TERT inhibition. Each ring represents an observation of an average of different fields of control or TERT inhibitor-treated cells. All the experiments were repeated 3 – 6 times. Statistically significant differences between groups are indicated (P values vs. control by Student’s t-test). The data are mean ± SEM.

TERT is required for mitochondrial responses to FSS

PGC1α has been known to play an important role in mitochondrial function and biogenesis ^61–63. Our data also showed a reduced mitochondrial staining in PGC1α knockout MLECs compare to WT MLECs (Supplemetal Figure. 7). TERT is known to have non-canonical functions that include translocation to the mitochondria and mitochondrial stabilization with oxidative stress ^{57, 64}. We probed the implications of TERT in the endothelial FSS response by exposing HAECs to oscillatory vs laminar FSS in the presence or absence of TERT inhibition. We found that laminar FSS produced elongated and branched mitochondrial staining with MitoTracker (Figure. 5A). In contrast, TERT inhibition produced punctate mitochondrial morphology that was reminiscent of oscillatory flow (Figure. 5A). Composite data also indicated that TERT inhibition with laminar FSS produces an endothelial response that mimics oscillatory flow with respect to mitochondrial mass by MitoTracker staining and network formation by mitochondrial length (Figure. 5B). To determine the role of TERT on mitochondrial ROS, we stained laminar FSS exposed HAECs with MitoSOX as a function of TERT inhibition. Compared to vehicle-treated cells, TERT inhibition enhanced the mitochondrial ROS signal (Figure. 5C). Next, we performed immunostaining for TERT and the mitochondria marker, CoxIV. We found that during FSS, TERT colocalizes with the mitochondria in endothelial cells (Supplemetal Figure. 8).

TERT has been implicated in telomere length maintenance. Although our experiments were short-term, we examined telomere length over the time course of our experiment and found that none of the conditions used for TERT inhibition had any impact on telomere length (Supplemetal Figure. 9). These data imply that TERT plays an important role in the endothelial mitochondrial response to laminar FSS that is independent of its canonical function on telomere length. To determine if these findings play a role in vivo, we used TERT knockout mice. En face staining of endothelial borders with β-catenin in laminar flow segments, revealed qualitatively impaired flow alignment in TERT knockout vs wild-type mice (Figure. 5D). Composite data of the length:width ratio revealed significantly reduced endothelial elongation in TERT knockout aorta (Figure. 5E). These data strongly suggest that TERT is required for optimal endothelial response to laminar FSS in vivo.

TERT is required for normal laminar FSS-induced HMOX1 expression

Our data showed that similar to the KLF4, HMOX1 was also upregulated with laminar FSS (Supplemental Figure 1). Since our data indicated that both PGC1α (Figure 1E) and ERRα (Figure 3A) participate in laminar FSS-induced HMOX1 expression, we tested the role of PGC1α, ERRα and TERT in this process. First we looked at the expression of HMOX1 in our PGC1α overexpressing transgenic mice. Endothelial overexpression of PGC1α increased HMOX1 protein expression under static conditions (Figure. 6A); whereas, siRNA against ERRα attenuated HMOX1 protein expression during laminar flow (Supplemental Figure. 5A).

Consistent with these findings, exercise-induced FSS upregulated HMOX1 expression in the aorta as well as in MLECs isolated from exercised mice in a PGC1α-dependent manner (Figure. 2A, 6B). In terms of TERT, we observed that laminar FSS upregulated HMOX1 protein compared to oscillatory FSS (Figure 6C). Furthermore, our data showed that this upregulation of HMOX1 during laminar flow was dependent upon TERT activity as it was inhibited by a TERT specific inhibitor, BIBR1532 (Figure. 6C). HMOX1 expression appears to be downstream of PGC1α/TERT/KLF2 as HMOX1-directed shRNA had no impact on laminar FSS-mediated expression of these genes (Figure. 6D). Next, we performed immunostaining for HMOX1 with mitochondria. Similar to TERT, we found that during FSS HMOX1 colocalizes with mitochondria in HAECs (Supplemental Figure. 10).

HMOX1 is required for the endothelial cell response to the laminar flow

To examine if HMOX1 dictates endothelial FSS responses, we used a pharmacological approach with two structurally distinct HMOX1 inhibitors, zinc (II) protoporphyrin IX (ZnPP) and 1-[[2-[2-(4-Bromophenyl)ethyl]-1,3-dioxolan-2-yl]methyl]-1H-imidazole hydrochloride (OB 24 hydrochloride)^65–67 using copper protoporphyrin IX (CuPP) as a control. CuPP is a similar compound to ZnPP but does not inhibit HMOX1 ⁶⁵. Both HMOX1 inhibitors impaired HAECs flow alignment in response to laminar FSS compared to control cells treated with CuPP (Figure. 6E). Similarly, HMOX1 inhibition impaired mitochondrial network formation in response to laminar FSS, leaving HAECs with punctate mitochondria as determined by MitoRed staining (Figure. 6F).

TERT can rescue the expression of HMOX1 in PGC1α knockout MLECs

Since, both PGC1α and TERT are required for endothelial alignment with laminar flow (Figure. 1 and 4), we sought to understand their effects via complementation studies using PGC1α-ECKO MLECs expressing either control plasmid or TERT-WT plasmid during FSS. We found that TERT can rescue the expression of the downstream target HMOX1 in PGC1α-ECKO MLECs during laminar flow (Supplemetal Figure. 11).

PGC1α-TERT-HMOX1 pathway regulates endothelial to mesenchymal transition during FSS

Exposure to low flow rate or disturbed flow causes induction of endothelial to mesenchymal transition, resulting in endothelial dysfunction^14–16. To determine the role of the PGC1α -TERT- HMOX1 pathway in the endothelial to mesenchymal transition, we used RT-qPCR for the expression of mesenchymal markers in endothelium isolated from PGC1α-ECTG mice as well as HAECs treated with TERT and HMOX1 siRNA. We found that forced overexpression of PGC1α in MLECs suppresses the mesenchymal markers SM22α (Tagln), Sca1 (Atxn1), MMP2 and ICAM1 (Supplemetal Figure. 12A). Similarly, mesenchymal markers were upregulated in HAECs treated with TERT siRNA or HMOX1 siRNA when compared to control (Supplemetal Figure. 12B–C). These data support a cascade pathway, depicted in Figure. 6G, whereby PGC1α/ERRα contributes to laminar FSS-induced TERT expression that is needed for the HMOX1 upregulation required for full endothelial FSS responsiveness.

Discussion

The data presented here indicates that PGC1α is a key element of the endothelial response to fluid shear stress. In the absence of PGC1α, endothelial cells exhibited an impaired ability to align and elongate as well as upregulate key genes characteristic to the response to laminar FSS. These findings appear relevant in vivo, as aortic segments with laminar FSS exhibited PGC1α upregulation; whereas, animals lacking endothelial PGC1α demonstrated impaired flow alignment in the laminar flow sections of the aorta. Similarly, exercise-induced FSS was associated with improved aortic endothelial function, antioxidant defenses, and reduced ROS which were all impaired in the absence of endothelial PGC1α. We found that PGC1α exerted its influence via a complex, including ERRα and KLF4, that affords PGC1α binding to the TERT promoter to increase TERT expression. In keeping with these findings, TERT upregulation was required for the action of PGC1α as TERT loss-of-function recapitulated the endothelial PGC1α loss-of-function phenotype, including endothelial elongation, flow alignment, mitochondrial stabilization, and HMOX1 upregulation with laminar FSS. With regard to the latter, we found that HMOX1 loss-of-function was a key downstream mediator of endothelial PGC1α as HMOX1 inhibition reproduced the endothelial PGC1α-null phenotype with regards to including flow alignment, elongation, and mitochondrial stabilization. Taken together, these data identify a PGC1α-TERT-HMOX1 axis as a key element of endothelial cell responses to laminar FSS.

Endothelial NOS (eNOS; NOS3) is a key component of endothelial function. It is known that shear stress can upregulate and activate eNOS in the endothelium ^{49–51, 68, 69}. In this study we have clearly shown that eNOS is a key component for shear-induced PGC1α-TERT-HMOX1 upregulation in the endothelium. One earlier study has implicated the role of eNOS in TERT regulation ⁵². These observations are generally consistent with prior literature which implicated the role of nitric oxide, via eNOS, is important for mitochondrial biogenesis ^{70, 71} and, in some instances, PGC1α activation ⁷⁰. Initial reports on the mechanism of this response pointed to NO^• -mediated cGMP production as a key element ⁷⁰. However, most prior studies investigated eNOS-mediated control of mitochondria in non-endothelial tissues. Whether or not similar mechanisms apply to the endothelial cell per se, are not yet clear.

In response to laminar FSS, we found that upregulation of TERT was required for endothelial elongation, flow alignment, and nuclear polarization. These observations identify TERT as another FSS-sensitive factor that dictates, in part, the endothelial phenotype in response to FSS. The classical function of TERT is restricted to the telomerase complex that prevents cellular senescence typical of aging, particularly in highly replicating tissues ⁷². However, we found no change in endothelial telomere length over the time course of our study, suggesting that the role of TERT in FSS is independent of its nuclear telomerase function. This contention is supported by studies identifying non-canonical functions of TERT in more quiescent tissues. For example, mice lacking TERT exhibit profound repression of PGC1α expression in the liver, that mediates impaired mitochondrial biogenesis and function, including gluconeogenesis ⁷³. Endothelial TERT undergoes nuclear export in response to oxidative stress ⁷⁴, and localizes in mitochondria to protect the mitochondrial DNA ^{57, 64}. This latter effect may be a consequence of its reverse transcriptase activity towards mitochondrial RNAs ⁷⁵. More recently, TERT has been implicated in the switch between NO^•-mediated and ROS-mediated vasodilation in the human microcirculation ^{58, 76} characteristic of both aging and coronary disease. Whether the functions of TERT in our system are related to its nuclear or mitochondrial localization is not yet clear and will require further study. One might speculate the latter seems plausible since we found TERT was required for the endothelial mitochondrial responses to laminar FSS.

Our study identified TERT upregulation as a key component for the upregulation of HMOX1 with laminar FSS. This is a novel observation that adds to the longstanding knowledge that laminar FSS upregulates HMOX1 and other genes under the control of the antioxidant response element (ARE) ⁷⁷ via activation of the NRF2-KEAP1 system ⁷⁸. These data align with the prior knowledge that one stabilizing component of laminar flow on the endothelium is the promotion of an antioxidant state and that HMOX1 upregulation is a key element in this process ⁷⁹. Our data prompts new speculation that TERT could be an important component of the overall cellular ARE phase II detoxification responses. This speculation is consistent with prior observations of TERT nuclear export and activation in response to endothelial oxidative stress ⁸⁰. The precise element(s) in the NRF2-ARE pathway that are sensitive to TERT-mediated regulation, however, are not yet clear, although cooperativity between KLF2 and NRF2 has been described in the endothelium ⁸¹.

An important element of our study is the observation that HMOX1 upregulation is key to endothelial flow alignment and elongation in response to laminar FSS. This is a profound observation in that it adds a completely new consequence of HMOX1 in endothelial cell phenotype. The classic function attributed to HMOX1 is heme degradation, whereas, with laminar FSS, HMOX1 is thought to promote the antioxidant phenotype indicative of quiescent endothelium in atherosclerosis-resistant vascular sites ⁷⁹. We observed co-localization of TERT controls HMOX1 in the endothelium, suggesting some level of cooperativity. Both proteins are known to associate with the nucleus and mitochondria ^{75, 82}. Additionally, TERT association with the latter appears to enhance endothelial resistance to oxidative stress via reverse transcriptase activity ⁸³. HMOX1 localization in the mitochondria is in association with biliverdin reductase and cytochrome P-450 reductase, suggesting its local function is in heme degradation ⁸². In contrast, HMOX1 nuclear localization has been associated with the upregulation of genes important for oxidative stress ⁸⁴. Our data indicate that HMOX1 inhibition did not impact the transcription of genes for PGC1α, KLF2, or TERT; however, some transcriptional role of HMOX1 in other laminar FSS-responsive genes cannot yet be ruled out. Nevertheless, our data provide a new role for HMOX1 in the endothelium though the definition of the specific mechanism(s) involved will require further studies.

Highlights.

• Peroxisome proliferator gamma coactivator-1α (PGC-1α) is a flow-responsive gene required for laminar flow induced endothelial cell alignment in vitro and in vivo.

• PGC-1α regulates laminar fluid shear stress (FSS)-induced expression of telomerase reverse transcriptase (TERT) in vitro and in vivo.

• PGC1α forms a complex with estrogen-related receptor alpha (ERRα) and Kruppel-like factor 4 (KLF4) on the TERT promoter.

• PGC1α-TERT axis is vital for expression of heme oxygenase-1 (HMOX1), a gene required for endothelial alignment during laminar FSS.

Nonstandard Abbreviations and Acronyms

HUVECs

human umbilical vein endothelial cells

HAECs

human aortic endothelial cells

FSS

fluid shear stress

TERT

telomerase reverse transcriptase

HMOX1

heme oxygenase-1

KLF4

Kruppel-like factor 4

ERRα

estrogen-related receptor alpha

PGC1α

peroxisome proliferator gamma coactivator-1α