Reduced Proteolytic Cleavage of von Willebrand Factor Leads to Aortic Valve Stenosis and Load-Dependent Ventricular Remodeling.

We hypothesized that excess endothelial-associated von Willebrand factor (vWF) and secondary platelet adhesion contribute to aortic valve stenosis (AS). We studied hyperlipidemic mice lacking ADAMTS13 (LDLR -/- AD13 -/- ), which cleaves endothelial-associated vWF multimers. On echocardiography and molecular imaging, LDLR -/- AD13 -/- compared with control strains had increased aortic endothelial vWF and platelet adhesion and developed hemodynamically significant AS, arterial stiffening, high valvulo-aortic impedance, and secondary load-dependent reduction in LV systolic function. Histology revealed leaflet thickening and calcification with valve interstitial cell myofibroblastic and osteogenic transformation, and evidence for TGFβ1 pathway activation. We conclude that valve leaflet endothelial vWF-platelet interactions promote AS through juxtacrine platelet signaling.

Aortic valve stenosis (AS) occurs from a broad array of processes including the generation of reactive oxygen species, cytokine and growth factor signaling, abnormal regulation of extracellular matrix content, and osteogenic signaling., This knowledge has not translated to any successful medical therapy to slow AS progression in patients, largely because the modifiable triggers that initiate and promote progression of AS are incompletely understood.

Platelets are a source of growth factors, chemokines, and cytokines that can potentially alter the cellular and matrix composition of heart valves. Yet, the role of platelets in AS is uncertain. In humans, associations between platelet activation and severity of AS are modest, and may not reflect a pathogenic role of platelets but rather the effect of valve shear on platelet activation. Moreover, the activation status of circulating platelets does not reflect their full pathogenic potential. Platelet adhesion mediated by interaction of constitutively-expressed glycoprotein-Ibα (GPIbα) component of the GPIb/V/IX complex, and endothelial-associated von Willebrand Factor (vWF) can activate vascular cells through chemokine and growth factor signaling and inflammasome activation.4, 5, 6, 7 These processes have been recognized to be proatherogenic., Shear-mediated self-association of vWF and exposure of the A1 binding domain for platelet GPIbα could similarly contribute to AS through platelet binding and juxtacrine or paracrine signaling through factors such as transforming growth factor-β1 (TGFβ1) that are known to cause myofibroblastic or osteogenic activation of valvular cells., Platelet adhesion to the surface of aortic leaflets has been described in mice with hyperlipidemia and on explanted leaflets from humans with AS., Moreover, platelet-specific deficiency of TGFβ1 has been shown to reduce aortic valve thickening in hyperlipidemic mice, and activated platelet products have been shown to produce pro-osteogenic profiles of cultured valve interstitial cells (VICs).

In the current study, we hypothesized that excess endothelial-associated vWF contributes to not only aortic valve thickening, but also the development of AS through leaflet adhesion of platelets or platelet microvesicles (PMVs) and subsequent myofibroblastic and osteogenic transformation of VICs. We studied mice with hyperlipidemia produced by Western-style diet (WSD) and genetic deficiency of the low-density lipoprotein receptor (LDLR) that were also deficient for ADAMTS-13, the enzyme responsible for proteolytic cleavage and removal of ultra-large self-associated multimers of vWF from the endothelial surface. Comprehensive noninvasive in vivo imaging data were integrated with tissue histology to examine the effect of excess valve endothelial cell (VEC) vWF on leaflet platelet adhesion, aortic valve morphology, VIC transformation, valve hemodynamics, and load-dependent changes of the left ventricle (LV).

Methods

A detailed description of all methods is provided in the Supplemental Appendix. A general description of study design follows. The study was approved by the Animal Care and Use Committee of the Oregon Health & Science University. We studied wild-type C57BL/6 mice, LDLR mice on a C57BL/6 background, and LDLR mice deficient for ADAMTS13 (LDLRAD13) generated by crossing LDLR mice with AD13 mice. At age 14 weeks, LDLR and LDLRAD13 mice were switched from chow diet to WSD containing 21% fat by weight (15% saturated fat), representing 42% of caloric intake. At age 50 weeks, comprehensive ultrasound evaluation of aortic hemodynamics and LV structure and function was performed, and plasma vWF and ADAMTS13 activity were measured. Also at 50 weeks, histological evaluation of the aortic valve and aorta by light microscopy, immunofluorescence, and transmission electron microscopy was performed for characterization of valve leaflet thickening, leaflet vWF, platelet adhesion, matrix and matrix metalloproteinase-9 (MMP-9), calcification, VIC myofibroblastic transformation, phosphorylated SMAD2 (pSMAD2) and β-catenin as a marker of TGFβ1 signaling, and myocyte hypertrophy. In situ hybridization with RNAscope was performed for transcriptomic increases in osteocalcin as a marker for osteogenic transformation. Computed tomography angiography was performed for coronary anatomy. At 30 weeks of age, representing a midpoint between initiation of WSD and final evaluation, the intended phenotype for the murine groups was evaluated by measurement of plasma lipids, complete blood counts, ultrasound molecular imaging of platelet adhesion and endothelial-associated vWF, and histology for vWF.

Results

Abnormal endothelial phenotype and hypertension in mice lacking ADAMTS13

Plasma lipids, blood pressure, and endothelial molecular imaging were performed at age 30 weeks, selected as an intermediate time point between initiation of WSD and the final evaluation of valve status to assess differences between the murine strains during disease development. At this age, LDLR and LDLRAD13 mice on WSD had similar degrees of hyperlipidemia compared with wild-type mice, manifesting as elevated total cholesterol, LDL cholesterol, non-LDL/high-density lipoprotein cholesterol, and triglycerides (Figures 1A to 1D). Awake systolic and diastolic blood pressure were elevated in both LDLR and LDLRAD13 mice on WSD compared with wild-type mice (Figures 1E and 1F). Contrast ultrasound molecular imaging of the aortic root was performed using validated protocols with microbubble probes that remain entirely within the vascular compartment and that detect endothelial-associated vWF by targeting the shear-exposed A1 domain binding domain, and platelet adhesion by targeting GPIbα of the GPIb-V-IX complex., Both LDLR and LDLRAD13 mice on WSD had higher signal compared with wild-type mice for endothelial-associated vWF A1 domain and platelet GPIbα, and the signal for both of these targets was 2- to 3-fold higher in LDLRAD13 mice than LDLR mice (Figures 1G to 1I, Supplemental Figure 1). Molecular imaging data for vWF were corroborated by immunofluorescent histology at age 30 weeks showing enhanced endothelial vWF not only in regions of plaque formation in the aortic sinuses, but also on both the ventricularis and fibrosa (aortic) aspects of the aortic valve leaflets (Supplemental Figure 2).

Figure 1

Characterization of the Murine Models at Age 30 Weeks

(A to D) Mean ± SEM plasma lipid concentrations including total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides. (E and F) Mean ± SEM systolic and diastolic awake blood pressure (BP) measurements. (G and I) Mean ± SEM signal intensity on contrast ultrasound molecular imaging signal from the aortic root for endothelial-associated vWF by microbubble targeting of the A1 binding domain, platelet adhesion by targeting GPIbα, and control microbubble contrast agent. P values are adjusted for multiple comparisons. Examples of contrast ultrasound molecular imaging of the aortic root from all 3 murine groups are illustrated in Supplemental Figure 1. ∗P < 0.05; ∗∗P < 0.001 (adjusted). HDL = high-density lipoprotein; LDL = low-density lipoprotein; vWF = von Willebrand factor; WSD = Western-style diet; WT = wild type.

ADAMTS13 activity, circulating vWF, and platelet count

At 50 age weeks, group-wise differences in plasma ADAMTS13 activity and vWF concentration were evaluated (Supplemental Table 1). Results were similar between wild-type and LDLR mice on WSD; whereas LDLRAD13 mice uniformly had no detectable ADAMTS13 activity and had significantly lower circulating plasma vWF concentrations compared with the other groups, a finding that is consistent with impaired ability to proteolytically cleave vWF from the endothelial surface and into the circulating plasma pool. Platelet counts tended to be higher in LDLRAD13 mice, although this did not meet statistical significance (Supplemental Table 1).

Blood pressure and aortic mechanical properties at age 50 weeks

At 50 age weeks, aortic hemodynamics were again assessed to fully evaluate the combined hemodynamic influence of blood pressure, vascular stiffness, and aortic valve disease in the different strains. Awake systolic and diastolic BP were higher in LDLR and LDLRAD13 mice on WSD compared with wild-type mice (Figures 2A and 2B). Systolic BP was highest in LDLRAD13 mice. Aortic distensibility (Figure 2C), representing the aortic strain-pressure relation, was reduced for LDLR and LDLRAD13 mice on WSD compared with wild-type mice. Differences in aortic distensibility between LDLR and LDLRAD13 mice did not reach statistical significance (adjusted P = 0.07). Arterial elastance, which takes into account differences in stroke volume and reflects nonpulsatile load and arterial compliance, was significantly higher in LDLRAD13 mice compared with the other strains, consistent with increased peripheral arterial afterload (Figure 2D).

Figure 2

Systemic Arterial Hemodynamics at Age 50 Weeks

Mean ± SEM values are shown for (A) systolic BP, (B) diastolic BP, (C) aortic distensibility, and (D) arterial elastance. P values are adjusted for multiple comparisons. ∗P < 0.01; ∗∗P < 0.001 (adjusted). Abbreviations as in Figure 1.

Hemodynamic Aortic Valve Stenosis occurs in LDLRAD13 mice

On echocardiography at 50 age weeks, there was a stepwise increase in peak and mean aortic gradients between wild-type LDLR mice on WSD and LDLRAD13 mice on WSD, although differences between LDLR and LDLRAD13 mice did not meet statistical significance only for the mean gradient (Figures 3A to 3C). Compared with other groups, LDLRAD13 mice had significantly greater aortic valve resistance (Figure 3D), which accounts for potential differences in stroke volume. Valvulo-aortic impedance, reflecting the net afterload placed on the LV from both aortic valve resistance and systolic hypertension, was 2- to 3-fold higher in LDLRAD13 mice compared with the other groups (Figure 3E). Doppler studies did not detect more than trivial to mild aortic regurgitation in any animal.

Figure 3

Echocardiographic Data on Aortic Valve Hemodynamics at Age 50 Weeks

(A) Examples of pulsed-wave Doppler at the aortic valve with angle-correction alignment and corresponding spectral Doppler measurements from each of the murine strains (note different velocity scales for spectral Doppler recordings). Graphs depict mean ± SEM) 2-dimenisonal ultrasound and Doppler-derived (B) peak aortic gradient, (C) mean aortic gradient, (D) aortic valve resistance, and (E) valvulo-arterial impedance (Zva). P values are adjusted for multiple comparisons.∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001 (adjusted). Abbreviations as in Figure 1.

Aortic leaflet morphology, endothelial-associated vWF, and platelet adhesion

Histology at age 50 weeks revealed plaque that was rich in Mac-2 staining in the aortic sinuses in both LDLR and LDLRAD13 mice on WSD (Figure 4A). In LDLRAD13 mice, plaque often extended into the base of the leaflets. Plaque-like structures on the leaflets were also observed in some animals. On quantitative analysis, LDLRAD13 mice had 2-fold greater aortic sinus plaque than LDLR mice (Figure 4B). Valve thickening by Masson’s trichrome staining was present in both LDLR and LDLRAD13 mice, but was far greater in LDLRAD13 mice in which both focal eccentric thickening and diffuse thickening were observed (Figures 4A and 4C). Leaflet thickening was associated with disarray of the normal laminar organized layers, a transmural increase in valve matrix including collagen, and increases in valve cellularity and area staining for Mac-2 (Figures 4A, 4D, and 4E). The majority of plaque thickening was not from Mac-2-positive plaque-like structures but from the increased matrix and cellularity. On calcium staining with alizarin red, only LDLRAD13 mice exhibited any positive leaflet staining manifest either as dense focal calcific nodules or as lower-intensity staining of VICs consistent with osteogenic transformation (Figures 4F and 4G).

Figure 4

Aortic Valve Histology at Age 50 Weeks

(A) Examples of aortic root histology at the level of the valve for wild-type mouse, LDLR mouse on WSD, and 2 different LDLRAD13 mice on WSD at age 50 weeks illustrating Masson’s trichrome at low and high magnification (top 2 rows), Mac-2 (galectin-3) immunofluorescence with positive staining in the green fluorescent range (third row), and picrosirius red staining visualized under polarized light (bottom row). Quantitative microscopy data (mean ± SEM) are provided for measurement of (B) plaque area, (C) average valve thickness, (D) valve leaflet area with positive Mac-2 staining, and (E) valve leaflet area with positive picrosirius staining. P values are adjusted for multiple comparisons. Alizarin red staining of the leaflets from LDLRAD13 mice showing (F) dense focal calcific nodules in the body of the leaflet; and (G) low-intensity staining of individual VICs (low magnification image of the leaflets shown, with inset illustrating high magnification of the area denoted by the dashed lines). ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001 (adjusted). Abbreviations as in Figure 1.

On immunofluorescence histology, vWF was observed overlying and within atherosclerotic plaque in the LDLR and LDLRAD13 strains as well as on the leaflet surface in all strains, where it was associated with valvular endothelial cells (Figure 5). Qualitatively, in wild-type and LDLR mice, much of the vWF appeared to be within intracellular granules, consistent with storage in Weibel-Palade bodies (Figures 5A and 5B). The extent of leaflet vWF was greatest in LDLRAD13 mice where it was observed on both the ventricularis and aortic aspects of the valve, and often appeared to be nongranular and to be associated with the luminal endothelial surface (Figure 5C). Immunofluorescence for CD41 indicated that there was almost no platelet adhesion to aortic leaflets in wild-type mice, and a modest number of single adhesion events in LDLR mice on WSD (Figures 6A and 6B). In LDLRAD13 mice, the extent of platelet CD41 staining was much greater and was found to be associated with atherosclerotic plaque, with the ventricular and aortic aspects of the aortic valve surface, and in some cases within the interior of the valve (Figures 6C and 6D). Platelets were occasionally observed to be associated with projections of vWF away from the leaflet surface consistent with self-associated ultra-large vWF fibers (Figure 6C). Dual staining revealed that platelet adhesion events occurred primarily at sites of intact endothelium, and not in regions of endothelial erosion (Figure 6E). Many of the CD41-positive entities seen on high-magnification imaging appeared submicron in size, consistent with adhesion of not only platelets but also PMVs. This finding was corroborated by transmission electron microscopy of valve tissue from LDLRAD13 mice, which detected that vesicles that were ≈400 nm in diameter were observed attached to valve endothelial cells (Figures 6F and 6G). On quantitative assessment, valvular vWF and valve-associated CD41-positive platelet or PMV staining was greatest in LDLRAD13 mice on WSD (Figures 6H and 6I).

Figure 5

Valve Endothelial Cell-Associated vWF

Examples of aortic valve immunofluorescence for vWF and CD31 at the level of the aortic valve at age 50 weeks from (A) a wild-type mouse, (B) an LDLR mouse on WSD, and (C) an LDLRAD13 mouse on WSD. Sequential panels from left to right represent incremental increases in magnification from the regions outlined by the dashed lines. Counterstaining for nuclei was performed with 4′,6-diamidino-2-phenylindole (DAPI). Examples of single immunofluorescent staining for vWF that were used for quantitative assessment are illustrated in Supplemental Figure 2. Abbreviations as in Figure 1.

Figure 6

Platelet Adhesion Patterns at Age 50 Weeks

(A to D) Examples of aortic valve dual immunofluorescence for vWF and platelet CD41 at age 50 weeks from (A) a wild-type mouse, (B) an LDLR mouse on WSD, and (C and D) 2 separate LDLRAD13 mice on WSD. Sequential panels from left to right represent incremental increases in magnification from the regions outlined by the dashed lines. Red arrow in C indicates evidence for a long vWF multimer on the valve surface with platelets at the distal end (yellow staining indicating dual CD41 and vWF staining). (E) Two examples of aortic valve dual immunofluorescence at high magnification for CD31 and platelet CD41 at age 50 weeks from LDLRAD13 mice on WSD. Arrows indicate the evidence for platelet staining for particles that appear <1 μm in size. (F and G) Transmission electron microscopy showing evidence of extracellular microvesicles (arrows) directly attached to the valvular endothelial cells. The structure denoted by the arrowhead in G may represent either microvesicle with contents but is also consistent with intermediate secretory pods that have been described after Weibel-Palade body (WPB) cell membrane fusion. Graphs depict mean ± SEM immunofluorescent positive area for (H) valve leaflet vWF and (I) valve leaflet platelet CD41 staining. ∗P < 0.01; ∗∗P < 0.001 (adjusted). Abbreviations as in Figure 1.

Disease-related changes in aortic valve interstitial cells and endothelial cells

Because platelets contain factors that are recognized to directly or indirectly induce myofibroblastic transformation of VICs, immunofluorescence was performed for α-smooth muscle actin (SMA) as a marker of myofibroblastic transformation.,, The number of cells positive for alpha smooth cell actin (αSMA) and the intensity of staining were both greater in LDLRAD13 mice (Figure 7A, Supplemental Figure 3). The αSMA-positive VICs were largely responsible for the increase in cellularity seen on Masson’s trichrome staining. Staining for MMP-9, a marker for active matrix remodeling that has been associated with progressive AS in humans and with myofibroblastic VICs, was intense within the leaflets of LDLRAD13 mice, particularly in areas with focal valve thickening (Figures 7B to 7D). In situ hybridization for osteopontin (SPP1) transcription as a marker for the early stages of VIC osteogenic transformation revealed dense transcription in aortic sinus plaques for both LDLR and LDLRAD13 mice. However, leaflet staining was only observed only in LDLRAD13 mice (Figure 7E). Osteopontin was not seen in any wild-type mice.

Figure 7

Valve Interstitial Cell Myofibroblastic and Osteogenic Transformation

(A and B) Examples of aortic valve immunofluorescence for (A) α-smooth muscle actin (αSMA) and ( matrix metalloproteinase (MMP)-9 from each murine strain. High-magnification images of the area defined by dashed lines are provided below each image. Additional examples of αSMA staining are shown in Supplemental Figure 3. Graphs depict mean ± SEM immunofluorescent-positive area MMP-9 expressed as (C) total area and (D) proportion of total leaflet area. P values are corrected for multiple comparisons. (D) In situ hybridization for osteopontin mRNA expression (red staining) illustrating positive staining for aortic plaque in both LDLR mice on WSD and LDLRAD13 mice on WSD, but leaflet staining (arrows) only in the latter. Examples of dual immunofluorescent staining for: (E) pSMAD2 and platelet CD41, and (F) β-catenin and platelet CD41 from w LDLRAD13 mice on WSD, with high-magnification images shown to the right. Additional examples from all 3 strains are shown in Supplemental Figure 4. ∗P < 0.01; ∗∗P < 0.001 (adjusted). Abbreviations as in Figure 1.

Histologic evaluation was performed for pSMAD2 and for β-catenin as a marker for activation of the canonical Wnt pathway. These represent 2 separate signaling pathways that have been thought to contribute to aortic valve disease and VIC myofibroblastic differentiation through TGFβ1 signaling.,, The intensity and extent of expression for both pSMAD2 and β-catenin was greater in LDLRAD13 mice than other groups, and both markers were present in VICs and VECs (Figures 7F and 7G, Supplemental Figure 4). Platelets also occasionally were found to express pSMAD2. The percentage of β-catenin staining in LDLRAD13 mice that was nuclear in location was 23 ± 9%. Staining of VECs and VICs for both pSMAD2 and β-catenin was often spatially associated with regions where there was also platelet or PMV adhesion on CD41 staining. Staining for NOTCH-1 was present in all strains but was similar between strains (Supplemental Figure 5).

Load-dependent changes in LV function

On echocardiography, LDLRAD13 mice on WSD had increased LV mass and smaller LV end-diastolic volumes (Figures 8A to 8C). In LDLRAD13 mice, indexes of systolic function that are known to be afterload-dependent were all impaired including LV ejection fraction, SV, SVi, global longitudinal and circumferential strain, and strain rate (Figures 8D to 8K). Because increased contractility is a compensatory mechanism for maintaining cardiac output in response to increased impedance, cardiac work was evaluated and found to be highest in the LDLRAD13 mice on WSD (Figure 8L), arguing against a decompensated state. Because LDLRAD13 mice on WSD had a far greater degree of aortic plaque formation, coronary computed tomography angiography was performed to exclude obstructive coronary artery disease as a cause for reduced LV ejection fraction and stroke volume. These studies failed to detect any epicardial coronary stenosis (Supplemental Figure 6).

Figure 8

Echocardiographic Data on LV function and Morphology at Age 50 Weeks

Graphs depict mean ± SEM. (A) Left ventricular (LV) mass, (B) LV mass indexed to body mass, (C) left ventricular end-diastolic volume index (LVEDVi), (D) left ventricular ejection fraction (LVEF); (E) stroke volume, (F) stroke volume index (SVi), (H) single-plane longitudinal strain (GLS), (I) single-plane longitudinal strain rate (SR), (J) single-plane circumferential strain (GCS), (K) single-plane circumferential SR, and (L) cardiac work index. (G) Representative images of the LV in the parasternal long- and short-axis (top) used for many of the measurements, and of circumferential strain curves (bottom) for a 6-segment model in the short-axis. ∗P < 0.05; ∗∗P < 0.01; ∗∗∗P < 0.001 (adjusted).

Discussion

The primary aim of this study was to examine how reduced proteolytic cleavage of vWF contributes to the development of AS. vWF is a large, multimeric glycoprotein involved in hemostasis that is synthesized by endothelial cells and megakaryocytes. Much of the vWF secreted by endothelial cells remains associated with the endothelial luminal surface in large self-associated form until it is removed by ADAMTS13. High fluid shear stress results in conformational change of vWF, thereby exposing the A1 domain, which mediates binding to platelet GPIbα, and also the otherwise cryptic A2 domain for ADAMTS13-mediated proteolysis or self-association., Failure of the proteolytic process, which can occur from a variety of acquired causes including oxidative stress and lipid disorders,, leads to excess self-associated vWF on the endothelial surface. Subsequent platelet adhesion to vWF in areas of high shear is known to influence vascular tissues through the direct recruitment of innate immune cells, release of platelet-derived signaling molecules, and membrane transfer of adhesion molecules or chemokines from platelets to other cells.,,

The underlying hypothesis that excess endothelial-associated vWF and secondary platelet adhesion accelerates aortic valve thickening was based on several lines of reasoning. Development of AS involves a cascade of cellular and matrix events including inflammatory activation as well as growth factor and osteogenic signaling through a wide variety of pathways that induce myofibroblastic and osteoblastic transformation of VICs.,,, Platelets have the potential to stimulate these processes through either the recruitment of inflammatory cells, or signaling through platelet factors including TGFβ1 and platelet-derived growth factor, which are known activators of the SMAD and Wnt/β-catenin pathways, and through specific bioactive phospholipids., The idea that platelets contribute to AS is further supported by histological observations of platelet adhesion directly to aortic VECs in hyperlipidemic mice that develop modest valve thickening and in humans with AS.,

We evaluated hyperlipidemic mice deficient for LDLR fed a WSD that not only develop atherosclerosis but also have increased endothelial-associated vWF and adhesion of platelets to the arterial wall., The additional genetic deletion of ADAMTS13 in these mice has been shown to markedly increase vWF-mediated platelet adhesion in the aorta and the development of large vessel atherosclerosis., In the current study, LDLRAD13 mice on WSD demonstrated extensive vWF not only in atherosclerotic lesions, but also on VECs. Most of these mice developed severe thickening of the aortic leaflets with increased transvalvular gradients in the traditionally stenotic range, representing a true model of AS, despite being studied at less than one-half the average lifespan of the background strain. The net increase in afterload from valve stenosis and arterial stiffness in LDLRAD13 mice was reflected by substantial elevations in valvulo-aortic impedance. As a consequence, these mice demonstrated afterload-related reduction in LV function. The high cardiac work index in LDLRAD13 mice indicated that mice were not in a decompensated state and that the decreases in LV function were all secondary to increased afterload. Accordingly, we believe that the hemodynamic measurements that account for differences in load-dependent stroke volume (eg, valve resistance) best reflect the true extent of AS in LDLRAD13 mice.

Our findings provide firm evidence that failure of ADAMTS13 cleavage of vWF predisposes to AS in a multifactorial way. In animal models of hyperlipidemia, vWF-mediated platelet adhesion is known to accelerate atherosclerosis and to promote the recruitment of cells of the innate immune system to areas of plaque formation.,, Accordingly, the increase in aortic root plaque formation seen in LDLRAD13 mice was not unexpected and, in many cases, there appeared to be an extension of plaque onto the valve leaflet surface. Yet, most of the valve thickening in LDLRAD13 mice was attributable not to inflammatory plaque formation, but rather to extensive proliferation of VICs with evidence for myofibroblastic and osteogenic transformation, an expansion of collagen-rich matrix, MMP-9 up-regulation, and microcalcification.

We believe that deficiency of ADAMTS13 exerted its effects on the aortic valve through enhanced vWF-mediated adhesion of platelets and PMVs. This assumption is reasonable based on mechanistic studies on the effects of ADAMTS13 activity on platelet adhesion in atherosclerosis, and on our findings of colocalization of platelet adhesion and signaling pathways known to result in VIC myofibroblastic transformation. Platelet-derived factors, including TGFβ1, can induce VIC myofibroblastic transformation and osteogenic differentiation.,, Although TGFβ1 can originate from a variety of vascular cell types, platelet-specific deletion of TGFβ1 has recently been shown to reduce myofibroblastic VICs. We provided evidence for TGFβ1 signaling by histological examination for 2 separate markers, including pSMAD2, which has been associated with valve fibrosis and is even found in platelets,,, and β-catenin, which in some, but not all, studies has been linked to valve thickening and osteogenic signaling.,, We observed a clear increase in both pSMAD2 and β-catenin in LDLRAD13 mice, which was particularly prominent in VECs and VICs that were in close proximity to platelet and PMV adhesion, suggesting juxtracrine signaling. Adherent platelets were also frequently positive for pSMAD2, consistent with studies that have defined the role of TGFβ1 in platelet activation. These findings highlight the accepted concept that events on the valve surface can influence VICs and valve matrix on the interior through paracrine signaling. It should also be noted that TGFβ1 can also cause mesenchymal transformation of VECs with subsequent expression of αSMA. We did not attempt to correlate the degree of platelet adhesion with the degree of leaflet thickening in LDLRAD13 mice based on the notion that leaflet thickening is a process that occurs over a long timespan, whereas histological evidence for platelet adhesion represents a single snapshot in time. LDLRAD13 mice tended to have higher platelet counts, which may have contributed further to platelet-mediated valve changes. This finding also argues against a thrombotic thrombocytopenic state, which usually requires some triggering stimulus in this strain.

With respect to clinical implications, it is possible that endothelial vWF-related events that predispose to aortic valve disease can be modified by drug therapy. Although antiplatelet therapies traditionally used in atherosclerosis can reduce platelet activation-related signaling, their effect has not been rigorously tested in native valve disease, and they are largely used to inhibit aggregation and not adhesion through interaction between vWF and GPIbα. The pathways we have defined can be linked to bleeding tendency in patients with severe AS (Heyde’s syndrome) that occurs from loss of ultralarge vWF multimers through valve shear-dependent exposure of the A2 cleavage site. Similar shear-dependent conformational changes exposes the A1 binding site for platelet GPIbα, thereby representing a mechanism for transition from mild to severe stenosis. It has been reported that proatherogenic abnormalities in cholesterol metabolism can impair vWF proteolysis, which may partially explain the excess endothelial-associated vWF and arterial platelet adhesion in hyperlipidemic animals. Hence, our findings provide additional nonatherosclerotic mechanisms for the association between dyslipidemia and AS.

Study limitations

There are limitations of the study in addition to those mentioned in the previous text that deserve mention. First, we have not yet performed interventional studies examining whether inhibiting GPIbα-vWF interaction suppresses AS development. Our study design was not intended to provide a comprehensive evaluation of all potential signaling pathways through which vWF-mediated platelet adhesion can act. Instead, we focused on TGFβ1 based on the strength of evidence for its role in AS and its role as a major platelet signaling pathway. It should be also mentioned that when evaluating abnormal aortic valve gradients, it was not possible to separate the relative contributions of valve thickening and aortic root plaque. It could be argued that LDLR mice on a chow diet would have also been an appropriate control group to study. Although this group was not prospectively included in our study design, we have generated data from a small number (n = 4) of age-matched LDLR mice on chow diet that indicate that they are similar to wild-type mice, with very little endothelial vWF and platelet adhesion on molecular imaging at 30 weeks, normal blood pressures, and no evidence for aortic valve disease on echocardiography at 50 weeks. Finally, we did not study mice at a more advanced age when strain-related differences may have been even more exaggerated.

Conclusions

Our data indicate that excess endothelial surface-associated vWF on the aortic leaflets leads to platelet adhesion, accelerated valve thickening, and AS at an advanced age, with secondary afterload-related changes in LV function. The valvular abnormalities that are associated with vWF-mediated platelet adhesion are multifaceted, involving altered matrix collagen, proliferation VICs, VIC myofibroblastic and osteogenic transformation, and to a lesser extent, inflammatory changes. These findings will likely stimulate new investigation into new approaches for mitigating progression of AS.

COMPETENCY OF MEDICAL KNOWLEDGE: AS involves the activation and transformation of VICs and possibly VECs, resulting in alteration of the valve matrix and calcification. Platelets are capable of triggering these processes through platelet-derived growth factors and cytokines.

TRANSLATIONAL OUTLOOK: Excess vWF on the valve endothelial surface can promote the adhesion of platelet and secondary paracrine signaling, representing a potential therapeutic target to prevent the progression from aortic sclerosis to AS.

Funding Support and Author Disclosures

Study protocols were supported by the Oregon Health & Sciences University Multiscale Microscopy Core and the Oregon National Primate Research Center Integrative Pathology Core. Dr Ozawa is supported by fellowship grants from the Japanese Society for the Promotion of Research, the Manpei Suzuki Diabetes Foundation, and the Japanese Society of Echocardiography. Dr López is supported by grant R35-HL145262 from the National Institutes of Health. Dr Lindner is supported by grants R01-HL078610, R01-HL130046, and P51-OD011092 from the U.S. National Institutes of Health; and grant 18-18HCFBP_2-0009 from the U.S. National Aeronautics and Space Administration. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.