Aortic valve disease in diabetes: Molecular mechanisms and novel therapies

Abstract Valve disease and particularly calcific aortic valve disease (CAVD) and diabetes (DM) are progressive diseases constituting a global health burden for all aging societies (Progress in Cardiovascular Diseases. 2014;56(6):565: Circulation Research. 2021;128(9):1344). Compared to non‐diabetic individuals (The Lancet. 2008;371(9626):1800: The American Journal of Cardiology. 1983;51(3):403: Journal of the American College of Cardiology. 2017;69(12):1523), the diabetic patients have a significantly greater propensity for cardiovascular disorders and faster degeneration of implanted bioprosthetic aortic valves. Previously, using an original experimental model, the diabetic‐hyperlipemic hamsters, we have shown that the earliest alterations induced by these conditions occur at the level of the aortic valves and, with time these changes lead to calcifications and CAVD. However, there are no pharmacological treatments available to reverse or retard the progression of aortic valve disease in diabetes, despite the significant advances in the field. Therefore, it is critical to uncover the mechanisms of valve disease progression, find biomarkers for diagnosis and new targets for therapies. This review aims at presenting an update on the basic research in CAVD in the context of diabetes. We provide an insight into the accumulated data including our results on diabetes‐induced progressive cell and molecular alterations in the aortic valve, new potential biomarkers to assess the evolution and therapy of the disease, advancement in targeted nanotherapies, tissue engineering and the potential use of circulating endothelial progenitor cells in CAVD.

incidence 7,8 of aortic stenosis, that progresses rapidly from mild to severe. 9 Histopathological assessments showed a high degree of calcification in diseased aortic valves of DM patients compared to non-diabetic patients. 10 Clinical, histological and animal model experiments revealed the complexity of the processes leading to CAVD. Using an original experimental model of hyperlipidemia and hyperglycaemia, we have shown that the aortic valve was the first vascular territory greatly affected by these two aggressors. 11  To date, there are no pharmacological treatments available to reverse or retard the progression of CAVD. Traditional cardiovascular drugs like cholesterol-lowering therapies (statins) and reninangiotensin system blocking drugs have proven to be unsuccessful in slowing the progression of CAVD in clinical trials. [12][13][14][15] These findings suggest that despite the similarity of the risk factors that induce calcification of vascular and valvular structures, different mechanisms underlie the development and progression of their mineralization. [16][17][18] Aortic stenosis (AS), the second most common indication for cardiac surgery, is most often treated by open-heart or transcatheter aortic valve replacement 19 ; both approaches are associated with a high risk of adverse events and substantial healthcare costs. 20 Given that the burden of diabetes and CAVD will continue to increase worldwide in the coming decade and that currently, there is no reliable method of determining the optimal timing of intervention for a patient with asymptomatic AS or predicting when a patient will become symptomatic, a pharmacological method to reverse or slow the progression of CAVD is greatly needed. To this purpose, it is critical to continue to uncover the specific mechanisms of valve disease progression, to reveal accurate biomarkers for diagnosis and find new targets for therapies. Despite significant advances in the field, to our knowledge, there are no specific targets or targeted therapies for the treatment of aortic valve disease in DM, urging for the need of new insights into the underlying cellular and molecular mechanism(s) on the basis of which new biomarkers, innovative bioengineering approaches and therapies for diabetes-associated valve disease could be developed.
We present here an update on diabetes-induced progressive changes in aortic valve structure and function, the advances in revealing molecular signatures and the new putative up-regulated molecules that could be potential accurate biomarkers for diagnosis and/or targets for CAVD in diabetes. Novel therapeutic strategies, i.e. targeted nanotherapy addressed to the diseased cells, stem cell therapy and recent advancements in valvular tissue engineering are discussed. The recent advances on CAVD in diabetes bring confidence for the prospective progress in the treatment of this disorder and for the reduction of valve replacement surgery.

| Normal aortic valve structure and function
The aortic valve is a tricuspid valve responsible for maintaining the unidirectional blood flow from the left ventricle into the aorta. The valve consists of three thin semilunar cusps (thickness of less than 1 mm in humans) i.e. left, right and non-coronary cusps, that are attached to a crown-shaped annulus at the base. 21 The thickness of the leaflets is not uniform but increases toward the free cusps margins. 22 During the cardiac cycle, the cusps are exposed to various stresses, including pressure, tension and bending forces. 23 The peak velocity of the blood flow through aortic valves during each cardiac systole is approximately 1.35 m/s, 24 but in a calcified and thickened valve, the velocity may exceed 4 m/s, which eventually leads to progressive valvular stenosis. Biomechanical stimuli induce phenotypic and gene expression profile changes in valvular cells. The ventricular surface of the aortic cusps is exposed to unidirectional shear stress, while the aortic side is exposed to oscillatory shear stress 25 ; as a consequence, the valvular endothelial cells (VECs) lining the two sides of the leaflets have distinct phenotypes and gene expression profiles.
Each leaflet has a trilaminar structure (fibrosa, spongiosa, and ventricularis), vital for the biomechanical properties of the aortic valve. 26 The fibrosa, facing the aorta, is exposed to low shear stress secondary to diastolic, low velocity and disturbed blood flow. The fibrosa, containing mainly type I and type III collagen fibres, provides most of the tensile strength to the valve. The central layer, spongiosa, which represents about 60%-70% of the thickness of the cusp is rich in glycosaminoglycans (GAGs), components of proteoglycans that are highly hydrated, and act as "shock absorbers" during the cardiac cycle. The ventricularis, the layer facing the left ventricle, contains collagen and elastin fibres, providing more compliance; it grants the apposition of free edge leaflet regions and prevents the backward blood flow into the left ventricle during diastole. 27,28 The ventricular side of the cusps is exposed to high-shear stress due to a systolic, high velocity and laminar blood flow. The normal valve leaflet is avascular 29 but innervated by afferent and efferent nerves which contribute to valve function. 30 As the entire cardiovascular system, the aortic valve surfaces are lined by endothelial cells that regulate vascular tone, inflammation, thrombosis and remodelling. They sense changes in shear stress and translate these mechanical stimuli into biological responses. 31 There is evidence that VECs express von Willebrand factor, exhibit angiotensin converting enzyme activity and synthesize a rich extracellular matrix 32 ; moreover, the cells are coupled by functional communicating (gap) junctions. VECs produce fibronectin, prostacyclin, hyaluronic acid and heparin-like GAGs and are metabolically active taking up LDL. 33,34 Although VECs were shown to share many functions similar to endothelial cells (ECs) from other locations, 33 they also display valve-specific alignment, with a perpendicular orientation to flow, as compared to aortic ECs which respond to flow by aligning parallel to the direction of flow. Moreover, VECs response to shear stress was shown to be different when compared to aortic EC, 35,36 and the responses were found to be dependent on cytoskeletal reorientation. Interestingly, VECs display a different transcriptional profile compared to aortic EC. 31 The predominant cell population that resides in the valve interstitium are the valvular interstitial cells (VICs), which serve to maintain tissue homeostasis and structural integrity. As shown in Figure 1, these cells are embedded within the valve extracellular matrix (ECM), lack an organized basal lamina and present numerous slender extensions that establish contact with VECs and the neighbouring VICs. 37 These cells are responsible for the generation, maintenance and repair of the ECM which is mainly composed of elastin, collagen and proteoglycans. 26,38 We have reported that VICs display both fibroblasts and vascular smooth muscle cells characteristics. 39 Interactions between mechanical forces, valvular cells and the ECM influence remodelling potential and therefore durability of heart valves.
In addition to VICs, the valvular stroma is populated with some resident macrophages and very rare smooth muscle cells. 40 A population of resident stem cells lying within the cusps has recently been identified. 41 They appear to originate from the mobilization of hematopoietic-derived stem cells towards cardiac valves, 42 potentially contributing to normal or abnormal valve repair.

| Diabetes-induced changes in the aortic valve structure and function
Experimental and clinical studies have shown that the aortic valve root is a lesion-prone area for atherosclerotic plaque development one of the first and faster affected areas in animal models of combined diabetes and atherosclerosis. 11 Notably, the changes were most consistently observed on the aortic aspect of the valve (fibrosa), the surface exposed to low shear forces and high hydrostatic pressures. In these areas, VECs acquire a synthetic phenotype and these changes are detected as early as 2 weeks after the onset of experimental diabetes. They are associated with a significant hy- Recently, we have reported that early diabetes induces aortic valve dysfunction in diabetic ApoE−/− mice fed with a hyperlipemic diet as detected by echography during the first week after the onset of diabetes. As importantly, we detected changes in the expression of molecules associated with inflammation, remodelling and osteogenesis. Remarkably, the peak aortic jet velocity (a marker of valve dysfunction) was highly correlated with an inflammatory biomarker (VCAM-1), with pro-osteogenic markers osteocalcin and alkaline phosphatase (ALP) with remodelling enzymes (MMP9) and with the myofibroblastic marker αSMA. These findings suggest that valvular dysfunction could develop even before clinical signs of AS and may be highly correlated with the specific molecular changes in valvular F I G U R E 1 Electron micrograph depicting a fragment of a bicuspid valve (hamster). Valvular endothelial cells (VEC) with thick and thin segments line the ventricular side of the valve. Within the rich extracellular matrix, a valvular interstitial cell (VIC) lacking a basal lamina, exhibits numerous extensions that establish focal close apposition with VECs (arrowhead) and with neighboring VIC (arrows). x13,500. By permission from Circulation Research, 59, 3, 1986, p.13, Figure 2 tissues. These highly expressed molecules may possibly become accurate biomarkers for diagnosis and targets for therapy. 46 A comparison of the evolution of CAVD between diabetic and non-diabetic subjects 10 showed that the aortic valve leaflet macrocalcification was significantly enhanced in diabetic patients, whereas inflammation was similar in diabetic and non-diabetic individuals. Moreover, Runx2 and ALP were detected to be significantly higher in diabetic patients suggesting that many valvular cells may undergo osteogenic differentiation. These studies, however, were restricted to surgically removed valve leaflets from patients with end-stage disease and only highlight a snapshot in time of late-stage events. 47,48 Future investigations involving comparative proteomics and transcriptomics analyses with larger sample sizes are warranted.
In another recent study, 49 the role of small leucine-rich proteoglycans in degenerative aortic valve disease and the influence of diabetes and hyperglycaemia on human aortic valves and valvular interstitial cells were examined. The results showed that biglycan, but not decorin or lumican was upregulated in degenerated human aortic valve cusps, hypothesizing that biglycan represents a potential link between degenerative aortic valve disease and diabetes. 49 It was also shown that in patients diabetes is associated with increased valvular inflammation, measured by C-reactive protein expression in patients' valvular tissue. 50 It was also demonstrated in diabetic patients that advanced glycation end products (AGEs) and AGE receptors (RAGE) accumulation is associated with AS severity; indeed, AGE-related valvular collagen cross-linking leads to enhanced inflammation, oxidative stress and calcification of the leaflets. 51 Lately, a prospective cohort study has confirmed that DM is associated with an increased risk of AS. 52,53 However, other studies failed to demonstrate an association between AS progression and metabolic syndrome or diabetes during 3 years follow-up; it was claimed that, in AS patients with well-controlled DM, the effect of hyperglycaemia on AS severity is minor. 54 More extensive clinical trials are needed to clarify this issue.

| In early diabetes valvular endothelial cells switch to a secretory and adhesive phenotype
As mentioned above, in a diabetic animal model, pathological changes occur particularly on VECs lining the aortic side of the valve, which is exposed to low shear forces and high hydrostatic pressure. 11 Two weeks after onset of diabetes, VECs switched to a secretory phenotype, exhibiting an increased number of the rough endoplasmic reticulum elements, Golgi apparatus, and caveolae, features that correlate well with the progressive development of a multilayered basal lamina. Moreover, VECs exhibit a dramatic abundance of microfilaments, microtubules, centrioles and Weibel-Palade bodies. 11 This is followed by the VECs switch to an adhesive (pro-inflammatory) phenotype, a process characterized by the expression of more and new surface adhesion molecules, that attract and induce adherence followed by diapedesis of blood monocytes ( Figure 2). We have reported that short exposure of cultured VEC (24 h) to high glucose induces enhanced monocyte adhesion by mechanisms involving ICAM-1, VCAM-1, E-selectin and CD18. 55 Interestingly, the adhesivity of VEC for monocytes was higher than that of aortic ECs, results which may explain, in part, the propensity of cardiac valves for accelerated atherosclerosis in diabetes. 55

| Valvular endothelial cells undergo endothelial-to-mesenchymal transition
In diabetes in the aortic valve disease, an early event occurring in VECs in the aortic valve disease is the hyperglycaemia-induced increased number of intermediary filaments and microtubules and F I G U R E 2 Early-stage ultrastructural modifications of the aortic valve lesion occurred in a hyperlipemic/ diabetic hamsters. Under a continuous endothelium (E) having thin areas intercalated within zones in which the cell is highly enriched in biosynthetic organelles, there is a characteristic hyperplasic, multilayered basal lamina (BL). The proliferated matrix contains numerous calcification cores (arrow). A plasma monocyte (M) insinuates a pseudopod between two valvular endothelial cells. (AL), aortic lumen. x7000. By permission from American Journal of Pathology, 148, 3, 1996, p. 1004, Figure 8 attenuation of intercellular junctions. 11 These features are associated with a switch of endothelial cells to a mesenchymal phenotype.
Endothelial-to-mesenchymal transition (EndMT) is a process by which the endothelial cells progressively acquire the phenotypic and functional characteristics of mesenchymal cells and express both endothelial and mesenchymal cell markers. In the heart valves development, EndMT is a physiological process. However, in the adult organisms when activated in the adult organisms, EndMT contributes to the progression of different diseases including diabetic nephropathy, diabetic renal fibrosis, cardiac fibrosis and atherosclerosis [51][52][53][54][55][56][57][58][59] and has been shown to play a role in the pathogenesis of CAVD. 64 Numerous in vivo and in vitro studies identified the EndMT-related stimulants, such as inflammatory cytokines (TNF, IL-6 and TGF), cellular transition features and underlying signalling pathways. 60,61 Although evidence of EndMT in VEC accumulated with time, 65 the precise role of this process in the early-and end-stage phases of CAVD is still unclear. Current data suggest that EndMT precedes osteogenic changes in VECs 61 but more studies are needed to uncover the links between the process of EndMT and calcification.
Hyperglycaemia was shown to induce EndMT in several endothelia such as aortic EC, HUVECs, human retinal EC and glomerular EC. The process involves various mediators, for instance Ephrin B2, TGFβ, angiotensin II, miR-328 and signalling pathways, as FAK pathway, MAPK pathways and ROS/ERK1/2/MAPK-dependent mechanisms. 62,63 Future, intensive investigations should focus on the role of these potential risk factors in mediating EndMT in CAVD.

| Chronic hyperglycaemia induces an inflammatory phenotype of valvular endothelial cells
In search for the mechanism(s) of CAVD in diabetic environments, we developed a 3D model of human aortic valve leaflet, based on methacrylate gelatin populated with human VECs and VICs. The construct was exposed to chronic high glucose (HG) for 7 and 14 days, and the phenotypic changes of VECs and VICs were assessed. 64,65 As shown in Figure  and E-selectin and of integrins αv, α4, β3 and β5. 65 Interestingly, at this time, from the TGF-beta family, only BMP-4 expression is increased in VECs. 64 Collectively, these data suggest that chronic HG induces in VEC mainly an inflammatory phenotype, but could have the capacity to adjust and control the expression of several pro-inflammatory and pro-osteogenic molecules. More studies will reveal the involvement of these cells in CAVD during the progression of diabetes.

| Chronic hyperglycaemia induces an inflammatory and osteoblastic phenotype in valvular interstitial cells
As mentioned above, VICs, the predominant cell population found within the rich ECM of the aortic valve, are a heterogeneous population. Some VICs, like those found in porcine valves, exhibit a pericyte-like behavior. 66 Others, like VICs isolated from patients with CAVD, have angiogenic potential 67 and in pathological conditions differentiate into osteoblast-like cells and promote calcification. 39,67 Recently, using single-cell RNA sequencing for the high-throughput evaluation of heterogeneity in cells isolated from healthy human aortic valves, three subpopulations of VICs were defined by combining their developmental origin, localization, physical properties, morphology and molecular functions. 68 VICs highly expressed the previously confirmed markers, collagen type IA1 (COL1A1) and type IIIA1 (COL3A1), substantiating that VICs are the primary resident cells in aortic valve tissues.
Activation of VICs is a normal regenerative process in the heart valve, but under pathological conditions, i.e. hyperlipidemia, diabetes or atherosclerosis, their activation leads to the inception of CAVD. 51 There is evidence that elevated glucose levels, similar to those found in DM, enhance mineralization of cultured VICs. 69 To uncover the mechanisms and the molecules implicated in the aortic valve calcification in diabetes, interesting data were obtained on VICs cultured in a 2D and 3D systems. Employing a 2D culture system, VICs did not show morphological changes and did not acquire an osteogenic phenotype in hyperglycaemia or hyperinsulinaemia. 70 However, using a 3D model of human aortic valve, with VECs seeded on the surface of the construct previously encapsulated with VICs, we identified the inflammatory, remodelling and osteogenic changes induced by chronic HG in these cells. 64,65 As shown in Figure 5, VICs exposed for 7 days to HG exhibit an enhanced expression of pro-inflammatory and osteogenic molecules. Importantly, HG increases the expression of RUNX2 transcription factor and activation of SMAD 2/3 and 1/5/8. In addition, HG increases the remodelling activity in VICs, as shown by the en- Compared to controls, after 14 days of HG exposure, VICs exhibit an increased gene expression of cytokines, cell adhesion molecules, integrins, remodelling and pro-osteogenic molecules. In addition, the elevated level of the phosphorylated form of PKCα could be involved in the production of IL-1β ( Figure 5).
In summary, the above data indicate that chronic hg (that mimics diabetes conditions) induces in human vics co-cultured with human vecs (in a 3d system), a concomitant inflammatory and osteoblastic phenotype and an increase in their remodelling activity. exposure of vecs and vics to osteogenic conditions leads to the development of calcium deposits, indicating that the 3d model is suitable to study valve calcification in diabetes. 64

| NE W THER APEUTI C APPROACHE S FOR AORTI C VALVE D IS E A S E IN D IAB E TE S
Currently, there are no efficient pharmacological treatments to prevent or reverse CAVD. Lately potential medical approaches to circumvent this disease include nanotherapies, which are now beginning to be explored, stem cell therapies and more generally tissue engineering. Some of the advancement in these fields is briefly described below.

| Nanotherapies
Interventions aiming to stop or reverse the osteoblastic transition of VIC may represent a therapeutic option for CAVD. A master transcription factor implicated in osteoblast differentiation is Runtrelated transcription factor 2 (Runx2) that regulates transcription and determines the increased expression of osteogenic genes such as collagen I, alkaline phosphatase (ALP), osteopontin (OSP), bone sialoprotein (BSP) and osteocalcin (OCN). 71 Runx2 is not expressed in normal aortic valves, but its expression is induced in CAVD. [72][73][74] An increased mRNA Runx2 level was determined in the aortic tissues of mice with combined dyslipidemia and type 2 diabetes compared to non-diabetic and control mice. 45 We have also detected an increased expression of Runx2 in VICs exposed to HG in a 3D model of human aortic valve. 64 Thus, Runx2 is an important contributor to CAVD in diabetes/diabetic conditions. Considering that the osteoblastic differentiation of VICs leads to aortic valve calcification and the role of Runx2 in the process, we have developed a nanotherapeutic strategy targeted to prevent the phenotipic differentiation of human aortic VICs into osteoblast-like cells in diabetic and pro-osteogenic conditions. 75 We designed nanocarriers to silence Runx2, namely fullerene (C60)-polyethyleneimine (PEI)/short hairpin (sh)RNA-Runx2. We reported that these nanocar-

| Stem cell therapy
In general, stem cell therapy was developed to correct the dysfunctional recruitment and homing of progenitor cells. 25 The most studied sources of stem cells are the endothelial progenitor cells (EPCs) and the adipose-derived stem cells (ADSCs).
EPCs represent a small fraction of the circulatory cells that are involved in vascular repair and angiogenesis. There are data that support the involvement of the αVβ3 and αVβ5 integrins in the adherence of EPC to denuded vessels. 76 In patients with aortic stenosis, valvular endothelium regeneration is impaired not only by an increased senescence of VECs but also by a reduced number and function of circulating EPCs. It is generally accepted that diabetes reduces the number of EPCs and induces dysfunction in circulating EPCs, by mechanisms still uncovered.
In a recent study on streptozotocin-induced diabetes in ApoE−/− mice, we showed that early-stage diabetes superimposed on atherosclerosis generates alterations in EPC number, phenotype and homing. 76 Importantly, in atherosclerosis-prone mice, lower recruitment of EPCs in the aortic valve in early diabetes is the result of

| Valvular tissue repair or regeneration employing biomaterials
Since heart valve tissues cannot regenerate spontaneously, replacement with artificial biological or mechanical heart valves, repair via reconstructive surgery or interventional catheterization is the current treatment option for management of advanced heart valve diseases. 25 Artificial valves have a limited lifespan of ~10-15 years after implantation due to degeneration, calcification and thrombosis. 80 Moreover, durability of artificial valves is further reduced to only 5-7 years in diabetic patients 81

| FUTURE CHALLENG E S AND OPP ORTUNITIE S
First believed that the aortic valve calcification is a passive degenerative process, it is now recognized that is a cell-driven active pro-  Table 1.
The pioneering studies performed in experimental hyperlipidemia/hyperglycaemia by Dr Maya Simionescu's group identified the aortic valve as the first vascular territory affected by diabetes. 11 Recently, employing murine models and in vitro studies using 3D valve scaffolds we have shown that early diabetes induces almost concurrently inflammatory and osteogenic modifications of the valvular interstitial cells leading to increased calcium deposition and overall valve dysfunction. 46,64,65 Notably, our studies suggest that, in diabetic conditions, a significant inflammatory process occurs in the  Lately, miRNAs were proposed as innovative biomarkers and therapeutic strategies for aortic valve stenosis, as revealed in a groundbreaking pre-clinical study using inhibitors of miR-34a. [86][87][88] However, there are few data that identify biomarkers in CAVD in diabetes. Several differentially expressed "early" molecules which could serve as putative biomarkers for diagnosis and therapeutic targets in aortic valve disease in diabetes are presented in Table 1. Still, it remains a future challenge to identify accurate biomarkers to be employed for diagnosis and therapy for this disease.
Importantly, new nanocarriers aiming to block the shift of VICs to an osteoblastic phenotype were tested and found to be efficient in vitro; these results pave the way to the development of nanotherapies for CAVD. 75 Another potential promise for treatment of CAVD is the recently emerged stem cell therapy. In addition, tissue engineering of AV attains continuously new improvements. A challenge for the future of valve replacement is to provide a viable valve populated with cells capable of self-regeneration and growth, with greater life span, better biocompatibility and resistance to diabetes. [89][90][91][92] Meanwhile complex emerging technologies such as the nextgeneration 'omics' techniques allows the study of the miRNAome, transcriptome, proteome and secretome [93][94][95][96] ; these big data will require analysis by improved bioinformatics tools which will lead to identification of the most relevant targets and the best therapeutic option. Transcriptional analyses of human aortic valves all focused on aortic stenosis were recently reported. Notably, a cell-type transcriptome atlas of human aortic valves was elaborated showing the cell heterogeneity and the involvement of endothelial to mesenchymal transition in the evolution of calcific aortic valve disease. 73 The study is an opportunity to understand the implication and the interaction of different valvular cells in CAVD in diabetes.
Another emerging perspective to be explored for the diagnosis and therapy of CAVD in diabetes are the extracellular vesicles (EV).
There is evidence of the potential role of ADSC-derived EVs in different pathologic conditions, either as biomarkers or as a direct effector or as a delivery system to target miRNAs to cells. 79 Finally, to find relevant targets for therapy, new animal models 96,97 which could recapitulate the evolution of human AV disease need to be developed. Once obtained, the newly found therapeutical targets need to be further validated in 3D-human aortic valve models, 64,98 as well as in large animals, for evaluation of safety and efficacy, before testing in clinical trials. 99 These new data together with the present achievements in the biomedical science warrant a good perspective to find ways to prevent, slow-down or reverse CAVD in diabetes. 100

TA B L E 1 (Continued)
NIGMS (to DS) and the Harriet and Jerry Dempsey Endowment (to DS).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The dataset presented in this study is available from the corresponding author upon reasonable request.