Non-pathological Chondrogenic Features of Valve Interstitial Cells in Normal Adult Zebrafish

Abstract

In the heart, unidirectional blood flow depends on proper heart valve function. As, in mammals, regulatory mechanisms of early heart valve and bone development are shown to contribute to adult heart valve pathologies, we used the animal model zebrafish (ZF, Danio rerio) to investigate the microarchitecture and differentiation of cardiac valve interstitial cells in the transition from juvenile (35 days) to end of adult breeding (2.5 years) stages. Of note, light microscopy and immunohistochemistry revealed major differences in ZF heart valve microarchitecture when compared with adult mice. We demonstrate evidence for rather chondrogenic features of valvular interstitial cells by histological staining and immunodetection of SOX-9, aggrecan, and type 2a1 collagen. Collagen depositions are enriched in a thin layer at the atrial aspect of atrioventricular valves and the ventricular aspect of bulboventricular valves, respectively. At the ultrastructural level, the collagen fibrils are lacking obvious periodicity and orientation throughout the entire valve.

Introduction

Valve leaflets are the anatomical elements that bear the stress due to the pressure generated by forward and inhibited backward blood flow in the pumping heart. Cell matrix elements and their interactions assure the proper response to the constantly varying strains under physiological and pathological conditions allowing the valves to withstand changes in hemodynamic flow and pressure.^1–3 In this context, there is increasing evidence that signaling pathways and gene regulatory networks governing normal heart valve development also contribute to valve pathologies.^4,5 Here, cartilage-like transformations are observed in human and murine heart valves under aging and pathological conditions, such as endocarditis.^6–8 The cartilage-like transformations are considered to reflect broad differentiation potential to adapt to changing physical demands.⁴

The heart valves of different species and their locations within the heart show extensive similarities in development. Yet individual remodeling occurs postnatally to ensure optimal function with regard to the role in the specific cardiac cycle.^1–3 The two chambered zebrafish (ZF) heart possesses two valves, with leaflets being directly attached to the myocardium at the atrioventricular (AV) inlet (AV valve) and the bulboventricular (BV) outlet (BV valve) pole of the ventricle.⁹ The AV valve functions as a site of flow resistance that influences diastolic ventricular filling, whereas the BV valve influences the blood pressure in the aortic root.^9,10

Previous studies elucidated the morphogenetic and molecular events of endothelial to mesenchymal transition during ZF endocardial cushion formation and valve development up to 28 dpf.^11,12 Cells from diverse embryonic origins (endocardial/endothelial, epicardial, neural crest) may possibly contribute to the mature valve structures^11,13–15 as indicated by tracking labeled neural crest cells¹⁴ or by in situ hybridization analysis of the expression of extracellular matrix (ECM) components by epicardial and valve interstitial cells (VICs).¹⁵ Three separate developmental phases can be distinguished based on morphological findings. Proliferative expansion of the endocardial cells is followed by elongation of the endocardial cushions into linear luminal projections and finally by remodeling of the projections into plump valvular structures. Resulting adult heart valves are reported to consist of layered collagens, elastin, and glycosaminoglycan-rich connective tissue and, accordingly, share structural similarities with human and murine heart valves.^12,16

However, no study has yet analyzed the final morphological transition from the juvenile (30 to 89 dpf) to the adult (≥90 dpf to 2.5 years) state in ZF heart valves at the light (LM) and electron microscopic (EM) level in detail.^12,17 Therefore, the present study is aimed to characterize the distribution of ZF VICs and the organization of the ECM in these postembryonic ZF stages using immunohistochemistry and electron microscopy.

Materials and Methods

Wild-type AB strain postfertilization ZF were employed to study tissue architecture of AV and BV valve leaflets at the LM and EM level in the following developmental stages: juvenile (35 dpf), sexual mature (4 months), optimal reproductive (1 year), early reproductive senescence (2 year), and reproductive senescence (2.5 year).^18,19 Each staining and morphological assessment was performed at least three times for each age group to ensure that the reported results were timely, consistent, and accurate.

Mating of the parental generation was timed to guarantee assessment of developmental age within 1 hr. Offspring were raised at 28C in egg water for the first 6 days and maintained in system at a 14 hr light, and 10 hr dark cycle beyond day 6. At the time point indicated, fish were either sacrificed for immediate/direct heart dissection with 300 mg/l MS-222 (Tricaine methanesulfonate, Sigma-Aldrich, St Louis, MO) or sacrificed in toto in 300 mg/l MS-222 for 15 min.

The dissected hearts and fish bodies were either formalin or periodate-lysine-paraformaldehyde (PLP) fixed for LM or paraformaldehyde/glutaraldehyde fixed and epoxy resin embedded for EM analysis. All animal protocols were performed according to European guidelines on animal welfare and experiments (directive 2010/63/EU) and approved by the local ethics committee (Landesdirektion Sachsen, Germany, T19/16).

The ZF hearts were compared with murine heart specimens obtained from adult C57Bl/6N wild type mice from an unrelated study (Landesdirektion Sachsen, Germany, T24/16).

Light Microscopy

Sectioning

Complete and undamaged hearts removed from the body cavity and entire fish bodies were immediately fixed in 4% phosphate buffered formalin, then optionally decalcified in ethylenediaminetetraacetic acid (EDTA) and finally embedded in paraffin wax.

Alternatively, dissected hearts were immersed in PLP fixative (2% paraformaldehyde containing 0.075 M lysine and 0.01 M sodium periodate solution, pH 7.4 for 24 hr at 5C^20,21). The embedded hearts were subsequently sectioned to a thickness of 7 µm.

Histology

Sections were stained with hematoxylin-eosin (H&E) for general histological overview. Azan staining according to Heidenhain and Weigert’s resorcin–fuchsin solution was used to identify collagen fibers and elastic fibers, respectively. All these stainings were carried out according to the standard protocols described in details in Romeis Mikroskopische Technik, 19th edition.²²

Russel-Movat’s pentachrome kit (American MasterTech Scientific, Lodi, CA) according to the manufacturer’s instructions served to label ECM components, which constitute bone and cartilage tissue. Alcian blue staining was used to evaluate the proteoglycan/glycosaminoglycan content. For this purpose, sections were incubated in 0.1% Alcian blue 8GX in 1% HCl/70% ethanol for 15 min as previously described by Kaneko et al. for ZF whole mounts and tissue sections.²³

Stained sections were evaluated under an Axioplan 2 upright light microscope (Zeiss, Jena, Germany), photographed with a ProgRes C3 digital camera (Jenoptik, Jena, Germany), and documented with a digital recording system (ProgRes CapturePro 2.8.8; Jenoptik).

Immunohistochemistry

Detection of cellular and subcellular morphological features was performed with primary antibodies previously validated for use in ZF¹⁹ as listed in Table 1.

Table 1.

Primary Antibodies Used in Immunohistochemistry.

Antigen	Reactive Species	Host Species	Clonality	Commercial Source	Clone Name/Catalog Number	Antigen Retrieval	Dilution	Cross-reactive Zebrafish Reference	Positive Control Tissue
Aggrecan	Human	Mouse	Mono	Thermo Fischer	BC-3/MA3-16888	Microwave, SSC buffer (pH 6)	1:000	Govindan and Iovine24	Growth plates
Collagen type 2	Chicken	Mouse	Mono	DSHB	II-II6B3	Microwave, SSC buffer (pH 6)	1:100	Bensimon-Brito et al.25	Cartilaginous structures
Cytokeratin	Human	Mouse	Mono	BioGenex	AE1 and AE3/MU071-UC	Microwave, SSC buffer (pH 6)	1:100	Paquette et al.22	Intestinal epithelium
PCNA	Human	Rabbit	Poly	Abcam	ab15497	Microwave, SSC buffer (pH 6)	1:400	NA	Intestinal crypt cells
SOX-9	Zebrafish	Rabbit	Poly	GeneTex	GTX128370	Microwave, SSC buffer (pH 6)	1:100	Paul et al.26	Growth plates
Vimentin	Pig	Mouse	Mono	Milipore/Chemicon	V9/MAB3400	Microwave, SSC buffer (pH 6)	1:100	Cerdà et al.23	Dermis

Abbreviations: PCNA, proliferating cell nuclear antigen; SSC, saline-sodium citrate; DSHB, Developmental Studies Hybridoma Bank.

Antibodies for intermediate filaments were used to identify epithelial (cytokeratin, anticytokeratin cocktail [AE1 and AE3], BioGenex, Fermont, CA²⁶) and mesenchymal (vimentin, mouse antivimentin V9 monoclonal antibody, Millipore²⁴) cell types. Antibodies for SOX-9 (rabbit anti-SOX-9a polyclonal antibody, GeneTex, Irvine, CA²⁵), aggrecan (mouse antiaggrecan monoclonal antibody [BC-3], Thermo Fischer, Pierce, Rockford²⁷), and the alpha-1 chain of type 2 collagen (COL2A1, mouse anti-COL2A1 [II-II6B3]²⁸ hybridoma culture supernatant, Developmental Studies Hybridoma Bank, Iowa City) served to identify chondrogenic differentiation. An antibody for proliferating cell nuclear antigen (PCNA, rabbit anti-PCNA antibody, abcam, Cambridge Science Park, Cambridge, England) was used to assess VIC proliferation.

Immunohistochemistry was performed on deparaffinized rehydrated sections of formalin-fixed or, in the case of aggrecan staining, of PLP-fixed tissue. Before staining, sections were heated in sodium citrate buffer (pH 6.0, 20 min) for antigen retrieval, immersed in 30% H₂O₂ in 100% methanol (10 min, room temperature) to quench endogenous peroxidase activity, and covered with 5% normal serum from the same host as the biotinylated secondary antibody (30 min, room temperature) to reduce background from nonspecific interactions between the secondary antibody and the section surface. Primary antibodies were applied at optimized concentration previously determined on ZF control tissues (see Table 1 for details), 1:100 and 1:400 (PCNA), respectively, in antibody buffer (phosphate buffered saline [PBS], pH 7.4 containing 0.5% bovine serum albumin, and 0.3% Triton X-100) and were applied overnight at 4C in a humidified chamber. After rinsing with PBS containing 0.3% triton, primary antibody binding was detected with a species-matched biotin-conjugated secondary antibody (1 hr, room temperature). After further rinsing with PBS, antibody labeling was visualized by incubation (30 min, room temperature) with avidin-biotin-peroxidase complex (Vectastain, Burlingame, CA), followed by incubation with 3,3′-diaminobenzidine (DAB) and urea hydrogen peroxide (SigmaFast DAB tablets, Sigma-Aldrich, St Louis, MO) for 30 to 60 sec. The staining reaction was stopped by rinsing the sections with 0.05 M Tris HCl buffer, pH 7.4. Finally, the sections were counterstained with Mayer’s hemalum, dehydrated and embedded with Roti-Histokitt (Carl Roth, Karlsruhe, Germany). Immunostained sections were evaluated, photographed, and documented as described above for histological sections.

Omission of the primary antibody served as a control of nonspecific binding of the secondary antibody.

Electron Microscopy

Freshly prepared ZF hearts were immediately fixed overnight in a mixture of 4% paraformaldehyde and 1% glutaraldehyde. The hearts were then washed using ice-cold PBS and postfixed with 0.5% osmium tetroxide in PBS for 30 min. After rinsing in PBS, the hearts were dehydrated in graded ethanol (30%, 50%, 70%, 10 min each). The hearts were stained with uranyl acetate in 70% alcohol for 1 hr before passing them further in graded ethanol (80%, 90%, 95%, 2× 100%, 10 min per step) at room temperature. The hearts were then immersed twice in pure propylene oxide for 5 min and in a mixture of Durcupan (Sigma-Aldrich, Steinheim, Germany) and propylene (1:1). After evaporation of propylene oxide, the remaining Durcupan was replaced by fresh Durcupan followed by polymerization for 48 hr at 56C.

The resulting Durcupan blocks were then trimmed and semithin survey sections (500 nm) were prepared using an utramicrotome (Leica Microsystems, Wetzlar, Germany) equipped with a diamond knife and azure II-methylen blue stained as described by Richardson et al.²⁹ at 70C. Once the heart valves had been identified at the level of light microscopy, the block faces were further trimmed to the final region of interest with a razor blade. From this region, ultrathin sections (50 nm) were cut from the surfaces of the blocks, mounted on formvar-coated grids, and finally poststained with lead citrate (7 min, room temperature). Ultrastructural analysis was performed using a Zeiss SIGMA electron microscope equipped with a scanning transmission electron microscope (STEM) detector (Zeiss NTS, Oberkochen, Germany).

Results

Differences in the Microarchitecture of Heart Valve Leaflets Between Adult ZF and Mice

To characterize the tissue composition and microarchitecture of ZF heart valves, heart valve leaflets of adult ZF were cross-sectioned and compared with leaflets of adult mice. Strikingly, characteristic differences were observed in standard H&E stained sections. Although mouse AV valves were characterized at their base and mid-region by a uniform stroma of fibrous connective tissue with few spindle-shaped fibrocyte-like VICs in between, a central cell layer of densely packed and rather plump cells with oval nuclei dominated the base and mid-region of AV leaflets in ZF (Fig. 1).The cells were non-uniformly arranged, exhibited abundant cytoplasm, and were of polygonal shape with an eccentric nucleus. Although the mouse leaflets revealed in their base and mid-region a rich content of collagen bundles as indicated by yellow-green Russel-Movat’s pentachrome and blue Azan staining, AV leaflets of ZF rather showed a cartilage-like proteoglycan/glycosaminoglycan-rich stroma as indicated by turquoise-green Russel-Movat’s pentachrome and strong Alcian blue staining. Of note, Azan and Weigert’s resorcin-fuchsin staining revealed a prominent enrichment of collagen and elastin, respectively, in a thin layer beneath the valvular endothelium at the atrial (inflow) aspect of AV leaflets and, to a lesser extent, around the more central polygonal-shaped cells. Central cellularity and proteoglycan/glycosaminoglycan-rich tissue were also characteristics of the base and mid-region in adult BV ZF valves leaflets (Fig. S1). In comparison, semilunar valve leaflets in mice were largely governed by collagen-rich stroma in the base and mid-region (not shown).

Figure 1.

Atrioventricular (AV) valves of young adult (1-year old) zebrafish (ZF, left) and mice (right). AV valve leaflets are outlined by arrow heads. Large polygonal-shaped cells (triangles) are the main structural constituent of the base and mid-region of AV valves in ZF. In contrast, collagen-rich connective tissue without layering represents the basic structural unit in the basal aspects of AV valves in mice. In ZF, collagen and elastic material (arrows) is enriched in a thin layer on the atrial (inflow) aspect of the AV valve. An ubiquitous strong Alcian blue staining demonstrates presence of a proteoglycan/glycosaminoglycan-rich matrix. A trabecular band anchors to the ZF valve tip and extend down in the ventricle (star). Insets indicate the regions magnified on the further right photographs. Russel-Movat’s pentachrome stain (Movat), Azan stain, Weigert’s resorcin-fuchsin stain, and Alcian blue stain. Scale bar equals 25 µm in each photograph. Abbreviations: H&E, hematoxylin-eosin; a, atrium; v, ventricle.

The tip of adult AV (Fig. S2) and BV ZF leaflets differed in structure from the base and mid-region. It was composed of densely packed small cells. Turquoise-green Movat’s pentachrom and strong Alcian blue staining indicated a polysaccharide- and proteoglycan/glycosaminoglycan-rich matrix. Immunoreactivity for matrix proteins was almost absent in this region (see immunohistochemical stain results below). The tips of adult mouse leaflets frequently showed myxomatous changes. The myxomatous lesions were characterized by high cellularity and a proteoglycan/glycosaminoglycan-rich and collagen poor-matrix (Fig. S2).

Valvular Interstitial Cells With Cartilage-like Differentiation in Adult ZF Hearts

To determine whether the VICs in the base and mid-region of ZF valve leaflets were of mesenchymal or epithelial identity, we applied immunohistochemistry for cytokeratin and vimentin (Fig. 2). Importantly, the ZF VICs co-expressed both intermediate filaments, which was indicative for epithelial and mesenchymal properties. As the histology of base and mid-region in ZF valve leaflets showed features suggestive for cartilaginous differentiation, we attempted to further substantiate this suggestion by detection of cartilage-relevant differentiation markers such as SOX-9, aggrecan, and type 2a1 collagen (Fig. 2). SOX-9 staining displayed nuclear localization. Almost all of the detected type 2a1 collagen was intracellular, a finding attributable to newly formed pro-α(II)chains and type II procollagens and compatible with an early stage of chondroblast formation.^30,31 For aggrecan staining, the distribution was occasionally thickened into clumps, possibly reflecting larger aggrecan aggregates.

Figure 2.

Immunohistochemical characterization of the tissue composing the base and mid-region of atrioventricular (AV) valves in young adult (1-year old) zebrafish (ZF). Heart valve leaflets are outlined by arrow heads. Immunohistochemistry reveals co-expression of epithelial (cytokeratin) and mesenchymal (vimentin) markers, SOX-9 staining in the nuclei (circles), and type 2a1 collagen and aggrecan-based matrix formation by the central layer of large polygonal-shaped cells. In contrast, the endocardial cells of the valves show vimentin expression, whereas cytokeratin is lacking (arrows). Type 2a1 collagen shows more intracellular than extracellular staining. Aggrecan antibody staining is occasionally thickened into irregular interstitial clumps. Of note, immunoreactivity in particular for type 2a1 collagen and aggrecan is almost absent in the apical tip region of the valves (asterisk). Scale bar equals 25 µm in each photograph. Abbreviations: a, atrium; v, ventricle; COL2A1, type 2a1 collagen; No pmAb, no primary monoclonal antibody controls; No ppAB, no primary polyclonal antibody controls.

Structural Changes in ZF Valves During the Transition From the Juvenile to the Adult Age

ZF heart valves at the juvenile stage (35 dpf) showed high cellularity with small oval-shaped cytoplasm poor cells (Fig. 3). In young adult ZF heart valves (4 months old), a morphological shift from small oval-shaped to larger polygonal-shaped cells was visible. In older ZF, large polygonal-shaped cells dominated the basal aspect of the valves and were distinctive for their histomorphological appearance. Cells demonstrating PCNA positivity populated the subendocardial area (Fig. 3) in juvenile ZF heart valves. A comparable PCNA staining pattern to that of juvenile age was not seen in older valves. Cytokeratin detection at the juvenile age was rather discrete but indicated that the cells in the central cell layer retained part of their epithelial identity from previous stages (Fig. 3). Cartilage-relevant differentiation markers such as SOX-9 and type 2a1 collagen were first observed with the transition from small oval-shaped to large polygonal-shaped cells between 35 dpf and approximately 4 months old (Fig. 3) and were clearly present in 1-year old AV (Fig. 2) and BV (Fig. S3) valves, respectively.

Figure 3.

Immunohistochemical characterization of ZF atrioventricular (AV) valves (outlined by arrow heads) at different ages. The atrium is indicated by an “a.” At the juvenile age (35 dpf), AV valves are populated by small oval-shaped cells. At the young adult age (4 months old) a mixture of larger polygonal-shaped cells is visible. In even older AV valves (>1-year old), large polygonal-shaped cells dominate the basal and mid-region aspect of AV valves. PCNA positivity (arrows) is predominantly observed at the juvenile age along the subendothelial aspect of the AV valves. Discrete cytokeratin staining in the juvenile AV valves indicates that the VICs forming the central cells of adult valves may have retained epithelial characteristics from previous stages. SOX-9 and type 2a1 collagen staining is associated with the transition of the cells to the large polygonal shape. Squares indicate the regions magnified on the further down photographs. Scale bar equals 25 µm in each photograph. Abbreviations: ZF, zebrafish; PCNA, proliferating cell nuclear antigen; VIC, valve interstitial cell; H&E, hematoxylin-eosin; COL2A1, type 2a1 collagen.

Electron Microscopy of Adult ZF Heart Valve Leaflets

At an ultrastructural level, the cell-rich parenchyma of individual ZF heart valve leaflets in the base and mid-region showed loosely arranged organelle-poor cells, which were occasionally connected to each other with adherens-like junctions (Fig. 4A). The ultrastructure of the matrix layers covering the cell-rich parenchyma differed dramatically at the atrial and ventricular aspect of AV valves and the ventricular and bulbar aspect of BV valves, respectively (Fig. 4B). Although the majority of the ECM appeared to be rather amorphous with only few fibrillary structures, the atrial and ventricular aspect of AV and BV valves, respectively, was rich in disorganized fibrils with no obvious periodicity and a mean diameter of 14 nm. Thus, the histologically observed collagen layer and the electron microscopically observed fibrils composing the layer were unlikely to represent mature collagen fibrils. In contrast, the subendothelial layer of ECM at the ventricular and bulbar aspect of AV and BV valves, respectively, appeared to be rather thin and contained only small amorphous deposits, but with no fibrillar material. Importantly, we neither observed differences with regard to the VICs nor with regard to the organization of the ECM when comparing basal aspects of AV and BV valves at the EM level. Taken together, electron microscopy was in accordance with a poorly differentiated cartilaginous tissue as the main structural element in the base and mid-region of juvenile and adult ZF heart valves. The above immunohistochemical and electron microscopical findings collectively indicated the valvular skeleton of ZF heart valves to possess chondrogenic properties.

Figure 4.

Transmission electron microscopic analysis of the basal aspects of the atrioventricular (AV) (A) and bulboventricular (BV) (B) valves in an adult (1-year old) zebrafish. The atrium is indicated by an “a,” the ventricle by a “v,” and the bulbus by a “b.” The large polygonal-shaped cells of the central cell layer are loosely arranged and only occasionally linked by cell-cell contacts (square). The cells show a central nucleus and an organelle-poor cytoplasm with few mitochondria. The extracellular matrix (ECM) around the circumference of the central cell layer is structured differently at the atrial and ventricular aspects and ventricular and bulbar aspects of AV and BV valves, respectively. At the atrial and ventricular aspect of AV and BV valves, respectively, the ECM is rich in fibril-like structures (triangle) in agreement with the strong histological staining for type 2a1 collagen. At the ventricular and bulbar aspect of AV and BV valves, respectively, it is relatively homogeneous and essentially free from visible fibril-like structures (asterisk). Scale bar equals 1 µm in each photograph. Abbreviation: E, endocardial lining.

Discussion

Our study provides an analysis of the microarchitecture of the basal aspect of heart valve leaflets in adult ZF. It extends a previous study by Martin and Bartman¹² where the period from the end of endothelial cell formation (at approximately 6 dpf) through the development of luminal projections to the rearrangement into cusp-like valves (at approximately 28 dpf) was studied.

Unlike the human, adult ZF valve leaflets are not characterized by up to five molecularly distinct matrix layers, but rather by a base mid-region zone and a tip region zone along the proximal-distal axis. The tip region shows high cellularity of small-sized cells that characterize reduced immunoreactivity for the tested antigens.

The base and mid-region of adult ZF heart valve leaflets are composed of a surrounding endocardium underlined by a basement membrane, a thin peripheral layer of connective tissue, and a prominent central cell layer of large polygonal-shaped cells abutting one another. The striking finding of a central cell layer of large polygonal-shaped cells as the main histological constituent in the base and mid-region of ZF heart valve leaflets is in sharp contrast to highly organized fibrous tissue representing the main scaffolding in heart valve leaflets of mammals, exhibiting up to five distinct matrix layers in man, but not in mice.^2,8,32,33 The latter is indeed of particular interest since mouse heart valves have been historically viewed as being structured like human heart valves.³³

A stratification, though far less conspicuous than in human, is also present in the base and mid-region of ZF valve leaflets, if tissue architecture of the peripheral mesenchymal thin layer ensheathing the prominent central cell layer of large polygonal-shaped cells is taken into consideration. A collagen-like matrix prevails in the subendocardial layer at the atrial and ventricular aspect of AV and BV, respectively, whereas an amorphous material dominates the opposite aspect of the valves. It is interesting to note that in human AV and semilunar heart valve leaflets, the layer with the most collagen fibrils is the broad subendocardial fibrosa layer at the ventricular and pulmonary/aortal aspect, respectively, whereas the subendocardial atrial (AV)/ventriclular (BV) layer at the opposite aspect is dominated by elastic fibrils.⁸ The small spongiosa layer in between these two matrix layers is rich in glycosaminoglycans and proteoglycans, which gives it a loosely spongy appearance as judged by routine histological staining. Therefore, depending on these different microarchitectures, alternate biomechanical properties can be expected.

ZF AV and BV valves show similar LM and EM profiles in their base and mid-region, although the precise microstructural distribution and orientation of the ECM fibrils warrant further investigation. The similarity in structural organization of AV and BV in ZF is in sharp contrast to structural differences reported between AV and semilunar valves, and between tricuspidal and biscuspidal valves in mammals.^2,3 The most obvious gross anatomical difference between ZF and mammalian AV valves is the lack of a subvalvular tension apparatus on the ventricular aspect.^9,10 Although papillary bands or sheets anchor to the AV valves, and extend down in the ventricle in ZF, these structures have been reported to constitute only a primitive form of papillary system, which resist only some of the stress generated by ventricular contraction.^10,34 The lack of a similar system of papillary muscles and chordae tendineae to that observed in mammals may cause the biomechanical properties and functions of AV valves in ZF to be closer to those of BV valves in ZF.^34,35 Granted that the primitive form of the papillary system is by itself ineffective to prevent valve flaps from entering into atria and ventricles, a cartilage-like tissue, additionally rich in cytokeratin, could confer stiffness, and thus take over this function.³⁶ On the contrary, a less proteoglycan/glycosaminoglycan-rich matrix with solely vimentin in mice could meet the need of more flexibility due to leaflet size and displacement distance during cardiac motion.

Our finding that large polygonal-shaped cells with characteristics of luminal cells (cytokeratin expression and adherens-like junction) build the supporting structure in the base and mid-region of cardiac valve leaflets of ZF is in line with findings by Martin and Bartman that cardiac valve leaflets of ZF harbor cells with an epithelial identity.¹² However, the preserved cytokeratin expression seems to be inconsistent with ZF VICs finally converting to interstitial intermediate or mesenchymal cells between 12 and 16 dpf, which would favor vimentin expression instead.¹² This is mainly because in mammals, cytokeratin filaments, in contrast to vimentin filaments, are cytoskeleton constituents not normally expected to occur in mature VICs or cartilage cells.³⁷ However, in trout and ZF, cytokeratin filaments are present in certain gill and skull cartilages, respectively.^38,39 In addition, cytokeratin positive staining is detected in ZF notochord cells, which otherwise express cartilage and mesenchyme genes.⁴⁰ Furthermore, cytokeratin is even detected in certain stages of cartilage formation during mammalian embryogenesis, although this appears to be a transient phenomenon.³⁸

The large polygonal-shaped cells in the ZF valves are reminiscent of the first steps toward cartilage formation. In regard to this hypothesis, cartilage-specific staining and marker expression are observed in the cell parenchyma. The cartilage-specific staining and marker expression are in line with skeletal and heart valve development being influenced by overlapping regulatory pathways and biomechanical stimuli in ZF and in other vertebrate species such as chicken, mouse, and human.^41–43

Furthermore, the spongiosa layer of mature heart valves expresses skeletal elements-related proteins such as SOX-9, aggrecan, and type 2a1 collagen, indicating a skeletal elements-like feature of the VICs in this mature valve region.^44,45 SOX-9 has a permissive role in cartilage formation and a repressive role in bone formation.³⁶ It is highly expressed in immature chondrocytes when they establish a cartilaginous matrix phenotype and is downregulated in mature chondrocytes and osteoblasts. Chondrocytes are reported to establish a cell-adhesion surface as they divide, which may explain the adherens-like junctions between the large polygonal-shaped cells in the ZF valves.⁴⁶

The population of VICs with skeletal element potential is neural crest in origin, as studies in chick and quail suggest.⁴⁷ The presence of skeletal tissue in cardiovascular structures is a rare event in other teleost such as Sardina pilchardus, Trachurus trachurus, Coris julis, and Arnoglossus thori.⁴⁸ In these species, cartilaginous foci are occasionally found in the wall of the bulbus arteriosus in close association with the BV valves of the aortic outflow tract.⁴⁸ Cartilaginous and bone tissue can also develop in the cardiac skeleton of several amphibians, reptilian, avian, and mammalian species, including mouse and man.⁸ Cartilaginous and bone tissue appear even more common in the hearts of these species than in the hearts of teleosts and salmoni.⁴⁹ In particular, avian and hamster focal formation of cartilage in cardiac skeleton and aortic/pulmonary valves is a frequent event.^50,51 Because the cartilage appears early in development, the change is assumed to be associated with biomechanical stimulation and not necessarily with aging and disease.^52,53 In contrast, occurrence of cartilage and/or bone in mature murine and human heart valves is considered to be an aging and/or pathological change.^6–8,54 Emerging studies suggest a redeployment of developmental pathways in the context of cartilage and/or bone appearance and suggest that action could be taken to prevent the differentiation of quiescent VICs into chondroblastic and/or osteoblastic VICs to retain lifelong normal valve function.^4,5,45,55,56

Loss of canonical Wnt/β-catenin signaling in the chicken increases SOX-9 nuclear localization and permits chondrogenic differentiation of VICs during heart valve development in the adult chicken⁵⁷ In ZF, however, constitutive activation of the Wnt/β-catenin pathway leads to excessive SOX-9-dependent endocardial cushion-formation.^58,59 Thus, species-specific measures may be necessary to prevent, correct, and compensate for disadvantages linked to redeployment of developmental processes, in different vertebrates.

In conclusion, early heart valve development in ZF shows extensive similarities with heart valve development in other vertebrates. However, our study shows that ECM compartmentalization and cellular composition differ in postembryonic stages of ZF heart valves. Base and mid-region VICs are proposed to gain chondrogenic properties.