Background
Calcific aortic valve disease (CAVD) is the most common form of valvulopathy in the western world, with a prevalence of 13.3% in people above 75 years of age [
1,
2]. CAVD is associated with a 50% elevated risk of morbidity and mortality [
3]. Once severe aortic stenosis develops, valve replacement, either via surgical or transcatheter approaches, is the current treatment standard, as early detection, prevention, and mitigation strategies have yet to be clinically adopted [
1,
2,
4]. There is thus a need to develop better diagnostic and therapeutic approaches for CAVD.
The early pathogenic processes in CAVD are defined by the phenotypic transformation of valve endothelial (VEC) and interstitial (VIC) cells via cellular activation, osteogenic differentiation [
5‐
7], as well as maladaptive extracellular matrix (ECM) remodeling [
8,
9]. CAVD pathophysiology is marked by infiltration of cytokines, lipid deposition, and calcific nodule formation [
4,
10,
11]. These early pathogenic processes are simulated using two- and three-dimensional in vitro models and in vivo models that mimic the CAVD process and valve milieu to varying degrees of accuracy [
7,
12‐
15]. Wild-type diet-based mouse CAVD models have previously demonstrated deposition of monocyte-macrophages, lipids, and lipoproteins [
16,
17], but to our knowledge no such diet-based models exhibit early calcification as demonstrated by positive Alizarin Red S (ARS) staining, as well as positive osteopontin and Runt-related transcription factor 2 (RUNX2) expression [14,17]. Assmann and colleagues have shown that a diet regimen supplemented with vitamin D, cholesterol and dicalcium phosphate caused significant calcification of the aortic root in Wistar rats [
18]. Vitamin D in combination with cholesterol has been separately shown to increase vascular calcification in rats [
19]. Vitamin D is responsible for maintaining serum calcium levels in humans and both deficiency and excess of Vitamin D has been shown to promote cardiovascular calcification [
20‐
22]. Higher calcium phosphate and elevated serum calcium are also established risk factors for calcification progression [
21,
22]. We exploited a similar dietary regimen to develop a wild-type mouse model for early CAVD.
Multiple biochemical and imaging markers are associated with independent events such as endothelial dysfunction, inflammatory cytokine infiltration, matrix remodeling, and mineral deposition to assess either the presence or severity of calcification [
2,
4,
10,
11,
23]. However, there is a dearth of label-free imaging biomarkers to monitor the progression of calcification in the aortic valve, both in clinics and research laboratories [
23]. Clinically, a stenotic aortic valve usually is detected by auscultation of a systolic murmur and confirmed by echocardiography, and relies on late hallmarks of CAVD like hemodynamic malfunction and impaired geometry of the valve [
24]. Imaging techniques such as cardiac magnetic resonance imaging and cardiac computed tomography have also been shown to be sensitive to the late hallmarks of the disease; however, there is a specific lack of tools to detect early CAVD progression [
24‐
26]. Two-photon excited fluorescence (TPEF) microscopy has recently shown potential in providing label-free quantitative metrics that might associate with osteogenic differentiation [
27]. TPEF allows quantification of the endogenous fluorescence ratios of the cellular co-factors flavin adenine (FAD) and nicotinamide adenine (NADH) dinucleotides in their oxidized and reduced forms, respectively and is known as the optical redox ratio (i.e. FAD/(FAD + NADH)) [
27]. It was previously established in vitro that osteogenic differentiation of mesenchymal stem cells is associated with a reduction in TPEF-derived optical redox ratios [
27]. We have also demonstrated that this TPEF-based optical redox ratios in VICs was reduced as cellular stretching increased from normal to pathologic magnitudes [
13]. Additionally, optical redox ratio altered and correlated with temporal changes in VIC phenotype during osteogenic de-differentiation, in vitro [
28]. In addition to NADH and FAD, molecules such as collagen, calcium and lipids within tissues will contribute to the autofluorescence emission in the visible range [
27,
29‐
32]. Recently, TPEF has been used to identify endogenous fluorescence from calcium deposition that correlated with mineralization in ApoE
−/−mice and calcified human valves [
29].
In this study, based on the above prior work, we hypothesized that TPEF autofluorescence markers would coincide with traditional phenotypic markers that indicate early progression of CAVD. We tested this hypothesis using ex vivo imaging of valves from a diet-based wild-type C57BL/6J mouse model that demonstrated early CAVD progression with lipid and calcium deposition, and matrix remodeling. In our model, the aortic valve commissures showed lower TPEF autofluorescence ratios, which was concurrent with increased cell activation, inflammatory cytokine expression, cell proliferation and markers for osteogenic differentiation, and which was predictive of increased calcification at later time points.
Discussion
The most common small animal model for CAVD is the hypercholesterolemic mouse with apoE and/or low density lipoprotein receptor knockouts [
14]. Other approaches involve genetically-modified mouse models with NOTCH1
+/− (NOTCH1 heterozygous), Postn
−/− (periostin deletion), or NOS3
−/− (endothelial nitric oxide synthase deletion) phenotypes [
14,
46]. These genetic mouse models are typically combined with a high fat and high cholesterol western diet [
14,
46]. Most of these models have been shown to develop valve calcification in aged mice, or require twenty weeks or more of treatment [
41,
47]. Another prior CAVD mouse model was created using a guidewire inflicted aortic valve tissue injury and was shown to develop calcification by twelve weeks following injury [
41]. In the current study, we developed and validated a diet-based wild-type small animal model for early progression of valve calcification, without genetic manipulation which showed early hallmarks of CAVD by 4 and 16 weeks. Male mice were chosen due to the higher prevalence of CAVD in males [
48]. We observed positive ARS staining of calcification at the commissural walls, similar to what Assmann et al. observed in a rat model [
18]. However, unlike the rat model in Assmann et al
. we did not observe evidence of calcification in the aortic roots.
The mice in the pro-calcific diet groups also had evidence of lipid deposition near the commissural walls at 4 weeks as has been previously reported in mice on a high-fat diet with added cholesterol [
14]. Two previous studies reported valve mineralization in wild-type diet-based mouse models without genetic manipulation approximately 16 weeks after initiation of diet [
16,
17]. However, the diet employed in these studies incorporated 58.7% fat, which is significantly more than the western diet, and also contained additional carbohydrates. In addition, demonstration of CAVD in these earlier studies relied solely on positive von Kossa particulate staining, which can be confounded in a mouse model by melanocytes or lipofuscin-containing granules with appearance similar to that of von Kossa positive calcification [
17]. In our model, we have shown positive ARS staining concurrent with the location of increased activation, proliferation and osteogenesis, suggesting improved utility of our model [
14]. To our knowledge, this is the only wild-type diet-based mouse model for early CAVD that has shown ARS-positive calcium staining.
Subtle increases in TGFβ1, which have been known to influence VIC proliferation, differentiation and apoptosis, were observed in the pro-calcific mice at 4 and 16 weeks. This observation, together with increased Ki67, osteopontin, and RUNX2 expression suggest increased proliferation and osteogenic differentiation are occurring in the tissues, which are known hallmarks of CAVD [
42] and has been previously reported in various other mouse models [
5,
11,
14,
17,
18,
41,
46,
47]. Additionally, concurrent expression of TGFβ1, αSMA and Ki67 also suggests that increased TGFβ1 could have led to the increased proliferation of the myofibroblastic phenotype of VICs in our disease model [
49].
Via QPLI, we observed that the 4-week pro-calcific mice had more mature collagen fibers and a higher positive pixel density with minimal changes in collagen fiber orientation compared to controls, suggesting increased collagen remodeling [
14,
45,
47]. In the 16-week pro-calcific mice, we observed a decrease in collagen thickness and density with an increased variance in collagen orientation in the leaflets. This was corroborated by the negative correlation between average retardation and average local directional variance. This led us to speculate that by 16 weeks, there was either greater collagen degradation or de novo collagen synthesis, leading to thinner fibers and increased variance in fiber orientation. Collagen remodeling is typically regulated by multiple factors including matrix metalloproteinases (MMPs), cathepsins or other pro-fibrotic factors like TGFβ and twist-1 [
50]. In fact, increased MMP activity co-localized with calcification at the commissures in rats fed with a similar diet in other studies [
18]. ECM changes have been associated with changes in phenotype [
51]. Moreover, in our study, the pro-calcific mice demonstrated a more myofibroblastic phenotype, as evidenced by increased expression of αSMA, which could further contribute towards matrix remodeling. Pro-calcific mice also had a higher percentage of matured collagen fibers in diseased regions at 4 weeks, consistent with CAVD progression [
14,
45,
47].
Previously, TPEF microscopy has been used to demonstrate collagen remodeling through SHG imaging [
45] and calcification through autofluorescence imaging [
29] in the valve at the tissue level. TPEF microscopy can also be used to obtain an optical redox ratio ([FAD]/([FAD] + [NADH]) which reflects the metabolic state of the cell [
13,
27,
39]. It has been shown in vitro that mesenchymal stem cells undergoing osteogenic differentiation exhibit decreased optical redox ratios [
27]. We have also shown previously that VICs under pathological stretch have reduced optical redox ratios compared to unstretched cells [
13]. We have also demonstrated that when VICs are subjected to osteogenic conditions in vitro, their optical redox ratio decreased and correlated with gene expression of osteogenic markers like RUNX2, osteopontin and osteocalcin [
28]. In the current study, we measured the TPEF 755–860 ratio (A
860/525/(A
755/460 + A
860/525)) from ex vivo tissue sections by obtaining the autofluorescence intensities at 755 nm and 860 nm excitation, which correspond to the NADH and FAD excitation wavelengths. However, we did not term it an optical redox ratio, as tissue autofluorescence measured via the TPEF 755–860 ratio can be attributable to multiple molecules including collagen, lipids, and mineralized calcific deposits [
29,
30]. The TPEF 755–860 ratio could thus have been influenced by collagen and calcium in the tissue [
29]. We therefore assessed the TPEF Col-Cal ratio, where the individual A
810/460 and A
810/525 intensities correspond to collagen and calcium, respectively [
29]. The TPEF Col-Cal ratio increased significantly in aortic valve roots and commissures for the 16-week pro-calcific mice suggesting an increase in collagen and calcium autofluorescence.
In our study, we found that the TPEF autofluorescence ratio rather than the TPEF Col-Cal ratio was more sensitive to disease progression. While TPEF Col-Cal ratio did not show statistical significance between control versus pro-calcific mice, TPEF 755–860 ratio was sensitive to both time, presence of disease conditions, and their interaction. TPEF Col-Cal ratio only showed differences in the commissures, while TPEF 755–860 ratio showed significant changes in the leaflets, commissures and roots. We speculate the difference in sensitivity between TPEF 755–860 and TPEF Col-Cal ratios might be because of increased sensitivity of the TPEF 755–860 ratio to NADH. It should be noted, however, that both these TPEF autofluorescence ratios are influenced by collagen, lipids, mineralized calcific deposits, NADH, and FAD [
29,
30].
Additionally, we observed that TPEF 755–860 ratio correlated with RUNX2 and Ki67 negatively suggesting increased osteogenesis and proliferation was marked by reduced TPEF 755–860 ratio, which is consistent with previous studies [
13,
27,
32,
39,
52]. It is also interesting to note that A
810/460 and A
810/525 intensities positively correlated with RUNX2 expression leading us to believe that RUNX2-expressing diseased areas may exhibit higher collagen and calcium deposition. Indeed, A
810/460 and A
810/525 intensities also positively correlated with average directional variance and collagen positive pixel density. This suggested that the increase in A
810/460 and A
810/525 intensities indeed allude to increased collagen [
45] and calcium [
29] deposition during disease progression. TPEF 755–860 ratio which showed a decrease during disease progression correlated negatively with average directional variance further evidencing, thinner disorganized fibers during disease progression. This may suggest increased de novo collagen and calcium deposition with increased CAVD progression. Similar correlation between cell redox state and collagen synthesis is consistent with other studies [
27,
53,
54]. The ratio of blue:green (i.e. 460 nm:525 nm emission) TPEF autofluorescence at 800 nm excitation has been shown to be sensitive to mineralization and altered CAVD progression [
29]. Our current study is the first instance when a TPEF 755–860 ratio has been shown to demonstrate tissue level changes occurring during early CAVD progression in a mouse model that was more sensitive than the TPEF Col-Cal ratio. Like we had previously demonstrated in our in vitro studies [
13,
28,
52], in this study also we observed TPEF autofluorescence ratios decreased at 16 weeks in the pro-calcific mice commissures and correlated with increased calcification, osteogenic differentiation and proliferation. This is also similar to the findings of Quinn et al
. where the osteogenic differentiation of the mesenchymal stem cells correlated with the reduction in an optical redox ratio [
27] and Jones et al. where increased proliferation correlated with a reduction in redox ratio in skin wound healing [
39]. TPEF 755–860 ratio not only correlated with osteogenic progression and increased proliferation but also seemed to have a predictive quality for calcification. This further provides an incentive to explore TPEF 755–860 as an early biomarker for CAVD progression.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.