Differential cell-matrix responses in hypoxia-stimulated aortic versus mitral valves

Abstract

Tissue oxygenation often plays a significant role in disease and is an essential design consideration for tissue engineering. Here, oxygen diffusion profiles of porcine aortic and mitral valve leaflets were determined using an oxygen diffusion chamber in conjunction with computational models. Results from these studies revealed the differences between aortic and mitral valve leaflet diffusion profiles and suggested that diffusion alone was insufficient for normal oxygen delivery in mitral valves. During fibrotic valve disease, leaflet thickening due to abnormal extracellular matrix is likely to reduce regional oxygen availability. To assess the impact of low oxygen levels on valve behaviour, whole leaflet organ cultures were created to induce leaflet hypoxia. These studies revealed a loss of layer stratification and elevated levels of hypoxia inducible factor 1-alpha in both aortic and mitral valve hypoxic groups. Mitral valves also exhibited altered expression of angiogenic factors in response to low oxygen environments when compared with normoxic groups. Hypoxia affected aortic and mitral valves differently, and mitral valves appeared to show a stenotic, rheumatic phenotype accompanied by significant cell death. These results indicate that hypoxia could be a factor in mid to late valve disease progression, especially with the reduction in chondromodulin-1 expression shown by hypoxic mitral valves.

1. Introduction

Aortic and mitral valve leaflets are thin connective tissues that function to ensure unidirectional blood flow through the left side of the heart and the aorta during the cardiac cycle. Although small in size, valve leaflets have a complex, multi-layered extracellular matrix (ECM) structure that provides strength, flexibility and resistance to compression [1]. Valvular interstitial cells (VICs) regulate ECM turnover and are the major cell type in both valves [2]. Believed to be normally quiescent, VICs can transition to an activated state to enhance ECM synthesis and degradation [3,4]. Pathological dysregulation of ECM turnover is a major contributor to leaflet dysfunction and disease in both aortic and mitral valves [5,6].

The main disease of the aortic valve, calcific aortic valve disease (CAVD), is characterized by thickening, stiffening and calcification of the leaflets, eventually leading to stenosis [7]. The transition of aortic VICs (AVICs) to activated or osteoblast-like cells is one of the major factors in CAVD, and AVICs are capable of producing calcific nodules in culture, although their precise role in CAVD remains unclear [8]. Mitral valves experience a more diverse array of ECM remodelling diseases including myxomatous mitral valve degeneration (MMVD) and rheumatic mitral stenosis (MS). MMVD involves glycosaminoglycan and proteoglycan accumulation resulting in leaflet weakening and valve prolapse, whereas MS is associated with fibrotic remodelling, collagen accumulation and leaflet stiffening [9,10]. Angiogenic remodelling is a hallmark of both CAVD and MS, but it is not observed during MMVD [10–14]. During CAVD and MS, both valves also undergo significant fibrotic remodelling with a breakdown of the layered structure [6,15]. Despite their mutual ability to calcify, it has been proposed that aortic and mitral valves exhibit different osteogenic potentials in culture [16]. Interestingly, pathways involved in CAVD in aortic valves may lead to MMVD instead of MS in mitral valves [17]. These studies suggest that a complex array of factors, many of which remain to be examined, are influencing the remodelling and disease progression of aortic and mitral valves.

Hypoxia is a well-known promotor of angiogenesis in tumours [18,19] and has been hypothesized to be a regulator of valve angiogenesis during disease [20]. Studies have shown that hypoxia inducible factor 1-alpha (HIF-1α) is present in calcific nodules in CAVD [21,22]. Valve architecture, which greatly affects oxygen diffusion and transport, is different between the aortic and mitral valves. Adult mitral valve anterior leaflets are approximately 30% thicker than adult aortic valve leaflets and contain a significant vascular bed in the proximal half of the leaflet [23], whereas aortic leaflets are largely avascular with most vasculature located near the annulus [24]. Without a robust vascular network, aortic leaflets are likely to experience difficulty when responding to hypoxic stresses. To better study the impact of hypoxia on valve leaflets, this study uses a three-dimensional cell culture system and presents the creation of a static culture and pressurized oxygen diffusion system. These systems were used to apply hypoxia directly to leaflets or VICs to quantify changes in angiogenic and calcific factors.

2. Material and methods

2.1. Pressurized oxygen diffusion chamber

A pressurized oxygen diffusion system was created to measure the time to steady state following step changes in oxygen levels in the phosphate buffered saline (PBS) solution in a chamber below the valve leaflet. The system was assembled with a 2 l reservoir (Nalgene, Rochester, NY, USA), flexible tubing and a custom-fabricated polystyrene chamber. Vinyl-coated polyester fabric (McMaster Carr, Elmhurst, IL, USA) was used to secure the fresh valve leaflet between an oxygen-sensing Clark-type electrode (Harvard BioScience, Holliston, MA, USA) (top chamber) and PBS solution with adjustable oxygen levels (bottom chamber). The reservoir allowed large volumes of PBS to be set to a specific oxygen concentration and then rapidly transported to the bottom chamber by an aquarium pump. Measurements of diffusion through the belly region of the valve leaflet were taken at two pressure levels, one simulating a low-pressure leaflet state (4.5 kPa) and the other a highly stressed state (10.5 kPa). Valve thickness was determined by taking a cross section of the recording area of the valve leaflet and imaging that section under a dissecting microscope. These non-fixed valves were hydrated using PBS to maintain thickness. The oxygen diffusion coefficient was calculated using previously described methods [25]. From the diffusion coefficient equation, , partial pressure was given by leaflet thickness (x) squared, multiplied by the rate constant (k) derived from the time transient and divided by pi (π) squared. The diffusion coefficients for the belly region of aortic and mitral valve leaflets were calculated using the time transient determined by the Clark electrode measurements, individual relaxed thickness measurements and the stress–strain relationship measured by compression testing. Using a compression tester, aortic and mitral valve leaflets were subjected to unconfined compression from 0 to 10.5 kPa using a Bose Electroforce ELF 3200 mechanical testing machine equipped with a 1000 g load cell (TA Instruments, Eden Prairie, MN, USA) to establish a stress–strain relationship.

2.2. Oxygen diffusion computational models

Computational models based on Fick's second law of diffusion, , were created using the partial differential equation solver, NDSolve, from Mathematica (Wolfram, Champaign, IL, USA). NDSolve uses the finite-element method for spatial discretization and the method of lines to numerically solve partial differential equations of this type. The solver requires a discrete representation of a region, provided as a mesh, along with a set of boundary conditions. Here, NDSolve was used to evaluate Fick's second law of diffusion over two-dimensional radial sections of aortic and mitral valve leaflets. Aortic valve thickness data were obtained from previous studies on valves with no applied pressure [24,26–29] and valves under 10.5 kPa pressure [30]. Mitral valve anterior leaflet thickness data were obtained from four radially sectioned leaflets under no applied pressure and from a previous study of mitral valves mid-systole [31]. All measurements were combined to create a range-of-motion model for aortic and mitral valve leaflets. Using these measurements, two-dimensional meshes of valve radial sections were created in Mathematica. The resulting oxygen partial pressure (P, mmHg) was calculated over the area of a leaflet radial section and was displayed as a heatmap. The diffusion coefficient of oxygen (D, cm² s⁻¹) was measured experimentally. The tissue oxygen consumption rate (VO₂, ml O₂ per ml tissue per second) and the solubility (S, ml O₂ per ml tissue per mmHg) were obtained from previous studies on porcine valves [27,29]. Time (t, seconds) was set to ensure that the simulation would reach steady state. Boundary conditions for NDSolve were set to ensure that all leaflet edges experienced maximum oxygenation (based on the oxygen level in the media immediately surrounding the leaflet). Additionally, these boundaries acted as an infinite oxygen source in the model. The initial conditions set the entire leaflet to full oxygenation. Diffusion in the leaflet circumferential direction was not considered. Oxygen delivery from blood vessels inside the valve was also not considered. Oxygen consumption by leaflets was modelled as a zero-order reaction and was considered to be uniform throughout the leaflet, as suggested by the leaflet's uniform VIC density [32].

2.3. Valvular interstitial cells

Aortic and mitral valve leaflets were dissected from six-month-old pig hearts obtained from a local commercial abattoir (Animal Technologies, Tyler, TX, USA). VICs were isolated from leaflets using collagenase digests following previously described methods [33] and cultured in Dulbecco's modified Eagle medium (DMEM, 1 g l⁻¹ glucose) containing Ham's F12 (Hyclone, Logan, UT, USA), 10% bovine growth serum (Hyclone) and 1% antibiotic/antimycotic (Mediatech, Manassas, VA, USA) for two passages before use. The medium was changed every 2 days.

2.4. Whole leaflet cultures

Whole aortic and mitral anterior valve leaflets were dissected from six-month-old pig hearts obtained from a local commercial abattoir (Animal Technologies) and secured in six-well plates for static leaflet cultures. Before tissue harvest, six-well plates were coated with a 2 mm layer of polydimethylsiloxane (PDMS) created from a Sylgard 184 silicone elastomer kit (Dow Corning, Midland, MI, USA) to discourage cell attachment. The plates were heated for 12 h at 80°C to cure the PDMS coating. A 12 mm stainless steel insect pin was inserted through the centre of each valve leaflet and the pin tip was secured in the PDMS layer. Each well was filled with 8 ml of cell-culture medium to ensure that the leaflets were completely submerged. Leaflets were cultured for two weeks with a medium change every 4 days.

2.5. Hypoxic incubator

A gas cylinder (Matheson, Basking Ridge, NJ, USA) and a 2.4 l glass desiccator (Corning, Corning, NY, USA) were combined to create a hypoxic chamber modified from a previously described design [34]. A gas cylinder with custom gas mixes (13% oxygen or 5% oxygen, each with 5% carbon dioxide and fill nitrogen) provided controlled oxygen levels to a sealed glass desiccator that housed the leaflet or culture sample. The desiccator also had a water reservoir to preserve humidity. The desiccator was housed in an incubator to ensure that the temperature inside the chamber was held at 37°C. Custom gas mixtures were moved through the chamber continuously throughout the duration of each culture period. Oxygen levels inside the chamber were verified with a Fyrite oxygen measurement system (Bacharach, New Kensington, PA, USA).

2.6. Three-dimensional filter paper cultures

The design and culture of VICs in filter paper scaffolds has been described previously [35]. A wax printer was used to create a 48-well plate template with 6 mm diameter wells on a 200 µm thick sheet of Whatman grade 114 filter paper (GE Healthcare, Chicago, IL, USA). The wax template was melted through the paper at 150°C to create cell impermeable barriers between wells. The paper scaffolds were submerged in water and sterilized in an autoclave. VICs were mixed with 2 mg ml⁻¹ rat tail type I collagen solution (Corning) and seeded at 50 million cells ml⁻¹ into each well of the filter paper scaffolds. Gels in filter paper were cultured for two weeks in 150 mm tissue culture dishes with media changes every 4 days. The seeding density of 50 million cells ml⁻¹ (maximum seeding density for filter paper scaffolds) was selected after VIC density in fresh adult porcine aortic leaflets was determined to be approximately 60 million cells ml⁻¹. This density was quantified by sectioning a fresh aortic leaflet in both the radial and circumferential directions. All cells within a slice were counted and the total was adjusted according to tissue shrinkage. Using stereology, the three-dimensional leaflet cell density was estimated assuming a uniform cell density throughout [32].

2.7. Histology, immunohistochemistry and immunocytochemistry

Following culture and fixation, aortic and mitral valve leaflets were cut into 5 µm paraffin sections for histological staining. Movat modified pentachrome stain, which colours elastic fibres black, collagen fibres yellow, proteoglycans blue, muscle red and cell nuclei purple [1], was used to quantify changes in leaflet ECM. Streptavidin/biotin colorimetry and diaminobenzidine (DAB) detection were used to quantify hypoxic and angiogenic marker expression and were performed as previously described [36]. Antigen retrieval was performed using heat-mediated citrate buffer. Antibodies were directed against HIF-1α (Abcam, 1 : 50), vascular endothelial growth factor receptor 2 (VEGFR2) (Abcam, 1 : 50), runt-related transcription factor 2 (RUNX2) (Abcam, 1 : 50), NOTCH1 (Abcam, 1 : 100), chondromodulin-1 (Chm1) (Lifespan Biosciences, 1 : 50) and smooth muscle alpha actin (SmαA) (Abcam, 1 : 50) to determine protein expression and localization. Apoptosis in leaflet cultures was examined using the TACS^® 2 TdT-DAB In Situ Apoptosis Detection Kit (Trevigen, Gaithersburg, MD, USA) according to the manufacturer's specifications. Briefly, rehydrated paraffin sections of aortic and mitral valve leaflets were incubated in proteinase K solution, then quenched and immersed in 1 × TdT Labelling buffer and reaction mix for one hour in a humidified chamber. Tissues were then immersed in TdT stop buffer and incubated with an anti-BrdU antibody for 30 min at 37°C. After streptavidin-peroxidase treatment, colour development was observed with the DAB substrate. Methyl Green was used as a counterstain. For staining of filter paper scaffolds, the cells were labelled with a 1 h incubation in Calcein AM (4 µM in media, Thermo Fisher Scientific, Waltham, MA, USA), fixed with 4% paraformaldehyde and then stored free floating in PBS, as described previously [35]. Antibodies against HIF-1α, Chm1, SMαA, VEGFR2, RUNX2 and osteocalcin (OCN) were all used at a concentration of 1 : 50 to stain for angiogenic, hypoxic and calcific marker expression.

2.8. Analysis of staining

ImageJ (National Institutes of Health, Bethesda, MD, USA) and Matlab (MathWorks, Natick, MA, USA) were used to quantitatively analyse immunohistochemistry (IHC), immunocytochemistry (ICC), Movat pentachrome stains and live staining, as described previously [37]. The ImageJ colour thresholding function was used to analyse per cent area coverage of markers that stained ECM alone or both cells and ECM in whole leaflet sections. All sections were thresholded to a brightness value that excluded slide background while including all of the tissue area. Movat pentachrome stained sections were thresholded according to a range of hues corresponding to the ECM stain colours. In DAB IHC staining of cells and ECM, tissue sections were thresholded to hues corresponding to the brown DAB colour and the blue nuclear stain, haematoxylin. IHC that stained cells and ECM concurrently was quantified as per cent area coverage of the stain over the total area of the tissue. For IHC that stained only cells, the ImageJ analyse particles function was used to quantify the percentage of cells stained with a cell-specific marker versus the total number of cells in the whole leaflet section. HIF-1α positive cell counts were determined using a different method. Using ImageJ, a grid was placed over each tissue section image and cells were sampled and counted by a blinded observer at areas where grid lines crossed. Fluorescent images of the filter paper culture ICC were analysed using a method that differed from the prior two methods. A custom Matlab code provided by the Whitesides group [38] was used to quantitatively analyse the fluorescence of each filter paper well of the AVIC 3D cell cultures. The cumulative fluorescence of each well was recorded over a 20 min window for ICC and a 10 min window for Calcein AM, and those values were each reported in relative fluorescence units. The Calcein AM staining intensity corresponded to the number of living AVICs in the filter paper wells. ICC stain intensities were then divided by the Calcein AM intensities to give an ICC intensity per AVIC value. The three oxygen groups examined by each ICC marker were then normalized to the highest value for that group, giving a maximum value of 1 (normalized ICC intensity per AVIC).

2.9. Statistical analyses

The statistical analyses for Movat staining, intracellular and extracellular DAB tissue staining, and filter paper culture immunocytochemistry were performed using an ANOVA with a post hoc Tukey's test with significance set at p < 0.05. The analyses for the oxygen diffusion coefficient (DO₂) measurements were performed using a two-way ANOVA with a post hoc Tukey's test with significance set at p < 0.05. Leaflet compressive strain analyses were performed using an unpaired two-tailed Student's t-test with significance set at p < 0.05.

3. Results

3.1. Aortic and mitral valve leaflets have similar oxygen diffusion coefficients

The DO₂ can be used to predict normal oxygen diffusion profiles for healthy valves and areas of hypoxia in diseased valves or leaflet cultures. The DO₂ of aortic valve leaflets was reported previously in non-submerged, non-pressurized conditions and was determined to be 1.06 × 10⁻⁵ cm² s⁻¹ [29]. For the studies presented here, a pressurized diffusion chamber, depicted in figure 1a, was created to measure the DO₂ of the belly region of submerged leaflets under fluid pressures of 4.5 kPa and 10.5 kPa, corresponding to pressures of approximately 35 mmHg and 80 mmHg, respectively. Leaflet thickness was measured in a relaxed state without applied fluid pressure following DO₂ measurements. As a final element necessary for the calculation of the DO₂, the stress–strain relationship of whole leaflet belly regions was recorded to determine the change in leaflet thickness at 4.5 kPa and 10.5 kPa, shown in figure 1b. Aortic valves exhibited greater strain than mitral valves beyond 5.5 kPa. Leaflet DO₂ values were relatively similar between 4.5 kPa and 10.5 kPa and among aortic and mitral valves as given in figure 1c. Aortic valve oxygen diffusion values were 1.23 and 1.50 × 10⁻⁵ cm² s⁻¹ and mitral values were 1.31 and 1.36 × 10⁻⁵ cm² s⁻¹ for 4.5 kPa and 10.5 kPa pressures, respectively.

Figure 1.

Oxygen diffusion characteristics of aortic and mitral valve tissue. (a) A pressurized oxygen diffusion system was created to measure the time to a new steady state after a step change in oxygen level in the chamber below the valve leaflet. Pressure was increased in the system by elevating the reservoir above the sample chamber. (b) Valves were mechanically tested in compression to determine the stress–strain (equivalent to pressure-thickness) relationship for the belly region of the leaflets. Aortic valves experienced significantly higher strain than mitral valves for stresses exceeding 5.5 kPa. Each point represents mean ± s.e. (t-test: p < 0.05; n = 7). (c) Aortic and mitral leaflets had similar DO₂ values for each pressure condition. Bar represents mean ± s.d. (t-test: p < 0.05; n = 12–15). (d) Two-dimensional computational oxygen diffusion maps were created to model static radial sections of aortic and mitral valves. Increases in pressure corresponded to a substantial increase in oxygen delivery to the centre of aortic valve leaflets. Although mitral valve leaflets experienced increased oxygen delivery with increased pressure, the central regions of the leaflet remained largely hypoxic.

3.2. Fluid pressure coupled with leaflet stretch enhances leaflet oxygen diffusion

The oxygen diffusion profile for healthy valve leaflets has not been well studied but is essential for understanding changes in oxygen availability during valve disease. The studies presented here demonstrate the relationship between oxygen diffusion and valve thinning due to leaflet flexure. Previous computational modelling studies have indicated that the aortic valve leaflet receives insufficient oxygen in the absence of leaflet motion [27]. In agreement with the conclusions of those previous models, simulations presented here in figure 1d for relaxed aortic valve leaflet conditions showed large areas of 0% oxygen at 0 kPa pressure. Specifically, the models predicted that approximately 40% of the aortic leaflet area would experience nearly 0% oxygen in the absence of motion. As pressure on the leaflets was increased, the leaflets thinned and elongated, allowing more oxygen to reach the centre regions. The same trend was observed with mitral leaflets as pressure was increased. However, even under high pressure, the models predicted that some interior regions in the mitral leaflet would not receive adequate oxygen supply from diffusion alone.

3.3. Hypoxia affects leaflet remodelling, hypoxia inducible factor 1-alpha expression and cell density in statically cultured leaflets

Static leaflet cultures were used to evaluate the impact of hypoxia on VICs in their native ECM environment. Loss of leaflet motion has been predicted to substantially reduce oxygen transport to VICs within leaflets [39]. All static culture unpressurized leaflet computational models showed areas of severe hypoxia in figure 2a. The 20% hyperoxic group (20% oxygen in atmosphere compared with 13% oxygen found in vivo) experienced the least amount of hypoxia yet still did not receive an adequate oxygen supply in all leaflet regions. Overall, as external oxygen levels decreased, the internal PO₂ of the leaflets also decreased. Movat pentachrome staining was used to evaluate the ECM content of valves, and fresh porcine aortic valves (48% collagen) closely matched previously reported values for human valves (52% collagen) [40]. Fresh mitral valves had higher collagen content (70% collagen) than reported previously for human valves (60% collagen) [40], likely to be a result of measuring the anterior leaflet alone in this study without including data from the posterior leaflet. Following static leaflet culture, aortic leaflets displayed increased collagen content and decreased glycosaminoglycan and proteoglycan (GAG/PG) content compared with fresh valves, as shown in figure 2b. Even though all leaflets contained hypoxic areas, leaflets that were exposed to a lower oxygen environment showed lower collagen content when compared with the 20% oxygen group in figure 2c. The 20% and 13% oxygen mitral leaflet groups also showed increased collagen content over fresh valves in agreement with trends from a previous static culture study on mitral valve leaflets [41]. In figure 2d, HIF-1α was used to determine the hypoxic response from the leaflets and showed very low expression in fresh leaflets, as expected. HIF-1α expression was highest in the 13% oxygen group for aortic valves as shown in figure 2e. The hypoxic response in mitral valves was significantly different between the 13% and 5% oxygen groups. Interestingly, the 5% oxygen groups in both valves showed virtually no increase in HIF-1α expression over fresh tissue. In figure 2f, valve cell density was slightly changed in the cultured aortic leaflets with an increase in density in the 20% oxygen group compared with the 13% oxygen group. All aortic valve cell densities, however, remained close to fresh tissue values. Mitral valve cell densities all decreased to approximately half of that found in fresh tissue valves.

Figure 2.

Effect of hypoxia on ECM composition and HIF-1α expression. (a) Two-dimensional computational oxygen diffusion maps were created to model relaxed-state radial sections of aortic and mitral valves in different oxygen environments. As external oxygen decreased, the overall leaflet PO₂ also decreased. (b) Representative Movat pentachrome stain of aortic and mitral valve leaflets. (c) Quantification of leaflet ECM demonstrated that all cultured aortic valves had increased collagen content and decreased GAG/PG content compared with fresh tissue. The 20% oxygen group had the highest collagen and lowest GAG/PG content. In mitral valve cultures, the 13% oxygen group had significantly greater collagen content and less GAG/PG content compared with the 5% oxygen group. Bars represent mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 4–6). (d) HIF-1α stain of aortic and mitral valve leaflets fresh or following static culture. (e) The quantification of HIF-1α positive cells is shown as a per cent of the total leaflet cell population. In both aortic and mitral valves, the 13% oxygen group had significantly higher HIF-1α expression compared with the 5% oxygen group. The HIF-1α expression in the 13% oxygen group was also significantly greater than the 20% oxygen group in aortic valves. Bars are mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 8–9 aortic, n = 4–5 mitral). (f) Changes in overall cell density per leaflet in response to altered oxygen levels. All three aortic valve groups remained near fresh tissue cell density levels with a significant difference between the 20% and 13% oxygen groups. In mitral valves, cell density decreased in all treatment groups to approximately half of the original fresh tissue levels. Bars are mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 5). Asterisk indicates that bar is significant from other two bars and dagger indicates that bar is significant from the other bar as shown by lines above the two bars.

3.4. Mitral valve leaflets show altered pro- and anti-angiogenic factor expression in response to hypoxia in static culture conditions

The whole leaflet cultures were studied to determine if hypoxia altered the protein expression of pro- and anti-angiogenic factors associated with CAVD and MS. Aortic and mitral valves showed changes in VIC densities based on the leaflet region in figure 3a. Terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL), an apoptosis assay, showed that all mitral leaflets experienced high levels of apoptosis after two weeks of culture. In aortic leaflets, the 5% oxygen group showed significantly higher cell death than the 20% and 13% oxygen groups as depicted in figure 3b. Cell death in the leaflets was likely to be affected by oxygen availability, which was substantially reduced in the thick mitral leaflets. In addition to VIC apoptosis in the leaflets, there was a substantial loss of the leaflet endothelium in all culture conditions. See electronic supplementary material, figure S1 in appendix A, for immunofluorescent staining of the leaflet endothelium. In figure 3c, leaflet Chm1 expression, associated with anti-angiogenic activity [42], was decreased in the 13% oxygen mitral leaflet group. In figure 3d, the 13% oxygen mitral leaflet group showed an increase in RUNX2, the upregulation of which is associated with myxomatous disease progression in mitral valves [17]. VEGFR2 expression was also increased in the 13% oxygen mitral leaflet group as shown in figure 3e. When compared with aberrant angiogenic factor expression profiles in human rheumatic mitral valves, the 13% oxygen mitral group showed similar changes with a decrease in Chm1 expression and an increase in VEGFR2 expression [13]. See electronic supplementary material, figure S2 in appendix B, for co-staining with HIF-1α and Chm1 and with HIF-1α and VEGFR2. The NOTCH1 receptor, believed to be involved in CAVD [43], showed no difference among the three aortic oxygen groups. The 5% oxygen mitral leaflet group showed a decrease in the receptor in figure 3f. Hypoxia was also shown to affect the location of cells after two weeks of culture. Cell density was analysed in figure 3g to determine changes in cell populations throughout different areas of the leaflet. The 13% oxygen groups in both aortic and mitral valve leaflets showed the most significant changes in cell density among regions. In comparison, healthy, fresh valve leaflets have consistent cell densities throughout each layer of the leaflet [32].

Figure 3.

Aortic and mitral valve leaflets respond to hypoxia. (a) Fluorescent 4′,6-diamidino-2-phenylindole (DAPI) staining of leaflet VICs show changes in VIC density based on leaflet region. T (top), M (middle) and B (bottom) represent equal divisions of a leaflet radial section with top oriented towards the aorta and bottom directed over the heart left ventricle. (b) Apoptosis was present in more than 5% of cells in all mitral leaflets and the aortic valve 5% oxygen group at two weeks. Aortic valve 13% and 20% oxygen groups showed very low amounts of apoptosis. (c) Chm1 expression decreased in the 13% oxygen mitral leaflet group, indicating an increase in pro-angiogenic activity. (d) RUNX2 levels increased in the 13% oxygen mitral leaflet group. (e) VEGFR2 expression increased in the 13% oxygen mitral leaflet group, also reflective of an increase in pro-angiogenic activity. All bars in (a–e) are mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 5–6). (f) NOTCH1 expression decreased in the 5% oxygen mitral leaflet group. Bars are mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 5–6). (g) Location of cells within three different regions of the leaflets. Radial sections were divided into three regions of equivalent area: bottom, middle and top. Fresh sections have an equal distribution of cells in all regions. The 13% oxygen aortic and mitral valve groups, the 5% oxygen aortic group and the 20% oxygen mitral valve group showed significant changes in cell density. Bars are mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 4). (h) Three-dimensional filter paper collagen cultures with AVICs. The 5% oxygen VIC group showed enhanced pro-angiogenic activity with an increase in HIF-1α expression, a decrease in Chm1 (compared with the 13% oxygen group), an increase in SMαA (compared with the 20% oxygen group) and an increase in VEGFR2. Altered calcific activity was not apparent for these cultures. Bars are mean ± s.d. (ANOVA with post hoc Tukey's test: p < 0.05; n = 6–28). Asterisk indicates that bar is significant from other two bars and dagger indicates that bar is significant from other bar as shown by lines above bars.

3.5. Aortic valvular interstitial cells respond to hypoxia through an increase in hypoxia inducible factor 1-alpha expression and activation in three-dimensional type I collagen cultures

Three-dimensional filter paper-supported collagen cultures were used to examine AVIC behaviour without the influence of complex and diverse ECM or large oxygen gradients. Each sheet of filter paper was cultured in 20%, 13% or 5% oxygen. When compared with whole leaflet expression of the same markers, hypoxic AVICs in three-dimensional filter paper collagen culture showed increased activation and angiogenic factor expression in figure 3h. This angiogenic marker profile for the 5% oxygen AVICs was similar to the 13% oxygen mitral leaflet group. No changes in osteogenic potential, shown by OCN expression, in response to altered oxygen levels were detected in the cultures. Lower oxygen levels in filter paper cultures led to an increase in SMαA expression, as shown in our previous study on AVICs in filter paper [35].

4. Discussion

The role of angiogenesis in CAVD and MS remains unclear, though there is growing evidence of its involvement in calcification [10,14]. HIF-1α, a marker frequently associated with angiogenesis, has recently been identified histologically in CAVD [21,22]. The impact of hypoxia on HIF-1α activation in the context of valve disease has yet to be determined, however. This study sought to establish a basic understanding of the role of oxygen in valve protein expression and ECM composition and the impact of decreased oxygen levels on VIC behaviour. Aortic and mitral valves had comparable DO₂ values in fresh leaflets yet computational oxygen diffusion models suggested that the thickness of the mitral leaflet was too great for diffusion alone to meet leaflet oxygen demand. In vivo, leaflet motion is predicted to enhance oxygen diffusion, greatly reducing the need for vasculature within the leaflet [39]. The aortic valve, with thinner leaflets, was predicted to receive above 7% oxygen in the majority of the leaflet area while under high fluid pressure. When factoring in oxygen diffusion from leaflet motion and interstitial flow [39], this model helps to explain the observation that adult leaflets contain a low vessel density [24]. By contrast, mitral leaflets were predicted to experience extreme hypoxia, which would explain the presence of robust vasculature in these leaflets [23].

The mechanisms that regulate angiogenesis in aortic valves also remain unclear. Hypoxia has been hypothesized to contribute to angiogenesis in CAVD [20], and changes in leaflet thickness and ECM during valve disease could create a hypoxic environment within some areas of the leaflets. Studies on aortic calcific nodules have implicated hypoxia in calcification [22,44]. However, whole aortic valve leaflets cultured in this study showed no apparent angiogenic or calcific changes in AVIC protein expression. Cell–ECM interactions within the leaflets may have affected the VIC response to hypoxia and have previously been shown to regulate calcification in the aortic valve [45,46]. ECM changes during valve disease are also associated with upregulation of the receptor for hyaluronan mediated motility, a receptor linked to angiogenesis [47]. When AVICs were isolated from leaflets and grown in three-dimensional collagen gel cultures under varying oxygen conditions, the cells showed altered protein expression of pro- and anti-angiogenic factors. These results suggest that hypoxia does impact VIC angiogenic signalling yet this behaviour may be masked or even suppressed in whole leaflets. Additionally, the filter paper and collagen combination provided a much stiffer mechanical environmental than much of the aortic valve leaflet and could have elicited stronger activation and pro-angiogenic responses.

In both MMVD and MS, hypoxia and HIF-1α expression remain virtually unstudied. ECM remodelling during these diseases likely alters the oxygen environment, similar to diseases of the aortic valve. MMVD is characterized by an increase in GAGs and PGs, associated with a high DO₂, whereas MS is accompanied by an increase in collagen and a low DO₂ [27]. In MMVD, this enhanced oxygen diffusion could compensate for increased leaflet thickness. Interestingly, HIF-1α gene expression decreased in human valves during MMVD [17]. In this study, hypoxia elicited a substantial change in pro- and anti-angiogenic factor expression from mitral valves. Similar to a previous study of human MS, an increase in VEGFR2 expression and a decrease in Chm1 were observed [13]. Decreases in Chm1 have been linked to angiogenesis in heart valves [42]. Considering the significant impact of hypoxia on pro-angiogenic activity shown in this study, hypoxia could be an essential factor in the progression of MS.

For the improved investigation of oxygen diffusion in tissue, this study also included the design and implementation of an oxygen diffusion chamber for analysing oxygen in valves. The validation of this system provides a foundation for future studies to improve characterization of oxygen distribution in valve leaflets, examine other small tissues or even culture VICs for tissue engineering applications. As an example, a previous study altered oxygen concentration in culture to acquire the desired ECM composition of aortic valve engineered tissue [34]. Furthermore, inclusion of vasculature in tissue-engineered constructs is often essential to the survival of the construct. With the tools described here, the optimal vessel locations and densities can be determined in future studies to better design tissue-engineered heart valves.

5. Conclusion

There is growing evidence that hypoxia may be involved with leaflet angiogenesis and calcification. The results presented here demonstrate that altered oxygen levels have a significant impact on the expression profiles of hypoxic and angiogenic markers in aortic and mitral valve cells or leaflets. Experiments involving hypoxia are often hampered by the difficulty of inducing low oxygen levels and measuring the corresponding changes in tissue oxygenation and cell behaviour. The culture systems and chambers presented here provide a simple and efficient method for studying cells and tissues over long culture durations. With these systems, future studies on heart valves can continue to examine the relationship between oxygen diffusion and angiogenesis and further elucidate the role of hypoxia in valve disease.

Data accessibility

The datasets supporting this article have been uploaded as part of the electronic supplementary material.

Authors' contributions

M.C.S. recorded and calculated the leaflet diffusion coefficients, created the diffusion models, cultured the whole leaflets and filter paper constructs, and analysed the data. V.K.K. performed the IHC and Movat staining. D.S.P. collected and analysed the data for the leaflet stress–strain curve. S.B. constructed the leaflet diffusion profiles for the diffusion models and performed the HIF-1α quantification. G.F. assisted in the whole leaflet cultures and preparation of the tissue specimens. N.M. quantified the leaflet cell densities and the Chm1 area coverage. All authors contributed to the manuscript and gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

The authors gratefully acknowledge funding from the National Science Foundation CBET grant no. 1404008 and the American Heart Association predoctoral fellowship 15PRE25080316.