The tensile and viscoelastic properties of aortic valve leaflets treated with a hyaluronidase gradient

Abstract

Purpose

When diseased, aortic valves are typically replaced with bioprosthetic heart valves (BPHVs), porcine valves or bovine pericardium that are fixed in glutaraldehyde. These replacements fail within 10–15 years due to calcification and fatigue, and their failure coincides with a loss of glycosaminoglycans (GAGs). This study investigates this relationship between GAG concentration and the tensile and viscoelastic properties of aortic valve leaflets.

Methods

Aortic valve leaflets were dissected from porcine hearts and digested in hyaluronidase in concentrations ranging from 0–5 U/mL for 0–24 hours, yielding a spectrum of GAG concentrations that was measured using the uronic acid assay and confirmed by Alcian Blue staining. Digested leaflets with varying GAG concentrations were then tested in tension in the circumferential and radial directions with varying strain rate, as well as in stress relaxation.

Results

The GAG concentration of the leaflets was successfully reduced using hyaluronidase, although water content was not affected. Elastic modulus, the maximum stress, and hysteresis significantly increased with decreasing GAG concentration. Extensibility and the radius of transition curvature did not change with GAG concentration. The stress relaxation behavior and strain-rate independent nature of the leaflet did not change with GAG concentration.

Conclusions

These results suggest that GAGs in the spongiosa lubricate tissue motion and reduce stresses experienced by the leaflet. This study forms the basis for predictive models of BPHV mechanics based on GAG concentration, and guides the rational design of future heart valve replacements.

Introduction

The aortic valve (AV) is located at the junction of the left ventricle and aorta, and serves to maintain unidirectional blood flow out of the heart and into the systemic circulation. The valve consists of three leaflets, which flex and stretch during diastole to withstand the backflow of blood into the left ventricle.¹ When the valve fails, a common replacement is a bioprosthetic heart valve (BPHV), usually a porcine valve or bovine pericardium, that is fixed with glutaraldehyde to improve tissue stability and reduce tissue antigenicity.^2,3 These replacements commonly fail within ~15 years due to calcification and fatigue.^2–5

Studies of explanted BPHVs have demonstrated that the failure of these replacements coincides with the loss of glycosaminoglycans (GAGs), linear polysaccharide chains of repeating disaccharide units that are commonly attached to proteoglycans (PGs).^6–9 In particular, there is a significant loss of the GAGs chondroitin-6-sulfate (C6S) and hyaluronan (HA)⁶, as a result of both the fixation process^7,8, and enzymatic digestion.⁹ These findings indicate that there is a correlation between the loss of GAGs and BPHV failure. The removal of GAGs from porcine leaflets through digestion with hyaluronidase (HAse) and chondroitinase ABC (Cse-ABC), however, did not significantly increase calcium content in AV leaflets.¹⁰ Thus, it can be inferred that the mechanism through which GAG loss leads to valve failure is mechanical rather than biochemical.

Therefore, this study investigated the role of GAGs on the material properties of the AV leaflet through HAse treatment. The material properties that the aortic valve leaflet natively possesses are very unique. The leaflet has been classified as “quasi-elastic,” in that it is neither purely elastic or viscoelastic, but has characteristics from each classification that suit its functional needs.¹¹ Similar to other biological tissues, the leaflet exhibits a bilinear stress-strain curve in response to tensile loading, a result of crimped collagen fibers that extend in the “toe” and transition regions before bearing significant load in the stiffer linear region.^12,13 The leaflet is also anisotropic, as it is three times stiffer in the circumferential direction than in the radial direction, due to the circumferential orientation of collagen fibers.^13–16 The leaflet also has low hysteresis, allowing the leaflet to retract fully during systole; behaves in a strain-rate-independent stress-strain manner so the valve can close regardless of heart rate; and undergoes stress relaxation due to realignment of collagen fibers towards the directions of stress.^11,15,16 Together, these material properties allow the aortic valve to function competently for multiple cycles over a long period of time.

GAGs exist throughout the AV leaflet but are primarily located in the spongiosa, the middle of three layers in the aortic valve, either in the form of HA or chondroitin (CS) and dermatan sulfates (DS) attached to the proteoglycan (PG) versican.¹⁷ The spongiosa is believed to dampen vibrations from closing, confer flexibility to the leaflet, and lubricate shear between the outer layers, all functions which are derived from the ability of GAGs to retain water.¹⁸ Removing GAGs from the leaflet has been shown to increase flexural stiffness, possibly due to a lack of lubrication between the outer layers, which creates friction as the layers shear against each other.¹⁰ This interpretation is supported by the fact that dehydrated and glutaraldehyde-fixed leaflets are stiffer in planar shear.^19,20 In addition, HA exhibits shear thinning, as its viscosity decreases with increasing shear rate, which possibly explains the leaflet’s function independent of strain rate.²¹ Taken together, GAGs would appear to reduce leaflet stiffness and lubricate shear to make the leaflet elastic, and functionally independent of strain rate.

This study builds upon previous research to provide more in-depth information about the role of GAGs in aortic valve leaflet mechanics, specifically with regards to the tensile and viscoelastic behavior of the leaflet. GAGs were enzymatically removed from the leaflet by treatment with successively increasing concentrations of HAse to yield a spectrum of GAG concentrations. Digested leaflets containing different GAG concentrations were then tested in tension at different strain rates, as well as in stress relaxation.

Materials and Methods

Tissue harvest

Fresh porcine hearts were obtained from a local commercial abattoir (Fisher Ham and Meats, Spring, TX) within 6 hours of death. Aortic valve leaflets were dissected from the heart, rinsed and stored in phosphate buffered saline (PBS, pH ~ 7.4) at 4°C until use the next day.

GAG digestion

GAGs were digested from aortic valve leaflets using bovine testicular HAse (Worthington Biochemical, Lakewood, NJ) in PBS, which has specificity for HA and CS.²² Whole aortic valve leaflets were digested with varying concentrations of HAse between 0–10 U/mL and varying times between 0–24 h in a shaking incubator (200 rpm) at 37°C. The wet weights (WW) of the leaflets were taken before and after digestion to measure weight loss as a result of HAse digestion. In a parallel study, 5 mm circumferential x 5 mm radial squares were cut from the belly of leaflets and similarly digested with HAse, after which the dimensions were measured again to assess the effect of HAse digestion on specimen size.

The uronic acid assay was used to quantify the amount of GAGs remaining in the leaflets after HAse digestion (n=5).²³ After digestion and weighing, the leaflets were frozen and lyophilized overnight. The dry weight (DW) of the leaflet was measured, and water content was measured by normalizing the change in weight from drying to the WW after digestion. The dry tissue was rehydrated in 100 mM ammonium acetate (pH~7.0), minced, then digested in 100 μL proteinase-K solution (10 mg/mL in ddI H₂O, Calbiochem, La Jolla, CA) for 16 h at 60°C. The enzyme solution was then inactivated by heating for 30 min at 70°C. The digested solution was then centrifuged at 5000 rpm for 3 min, after which sample volumes of 100 μL were taken from the supernatant. Cold sulfuric acid tetraborate was added to the samples and heated to 100°C for 5 min. After cooling to room temperature, 10 μL of hydroxyphenyl reagent (0.15% m-phenylphenol in 0.5% NaOH, Sigma-Aldrich, St. Louis, MO) were added to each sample. The samples were read for absorbance on a spectrophotometer (SpectraMax M2, Molecular Devices, Sunnyvale, CA) at 532 nm with background subtraction at 750 nm. Raw absorbance data was compared to glucuronic acid standards (Sigma-Aldrich) to calculate the total amount of GAGs.

Histology

The removal of GAGs from the spongiosa was confirmed by histology. HAse treated leaflets were fixed overnight in 10% formalin, then dehydrated, embedded in paraffin, and sectioned according to standard procedures. Sections were stained with Alcian Blue (1% in 3% acetic acid, Sigma-Aldrich) for 5 min, washed, then dipped in alkaline alcohol (10% ammonium hydroxide in 95% ethanol) for 2 min at 80°C to render the stain insoluble.^10,24 After 20 min of washing under running water, the sections were then counterstained using Nuclear Fast Red (0.1% in dI H₂O, Sigma-Aldrich) for 5 min. The slides were then dehydrated and coverslipped. Images of the sections were taken on an upright microscope (DM LS2, Leica, Wetzlar, Germany) using a charge coupled device camera (DFC 320, Leica).

Tensile Testing

Mechanical testing of leaflets was performed on a uniaxial mechanical tester with a 22 N load cell (ELF 3220, Bose ElectroForce, Eden Prairie, MN). All testing was performed in a PBS bath warmed to 37°C. The leaflets were cut into 5 mm wide strips in either the circumferential or the radial direction. Both ends of the strip were glued to pieces of balsa wood on one side, where the sample was gripped with coarse sandpaper, leaving roughly 6 mm of tissue between the two pieces of balsa (Fig. 1). The thickness of the leaflet was measured using a stereomicroscope (MZ6, Leica Microsystems, Wetzlar, Germany) at 5 places along the leaflet length between the balsa wood pieces and averaged.

FIGURE 1:

Schematic of leaflet gripping. 5 mm wide strips of tissue were glued on one side to balsa wood, and then the sample was gripped with sandpaper from both sides.

For each digestion group, samples were pulled to 50% strain at 1, 5, and 10 mm/s in both the circumferential and radial directions (n=5 per strain rate and direction). Length and width of each sample was measured after loading onto the mechanical tester but prior to testing. Each sample was preconditioned during testing using a particular regimen based on direction and strain rate; each regimen started with repeated stretches to 10%, then 20% and 50%, until a final stretch to 50% strain, which was recorded (Fig. 2).

FIGURE 2:

Preconditioning cycles for (a) tension and (b) stress relaxation. Leaflets tested in tension were preconditioned by stretching to 10%, 20%, and 50% based on the strain rate. The number of cycles needed for each strain level was determined before testing each group as the number of cycles needed before the maximum load of one cycle was within 1% of the maximum load of the previous cycle. Leaflets tested in stress relaxation were preconditioned using a regimen from a previous study.²⁷

The final number of cycles needed was determined for each strain rate before each group was tested by finding the cycle in which the maximum load was within 1% of the maximum load of the previous cycle (Table 1). Data was acquired from the last loading and unloading curves (Fig. 3).

TABLE 1:

Number of cycles to each strain during preconditioning for each strain rate

	Strain Rate (mm/s)
	1	5	10
n 10	30	30	25
n 20	20	15	15
n 50	15	15	15

FIGURE 3:

Representative stress-strain curves of HAse-treated leaflets in the (a) circumferential and (b) radial directions, and (c) relaxation curves of HAse-treated leaflets in both the circumferential and radial directions.

From each load-displacement curve, several pieces of data were derived as previously described using MATLAB (MathWorks, Natick, MA).²⁵ First, gage length was calculated from the loading curve as the local minimum of a cubic function fit to the lower displacement region of the curve.^25,26 After conversion of load-displacement to stress-strain, the distinct regions of the loading curve were identified. The linear region of the curve was initially identified as the region of 10% of total strain and the maximum r² from a linear least-squares fit. The linear region was then extended and shortened point by point to identify the curve segment with the maximum r². The toe region was identified in a similar manner starting from the origin. The transition region was located between the toe and linear regions.

From the linear region, the elastic modulus ($E$, MPa) was derived as the slope of the linear fit. The maximum stress (σ_max, MPa) was also recorded. Extensibility (mm/mm) was calculated as the x-intercept of this fit. Hysteresis was calculated as the energy dissipated (area between the loading and unloading curves) as a percentage of the strain energy of loading (area underneath the loading curve).

The radius of transition curvature (RTC) was also calculated as a measure of the abruptness in transition between the toe and linear regions.²⁵ The stress-strain curve was normalized to σ_max and ultimate strain. A hyperbola was fit to the resulting curve in a second coordinate space, with the y-axis bisecting the linear fits of the toe and linear regions, and their intersection serving as the origin. The RTC was calculated as the reciprocal of the second derivative at the minimum.

Stress Relaxation

Samples in each digestion group also underwent stress relaxation testing in both directions (n=5 per direction). The preconditioning regimen was derived from a previous study on stress relaxation in aortic valve tissues.²⁷ Samples were first stretched to 50% strain for 25 cycles at 5 mm/s, then pulled to 50% strain at 10 mm/s and relaxed for 100 s. This relaxation cycle was repeated 7 times, with one triangle loading-unloading curve at 5 mm/s in between relaxation cycles (Fig. 2). The last relaxation cycle was collected.

The time-load curve was converted to time-stress and the final relaxed stress was calculated as a percentage of the original stress (Fig. 3). The relaxation curve was then fit to a two-phase decay equation:

$$\sigma ={\sigma }_0+A_0{e}^{-t/{\tau }_{fast}}+A_1{e}^{-t/{\tau }_{slow}}$$

where the ${\text{τ}}_{\text{fast}}$ and ${\text{τ}}_{\text{slow}}$, the time constants, were recorded.

Statistical Analysis

Statistical analysis was performed on the data using statistical analysis software (SigmaStat, Systat Software, San Jose, CA). One-way ANOVAs were performed for analysis of GAG digestion regimens. For tensile testing data, two-way ANOVAs were performed on the data in both the circumferential and radial direction for the effect of GAG concentration and strain rate. Interactions were analyzed as well if the effects of both GAG concentration and strain rate were significant. For stress relaxation data, one-way ANOVAs were performed. Significance was defined as p<0.05. If a significant effect was found, a post-hoc Tukey’s test was performed to compare between groups (see Supplemental Tables S1–10). All data is presented as mean ± standard deviation.

Results

GAG Digestion

Aortic valve leaflets were treated with HAse in varying amounts for various times to yield a spectrum of GAG concentrations, as demonstrated by the uronic acid assay. This reduction in GAGs was confirmed by histology, where the strong staining for GAGs in the spongiosa was progressively reduced with stronger enzyme digestion regimens (Fig. 4).

FIGURE 4:

Alcian Blue stains of AV leaflets (GAGs=blue, cell nuclei=red) digested in (a) 0, (b) 1, (c) 2, (d) 5 U/mL HAse for 8 hours, and (e) 10 U/mL HAse for 24 hours. Note the progressive reduction in staining strength of the GAGs in the center of the leaflet. Scale bar = 200 μm.

There was also a significant effect of digestion regimen on WW loss, with samples treated with HAse concentrations of 2 U/mL and higher losing a significant amount of WW after digestion (Fig. 5b). Total leaflet water content and leaflet size did not significantly change between digestion regimens (Fig. 5c-d).

FIGURE 5:

(a) GAG concentrations, (b) wet weight loss, (c) water content, and (d) size change of the digested leaflet. GAG concentrations and wet weight loss progressively decreased and increased, respectively, with increasing regimen strength, while water content and size did not change. Data presented as mean ± standard deviation. * indicates p<0.05 within brackets.

Tensile Testing

Leaflets digested in HAse were tested at varying strain rates (1, 5, and 10 mm/s) in both the circumferential and radial directions to elucidate the effects of GAG concentration on $E$, extensibility, radius of transition curvature, and hysteresis. Both $E$ and σ_max demonstrated a significant increase in both directions with decreasing GAG concentration, but there was no significant effect of varying strain rate (Fig. 6).

FIGURE 6:

$E$ (a,d), ${\text{σ}}_{\text{max}}$ (b,e), and hysteresis (c,f) of HAse-treated AV leaflets in the circumferential (a-c) and radial directions (d-f) at different strain rates. $E$, ${\text{σ}}_{\text{max}}$, and hysteresis in both directions varied significantly with GAG concentration. Hysteresis also varied with strain rate, but there was no significant interaction between strain rate and GAG concentration. Data presented as mean ± standard deviation. * indicates that there was a significant overall effect (p<0.05) of GAG concentration. # indicates that there was a significant overall effect (p<0.005) of strain rate.

Extensibility significantly decreased in both directions with strain rate, but varying GAG concentration did not significantly alter extensibility (Fig. 7). Both GAG concentration and strain rate had significant effects on hysteresis in both the circumferential and radial direction, although there was no significant interaction between the two effects (Fig. 6). There was no significant effect of either GAG concentration or strain rate on RTC (Fig. 7).^a

FIGURE 7:

Extensibilities (a,c) and RTCs (b,d) of HAse-treated AV leaflets in the circumferential (a,b) and radial (c,d) directions. Extensibility in both directions varied with strain rate, but not with GAG concentration. Neither GAG concentration nor strain rate had a significant overall effect on the radius of transition curvature. Data presented as mean ± standard deviation. # indicates that there was a significant overall effect (p<0.05) of strain rate.

Stress relaxation

Digested leaflets also underwent stress relaxation to further characterize the effects of GAG concentration on leaflet material properties. Leaflets were pulled to 50% strain at 10 mm/s and held at constant strain for 100 s. The time constants ${\text{τ}}_{\text{fast}}$ and ${\text{τ}}_{\text{slow}}$ from the double exponential fit, as well as the final relaxed stress as a percentage of original stress, were measured. There was no significant effect of GAG concentration on all three measures in each direction (Fig. 8).

FIGURE 8:

Results of stress relaxation in HAse treated AV leaflets: the time constants (a) ${\text{τ}}_{\text{fast}}$ and (b) ${\text{τ}}_{\text{slow}}$ from a double exponential decay fit, and (c) the final relaxed stress as a percentage of the original stress. There was no significant overall effect of GAG concentration found. Data presented as mean ± standard deviation.

Discussion

This study investigated the role of HAse treatment on the tensile and viscoelastic properties of aortic valve leaflets. With decreasing GAG concentration, leaflets became stiffer, and had greater hysteresis, in both the circumferential and radial directions. Removing GAGs from the leaflet did not change the strain-rate-independent nature of the leaflet, nor did it change the stress relaxation profile of the leaflet. The interpretation of these results is that GAG concentration in the leaflet plays a significant role in leaflet tensile behavior, but not viscoelasticity.

The increases in circumferential and radial $E$, ${\text{σ}}_{\text{max}}$ and hysteresis as a result of HAse treatment are supported by previous results in literature. Leaflets enzymatically digested with HAse and Cse-ABC were stiffer in flexure than native leaflets.¹⁰ Similarly, dehydrated leaflets were stiffer in planar shear and experienced increased hysteresis.¹⁹ Glutaraldehyde fixed valves, which lose GAGs during fixation^7,8, have also been shown to be less extensible and stiffer than the native leaflet, although this result can be primarily attributed to cross-linking during fixation.^20,28–30 Taken together, the results from this study regarding $E$, ${\text{σ}}_{\text{max}}$, and hysteresis support previous theories that the spongiosa serves as a lubricant of tissue motion.^18,19 The removal of GAGs from the leaflet would then lead to increased friction during tissue motion that translates to increased stiffness and hysteresis. These results further validate the importance of GAGs in valve mechanics and function.

The RTCs and extensibilities of the leaflet in both directions did not significantly change with varying GAG concentration. A possible explanation for this phenomenon is that these properties are more related to the structure of the collagen fiber network, which is oriented circumferentially and located in the outflow layer of the aortic valve, the fibrosa.^17,31 As the HAse does not degrade collagen, it is a reasonable finding that the HAse treatment in this study had no effect on these properties.

Interestingly, the stress relaxation behavior and strain-rate-independent tensile behavior of leaflets did not change with decreasing GAG concentration. This result contradicts previous studies on the effects of GAGs or dehydration on viscoelasticity in the aortic valve leaflet.^19,24,32 A possible explanation for the lack of an effect is that there was no change in water content with increasing enzyme regimen strength. One previous study showed that GAG digestion with HAse and Cse-ABC does reduce water retention as well as the ability of the leaflets to rehydrate.¹⁰ This could be a result of potential tandem enzyme activity from the HAse and Cse-ABC, which has specificity for CS, and works slowly on HA.³³ The result on water content in the AV leaflet presented here highlights the need for further examination into the mechanism of water retention in leaflets with lower GAG concentration.

The results of this study also differ from recent work on the effects of enzymatic digestion of GAGs on leaflet mechanical properties.^24,34 In one study, GAGs were removed using an enzyme cocktail consisting of 5 U/mL HAse, 0.1 U/mL Cse-ABC, and 0.15 U/mL keratanase, as demonstrated by Alcian Blue staining.²⁴ The resulting digested leaflets demonstrated significantly less stress relaxation while showing no change in the tensile properties of the leaflet,²⁴ which strongly contrasts the results of this study wherein leaflets showed no difference in stress relaxation, but significant increases in $E$, ${\text{σ}}_{\text{max}}$, and hysteresis. In another study, in which GAGs were removed from leaflets using 30 U/mL HAse and 0.6 U/mL Cse-ABC, no significant change was found in the tensile behavior or hysteresis of digested leaflets at peak physiological tension (90 N/m), while at low membrane tension (10 N/m), there was no difference in hysteresis from planar biaxial tension but significantly lower hysteresis from flexure as a result of GAG digestion.³⁴ In this study, GAG concentration significantly affected $E$, ${\text{σ}}_{\text{max}}$, and hysteresis at membrane tensions between 50–1000 N/m (see Supplemental Fig. S11), levels well past peak physiological loading. As the presented study demonstrates that an HAse-only treatment with higher specificity primarily removed GAGs from the spongiosa (Fig. 4), it is possible that the enzyme digestion regimens of the previous studies, with higher strength, broader specificity and potentially enhanced tandem enzyme activity, affected GAGs in the outer layers of the aortic valve leaflet to a much larger extent than this study. These outer layers are the collagenous fibrosa, and the elastic ventricularis, the thinner inflow layer containing sheets of elastic fibers, as well as collagen. GAGs in these layers are most commonly part of the small leucine-rich PGs (SLRPs) decorin and biglycan. These PGs are also closely associated with collagen fibrillogenesis.^17,25,35,36 Indeed, decorin has been shown to affect the mechanical behavior of tendons, particularly their viscoelasticity^37,38, and reduction in biglycan content correlates with the stiffening of neonatal Achilles’ tendons.³⁹ Interestingly, GAG removal in tendons demonstrated little effect on their viscoelastic properties, suggesting that decorin and biglycan may not necessarily influence mechanical properties through their GAG chains.⁴⁰ Further studies regarding the effect of GAGs on leaflet mechanical properties should focus on GAG type, the specific PGs to which these GAGs are attached, the specific region from which GAGs are removed, and the physiological loading regime within which the leaflets are tested.

In conclusion, leaflets treated with increasing concentrations of HAse show progressively and significantly higher $E$, hysteresis, and ${\text{σ}}_{\text{max}}$ than native AV leaflets, but similar stress relaxation behavior. Given previous results using different enzyme regimens, this present study should be considered in the appropriate context: the removal of GAGs from the spongiosa using HAse significantly increased stiffness but did not change the stress relaxation behavior of the overall leaflet. These results help paint a more complex and nuanced picture of the role of GAGs in leaflet material properties and overall valve function. The results of this study have many implications for aortic valve replacement. First, as the only study to look at the mechanical properties of AV leaflets based on a spectrum of GAG concentrations, the mechanical testing results in this study form the basis of predictive models of BPHV material properties based on GAG concentration. There are also implications in this study for BPHV design. Recent studies have focused on identifying fixation techniques that stabilize and preserve GAGs.^41–43 In addition, there are implications in this study for scaffolds for heart valve tissue engineering. GAG hydrogels, including HA^44–48 and CS-based⁴⁹ gels, are being extensively investigated for possible use as scaffolds for heart valve tissue engineering, in particular as part of a tri-layered scaffold as a middle GAG layer that mimics the native heart valve structure. Thus, for the rational design of a tri-layered heart valve scaffold using GAG hydrogels as the middle layer, it is critical to understand the effect of GAG concentration on overall tissue material properties. This study demonstrates the importance of GAGs in aortic valve function, and future research should focus on their role in greater detail.