Osteopontin mediation of disturbed flow–induced endothelial mesenchymal transition through CD44 is a novel mechanism of neointimal hyperplasia in arteriovenous fistulae for hemodialysis access

In dysfunctional arteriovenous fistulae (AVF) for hemodialysis access, neointimal hyperplasia (NH) is prone to occur in the region exposed to disturbed flow. We hypothesized that disturbed flow contributes to NH in AVF by inducing endothelial mesenchymal transition (EndMT) through activation of the osteopontin/CD44 axis. In rats with aortocaval fistula, a rodent model of AVF, we demonstrated development of EndMT and expression of osteopontin and CD44 specifically in the vicinity of the arteriovenous junction using immunostaining. Duplex scan confirmed this region was exposed to a disturbed flow. A mixed ultrastructural phenotype of endothelium and smooth muscle cells was found in luminal endothelial cells of the arteriovenous junction by electron microscopy ascertaining the presence of EndMT. Endothelial lineage tracing using Cdh5-Cre/ERT2;ROSA26-tdTomato transgenic mice showed that EndMT was involved in NH of AVF since the early stage and that the endothelial-derived cells contributed to 24% of neointimal cells. In human umbilical vein endothelial cells (HUVECs) in culture, osteopontin treatment induced EndMT, which was suppressed by CD44 knockdown. Exposure to low oscillatory wall shear stress using a parallel-plate system induced EndMT in HUVECs, also suppressed by osteopontin or CD44 knockdown. In AVF of CD44 knockout mice, EndMT was mitigated and NH decreased by 35% compared to that in wild-type mice. In dysfunctional AVF of patients with uremia, expressions of osteopontin, CD44, and mesenchymal markers in endothelial cells overlying the neointima was also found by immunostaining. Thus, the osteopontin/CD44 axis regulates disturbed flow-induced EndMT, plays an important role in neointimal hyperplasia of AVF, and may act as a potential therapeutic target to prevent AVF dysfunction. In dysfunctional arteriovenous fistulae (AVF) for hemodialysis access, neointimal hyperplasia (NH) is prone to occur in the region exposed to disturbed flow. We hypothesized that disturbed flow contributes to NH in AVF by inducing endothelial mesenchymal transition (EndMT) through activation of the osteopontin/CD44 axis. In rats with aortocaval fistula, a rodent model of AVF, we demonstrated development of EndMT and expression of osteopontin and CD44 specifically in the vicinity of the arteriovenous junction using immunostaining. Duplex scan confirmed this region was exposed to a disturbed flow. A mixed ultrastructural phenotype of endothelium and smooth muscle cells was found in luminal endothelial cells of the arteriovenous junction by electron microscopy ascertaining the presence of EndMT. Endothelial lineage tracing using Cdh5-Cre/ERT2;ROSA26-tdTomato transgenic mice showed that EndMT was involved in NH of AVF since the early stage and that the endothelial-derived cells contributed to 24% of neointimal cells. In human umbilical vein endothelial cells (HUVECs) in culture, osteopontin treatment induced EndMT, which was suppressed by CD44 knockdown. Exposure to low oscillatory wall shear stress using a parallel-plate system induced EndMT in HUVECs, also suppressed by osteopontin or CD44 knockdown. In AVF of CD44 knockout mice, EndMT was mitigated and NH decreased by 35% compared to that in wild-type mice. In dysfunctional AVF of patients with uremia, expressions of osteopontin, CD44, and mesenchymal markers in endothelial cells overlying the neointima was also found by immunostaining. Thus, the osteopontin/CD44 axis regulates disturbed flow-induced EndMT, plays an important role in neointimal hyperplasia of AVF, and may act as a potential therapeutic target to prevent AVF dysfunction. Translational StatementIn dysfunctional arteriovenous fistula (AVF) for hemodialysis access, neointimal hyperplasia is prone to develop in the juxta-anastomotic region, the inner side of which is exposed to disturbed flow. This study demonstrated in the rat model of AVF that disturbed flow induced endothelial mesenchymal transition (EndMT) in the juxta-anastomotic region through the osteopontin (OPN)/CD44 axis, which contributes substantially to neointimal hyperplasia. The clinical relevance was supported by the presence of EndMT and expression of OPN and CD44 in luminal endothelial cells overlying the neointima of dysfunctional AVF of uremic patients. The OPN/CD44 axis may act as a potential therapeutic target to prevent AVF dysfunction. In dysfunctional arteriovenous fistula (AVF) for hemodialysis access, neointimal hyperplasia is prone to develop in the juxta-anastomotic region, the inner side of which is exposed to disturbed flow. This study demonstrated in the rat model of AVF that disturbed flow induced endothelial mesenchymal transition (EndMT) in the juxta-anastomotic region through the osteopontin (OPN)/CD44 axis, which contributes substantially to neointimal hyperplasia. The clinical relevance was supported by the presence of EndMT and expression of OPN and CD44 in luminal endothelial cells overlying the neointima of dysfunctional AVF of uremic patients. The OPN/CD44 axis may act as a potential therapeutic target to prevent AVF dysfunction. Dysfunction of vascular access for hemodialysis is the most common comorbidity for patients with end-stage renal disease on maintenance hemodialysis.1Lok C.E. Huber T.S. Lee T. et al.National Kidney FoundationKDOQI Clinical Practice Guideline for Vascular Access: 2019 Update.Am J Kidney Dis. 2020; 75: S1-S164Abstract Full Text Full Text PDF PubMed Scopus (633) Google Scholar Arteriovenous fistula (AVF) is a common access with a lower rate of thrombosis.1Lok C.E. Huber T.S. Lee T. et al.National Kidney FoundationKDOQI Clinical Practice Guideline for Vascular Access: 2019 Update.Am J Kidney Dis. 2020; 75: S1-S164Abstract Full Text Full Text PDF PubMed Scopus (633) Google Scholar However, the incidence of AVF dysfunction remains high, predominantly caused by neointimal hyperplasia (NH) at the venous segment.2Badero O.J. Salifu M.O. Wasse H. et al.Frequency of swing-segment stenosis in referred dialysis patients with angiographically documented lesions.Am J Kidney Dis. 2008; 51: 93-98Abstract Full Text Full Text PDF PubMed Scopus (99) Google Scholar In dysfunctional AVF, migration of medial smooth muscle cells (SMCs) with subsequent active proliferation in the intima is recognized as the pathogenic mechanism of NH.3Mangrum A. Okusa M. Mechanisms underlying vascular access dysfunction.Drug Discov Today Dis Mech. 2007; 4: 147-151Crossref Scopus (6) Google Scholar It was demonstrated recently that endothelial mesenchymal transition (EndMT) represents another potential mechanism involved in NH.4Cooley B.C. Nevado J. Mellad J. et al.TGF-β signaling mediates endothelial-to-mesenchymal transition (EndMT) during vein graft remodeling.Sci Transl Med. 2014; 6: 227ra34Crossref PubMed Scopus (277) Google Scholar,5Ranchoux B. Antigny F. Rucker-Martin C. et al.Endothelial-to-mesenchymal transition in pulmonary hypertension.Circulation. 2015; 131: 1006-1018Crossref PubMed Scopus (370) Google Scholar Meanwhile, low wall shear stress (WSS) caused by disturbed flow was recently revealed to be an important mechanical regulator of EndMT.6Mahler G.J. Frendl C.M. Cao Q. et al.Effects of shear stress pattern and magnitude on mesenchymal transformation and invasion of aortic valve endothelial cells.Biotechnol Bioeng. 2014; 111: 2326-2337Crossref PubMed Scopus (92) Google Scholar, 7Mahmoud M.M. Kim H.R. Xing R.Y. et al.TWIST1 integrates endothelial responses to flow in vascular dysfunction and atherosclerosis.Circ Res. 2016; 119: 450-462Crossref PubMed Scopus (84) Google Scholar, 8Mahmoud M.M. Serbanovic-Canic J. Feng S. et al.Shear stress induces endothelial-to-mesenchymal transition via the transcription factor Snail.Sci Rep. 2017; 7: 3375Crossref PubMed Scopus (103) Google Scholar, 9Moonen J.R.A.J. Lee E.S. Schmidt M. et al.Endothelial-to-mesenchymal transition contributes to fibro-proliferative vascular disease and is modulated by fluid shear stress.Cardiovasc Res. 2015; 108: 377-386Crossref PubMed Scopus (155) Google Scholar Creation of AVF inherently generates disturbed flow in the juxta-anastomotic region, variably defined as the outflow venous segment 2 to 5 cm from the atrioventricular junction.10Ene-Iordache B. Remuzzi A. Disturbed flow in radial-cephalic arteriovenous fistulae for haemodialysis: low and oscillating shear stress locates the sites of stenosis.Nephrol Dial Transpl. 2012; 27: 358-368Crossref PubMed Scopus (126) Google Scholar,11Fitts M.K. Pike D.B. Anderson K. et al.Hemodynamic shear stress and endothelial dysfunction in hemodialysis access.Open Urol Nephrol J. 2014; 7: 33-44Crossref PubMed Scopus (47) Google Scholar Disturbed flow consequently exerts low oscillatory WSS at the inner side (the lesser curvature side) of the juxta-anastomotic region. Interestingly, this is the region that NH is prone to occur in AVF.12Asif A. Gadalean F.N. Merrill D. et al.Inflow stenosis in arteriovenous fistulas and grafts: a multicenter, prospective study.Kidney Int. 2005; 67: 1986-1992Abstract Full Text Full Text PDF PubMed Scopus (187) Google Scholar,13Nassar G.M. Nguyen B. Rhee E. et al.Endovascular treatment of the “failing to mature” arteriovenous fistula.Clin J Am Soc Nephrol. 2006; 1: 275-280Crossref PubMed Scopus (132) Google Scholar Accordingly, we hypothesized that disturbed flow induces EndMT at the juxta-anastomotic region of AVF and contributes to NH. Moreover, osteopontin (OPN), a master regulator of epithelial mesenchymal transition,14Kothari A.N. Arffa M.L. Chang V. et al.Osteopontin—a master regulator of epithelial-mesenchymal transition.J Clin Med. 2016; 5: 39Crossref PubMed Scopus (72) Google Scholar has been shown to be the most highly induced gene in the venous segment after creation of AVF.15Abeles D. Kwei S. Stavrakis G. et al.Gene expression changes evoked in a venous segment exposed to arterial flow.J Vasc Surg. 2006; 44: 863-870Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar,16Croatt A.J. Grande J.P. Hernandez M.C. et al.Characterization of a model of an arteriovenous fistula: the rat the effect of l-NAME.Am J Pathol. 2010; 176: 2530-2541Abstract Full Text Full Text PDF PubMed Scopus (51) Google Scholar We thus hypothesized that OPN regulates disturbed flow–induced EndMT in AVF. A total of 10 specimens of stenotic venous segments of dysfunctional radiocephalic fistulae resected during surgical revision were collected. We excluded the lesions with thrombosis. Five segments of cephalic veins of uremic patients were obtained during the creation of AVF to serve as controls. These samples were collected with the approval of the Institutional Review Board of the Chang Gung Memorial Hospital. Samples were fixed in 10% formalin prior to paraffin embedding and cut into 5-μm thick sections. The ACF was created in adult male Sprague–Dawley rats (300–350 gm in weight). After anesthetization, the abdominal cavity was opened and the inferior vena cava (IVC) and aorta were exposed and freed from the surrounding tissues. Both aorta and IVC were clamped with vessel clips at the infrarenal level and the level right above the iliac bifurcation. An 18-gauge needle was used to puncture the lateral wall of the lower abdominal aorta and then advance carefully to cross the opposite aortic wall and the neighboring wall of the IVC to create an arteriovenous communication. Care was taken to avoid puncturing the opposite wall of the IVC. Then the needle was withdrawn carefully and the entry point at the lateral aortic wall was sealed with cyanoacrylate glue (Vetbond 3M) after withdrawing the needle. Sham operations were performed by puncturing the lateral wall of the IVC opposite to the side adjacent to the aorta using a same gauge needle. For generating IVC-banded ACF rats, the ACF was created first as described above. Then a 2-0 silk suture was snared around the IVC right below the renal vein. Then, an 18-gauge needle was placed next to the IVC and the suture was snugly tied around the needle and the vessel. After ligation, the needle was removed. For the creation of ACF or sham operation in mice, the procedure was the same except that a 25-gauge needle was used for making the puncture. Cdh5-Cre/ERT2 (kindly provided by Prof. Ralf H. Adams, London, UK) mice were crossed with tdTomato reporter mice (B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J) to generate Cdh5-Cre/ERT2;ROSA26-tdTomato transgenic mice, where endothelial cells were labeled with tdTomato following tamoxifen (80 mg/kg/d i.p. for 4 days) injection. CD44 knockout mice (B6.129(Cg)-Cd44 tm1Hbg/J) were purchased from the Jackson Laboratory. For creation of carotid-jugular arteriovenous fistula in rats, a horizontal midline cervical incision across the lower neck area was made and the common carotid and external jugular vein were carefully dissected. The branches of the external jugular vein were divided to prepare an adequate length for anastomosis. The distal common carotid artery was clamped proximal to the bifurcation, and the proximal common carotid artery was clamped as close to the clavicle using a clip. All the animal experiments were performed according to the guidelines of the Committee on Animal Research at the Chang Gung Memorial Hospital. For collection of luminal ECs from IVCs, the IVC was dissected from the surrounding tissue before harvested to keep the outer surface of IVCs smoother. After being harvested, the IVCs were cut longitudinally and opened. Then the tissue was immersed in phosphate-buffered saline (PBS) with the luminal surface upward and fixed by 23-gauge needles at 4 corners on a sterile eraser. The surface of IVCs was scraped gently by a scalpel and washed with PBS. The cells scraped off in the PBS were collected. To confirm that only the intima tissues were scraped, 5 IVCs were fixed for frozen sections after scraping. Histologic examination using fluorescence microscopy to observe the autofluorescence of internal elastic lamina of IVCs showed that the internal elastic lamina was intact after scraping in all specimens (data not presented). This finding excluded the possibility that the cells of media layer were scraped off and included in the cells collected. Duplex scan (Acuson, Aspen) was performed to assess the flow patterns of IVCs. B-mode and Doppler images were obtained using a high-resolution linear transducer at a frequency of 15 MHz. After euthanization, laparotomy was performed and IVC was exposed. The transducer was placed over the IVC as lightly as possible to avoid compressing the vessel, as guided by the B-mode image. The IVC flow velocity was recorded along the entire IVC segment. For the preparation of immuohistochemistry samples, perfusion-fixed IVC specimens of rats or mice were snap frozen and cut into 8-μm thick sections. Human AVF specimens were rinsed briefly and fixed in 4% paraformaldehyde immediately after harvesting. Then, the tissues were embedded in paraffin and cut into 4-μm-thick sections. For immunocytochemistry study, human umbilical vein endothelial cells (HUVECs) were fixed with 3% paraformaldehyde and permealized with 0.1% Triton X-100. After blocking with 1% bovine serum albumin for 30 minutes, the tissue sections or cells were incubated with rabbit anti-human VE-cadherin (1:150; Thermo Scientific), monoclonal anti-human VE-cadherin (1:200; Santa Cruz Biotechnology), goat anti-mouse CD31 (1:200; R&D Systems Inc.), rabbit anti-human α–smooth muscle actin (α-SMA; 1:200; Abcam), monoclonal anti-human α-SMA (1:500; Sigma-Aldrich), rabbit anti-human osteopontin (OPN; 1:250; Abcam), or rabbit anti-human CD44 (1:200; Abclonal) antibodies and then with appropriate fluorescein isothiocyanate–, CY3-, or Cy5- (for triple staining) conjugated secondary antibodies. The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (blue). The expression of particular protein in HUVECs was assessed through quantifying the fluorescence intensity for 100 cells randomly selected from 3 independent experiments using the MetaMorph software. The IVC specimens were harvested with the animal perfusion-fixed with 3% paraformaldehyde in PBS at 100 mm Hg for 10 minutes. The adventitia of the IVCs was trimmed gently before and if needed after the perfusion fixation. After harvesting, the specimens were transected into the upper two-thirds and lower one-third segments. The vessel was opened longitudinally, and then the whole specimen was submitted for whole-mount immunofluorescence staining. The samples were first rinsed in PBS for 5 minutes and blocked in 1% bovine serum albumin in PBS for 30 minutes. The samples were incubated with goat anti-mouse CD31 (R&D Systems Inc.), rabbit anti-human α-SMA (Abcam), rabbit anti-human OPN (Abcam), or rabbit anti-human CD44 (Abclonal) antibodies, and then with appropriate fluorescein isothiocyanate– and CY3-conjugated secondary antibodies. During the immunolabeling process, all the steps were performed in 2-ml Eppendorf tubes to prevent injury to the endothelial surface. After immunoreactions, the IVC tissues were carefully placed on the slides with the luminal side facing upward. The specimens were then examined using a confocal laser-scanning microscope (Leica TCS SP8X; Stansted). The images were captured at original magnification ×400. Each recorded image was 1024 × 1024 pixels in size, and the projection views of consecutive optical sections were captured at 0.5 μm. All specimens were examined within 24 hours of immunolabeling. Perfusion-fixed IVCs were dissected, opened longitudinally, and post-fixed in 1% glutaraldehyde. Specimens were then incubated in 1% osmium tetroxide, dehydrated through a gradient series of ethanol, and finally embedded in Epon 812. Sections of 80 nm were cut and stained with uranyl acetate and lead citrate. The specimens were examined using a Hitachi HT7800 electron microscope. HUVECs were isolated and cultured up to passages 3–4. To study the effect of OPN on EndMT, cells were treated with OPN at a concentration of 100 ng/ml in medium for 48 hours. For RNA interference, cell cultures were transfected with small, interfering RNAs targeting OPN or CD44 purchased from Dharmacon using DharmaFECT 1 (Dharmacon), according to the manufacturer’s instructions. Nontargeting sequences small, interfering RNAs were used as control. HUVECs were exposed to different flow conditions using the Ibidi parallel-plate system (Ibidi). HUVECs were cultured on Ibidi gelatin-coated μ-slides (Ibidi GmbH) until confluent. Then the Ibidi pump system was applied to generate low (4 dyn/cm2), low oscillatory (±4 dyn/cm2), or high (70 dyn/cm2) WSS for 72 hours. The system was enclosed in a culture incubator maintained at 37 °C. RNA was extracted from ECs of IVCs or from cultured HUVECs using the RNeasy kit (Qiagen) according to the manufacturer’s instructions. RNA was reverse-transcribed using the Superscript II reverse transcriptase (Gibco-BRL, Life Technologies GmbH) following the manufacturer’s instructions. The expression levels of transcripts were assessed by quantitative real-time polymerase chain reaction using gene-specific primers (the primer sequences are listed in Supplementary Table S2) in triplicate. Relative expression of a particular gene was determined by comparing the number of thermal cycles that were necessary to generate the amount of PCR product required to reach the threshold. Fold changes were analyzed using the ΔΔCt method. Western blot analysis was performed as described previously (R). Briefly, equal amount of protein in SDS-PAGE sample buffer was sonicated and subjected to electrophoresis on 8% sodium dodecylsulfate–polyacrylamide gels. After transferring the proteins to the polyvinylidene fluoride membrane (Stratagene), proteins were incubated with primary antibodies against VE-cadherin (Abcam), twist family bHLH transcription factor 1 (TWIST1; Abcam), CD31 (Abcam), fibroblast-specific protein-1 (FSP-1; Abcam), α-SMA (Abcam), vimentin (Santa Cruz, Santa Cruz Biotechnology), collagen I (Abcam), CD44 (Abclonal), and glyceraldehyde-3-phosphate dehydrogenase (Santa Cruz). Signals were detected by using the ECL-detection kit (Amersham) and were quantified by densitometry. The expressions were normalized relative to glyceraldehyde-3-phosphate dehydrogenase. Metamorph software was used for morphometric analyses. For the measurement of neointimal area, sections of IVCs were stained with hematoxylin and eosin. We first determined the total luminal area by measuring the area within the perimeter of the internal elastic lamina. Then, we measured the residual luminal area surrounded by the neointima and the elastic lamina. The neointimal area of each section was determined by calculating the difference of these 2 values. The first sections of every 5 sections of the lower one-third of IVC were selected for evaluation of neointimal areas. The neointimal volume of IVC in each mouse was determined by summation of all the neointimal areas. For determining the proportion of GFP+ cells in the neointima, GFP+ and total cells in the neointima were counted manually under the counting grid on scanned confocal microscopic images captured at original magnification ×400. Total cell number was measured by 4′,6-diamidino-2-phenylindole staining. Images were analyzed in a blind manner. The data are presented as the number of GFP+ cells relative to the total number of neointimal cells. To determine the proportion of luminal cells with positive staining for OPN, CD44, α-SMA, or TWIST1 in luminal ECs, the total number of luminal ECs and positively stained cells were also counted manually under the counting grid on scanned confocal microscopic images at original magnification ×400. The total number of luminal EC was measured by anti-CD31 staining, and the number of luminal ECs with positive staining for each protein was measured by double staining using anti-CD31 and specific antibody against the corresponding protein. The data are presented as the number of cells with positive staining for each protein relative to the total number of luminal ECs. The duplex scan survey for the IVC of ACF at day 3 revealed a disturbed flow profile in the vicinity of the arteriovenous junction characterized by a particularly broad spectrum with minimal arterial phasicity. Downstream away from the arteriovenous junction, the flow profiles at the lower one-third of IVCs were of broad spectrum with arterial phasicity, which faded progressively while moving further downstream. At the upper two-thirds of IVCs, a normal central venous flow pattern was observed, reflecting atrial pulsation (Figure 1a). Gross observation revealed that no visible neointimal lesions were found until day 7. At day 7, neointimal lesions were observed around the puncture hole and extended 2 to 3 mm downstream at the inner side of the lower one-third of IVCs (Figure 1b). Microscopically, neointima was not found in IVCs of sham controls at any time point. In AVF rats, minimal lesions were observed at the lower one-third of IVCs at day 5, which appeared substantially at day 7 (Figure 1c). These observations demonstrated that early NH in this AVF model was localized to the segment exposed to disturbed flow. We next tested whether creation of AVF induced expression of mesenchymal markers in luminal ECs of IVCs. Luminal ECs of the whole IVC segments were harvested at day 3 by gentle scraping as described previously.17Passerini A.G. Polacek D.C. Shi C.Z. et al.Coexisting proinflammatory and antioxidative endothelial transcription profiles in a disturbed flow region of the adult porcine aorta.Proc Natl Acad Sci U S A. 2004; 101: 2482-2487Crossref PubMed Scopus (315) Google Scholar We collected ECs at day 3 because NH was not found in IVCs at this time point as described above and thus no neointimal cells would be contained in the cells collected. Quantitative polymerase chain reaction analysis showed increased gene expression of the fibroblast marker fibroblast-specific protein-1 (FSP-1) and early vascular α-SMA and smooth muscle (SM)22α in ECs of the AVF group. Expression of mature vascular SMC markers caponin and myosin-11 remained unaltered compared with sham controls. Additionally, expression of collagen I was found to be increased (Figure 2a). Then, we analyze the protein expression of transcription factors regulating EndMT in IVCs of rats with AVF or with sham operation at day 3. Because it requires enormous animals to collect enough luminal ECs for Western blotting analysis, we submitted whole IVC tissue for analysis. We demonstrated upregulation of TWIST1 but not snail family transcriptional repressor 2 (Slug) or snail family transcriptional repressor 1 (Snail) in the AVF group compared with the control group. Immunostaining of sectioned IVCs localized the anti-TWIST1 signals predominantly to luminal ECs of the lower half of IVCs, which were also colocalized with anti–α-SMA signals (Figure 2b). These findings indicated mesenchymal transition of luminal ECs of IVCs of AVF likely localized to the lower one-third. We then performed whole mount immunofluorescence staining for IVCs at days 3, 5, and 7 to precisely localize the distribution of EndMT. Compared with immunostaining of sectioned tissues, whole mount staining with en face observation allowed a thorough examination of all luminal ECs with the cells remained unsectioned, which assists precise identification and localization of ECs with mesenchymal transition. In IVCs of sham control rats, luminal ECs were characterized by expression of CD31 at the cell-cell junction. Expression of α-SMA was hardly detectable at all time points. In the upper two-thirds of IVCs of AVF rats, the expression patterns of CD31 and α-SMA were similar to those of sham controls. Conversely, in the lower one-third, scattered expression of α-SMA was found in ECs of the inner side (the side adjacent to the aorta) at day 3. Instead of expression at the cell-cell junction, cytosolic expression of CD31 was also found. At day 5, expression of α-SMA was further enhanced and characterized by a microfibrillar pattern. Absence of CD31 expression at cell-cell junction ECs was also frequently found. At day 7, thick neointimal lesions were found at the inner half around the arteriovenous junction. In luminal ECs overlying the neointima, more diffuse expression of α-SMA was observed. The absence of CD31 expression with concomitant strong expression of α-SMA was found in clusters of cells, indicating complete transition to the mesenchymal phenotype of ECs (Figure 2c). These findings demonstrated that creation of AVF leads to active EndMT specifically at the inner side of the lower one-third of IVCs, the region that is exposed to disturbed flow. To further clarify the effect of disturbed flow on EndMT, we performed banding of IVCs of AVF rats at the level below the renal vein. The duplex scan confirmed that the presence of disturbed flow at the site right distal to the banding segment (Supplementary Figure S1A). Whole mount immunostaining demonstrated strong expression of α-SMA in luminal ECs (Supplementary Figure S1B). Vascular injury caused by needle puncture during creation of ACF may potentially contribute to EndMT. However, in sham rats, puncture limited to the lateral wall of the IVC without causing fistula was not shown to induce EndMT (Figure 2c), suggesting that vascular puncture per se did not contribute to EndMT. To further demonstrate that creation of AVF without needle puncture also induce EndMT, we created carotid-jugular fistula in rats by anastomosing both vessels using a nonabsorbable monofilament similar to AVF creation in uremic patients. Similar to the findings in the ACF model, induced expression of α-SMA and TWIST1 in ECs of the jugular vein in the vicinity of anastomosis was shown by immunohistochemistry study (Supplementary Figure S1C). Taken together, these findings demonstrated that in the segment of the venous limb of AVF that is exposed to disturbed flow, mesenchymal transition was induced in luminal ECs. We next used electron microscopy to ascertain the phenotypic changes in luminal ECs and examined the involvement of EndMT in NH. At 4 weeks after creation of AVF, typical ultrastructural features of the endothelium characterized by the presence of abundant caveolae and Weibel-Palade bodies18Ochoa C.D. Wu S.W. Stevens T. New developments in lung endothelial heterogeneity: von Willebrand factor, P-selectin, and the Weibel-Palade body.Semin Thromb Hemost. 2010; 36: 301-308Crossref PubMed Scopus (53) Google Scholar,19Weibel E.R. Fifty years of Weibel-Palade bodies: the discovery and early history of an enigmatic organelle of endothelial cells.J Thromb Haemost. 2012; 10: 979-984Abstract Full Text Full Text PDF PubMed Scopus (32) Google Scholar (Figure 3b) was demonstrated in luminal ECs of the upper two-thirds of IVCs, similar to that observed in the sham group (Figure 3a). Medial SMCs were characterized by the presence of dense bodies and abundant myofilaments5Ranchoux B. Antigny F. Rucker-Martin C. et al.Endothelial-to-mesenchymal transition in pulmonary hypertension.Circulation. 2015; 131: 1006-1018Crossref PubMed Scopus (370) Google Scholar and were devoid of ultrastructural features of ECs (Figure 3b). In the lower one-third of IVCs of the AVF group, neointimal lesions were found with the overlying ECs exhibiting a mixed ultrastructural phenotype of the endothelium and SMCs as they had the Weibel-Palade bodies, high density of caveolae, and abundant microfilaments in the cytoplasm (Figure 3c). Outgrowth of luminal ECs toward the subintima was frequently observed, suggesting ongoing migration of luminal ECs toward the subintimal space (Figure 3d). Some neointimal cells were found to have the Weibel-Palade bodies in addition to dense bodies and abundant filaments, suggesting that they were of endothelial origin (Figure 3e). Then, we created ACF in Cdh5-Cre/ERT2;ROSA26-tdTomato transgenic m