Gene expression analysis in AVS and bAVS

Freshly explanted aortic valves of patients with severe AVS or bAVS were collected in the operating room, and valve leaflets were either processed for histological stainings or frozen for subsequent RNA isolation and bulk RNA sequencing (Supplemental Fig. 1A). In total, 64 aortic valves (75 leaflets) from 61 different patients were used (9% women). Images of explanted AVS, bAVS, and non-calcified aortic regurgitation (AR) control valves show calcifications typical for AVS and bAVS (Supplemental Fig. 1B).

Fig. 1

Bulk RNA-seq analysis of bAVS vs AR, and bAVS vs AVS: A volcano plot of the up- and downregulated genes in bAVS (n = 13) vs AR (n = 7). Genes with p < 0.05 and log2FC > 1 are highlighted in color. The 20 most significantly differentially expressed genes (DEGs) are labeled by name. B GO analysis of the upregulated genes. Categories related to angiogenesis, inflammation, extracellular matrix organization, and calcification are underlined. C volcano plot of the up- and downregulated genes comparing bAVS (n = 13) with AVS (n = 14). Genes with p < 0.05 and log2FC > 1 are color-coded (red: upregulated; blue: downregulated). The 21 most significant DEGs are labeled by name; DMP1, PRG4, and MMP12 are highlighted. D Venn diagram of DEGs of the different comparisons

To explore potential differences in the pathomechanisms of AVS and bAVS, we performed in-depth bulk RNA-seq analyses and compared the gene expression profiles of AVS (n = 14) and bAVS (n = 13) with those of AR controls (n = 7). Interestingly, this analysis revealed only a relatively small number of differentially expressed genes (DEGs) (58 in total; 41 up- and 17 downregulated) between AVS and AR controls (Supplemental Table 2). In contrast, when comparing bAVS with AR controls, a significantly higher number of DEGs (447 in total; 298 up- and 149 downregulated) was identified (Fig. 1A; Supplemental Table 3). Gene Ontology (GO) analysis of upregulated genes in the bAVS group yielded terms such as angiogenesis/vascular development, extracellular matrix (ECM) organization, and regulation of tumor necrosis factor superfamily cytokine production, suggesting that angiogenesis, immune response, extracellular matrix remodeling, and calcification are involved in the disease process (Fig. 1B; Supplemental Table 4). Next, we compared bAVS vs. AVS to obtain more information on potentially different disease mechanisms (Fig. 1C). We found 21 (17 up-, 4 downregulated) DEGs. In particular, genes involved in the calcification process, such as dentin matrix acidic phosphoprotein 1 (DMP1), proteoglycan 4 (PRG4), and matrix metallopeptidase 12 (MMP12), were strongly upregulated in bAVS (Supplemental Table 5). Additionally, we examined common genes that were differentially expressed in AVS and bAVS vs. AR, or were specific to either AVS or bAVS (Fig. 1D). Among the 40 jointly differentially expressed genes were integrin-binding sialoprotein (IBSP), various collagens (e.g., collagen type I alpha 1 chain (COL1A1), collagen type IV alpha 1 chain (COL4A1)), and serpine family E member 1 (SERPINE1). Notably, members of the matrix and ADAM metalloprotease families, which are involved in ECM processing, along with members of the Fc gamma receptors (FCGR) and interleukin/chemokine families, which are involved in inflammation, were only upregulated in bAVS vs. AR, but not in AVS. Therefore, we focused our further analyses on angiogenesis, inflammation, ECM composition, and calcification.

Processing of strongly calcified human aortic valves with Kawamoto’s film method

We also aimed to investigate protein expression and distribution using immunostainings. However, the standard decalcification procedure routinely used for cryosectioning calcified aortic valves often compromises the tissue sample integrity and staining quality, similar to the effects of chelating agents or acids. Because preserving the structural integrity of calcified aortic valve tissue can enhance our understanding of disease processes, we evaluated Kawamoto’s film method for this purpose. In this method, a cryofilm is applied to the specimen before sectioning with a tungsten carbide blade. This allows the freshly cut sections to adhere to the film, thereby potentially maintaining the structural integrity of even heavily calcified aortic valve samples. We first compared conventional cryosectioning with the Kawamoto method on calcified aortic valve samples. The hard tungsten carbide blade effectively cut through aortic valve tissue with significant calcifications, and the cryofilm prevented the calcified regions from crumbling and disrupting adjacent tissue. In contrast, conventional cryosectioning, whether or not decalcification was performed (Supplemental Fig. 2C,D) (decalcified valves n = 20, non-decalcified valves n = 20), often resulted in folding and a severe loss of structural integrity, which was also caused by displaced calcified material. By comparison, sections prepared with Kawamoto’s film method (Supplemental Fig. 2D) were well-preserved, even when processing heavily calcified human aortic valves (n = 58). This is particularly relevant for the endothelial cell layer, which surrounds healthy aortic valves and serves as an important barrier. Unlike the conventional method, Kawamoto sections preserved an intact CD31+ endothelial cell layer even in areas near heavily calcified regions in AVS and bAVS (Supplemental Fig. 2D, inset). These findings further highlight the excellent preservation of strongly calcified aortic valve tissue using Kawamoto’s film method, as even the delicate endothelial cell layer remained intact. In addition, the apotome device of the microscope was used in order to obtain sharper, high-contrast images (Supplemental Fig. 2B).

Fig. 2

Assessment of angiogenesis in aortic valves using immunostainings and bulk RNA-seq analysis: A–E bAVS sample stained for CD31 (red) and nuclei (blue); magnification of the boxed area in A. B–E Scale bar = 1000 µm for overview image, 50 µm for inset. F Quantitation of CD31+ vessels. *p < 0.05; analyzed by one-way ANOVA. G Heatmap of bulk RNA-seq data (AR: n = 7, AVS: n = 14, bAVS: n = 13)

Angiogenesis, inflammation, and endothelial barrier function in AVS and bAVS

Since Kawamoto’s film method greatly improved the histomorphological preservation of aortic valve samples, we next investigated angiogenesis by staining for CD31. This was also prompted by earlier studies suggesting that neovascularization is altered in bAVS compared with AVS [34]. Immunohistochemical analysis demonstrated numerous endothelial cells in bAVS samples (Fig. 2A–E), indicating angiogenesis. In fact, quantitation of CD31+ vessels revealed a significantly higher number in bAVS (n = 4) than in AR controls (n = 5; Fig. 2F). Although an increase was also observed when comparing bAVS with AVS, this difference did not reach statistical significance (p = 0.0621). To further explore the mechanism underlying increased angiogenesis, we examined our bulk RNA-seq data at the gene level, since GO analysis had already indicated increased angiogenesis. As illustrated in the heatmap (Fig. 2G), bAVS samples showed strong upregulation of genes involved in angiogenesis, including angiopoietin 2 (ANGPT2), emilin 2 (EMILIN2), and fibronectin (FN1). In contrast, expression of these angiogenic genes was highly variable across AVS samples, with no clear overall upregulation of angiogenesis (Fig. 2G), consistent with our immunostaining results.

Because angiogenesis is associated with immune cell infiltration and inflammation [47], both of which are thought to contribute to aortic valve stenosis [50, 52], we investigated these processes using immunostaining and bulk RNA-seq. Immune cell infiltration was assessed by staining with the pan-hematopoietic marker CD45. In AR samples (n = 5; Fig. 3A–C), only a moderate number of CD45 + cells were observed. In contrast, both AVS (n = 3) and bAVS samples (n = 4) exhibited a marked increase in CD45+ cells, primarily located near calcified areas (Fig. 3D–K). Quantitative analysis confirmed a significant increase in CD45+ cells in AVS (n = 3) and in bAVS (n = 4) compared to AR controls (n = 5; Fig. 3L). RNA-seq analysis revealed a significant upregulation of inflammatory genes including interleukins, namely interleukin 1 receptor antagonist (IL1RN), interleukin 7 receptor (IL7R), interleukin 6 (IL6), interleukin 11 (IL11), chemokines (C–C chemokine motif ligand 8 (CCL8), C-X-C motif chemokine ligand 5 (CXCL5), C–X–C motif chemokine ligand 8 (CXCL8), C–C chemokine motif receptor 1 (CCR1), C–C chemokine motif receptor 7 (CCR7)) and integrins (integrin subunit alpha 4 (ITGA4), integrin subunit alpha 4 (ITGA11), integrin subunit beta 2 (ITGB2), integrin subunit beta 5 (ITGB5)) along with their respective receptors in calcified bAVS (n = 13) compared to non-calcified AR controls (n = 7; Fig. 3M). Interestingly, earlier studies have also reported increased expression of chemokines, such as C–C chemokine motif ligand 21 (CCL21/CCR7), elevated IL6 levels in the serum of AVS patients, and higher CXCL8 levels in aortic disease [14, 23, 51]. Although these genes were also upregulated in AVS (n = 14) compared to AR controls, the increase was less pronounced. Overall, these findings indicate strong immune cell infiltration and heightened inflammation in bAVS, highlighting a potentially crucial role for these processes in valve calcification, as suggested in the literature [50, 52].

Fig. 3

Assessment of immune cell infiltration in calcified aortic valves using immunostainings and bulk RNA-seq analysis: CD45 staining (red) in non-calcified ARs (A–C), AVS (D–G), and bAVS (H–K); autofluorescence is shown in white. Arrows mark the aortic side of the valves. Orange boxes in overview images mark the locations of the insets shown in G for AVS and in K for bAVS. Scale bar = 1000 µm; scale bar for insets = 100 µm. L Quantitation of CD45+ cells in aortic valves. M Heatmap of chemokines and immune cell genes in AVS and bAVS vs. AR controls

Due to increased immune cell infiltration and inflammation, we also evaluated endothelial barrier function by co-staining for CD31 and albumin. As expected, albumin was visible on the endothelial cell layer in AR (n = 6), indicating an intact barrier that prevents albumin from passing transendothelially (Supplemental Fig. 3A–C). In clear contrast, in calcified AVS (n = 9) and bAVS (n = 13) samples, albumin was detected within the valve tissues beneath the endothelial cell layer, indicating disruption of the endothelial barrier function (Supplemental Fig. 3D–I), as previously suggested for AVS [30].

Extracellular matrix and calcifications in AVS and bAVS

Given the importance of fibrosis and calcification in aortic valve stenosis, we conducted a detailed analysis of these processes. We found that many genes significantly upregulated in bAVS (n = 13) compared to AR controls (n = 7) are involved in calcification and ECM biology. These include genes related to mineralization, bone formation, and calcium metabolism, such as the proteoglycan aggrecan (ACAN), stanniocalcin (STC1), DMP1, hypoxia inducible factor 1 subunit alpha (HIF1A), and also secreted phosphoprotein 1 (SPP1), which encodes osteopontin (Fig. 4A). Although the upregulation of these genes was less pronounced in the AVS group (n = 14), genes associated with calcification, such as IBSP and various collagens (e.g., COL1A1, collagen type IV alpha 2 chain (COL4A2)), were also differentially expressed in AVS vs. AR (Fig. 4A), which aligns with the presence of calcifications in these valves. Notably, as noted above, genes encoding proteins known to play important roles in ECM remodeling and calcification, namely DMP1, MMP12, and PRG4, were significantly more highly expressed in bAVS than in AVS, suggesting a potentially important pathophysiological role in this disease. Osteopontin is a secreted phosphoprotein expressed in osteoblasts and serves as a marker for bone formation and bone-like structures [33]. To further investigate the localization and extent of calcification, we performed osteopontin staining in AVS (n = 6) and bAVS (n = 4) samples. Notably, in both groups, osteopontin expression was primarily observed in areas surrounding acellular calcifications, specifically in the outer layers of the calcium deposits (Fig. 4B–E). We also tested various methods to visualize calcifications in aortic valves, and found that background fluorescence in Kawamoto’s film sections provided the best results (Fig. 4C).

Fig. 4

RNA-seq analysis of calcification markers, and assessment of osteopontin expression in calcified aortic valves and controls using immunostainings: A Heatmap showing ECM-related gene expression in AR, AVS, and bAVS samples; SPP1 encodes osteopontin. B, C Staining for osteopontin (green), CD31 (red), and autofluorescence (white) labeling endothelial cells and calcifications in an AVS sample. The white arrow in C points to the aortic side of the valve, the boxed area labels the section shown at higher magnification in the inset. D, E Stainings as in B, C in a bAVS sample. Scale bar = 500 µm; scale bar for inset = 50 µm

A key cellular component of aortic valves, besides VECs, is VICs. VICs are a heterogeneous group of cells that display characteristics of fibroblasts, myofibroblasts, and smooth muscle-like cells [13]. When analyzing the DEGs and their associated expression patterns, we observed upregulation of many genes involved in cell structure, cell adhesion, and cytoskeletal organization in bAVS (n = 13). The most strongly upregulated genes included several myosins [myosin VA (MYO5A), myosin X (MYO10), myosin IX B (MYO9B), myoferlin (MYOF)], MMP12, matrix metallopeptidase 1 (MMP1), and desmoplakin (DSP), consistent with the prominent calcifications observed in these valves (Fig. 5A). Although these genes were also upregulated in AVS (n = 14), the increase was less pronounced compared to AR (n = 7; Fig. 5A), with only DSP showing significant upregulation in AVS relative to AR.

Fig. 5

RNA-seq analysis of cellular structure markers, and α-SMA and vimentin expression in calcified aortic valves and controls using immunostainings: A Heatmap shows expression of structural proteins and cytoskeletal markers in AVS and bAVS compared to ARs. Staining for vimentin (green) and α-SMA (red) in non-calcified ARs (B-E), AVS (F-M), and bAVS (N-Y), autofluorescence is shown in white, and nuclei are stained with Hoechst. White arrows in the merged images mark the aortic side of the valves. Red boxes with numbers in overview images identify areas selected for insets and their corresponding insets. Scale bar = 500 µm for overview images, 50 µm for insets

Next, we performed immunostaining on Kawamoto-processed aortic valves for vimentin, a common VIC marker, and α-SMA. In non-calcified aortic valves, α-SMA is typically localized to the ventricular side, but earlier studies have reported abnormal localization in AVS, with α-SMA expressed on both sides [26]. Accordingly, we found that in AR (n = 6), α-SMA was mainly confined to the ventricular side (Fig. 5B–E). In contrast, in both AVS (n = 7) and bAVS (n = 7), α-SMA was detected on both sides of the valve (Fig. 5F–Y) and in regions adjacent to calcifications. We also examined the distribution of VICs, focusing on calcifications and their surrounding regions. In non-calcified AR, vimentin was widely distributed throughout the valve, with more prominent expression on the ventricular side. In both AVS and bAVS, vimentin expression was confined to soft tissue, with no vimentin+ cells observed in heavily calcified regions. However, vimentin+ cells were found between areas of early, non-compact calcifications (Fig. 5J–M, R–Y) and also between regions of dense calcified material (Fig. 5H, P). Evidence suggests that in AVS, VECs of the aortic valve undergo EndoMT during calcification [29]. However, our co-staining experiments for α-SMA or vimentin with CD31 (data not shown) showed no evidence of EndoMT.

Correlating CT scan, histomorphological, and clinical data in AVS and bAVS patients

The excellent preservation of Kawamoto-processed AVS and bAVS samples allowed us to compare the histomorphological features of heavily calcified aortic valves in patients both before and after surgical removal. To this end, we correlated clinical CT scans with micro-CT, histological, and immunostaining data from the same-valve leaflets. Micro-CT analysis enabled a direct comparison of the clinical severity of valve stenosis with the degree of leaflet calcification in patients with severe aortic valve stenosis.

Micro-CT data highlighted well-preserved morphological features in both AVS (Supplemental Fig. 4A, B) and bAVS (Supplemental Fig. 4E, F). Notably, the micro-CT images of the AVS (Supplemental Fig. 4A) and bAVS (Supplemental Fig. 4E) closely matched the corresponding clinical CT scans (Supplemental Fig. 4B, F, blue-colored leaflet) acquired before surgery. We further correlated micro-CT findings from explanted valve leaflets with histomorphological data and assessed the extent and distribution of the calcifications [41]. In addition, 3D reconstruction of micro-CT data enabled visualization and quantitation of the calcification volume in AVS (Supplemental Fig. 4C) and bAVS (Supplemental Fig. 4G). The valve leaflets are shown from a top view in the clinical CT scans, with the same leaflets examined by micro-CT and immunohistology highlighted in blue for AVS (Supplemental Fig. 4D) and bAVS (Supplemental Fig. 4H). Calcification in the micro-CT slice of the AVS (Fig. 6A) appears as a bright white area with a faint signal (dashed line) delineating the surrounding soft tissue. Quantitative analysis revealed that calcification occupied 32.6% of the area and 14.4% of the volume within the leaflet (Fig. 6A–D). To correlate micro-CT findings with histological features, the leaflet was cryosectioned along the same axis and further analyzed using von Kossa (Fig. 6B) and Movat-Pentachrome (Fig. 6C) stainings. Von Kossa staining (Fig. 6B) demonstrated extensive calcifications on the aortic side of the leaflet, which was corroborated by Movat-Pentachrome staining. According to the literature [2], the different colors in Movat-Pentachrome staining allow identification of the ventricular and aortic sides of the aortic valve and distinguish between stages of calcification, as mineralized cartilage or bone tissue stains dark blue, whereas newly formed osteoid-like structures appear dark red. In the analyzed AVS, only a single stage of calcification was detected, as the valve leaflet stained uniformly red (Fig. 6C). We also performed immunostainings for vimentin and α-SMA on consecutive sections of the leaflets (Fig. 6D) and found that vimentin was widely expressed throughout the valve, except in areas of dense calcifications. Again, aberrant ɑ-SMA expression was observed on the aortic side of this AVS (Fig. 6D). In contrast, calcification in the bAVS (Fig. 6E–H), as measured by micro-CT (Fig. 6E), accounted for 58.7% of the total valve leaflet area and 42.1% of the volume. Von Kossa staining (Fig. 6F) highlighted extensive calcifications on both sides of the aortic valve, while Movat-Pentachrome (Fig. 6G) staining showed different stages of calcification, a mineralized stage (dark blue) on the ventricular side, and a more recently formed stage (red) on the aortic side. To quantify calcified areas in valve leaflets, they were normalized to the total leaflet area in von Kossa stainings. Our analysis yielded a significantly higher degree (1.7-fold) of calcification in bAVS (46.6 ± 15%, n = 13) compared to AVS (27.9 ± 11,6%, n = 12, p = 0.0025, Fig. 6I). Thus, gene expression, micro-CT, and histological analyses indicate more extensive calcifications in bAVS compared to AVS.

Fig. 6

Micro-CT and histomorphological characterization of calcifications in aortic valves: Micro-CT and histomorphological analysis of a calcified AVS (A–D) and bAVS (E–H). A, E Quantitative assessment of the calcified area percentage in valve slices, tissue borders are marked with dashed lines. B, F Detection of calcifications using von Kossa stainings. C, G Movat-Pentachrome stainings in adjacent sections of the von Kossa stainings. D, H Calcifications visualized in the autofluorescence channel (white). I Quantitation of calcifications in AVS (n = 12) and bAVS (n = 13) samples using von Kossa staining. Scale bar = 1000 µm

Given the significant differences in calcification levels between AVS and bAVS, we conducted a retrospective analysis of 1108 patients with severe AVS or bAVS undergoing TAVI. Both groups showed similar baseline clinical characteristics (Supplemental Table 6) and had comparable mean ages. However, patients with bAVS had slightly lower body mass index (BMI), EuroSCORE II, STS-PROM, and left-ventricular ejection fraction (EF) than those with AVS. Additionally, the prevalence of arterial hypertension was lower in the bAVS cohort (Supplemental Table 6). A detailed analysis of pre-interventional CT scans showed, consistent with earlier studies, that bAVS patients had significantly larger anatomical dimensions across all measured parameters (Fig. 7A). Importantly, the total valvular calcification burden was significantly higher in bAVS patients compared to AVS patients. Notably, the relative valvular calcification (defined as total calcium load/total valve area) was also significantly higher (1.6-fold) in bAVS patients, despite their lower levels of iliac artery calcification, a marker of systemic vascular calcification (Fig. 7B). These findings align with our histological data, which show that bAVS is characterized by markedly increased localized valvular calcification, suggesting distinct pathophysiological mechanisms.

Fig. 7

Retrospective analysis of clinical aortic valve parameters in AVS and bAVS patients before TAVI (n = 1108). A CT-based diameters of the aortic and aortic valve annulus areas in bAVS (left) and AVS (right) patients. B Statistics of CT-based assessment of relative valvular and total vascular (iliac arteries) calcification