High Fluid Shear Stress Inhibits Cytokine‐Driven Smad2/3 Activation in Vascular Endothelial Cells

Abstract

Background

Atherosclerosis occurs preferentially in regions of low and disturbed fluid shear stress (FSS) but is limited in regions of high laminar FSS as a result of inhibition of endothelial inflammatory pathways. Recent work has identified endothelial to mesenchymal transition (EndMT) driven by TGFβ2 (transforming growth factor beta 2)–Smad2/3 (mothers against decapentaplegic) signaling as a critical component of atherogenesis. However, interactions between FSS and EndMT in this context have not been investigated.

Methods and Results

Endothelial cells were treated with TGFβ2 and inflammatory cytokines (interleukin 1β and tumor necrosis factor alpha) with or without high FSS in a parallel plate flow chamber. Smad2/3 nuclear translocation and target gene expression, assayed by immunofluorescence and quantitative polymerase chain reaction, revealed that high FSS blocked the Smad2/3‐EndMT pathway. In vivo, mice were injected with TGFβ2 and inflammatory cytokines, then regions of the aorta under low versus high FSS were examined. TGFβ2 and inflammatory cytokine treatment stimulated Smad2/3 nuclear translocation and target gene expression predominantly in regions of low FSS with little effect in regions of high FSS.

Conclusions

High FSS inhibits endothelial Smad2/3 activation and EndMT in response to inflammatory mediators, resulting in selective EndMT at athero‐susceptible, low FSS regions of arteries.

Nonstandard Abbreviations and Acronyms

EC

endothelial cell

EndMT

endothelial to mesenchymal transition

FN1

fibronectin

FSS

fluid shear stress

HUVEC

human umbilical vein endothelial cell

ICAM1

intercellular adhesion molecular 1

ID1

inhibitor of DNA‐binding protein 1

IL1β

interleukin 1 beta

KLF2

Krüppel‐like factor 2

MEKK

mitogen‐activated protein kinase/extracellular signal‐regulated kinase kinase kinase

MMP2

matrix metalloproteinase 2

NF‐kB

nuclear factor kappa B

Smad2/3

mothers against decapentaplegic

Snail2

snail family transcriptional repressor 2

TGFβ

transforming growth factor beta

TNFα

tumor necrosis factor alpha

Clinical Perspective.

What Is New?

• High fluid shear stress prevents TGFβ (transforming growth factor beta) and inflammatory cytokine‐induced Smad2/3 (mothers against decapentaplegic) nuclear translocation and target gene expression in vitro and in vivo.

• Region‐selective activation of endothelial Smad2/3 by TGFβ and inflammatory cytokines are thus likely to contribute to formation and progression of atherosclerotic lesions.

What Are the Clinical Implications?

• Identification of high fluid shear stress as a potent inhibitor of Smad2/3 activation in endothelial cells implies that activation of elements in the high fluid shear stress pathway is a potential strategy for combating atherosclerosis.

Endothelial cells (ECs) lining the inner surfaces of blood vessels sense fluid shear stress (FSS) from flowing blood to regulate the expression of thousands of genes and profoundly affect EC phenotype. Physiological levels of laminar FSS promote the expression of anti‐inflammatory and antioxidative genes, protecting the vessel wall from inflammatory and metabolic risk factors. Atherosclerotic lesions thus form near branch points and curvatures, where FSS is of lower magnitude and shows complex changes in direction, commonly called disturbed or atherogenic shear stress. ¹ , ²

A major paradigm in atherosclerosis is the synergy between soluble inflammatory mediators and shear stress, such that high FSS blocks activation of inflammatory transcriptional pathways such as NF‐kB (nuclear factor kappa B) and c‐Jun N‐terminal kinase, while low shear stress permits or enhances these responses. ³ , ⁴ These interactions between FSS and inflammatory pathways are thought to underlie the site specificity of atherosclerosis. ¹ The transcription factor KLF2 (Krüppel‐like factor 2) is strongly induced by high FSS and is thought to mediate a large fraction of its antiatherosclerotic effect. ⁴

More recent studies have found that endothelial to mesenchymal transition (EndMT), driven by both increased TGFβ (transforming growth factor beta) secretion and increased sensitivity of ECs to TGFβ, is a major contributor to atherosclerosis. ⁵ , ⁶ , ⁷ , ⁸ ECs that undergo complete EndMT contribute to the neointima, while ECs that have undergone partial EndMT acquire an inflammatory phenotype, secreting multiple inflammatory mediators and recruiting or activating leukocytes in the vessel wall. ⁵ , ⁹ Importantly, cytokines such as TNFα (tumor necrosis factor alpha), interleukin 1 beta (IL1β), and γ interferon sensitize ECs to TGFβ and thus promote EndMT. ⁸ , ¹⁰ These regulatory networks thus create a positive feedback loop that drives disease progression. ⁵ , ¹⁰ Indeed, endothelial‐specific knockout of TGFβ receptors strongly reduces plaque size and even induces plaque regression, ⁵ , ¹⁰ supporting a role for TGFβ signaling in initiation and progression of atherosclerosis.

Our previous work showed that the activation and nuclear translocation of Smad2/3 (mothers against decapentaplegic) revealed a distinct maximum at low FSS, attributable to a sensitive activation mechanism that is triggered by shear at ≈1 dyne/cm² and suppressed by high FSS >15 dynes/cm². ¹¹ However, the wider significance of these effects in the context of inflammation and atherosclerosis has not been explored. We therefore investigated the role of high FSS in regulating the TGFβ‐Smad2/3‐EndMT pathway in response to soluble atherogenic factors.

METHODS

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

Animals

Twelve‐week‐old C57BL/6J mice from The Jackson Laboratory (Strain #: 000664) were used in this study. All mouse protocols and experimental procedures were approved by the Yale University Institutional Animal Care & Use Committee.

Cell Culture and Small Interfering RNA Transfection

HUVECs (human umbilical vein endothelial cells) were obtained from the Yale Vascular Biology & Therapeutics Program tissue culture core laboratory at passage 1. HUVECs were maintained in EGM2 Endothelial Cell Growth Medium‐2 (Lonza) and used for experiments between passage 2 and passage 5. Small interfering RNA (siRNA) transfection was performed using Opti‐MEM medium (ThermoFisher) and Lipofectamine RNAiMAX (Invitrogen). Cells were used for experiments 3 to 4 days after transfection. IL1β and TNFα were purchased from PeproTech. TGFβ2 was purchased from R&D Systems. ON‐TARGET Plus Smartpool siRNA from Dharmacon were used for knockdown of human Smad2 (L‐003561‐00‐0005) and human Smad3 (L‐020067‐00‐0005).

In Vitro Stimulation

Cultured HUVECs in basal medium, with or without the indicated shear stress patterns, were stimulated by addition of 1 ng/mL TGFβ, 10 ng/mL IL1β, or 10 ng/mL TNFα to the medium. A preliminary experiment (Figure S1) determined the optimal times for Smad2/3 nuclear translocation: 6 hours for TGFβ and 12 hours for IL1β and TNFα. Cells were then fixed and stained for Smad2/3 and nuclear translocation quantified.

Shear Stress

HUVECs were seeded on tissue culture plastic slides coated with 20 μg/mL of fibronectin for 2 hours at 37 °C and grown to confluence. Shear stress with a calculated intensity of 25 dynes/cm² was applied in parallel flow chambers as described. ¹²

Immunohistochemistry

For in vitro experiments, cells were washed with PBS and fixed in 4% paraformaldehyde (Electron Microscopy Sciences) at room temperature for 10 minutes. Mice were intravenously injected with PBS or cytokines at indicated times then killed by isofluorane overdose. Aortas were perfused in situ with PBS, then perfusion‐fixed with 4% paraformaldehyde. Aortic arches were isolated, and the outer and inner curvatures were separated for staining. For aorta en face staining, samples were incubated in blocking buffer (5% donkey serum, 0.2% BSA, 0.3% Triton X‐100 in PBS), followed by incubation with primary and secondary antibodies diluted in blocking buffer. Negative controls were performed with nonimmune species– and isotype‐matched IgG, as presented in Figure S2. Images were taken using an SP8 confocal microscope (Leica).

RNA Isolation and Quantitative Real‐Time Polymerase Chain Reaction

RNA was extracted from cells with RNeasy Plus Mini Kit (QIAGEN) according to the manufacturer's instructions, and reverse transcription performed with the iScript Reverse Transcription Supermix for real‐time quantitative polymerase chain reaction (PCR) (BIO‐RAD). cDNA was then amplified by real‐time PCR with iQ SYBR Green Supermix (BIO‐RAD). The expression of target genes was normalized to expression of the housekeeping gene GAPDH or 18s RNA. Primers for quantitative PCR are listed in Table S1.

Antibodies

We used the following antibodies for immunohistochemistry: Rat anti‐Mouse CD31 (BD 550274; 1:200), Smad2/3 (Cell Signaling 8685S; 1:500), ICAM1 (Abcam ab179707; 1:800), Fibronectin (BD 610078; 1:400), Rabbit IgG (Vector Laboratories I‐1000; 1:400), Mouse IgG (Vector Laboratories I‐2000; 1:400). Alexa Fluor secondary antibodies used for immunohistochemistry were from Invitrogen.

Statistical Analysis

Statistical analysis was performed using GraphPad Prism software (GraphPad Software). Data were analyzed for normality and equal variance using the Shapiro–Wilk test and Brown‐Forsythe test, respectively. If both tests were passed, statistical significance was further analyzed by unpaired t test for 2 groups comparison or 1‐way ANOVA with Tukey post hoc test for multiple groups comparison. Statistical significance between 2 groups plus treatment was calculated by 2‐way ANOVA with Tukey or Bonferroni multiple comparison tests. A P value <0.05 was considered significant (*P<0.05, **P<0.01, ***P<0.001).

RESULTS

High FSS Inhibits TGFβ Signaling In Vitro

HUVECs in basal medium were treated with or without TGFβ2 under FSS at 4 dynes/cm² or 25 dynes/cm², or oscillatory shear stress at 0.5±4 dynes/cm² for 6 hours. TGFβ2 alone induced strong Smad2/3 nuclear translocation as expected, which was further increased under low FSS or oscillatory shear stress but nearly completely blocked by high FSS (Figure 1A and 1B; Figure S3). We next measured Smad2/3 target genes that are associated with the mesenchymal phenotype, specifically FN1 (fibronectin), N‐cadherin, ID1 (inhibitor of DNA‐binding protein 1), Snail2 (snail family transcriptional repressor 2), and MMP2 (matrix metalloproteinase 2). We also examined intercellular adhesion molecular 1 (ICAM‐1), which mediates leukocyte recruitment into artery walls in atherosclerosis and is induced by Smad2/3 as well as by NF‐kB. ¹⁰ , ¹³ , ¹⁴ Assaying mRNA levels by quantitative PCR showed that the response to TGFβ was modestly increased by low shear, whereas high FSS essentially eliminated TGFβ‐induced Smad2/3 target genes expression (Figure 1C). Thus, high FSS inhibits TGFβ signaling.

Figure 1. High fluid shear stress (FSS) prevents TGFβ2 (transforming growth factor beta 2)‐induced Smad2/3 (mothers against decapentaplegic) signaling in vitro.

A, HUVECs (human umbilical vein endothelial cells) were treated without or with 1 ng/mL TGFβ2 and FSS at 4 dynes/cm² or 25 dynes/cm² for 6 hours. Cells were fixed and stained for Smad2/3. Scale bar: 25 μm. B, Smad2/3 nucleus/cytoplasm intensity ratio was quantified. n=100 cells for each group from 3 independent experiments. C, HUVECs were treated without or with 1 ng/mL of TGFβ2 in the absence or presence of FSS at 4 dynes/cm² or 25 dynes/cm² for 24 hours. n=4 experiments. Quantitative polymerase chain reaction analysis of Smad2/3 target genes, FN1, N‐cadherin, ID1 (inhibitor of DNA‐binding protein 1), Snail2 (snail family transcriptional repressor 2), MMP2 (matrix metalloproteinase 2), and ICAM1 (intercellular adhesion molecular 1). Data represent mean±SEM. *P<0.05, **P<0.01, ***P<0.001, calculated by 1‐way ANOVA with Tukey multiple comparison tests. DAPI indicates the nuclear stain 4′,6‐diamidino‐2‐phenylindole.

High FSS Inhibits Activation of Smad2/3 by IL1β and TNFα

Inflammatory mediators play an important role in activating EC Smad2/3 in atherosclerosis, mainly by sensitizing the cells to TGFβ. ⁵ HUVECs were treated with the inflammatory cytokine IL1β with or without FSS at 4 dynes/cm² or 25 dynes/cm², or with oscillatory shear stress at 0.5±4 dynes/cm² for 12 hours. IL1β alone induced Smad2/3 nuclear translocation, which again was higher in ECs under low shear or oscillatory shear stress but essentially completely inhibited by high FSS (Figure 2A and 2B; Figure S4). Nearly identical results were observed with TNFα (Figure 3A and 3B; Figure S5). Assays of Smad2/3 target genes confirmed these findings (Figures 2C and 3C). Thus, inflammation‐induced Smad2/3 signaling is also inhibited by high FSS.

Figure 2. High fluid shear stress (FSS) prevents interleukin 1 beta (IL1β)–induced Smad2/3 (mothers against decapentaplegic) signaling in vitro.

A, HUVECs (human umbilical vein endothelial cells) were treated without or with 10 ng/mL IL1β and FSS at 4 dynes/cm² or 25 dynes/cm² for 12 hours. Cells were fixed and stained for Smad2/3. Scale bar: 25 μm. B, The Smad2/3 nucleus/cytoplasm intensity ratio was quantified. n=100 cells for each group from 3 independent experiments. C, HUVECs were treated without or with 10 ng/mL IL1β in the absence or presence of FSS at 4 dynes/cm² or 25 dynes/cm² for 24 hours. n=4 experiments. Quantitative polymerase chain reaction analysis of Smad2/3 target genes, FN1, N‐cadherin, ID1 (inhibitor of DNA‐binding protein 1), ICAM1 (intercellular adhesion molecular 1), VCAM1 (vascular cell adhesion molecular 1), and IL‐6 (interleukin 6). Data represent mean±SEM. *P<0.05, **P<0.01, ***P<0.001, calculated by 1‐way ANOVA with Tukey multiple comparison tests. DAPI indicates 4′,6‐diamidino‐2‐phenylindole.

Figure 3. High fluid shear stress (FSS) prevents tumor necrosis factor alpha (TNFα)–induced Smad2/3 (mothers against decapentaplegic) signaling in vitro.

A, HUVECs (human umbilical vein endothelial cells) were treated without or with 10 ng/mL TNFα and FSS at 4 dynes/cm² or 25 dynes/cm² for 6 hours. Cells were fixed and stained for Smad2/3. Scale bar: 25 μm. B, The Smad2/3 nucleus/cytoplasm intensity ratio was quantified. n=100 cells for each group from 3 independent experiments. C, HUVECs were treated without or with 10 ng/mL TNFα in the absence or presence of FSS at 4 dynes/cm² or 25 dynes/cm² for 24 hours. n=4 experiments. Quantitative polymerase chain reaction analysis of Smad2/3 target genes, FN1, N‐cadherin, ICAM1 (intercellular adhesion molecular 1), VCAM1 (vascular cell adhesion molecular 1), and interleukin 6 (IL‐6). Data represent mean±SEM. *P<0.05, **P<0.01, ***P<0.001, calculated by 1‐way ANOVA with Tukey multiple comparison tests. DAPI indicates 4′,6‐diamidino‐2‐phenylindole.

High FSS Inhibits TGFβ2‐Induced Smad2/3 Responses In Vivo

To validate these results in vivo, we next examined responses of the endothelium in high FSS/atheroprotected and low FSS/atherosusceptible regions of arteries in mice injected with TGFβ, using the outer and inner curvature of the aorta as regions of high and low/oscillatory FSS, respectively. TGFβ strongly induced Smad2/3 nuclear translocation in the inner curvature but had only weak effects in the outer curvature (Figure 4A and 4B). Examination of the Smad2/3‐inducible genes, fibronectin and ICAM‐1, showed that TGFβ similarly induced their expression in the inner but not the outer curvature (Figure 4C through 4F). Thus, effects in vivo correlate with those seen in vitro.

Figure 4. High fluid shear stress (FSS) inhibits TGFβ (transforming growth factor beta) signaling in vivo.

A, Mice were intravenously injected with PBS or 10 ng of TGFβ2. After 4 hours, aortas were isolated and the outer and inner curvature of the arches separated, processed as described in the Methods section and stained for Smad2/3, CD31, and 4′,6‐diamidino‐2‐phenylindole (DAPI) to label the nucleus. B, The Smad2/3 nucleus/cytoplasm intensity ratio was quantified. N=6 mice (3 male and 3 female) per group. C through F, Mice were intravenously injected with PBS or 10 ng/mL of TGFβ2. After 12 hours, the aortic arch was isolated and the outer and inner curvature separated, processed, and stained for fibronectin (C and D) and ICAM1 (intercellular adhesion molecular 1) (E and F). Fibronectin and ICAM1 intensity were quantified and normalized to the aorta outer curvature PBS group. N=6 mice (3 male and 3 female) per group. Scale bar: 25 μm. Data represent mean±SEM. *P<0.05, **P<0.01, ***P<0.001, calculated by 2‐tailed unpaired t tests.

High FSS Inhibits Inflammatory Cytokine‐Induced Smad2/3 Signaling In Vivo

We similarly sought to validate the effects of IL1β and TNFα in vivo. Mice were injected with inflammatory cytokines (IL1β and TNFα) and subsequent Smad2/3 nuclear translocation examined in different regions of the aorta. Both IL1β and TNFα induced Smad2/3 nuclear translocation in the inner curvature, but not in the outer curvature (Figures 5A, 5B and 6A, 6B). Staining artery segments revealed that IL1β and TNFα induced the Smad2/3‐inducible proteins fibronectin and ICAM‐1 in the inner but not in the outer curvature (Figures 5C through 5F and 6C through 6F). Thus, activation of Smad2/3 signaling in vivo correlates with regions of low FSS and is essentially undetectable in regions of high FSS.

Figure 5. High fluid shear stress (FSS) prevents interleukin 1 beta (IL1β)–induced Smad2/3 (mothers against decapentaplegic) signaling in vivo.

A, Mice were intravenously injected with PBS or 100 ng of IL1β. After 6 hours, the aortic arches were isolated, the outer and inner curvatures separated, and then processed and stained with 4′,6‐diamidino‐2‐phenylindole (DAPI) and antibodies to Smad2/3 and CD31. B, The Smad2/3 nucleus/cytoplasm intensity ratio was quantified. N=6 mice (3 male and 3 female) per group. C through F, Mice were intravenously injected with PBS or 100 ng/mL of IL1β. After 12 hours, aortic arches were isolated, the outer and inner curvature separated, and the samples processed and stained for FN1 (fibronectin 1) (C and D) and ICAM1 (intercellular adhesion molecular 1) (E and F). FN1 and ICAM1 intensity were quantified (normalized to aorta outer curvature PBS group). N=6 mice (3 male and 3 female) per group. Scale bar: 25 μm. Data represent mean±SEM. *P<0.05, **P<0.01, ***P<0.001, calculated by 2‐tailed unpaired t tests.

Figure 6. High fluid shear stress (FSS) prevents tumor necrosis factor alpha (TNFα)–induced Smad2/3 (mothers against decapentaplegic) signaling in vivo.

A and B, Mice were intravenously injected with PBS or 100 ng of TNFα. At 6 hours, aortic arches were isolated, the outer and inner curvatures separated, and then were processed and stained for Smad2/3 and CD31 and with 4′,6‐diamidino‐2‐phenylindole (DAPI) to label the nucleus (A). Smad2/3 nucleus/cytoplasm intensity ratio was quantified (B). N=6 mice (3 male and 3 female) per group. C through F, Mice were intravenously injected with PBS or 100 ng/mL of TNFα. At 12 hours, the aortic arches were isolated, the outer and inner curvatures separated, and the specimens were then processed and stained for fibronectin 1 (C and D) and ICAM1 (intercellular adhesion molecular 1) (E and F). Fibronectin and ICAM1 intensity were quantified (normalized to aorta outer curvature PBS group). N=6 mice (3 male and 3 female) per group. Scale bar: 25 μm. Data represent mean±SEM. *P<0.05, **P<0.01, ***P<0.001, calculated by 2‐tailed unpaired t tests.

Smad2/3 Mediate Cytokine‐Induced Genes Expression In Vitro

To confirm that cytokine‐induced gene expression occurs through Smad2/3, we silenced Smad2/3 in HUVECs using siRNA‐mediated knockdown. ECs were then treated with TGFβ2 or IL1β under FSS at 4 dynes/cm² or 25 dynes/cm², and gene expression was analyzed by quantitative PCR. Smad2/3 knockdown blocked the TGFβ2‐ and IL1β‐induced increase in FN1 and N‐cadherin under all conditions (Figure 7A and 7B). Expression of these genes in ECs is thus Smad2/3 dependent.

Figure 7. Smad2/3 knockdown blocks TGFβ2 (transforming growth factor beta 2) or interleukin 1 beta (IL1β)–induced genes expression in vitro.

A and B, HUVECs (human umbilical vein endothelial cells) were transfected with control (siCtrl) or Smad2/3 (siSmad2/3) small interfering RNA (siRNA) for 3 days, then treated with 1 ng/mL of TGFβ2 (A) or 10 ng/mL of IL1β (B) in the absence or presence of fluid shear stress (FSS) at 4 dynes/cm² or 25 dynes/cm² for 24 hours. Quantitative polymerase chain reaction analysis of Smad2, Smad3, FN1, and N‐cadherin, normalized to 18sRNA. n=4 samples per group. Data represent mean±SD. *P<0.05, **P<0.01, ***P<0.001, calculated by 2‐tailed unpaired t tests.

DISCUSSION

Atherosclerosis is a complex disease involving multiple inflammatory, metabolic, and biomechanical risk factors, with interactions among these factors governing the site‐specificity and progression of disease. EndMT, driven by inflammatory factors that enhance both secretion and sensitivity to TGFβ, establishes an inflammatory phenotype with expression of multiple inflammatory mediators. Interactions between inflammatory pathways and Smad2/3 signaling/EndMT thus establish a feedforward loop that promotes disease progression. The potent reversal of established lesions after blockade of TGFβ signaling argues for a key role of this pathway in the progression of atherosclerotic lesions. ¹⁰

The results of this study demonstrate that low and oscillatory flow potentiate, whereas high FSS inhibits Smad2/3 signaling and expression of mesenchymal and inflammatory genes in response to TGFβ and inflammatory cytokines (IL1β and TNFα). These results thus update and extend the major paradigm that high flow inhibits atherogenesis via its effects on inflammatory pathways to include the more recently elucidated TGFβ‐Smad2/3‐EndMT mechanism. For the underlying mechanisms, we previously reported that high FSS suppresses Smad2/3 through a mitogen‐activated protein kinase/extracellular signal‐regulated kinase kinase kinase (MEKK3)‐KLF2‐cyclin‐dependent kinase 2 pathway ¹¹ ; indeed, blocking this pathway via EC‐specific deletion of MEKK3 greatly accelerates development of atherosclerosis in mice. ⁶ Together, these results elucidate a novel mechanism that governs site‐selectivity of atherosclerotic lesions. Whereas low and disturbed FSS amplify Smad2/3 activation by cytokines, high flow potently inhibits. These effects thus contribute to preferential formation of atherosclerotic plaque in regions of low and disturbed flow, where positive feedforward between inflammation and EndMT promotes disease progression.

Sources of Funding

This work was supported by National Institutes of Health grant R01 HL135582 (Schwartz).

Disclosures

None.

Supporting information

Table S1

Figures S1–S5