Model studies of advanced glycation end product modification of heterograft biomaterials: The effects of in vitro glucose, glyoxal, and serum albumin on collagen structure and mechanical properties

Abstract

Glutaraldehyde cross-linked heterograft tissues, bovine pericardium (BP) or porcine aortic valves, are the leaflet materials in bioprosthetic heart valves (BHV) used in cardiac surgery for heart valve disease. BHV fail due to structural valve degeneration (SVD), often with calcification. Advanced glycation end products (AGE) are post-translational, non-enzymatic reaction products from sugars reducing proteins. AGE are present in SVD-BHV clinical explants and are not detectable in un-implanted BHV. Prior studies modeled BP-AGE formation in vitro with glyoxal, a glucose breakdown product, and serum albumin. However, glucose is the most abundant AGE precursor. Thus, the present studies investigated the hypothesis that BHV susceptibility to glucose related AGE, together with serum proteins, results in deterioration of collagen structure and mechanical properties. In vitro experiments studied AGE formation in BP and porcine collagen sponges (CS) comparing ¹⁴C-glucose and ¹⁴C-glyoxal with and without bovine serum albumin (BSA). Glucose incorporation occurred at a significantly lower level than glyoxal (p<0.02). BSA co-incubations demonstrated reduced glyoxal and glucose uptake by both BP and CS. BSA incubation caused a significant increase in BP mass, enhanced by glyoxal co-incubation. Two-photon microscopy of BP showed BSA induced disruption of collagen structure that was more severe with glucose or glyoxal co-incubation. Uniaxial testing of CS demonstrated that glucose or glyoxal together with BSA compared to controls, caused accelerated deterioration of viscoelastic relaxation, and increased stiffness over a 28-day time course. In conclusion, glucose, glyoxal and BSA uniquely contribute to AGE-mediated disruption of heterograft collagen structure and deterioration of mechanical properties.

1. INTRODUCTION

Heart valve disease is a worldwide problem, affecting millions[1]. At present, there is no effective medical therapy. Patients requiring treatment must undergo either attempted repair of the diseased valves or replacement with a prosthesis. The current preferred replacement valves in adults are bioprosthetic heart valves (BHV) fabricated from glutaraldehyde-fixed xenografts, either bovine pericardium (BP) or porcine aortic valves[1]. BHV are relatively non-thrombogenic compared to mechanical heart valve prostheses; this reduces the need for anticoagulants[2]. However, BHV functional lifespans are limited due to the progressive development of structural valve degeneration (SVD), most often involving leaflet calcification[3,4]. Much of the prior research concerned with mitigating SVD focused on the inhibition of calcification as the primary mechanism[5–9]. However, this yielded only incremental improvements in BHV lifespan[10–13]. Prior studies of explanted BHV showed that the degree of calcification does not completely explain SVD and that approximately 25% of failed valves lack any significant calcification[14].

The present studies investigated the hypothesis that advanced glycation end products (AGE) contribute to the pathophysiology of SVD. AGE result from the post-translational, non-enzymatic modification of proteins by non-enzymatic modification of proteins by hexoses, pentoses, or their breakdown products from Amadori reactions or other oxidative related mechanisms[15–17]. AGE both modify proteins involved in normal physiologic functions, such as serum albumin and hemoglobin (HA1C) and are associated with the pathophysiology of a number of important diseases including diabetes[18,19] and Alzheimer’s Disease[20–22]. The importance of AGE for SVD is incompletely understood. A prior study by our group evaluated a clinical cohort of 45 explanted BHV with SVD and demonstrated via immunohistochemistry (IHC) that AGE and human serum albumin were present in all explants but were undetectable in unimplanted BHV[23]. This BHV explant study was complemented by ex vivo pulse duplicator experiments demonstrating that exposure to both glyoxal, a common AGE intermediary, and serum albumin significantly impaired the hydrodynamic performance of trileaflet, clinical grade BHV[23].

The present studies investigated the hypothesis that BHV are highly susceptible to AGE modification due to both glucose and glyoxal addition and serum albumin uptake, and that these events disrupt the structure of BHV leaflets, adversely affecting biomechanical performance. To test this hypothesis, experiments were performed to determine the following: the glycation kinetics of BHV tissues, the effects of glutaraldehyde-fixation on glycation, the impact of serum albumin exposure on glycation, glycation’s capacity to alter BHV collagen microarchitecture, and glycation’s capacity to alter glutaraldehyde-fixed collagen’s linear elastic and viscoelastic properties.

2. MATERIALS AND METHODS

2.1. Materials

Bovine serum albumin (BSA, protease-free, >98% purity), D-glucose, glyoxal, sodium azide, sodium chloride, sodium borohydride, and HEPES were purchased from Sigma-Aldrich (St. Louis, MO). Fresh BP were shipped on ice from Animal Technologies (Tyler, TX). Surgifoam® hemostatic collagen sponges (CS) composed of gelatin purified from porcine skin were purchased from Ethicon (Somerville, NJ). The ¹⁴C-radiolabeled D-glucose (5mCi/mmol) and glyoxal (110mCi/mmol) were purchased from American Radiolabeled Chemicals (St. Louis, MO). Biosol and Bioscint were purchased from National Diagnostics (Atlanta, GA). All other sources of materials have been indicated in individual methods.

2.2. Glutaraldehyde Fixation of Bovine Pericardium and Collagen Sponges

BP in a fresh state, shipped on ice, were rinsed in saline (0.9% NaCl) and any residual fatty or muscular tissue was removed by dissection. The BP and CS were immersed for 7 days at room temperature 0.6% glutaraldehyde (Polysciences Inc.; Warrington, PA) in HEPES buffer solution (50mM HEPES, 0.9% NaCl, pH 7.4). The samples were then rinsed in fresh HEPES buffer solution for 1 hour before transferring to storage solution of 0.2% glutaraldehyde in HEPES buffer solution and stored at 4°C. Prior to any experiments, samples were exhaustively rinsed with phosphate-buffered saline (PBS) to remove the storage solution.

2.3. Radiolabeled Glycation Assays

2.3.1. Glycation of Glutaraldehyde Fixed Collagenous Tissues

To quantify the glycation kinetics of the model substrates, samples (n=5) were punched from glutaraldehyde-fixed BP and CS using an 8mm biopsy punch. These samples were incubated for 1, 3, 7, 14, or 28 days at 37°C shaking at 110RPM in solutions (1ml per sample) of PBS with either: glucose (100mM), glucose (100mM) + BSA (5%), glyoxal (50mM), or glyoxal (50mM) + BSA (5%). All incubations contained sodium azide (0.1%) to maintain sterility. The glucose media contained ¹⁴C-labeled glucose (0.544μCi/ml). The glyoxal media contained ¹⁴C-labeled glyoxal (0.356μCi/ml). To calibrate the radioactive signal and confirm sufficient excess of radioactive reagents, a 500μl aliquot was taken from each incubation medium at the start and end of incubation, and radioactivity levels determined. The radioactivity of each medium was measured using a Beckman LS 6000 (Beckman Coulter; Brea, CA). Following incubations, the samples were extensively rinsed with deionized water and lyophilized for at least 48 hours. The dried samples were weighed and then digested using Biosol and combined with Bioscint for scintillation counting, which was used to calculate glucose or glyoxal incorporation into the substrate normalized by dry weight. Additional BP samples (n=5) were subjected to 28-day incubations and subsequent rinsing to quantify glucose and glyoxal dissociation. The samples were rinsed in deionized water (1ml per sample) for 28 days at 37°C shaking at 110RPM. The rinses were collected and replaced at days 1, 3, 7, 14, and 28 and then combined with Bioscint for scintillation counting. The amount of glucose and glyoxal in the rinse was normalized to the dry weight and expressed as fraction dissociated based off the 28-day incorporation data.

2.3.2. Glycation Kinetics of Bovine Serum Albumin

To study the glycation kinetics of albumin in vitro, BSA (5%) was incubated with either glucose (100mM) or glyoxal (50mM) for 28 days at 37°C shaking at 110 RPM. The solutions contained either ¹⁴C-labeled glucose (0.544μCi/ml) or ¹⁴C-labeled glyoxal (0.356μCi/ml) like Methods 2.3.1 with samples taken at the start to confirm baseline activity. To maintain sterility, sodium azide (0.1%) was added to all solutions. At 1 day, 3 days, 7 days, 14 days, and 28 days samples were taken from the reacting mixture and passed through Econo-PAC® 10DG desalting columns (BIO-RAD; Hercules, CA) to separate glycated BSA from unbound glucose or glyoxal. To estimate the degree of glycation, each of triplicate samples was mixed with Biosol and measured using a liquid scintillation counter. The amount of bound glucose or glyoxal was quantitated and expressed as a molar ratio of glucose or glyoxal per mg BSA.

2.3.3. Glycation Comparison of Fresh and Fixed Bovine Pericardium

To compare the glycation capacity of non-fixed BP relative to glutaraldehyde-fixed BP, samples (n=5) were punched from fresh BP stored at −80°C and glutaraldehyde-fixed BP using an 8mm biopsy punch. Samples were rinsed with PBS and then incubated for 28 days in PBS with either: glucose (100mM), glucose (100mM) + BSA (5%), glyoxal (50mM), or glyoxal (50mM) + BSA (5%). Sterility was maintained by adding sodium azide (0.1%) to each incubation. Prior to incubation, radiolabeled reagents were added and activity was measured before and after incubation, as above. To measure each aliquot’s radioactivity, each aliquot had 5ml of Bioscint added, was shaken till the solution turned clear, and then underwent scintillation counting. At the conclusion of the incubations, samples were exhaustively rinsed with deionized water and underwent a minimum of 48 hours of lyophilization. The lyophilized samples were weighed and then digested in Biosol as above.

2.4. Serum Albumin Uptake by Bovine Pericardium

Serum albumin uptake kinetics were quantified by measuring the dry and wet mass change of BP samples following BSA exposure. Square samples (n=5, ca. 20mm by 20mm, >18mg dry weight) were cut from glutaraldehyde-fixed BP. The samples were exhaustively rinsed in deionized water to remove any residual salts. Samples were lyophilized for at least 48 hours and their dry weight was measured. The samples were subsequently immersed in 5 ml deionized water per sample for 24 hours gently shaking at 4°C to rehydrate. The wet weight of each sample was measured by blotting each side with Whatman paper (Maidstone, United Kingdom) to remove excess water. The samples were next incubated for 1, 3, 7, 14, or 28 days in solutions (5ml per sample) of PBS with either: BSA (5%), BSA (5%) + glucose (100mM) or BSA (5%) + glyoxal (50mM). All incubations had sodium azide (0.1%) added for sterility. Following incubations, the samples were again exhaustively rinsed with deionized water and lyophilized for 48 hours to calculate dry weight. Wet mass was recalculated as per the previous method.

2.5. Morphological Studies

2.5.1. Sample preparation

To prepare samples for immunohistochemistry (IHC) and two-photon microscopy endpoints, samples (n=5) were punched from glutaraldehyde-fixed BP using an 8mm biopsy punch. The BP samples were incubated for 28 days in (1ml per sample) PBS or PBS with either: BSA (5%), glucose (100mM), glucose (100mM) + BSA (5%), glyoxal (50mM), or glyoxal (50mM) + BSA (5%). A group of 5 BP samples were set aside after 24 hours of each incubation to provide a baseline samples for two-photon microscopy. Sodium azide (0.1%) was added to maintain sterility. BP samples were then exhaustively rinsed with PBS prior to follow-up protocols.

2.5.2. Immunohistochemistry Assays

Tissue designated for IHC was fixed in 10% neutral buffered formalin at 4°C for 48 hours. The tissue was then gradually dehydrated and embedded in paraffin. The paraffin blocks were sectioned at 6μm and mounted on Histobond (VWR; Radnor, PA) slides. The slides were then heated in an oven and re-hydrated in successive xylene to ethanol baths. The slides were then incubated overnight in 60°C citrate buffer (ThermoFisher; Waltham, MA) for antigen retrieval. Following antigen retrieval, slides were rinsed then incubated with primary antibody (αAGE, 0.4 μg/ml | Abcam, Cambridge, United Kingdom; αCML, 0.12 μg/ml | Abcam; αGlucosepane, 4.5μg/ml | David Spiegel laboratory, Yale University, New Haven, CT, per material transfer agreement; αBSA, 0.05 ug/ml| Abcam) overnight at 4°C. Samples to be stained using α-glucosepane were first incubated overnight at room temperature with NaBH₄ (80mM) in order to reduce glutaraldehyde reaction products that could cross-react with the antibody [23]. Samples were then were incubated with primary antibody as previously described. After incubation with primary antibodies, slides were washed and incubated with H₂O₂ (3%) for 10 minutes. Slides were then rinsed and incubated for 1 hour at room temperature with the appropriate horseradish peroxidase polymer-conjugated secondary antibody (Abcam; Cambridge, United Kingdom). Slides were then rinsed and incubated for 8 minutes at room temperature with 3,3’ Diaminobenzidine substrate (Abcam; Cambridge, United Kingdom). Slides were then counter-stained using regressive hematoxylin staining, dehydrated, and coverslipped.

2.5.2. Two-photon Microscopy

All two-photon microscopy scans were performed using a custom Prairie Technologies Ultima Multiphoton Microscopy on Olympus BX-61 upright microscope (Olympus; Tokyo, Japan). This system is equipped with GaAsP photomultiplier tubes and a tunable femtosecond laser (Spectra Physics, MaiTai DeepSee). The system is capable of providing multi-color imaging including second harmonic generation (SHG) imaging. Tissues designated for two-photon microscopy scans were mounted on chamber slides and immersed in PBS. Scans were performed with excitation at 980nm and scanning from the surface to as deep as could be resolved at 5μm steps. Crimp period was quantified by pixel measurement of distance between crimp bands using custom Matlab (Mathworks; Natick, MA) scripts.

2.6. Uni-axial Testing and Related Data Analyses of Collagen Sponges

Dog bone shaped samples (8mm wide that narrows to 4mm wide in the middle, 25mm long) were punched from the glutaraldehyde-fixed CS for the mechanical tests. The samples were each incubated for 7 days or 28 days at 37°C shaking at 110RPM in 5ml per sample of PBS or PBS with either: glucose (100mM), glyoxal (50mM), BSA (5%), glucose (100mM) + BSA (5%), or glyoxal (50mM) + BSA (5%). All incubations contained sodium azide (0.1%) to maintain sterility. Samples were extensively rinsed with PBS and stored in PBS at 4°C until the mechanical tests. Prior to testing, samples’ cross-sectional area was calculated by measuring the cross-section’s length and width using a micrometer. The mechanical testing consisted of: preloading to 0.03N (~3x the sample wet weight), 10 cycles at 1Hz of preconditioning going from 0% to 10% strain, a relaxation test rapidly (14mm/s) extending to 20% strain and holding for 60s, a 2-minute recovery period at 0 strain, then a slow extension (0.05mm/s) to failure. All mechanical testing was performed at the Penn Center for Musculoskeletal Disorders (University of Pennsylvania; Philadelphia, PA) on an Instron® 5542 system (Instron; Norwood, MA). Load (N) and extension (mm) measurements were recorded alongside images of samples following preloading. Samples that did not break cleanly in the center or that slipped during testing were excluded.

The time-load-extension data were analyzed using custom Matlab (Mathworks; Natick, MA) scripts to extract viscoelastic, linear elastic, and failure mechanical properties. Zero strain length was determined by pixel measurement of clamp to clamp distance following preloading. Viscoelastic metrics calculated were the degree of relaxation (the fraction of stress attenuation during relaxation, 1 - σ_equilibrium/σ_peak) and time constant of relaxation curve. Linear elastic metrics calculated were the elastic modulus (slope of the stress-strain curve in the linear loading region, expressed as MPa) and the engineering strain at start of the linear loading region of the stress-strain curve. Failure metrics calculated were the ultimate tensile strength (the stress at mechanical failure, expressed as MPa) and the engineering strain at failure. The mechanical data were normalized to the PBS incubation data for each repetition before comparisons.

2.7. Statistical Methods

The significance of the effects of BSA presence on 28-day glucose and glyoxal incorporation and dissociation was determined by two-sample t-test (2.3.1). A two-sample t-test was also used to compare the 28-day incorporation of glucose and glyoxal into BSA (2.3.2) and the 28-day incorporation of glucose and glyoxal into fresh and glutaraldehyde fixed bovine pericardium (2.3.3). It was also considered whether the assumptions of the t-test are warranted and the nonparametric analogue, Wilcoxon rank sum, was used to verify significance. To evaluate the significance of the change in dry mass at 28 days, a paired t-test determined the significance of the mass change from the initial mass for each group, Dunnett’s method was used to compare all BSA groups to the PBS control, and Tukey’s honestly significant difference (HSD) test compared each BSA incubation to one another (2.3.3). Tukey’s HSD test compared the crimp period from each incubation to one another (2.5.2). Dunnett’s method was used to determine the significance of the changes from the PBS control for each of the metrics from the mechanical tests on the CS (2.6). A two-sample t-test was used to compare the 7-day PBS mechanical properties to the 28-day mechanical properties (2.6). For all statistical tests, p<0.05 was considered significant. All data are expressed as mean ± standard deviation.

3. RESULTS

3.1. Model glycation studies

These experiments sought to characterize the kinetics and extent of glucose and glyoxal incorporation into and dissociation from BP and CS in order to model AGE formation in these materials. BP was chosen for use in these experiments because of its use in BHV leaflets, and CS was investigated as a model collagenous material, hypothetically comparable to BP, and better suited for the mechanical studies presented later in this paper. The approach for these studies was to assess the overall potential for AGE formation using both ¹⁴C-glucose and ¹⁴C-glyoxal incorporation as exemplary glycation reagents. The rationale for these studies was based on established AGE formation reactions involving glyoxal, a reactive intermediary derived from glucose that specifically reacts with lysine and arginine residues[15]. Furthermore, glyoxal is involved in the formation of carboxy-methyl-lysine (CML), an AGE involved in the pathophysiology of diabetes and other diseases[24–26]. Serum proteins are AGE modified[27,28]; thus, their influence on ¹⁴C-glucose and ¹⁴C-glyoxal incorporation was modeled in these studies of BP and CS using serum albumin, the most abundant serum protein, at physiologic concentrations in the specific protocols.

Glucose incorporation, without BSA, was observed after 24 hours to reach 11% of the total in BP (Figure 1A) and 8% in CS (Figure 1C). Glucose incorporation continued to increase through 28 days in BP (Figure 1A) and CS (Figure 1C). Glucose incorporation in the presence of BSA while reduced compared to incubations without BSA (Figure 1A) had comparable kinetics to glucose alone in both BP and CS. Co-incubation with BSA resulted in 26% less glucose content in BP (Figure 1A) and 24% less in CS at 24 hours (Figure 1C). After 28 days, the attenuation in glucose levels for the co-incubation with BSA increased to reach 31% less glucose in BP (Figure 1A) and 54% less glucose in CS (Figure 1C). Glucose dissociation from BP previously incubated in ¹⁴C-glucose with or without BSA was similar with or without BSA up to day 7, where the glucose only incubation demonstrated diminished dissociation over time, while the glucose-BSA co-incubation maintained a higher dissociation rate (Figure 1C). By 28 days, the glucose-BSA co-incubation showed a significantly greater relative dissociation relative to the glucose only incubation (Figure 1B). The glycation of BSA by glucose was a relatively faster process than of BP or CS with three times as much glucose incorporated after 28 days (Figure 1D). While the incorporation rate attenuated slightly after 7 days, there is no indication of approaching saturation (Figure 1D).

Figure 1. Glycation kinetics of glutaraldehyde pretreated bovine pericardium (BP) compared to glutaraldehyde pretreated collagen sponges (CS) and bovine serum albumin (BSA): ¹⁴C-glucose and ¹⁴C-glyoxal incorporation studies with and without the presence of BSA to simulate serum protein conditions.

A) ¹⁴C-glucose (100mM) uptake by BP with and without the presence of BSA (5%). BSA was associated with significant decrease in incorporation for all days > 0 (p<0.02).

B) ¹⁴C-glucose (100mM) dissociation from BP with and without the presence of BSA (5%).

C) ¹⁴C-glucose (100mM) uptake by CS with and without the presence of BSA (5%). BSA caused significant decrease in incorporation for all days > 0 (p<0.02).

D) ¹⁴C-glucose (100mM) uptake by BSA (5%).

E) ¹⁴C-glyoxal (50mM) uptake by BP with and without the presence of BSA (5%). BSA caused significant decrease in incorporation for all days > 0 (p<0.001).

F) ¹⁴C-glyoxal (50mM) dissociation from BP with and without the presence of BSA (5%).

G) ¹⁴C-glyoxal (50mM) uptake by CS with and without the presence of BSA (5%). BSA caused significant decrease in incorporation for all days > 0 (p<0.005).

H) ¹⁴C-glyoxal (50mM) uptake by BSA (5%).

Data shown are means of 10 replicates for 1A and 1E and means of 5 replicates for 1B-D and 1F-H. Error bars when possible to display indicate standard deviation. Significance was determined by two sample t-tests comparing the time points indicated.

Glyoxal incorporation at 24 hours was 56% of the 28-day amount in BP (Figure 1E) and 60% of the of the 28-day amount in CS (Figure 1G). Glyoxal incorporation leveled off after 7 days and plateaued at 28 days in both BP and CS (Figure 1E&1G). Glyoxal incorporation in the presence of BSA appeared to also level off by 7 days and plateau at 28 days in both BP and CS (Figure 1E&1G). Co-incubation with BSA resulted in 33% less glyoxal in BP (Figure 1E), and 29% less glyoxal in CS at 24 hours (Figure 1G). The plateau levels reached at 28 days were 31% lower in BP (Figure 1E) and 20% lower in CS when co-incubated with BSA (Figure 1G). Similar to glucose, glyoxal fractional dissociation was similar with or without BSA up to day 7 at which point the glyoxal incubation starting to level off while the glyoxal-BSA did not (Figure 1F). At 28 days, the glucose-BSA co-incubation showed a significantly greater fraction of glyoxal dissociated relative to the glyoxal only incubation (Figure 1F). Glycation of BSA by glyoxal was a relatively more rapid process, similar to BP and CS, with 50% of the glyoxal added after 24 hours (Figure 1H). Glyoxal incorporation reached saturation by day 28 (Figure 1H).

3.2. Glycation of bovine pericardium occurs regardless of glutaraldehyde pretreatment

Glutaraldehyde crosslinking is the universally used pre-treatment step for preparing bioprosthetic heart valves. However, the effects of glutaraldehyde pretreatment on AGE formation have not been previously studied. Glutaraldehyde reacts with primary amines, principally lysine, to form Schiff bases and heterocyclic crosslinks. This could hypothetically block glycation sites arising from lysyl amine reactions. To assess the effect glutaraldehyde-crosslinking has on glycation capacity, experiments were performed to compare ¹⁴C-glucose and ¹⁴C-glyoxal incorporation into fresh BP and glutaraldehyde-fixed BP.

Non-crosslinked BP’s 28-day glucose incorporation was 61% greater than the glutaraldehyde-fixed BP’s 28-day glucose incorporation (Figure 2A). In the co-incubation with glucose plus BSA, glucose had 103% greater incorporation into the non-fixed BP than observed with glutaraldehyde-fixed BP (Figure 2A). Conversely, non-fixed BP had significantly less 28-day glyoxal incorporation than the glutaraldehyde-fixed BP; 9% less without BSA co-incubation, and 13% less with BSA (Figure 2B).

Figure 2. Glycation of non-crosslinked bovine pericardium (BP) compared to glycation of glutaraldehyde-crosslinked BP: ¹⁴C-glucose and ¹⁴C-glyoxal incorporation studies.

A) ¹⁴C-glucose (100mM) incorporation into non-fixed or glutaraldehyde pretreated BP, with or without co-incubation in BSA (5%), was studied as above (Figure 1). Glutaraldehyde pretreatment significantly attenuated glucose incorporation both with and without BSA co-incubation (p<0.001).

B) ¹⁴C-glyoxal (50mM) incorporation with or without BSA (5%) into non-fixed BP versus glutaraldehyde-fixed BP was also studied. Glutaraldehyde pretreated BP demonstrated a small but significant increase in glyoxal incorporation compared to non-fixed BP.

Significance was determined by two sample t-tests.

3.3. Bovine serum albumin mass uptake by bovine pericardium

BP samples were incubated in BSA, BSA with glucose, and BSA with glyoxal to model BHV serum protein exposure under different glycation conditions with the resulting BP mass change quantitated. Albumin addition was quantified as the percent change in the dry (Figure 3) and wet weight (not shown) of BP samples from the dry and wet weight before a 1- to 28- day incubation. The dry weight data showed a cumulative loss of mass in the PBS incubation by day 28 (Figure 3A). Incubation in BSA offset this loss of dry weight, causing a net increase by day 28 (Figure 3A). Glucose presence in the glucose-BSA co-incubation demonstrated no significant effect on the dry mass change relative to BSA by itself (Figure 3B). By contrast, glyoxal presence in the glyoxal-BSA co-incubation significantly increased the dry weight gain relative to both the PBS incubation and the BSA-only incubation after 28 days (Figure 3B). The change in wet weight data was inconsistent, with too great a variance for statistically significant differences between treatments to be observed.

Figure 3. The changes in mass of glutaraldehyde-crosslinked bovine pericardial samples (BP) using glycation conditions in the presence of bovine serum albumin (BSA) to simulate serum protein exposure.

Change in dry mass of BP samples over 28-day time course incubating in either: PBS (control), BSA (5%), BSA (5%) with glucose (100mM), or 5% BSA (5%) with glyoxal (50mM). Data shown are percentages of the change in weight from the starting measures as well as calculating the corresponding nmol BSA per mg added. Data shown are means of 5 replicates. Error bars indicate standard deviation.

3.4. Protein glycation as demonstrated by immunostaining of bovine pericardium

These studies sought to characterize the morphologic distribution of protein glycation resulting from the incubation conditions described above. Glycation precursors, such as glucose and glyoxal, have a large family of intermediaries, including Amadori products, on the pathway to forming the mostly irreversible AGE[15]. The antibodies used in these experiments were specific for: CML, an AGE derived from glyoxal[15,28]; general AGE formation; and glucosepane, the most common physiological crosslinking AGE[29]. The BP samples were incubated for 28 days to correspond with the radioactive assays (Figure 1) and two-photon microscopy scans (Figure 5).

Figure 5. Two-photon microscopy images along with crimp period distance of glutaraldehyde-fixed bovine pericardium following in vitro incubations:

Samples were incubated for 24 hours (A-F) or 28 days (G-L) in the following media: A&G) PBS, B&H) bovine serum albumin (BSA, 5%), C&I) glucose (100mM), D&J) glucose (100mM) plus BSA (5%), E&K) glyoxal (50mM), and F&L) glyoxal (50mM) plus BSA (5%). Scale bar is 50μm. Arrows indicate typical crimp spacing where bands are discrete. M) Comparison of crimp period [n=5]. Error bars indicate standard deviation. Significance was determined using Tukey’s honestly significant difference.

An array of representative micrographs of immunohistochemistry-stained incubated BP following a 28-day incubation are shown in Figure 4. The α-CML antibody lead to moderate staining in the glyoxal incubated sample (Figure 4E) and heavy staining in the glyoxal-BSA co-incubated sample (Figure 4F) relative to the PBS control (Figure 4A). The α-AGE antibody prompted increased staining only in the glucose incubated sample (Figure 4I) relative to the PBS control (Figure 4G). Likewise, the α-glucosepane antibody demonstrated increased staining over the PBS control only in the glucose incubated samples (Figure 4O vs 4M). The samples exposed a BSA incubation (Figure 4T, 4V and 4X) each showed significant staining with the α-BSA antibody relative to the PBS control (Figure 4S) with the darkest staining occurring in samples from the glyoxal-BSA co-incubation (Figure 4X).

Figure 4. Formation of advanced glycation end products (AGE) in glutaraldehyde-crosslinked bovine pericardium following in vitro incubation: glycation-specific immunohistochemistry micrographs of glutaraldehyde-crosslinked bovine pericardium incubated for 28 days using the conditions indicated:

A-F) Staining for carboxy-methyl-lysine (CML); G-L) Staining for general AGE formation. M-R); Staining for glucosepane formation. S-X); Staining for bovine serum albumin (BSA). Immunoperoxidase staining was used with a substrate of 3,3'Diaminobenzidine; Incubation concentrations were: bovine serum albumin (BSA, 5%), glucose (100mM), and glyoxal (50mM). Original magnification 100x. 200μm scale bar.

3.5. The effects of protein glycation on BP collagen structure—two photon microscopy results

To study the effects of glycation and serum protein exposure on BHV’s collagen microarchitecture, two-photon microscopy scans were performed on BP samples following in vitro incubations. BP samples after 24 hours of incubating consistently demonstrated collagen fibers were aligned with a tight crimp period and distinct crimping bands under all conditions (Figure 5A–F). Incubating BP in PBS for 4 weeks caused some alterations in the structure, but no significant loss of alignment nor a change in crimp period relative to 24 hours of PBS incubation (Figure 5G & 5M). After 4 weeks of BSA exposure, there is a significant loss of identifiable crimp and an increase in the crimp period relative to the image at 24 hours and 4 weeks of PBS, however, the fibers retain the bulk of the orientation and still have distinct crimp bands (Figure 5H & 5M). Glucose exposure, by itself, had relatively mild effects that were similar to BSA exposure, some loss of crimp, a significant increase in crimp period relative to the 24 hours scans and 4 weeks of PBS exposure, but the BP retained collagen alignment and banding (Figure 5I & 5M). However, glucose-BSA co-incubation produced a dramatic disruption in the collagen microarchitecture, almost completely eliminating any crimp bands with significant collagen misalignment (Figure 5J). After 4 weeks of glyoxal exposure, the fibers are significantly maligned and the crimp bands are completely lost (Figure 5K). The glyoxal-BSA co-incubation produced the greatest modification of the structure after 4 weeks with compete loss of any crimp banding (Figure 5L).

3.6. The effects of glycation on the mechanical properties of collagen sponges

To evaluate the effects glycation and serum proteins have on glutaraldehyde-fixed collagen’s mechanics, uni-axial mechanical tests were performed on CS samples following 7-day and 28-day in vitro incubation. CS was selected as there was no dominant directionality to the collagen fibers to complicate uni-axial tests. Furthermore, CS was found to have highly consistent and uniform mechanical and chemical properties in pilot studies. The mechanical tests consisted of a relaxation test, followed by a recovery period, then a slow extension to failure.

The effects of each treatment on the viscoelastic, linear elastic, and failure mechanical properties of the collagen sponges are expressed as changes from the PBS incubation (Figure 6). By 7 days, glucose by itself caused a significant loss in relaxation (Figure 6B) with no significant changes in any other metric. Glyoxal exposure by itself produced a significant loss of relaxation (Figure 6B), a significant decrease in the strain to reach failure (Figure 6C), and a significant increase in stiffness (Figure 6D). BSA by itself caused significant increase in elastic modulus (Figure 6D) and increase in ultimate tensile strength (Figure 6E). The glucose-BSA co-incubation was similar to glucose alone with the only significant change being a loss of relaxation (Figure 6B). The glyoxal-BSA co-incubation had effects similar to the BSA-only incubation with a significant increase in stiffness (Figure 6D) and a marginal increase in ultimate tensile strength (p=0.22, Figure 6E) relative to the PBS control.

Figure 6. Changes of mechanical properties determined with uniaxial testing of collagen sponge samples following 7-day or 28-day incubations. Results are normalized relative to phosphate buffered saline (PBS) controls.

A&F) Consolidated stress-strain curves during failure loading. B&G) Degree of relaxation, relative stress attenuation during relaxation. C&H) Strain at mechanical failure. D&I) Elastic modulus. E&J) Ultimate tensile strength.

Incubation conditions were: PBS (control) [7-day n=46, 28-day n=25], glucose (100mM) [7-day n=21, 28-day n=23], glyoxal (50mM) [7-day n=19, 28-day n=15], bovine serum albumin (BSA, 5%) [7-day n=38, 28-day n=23], glucose (100mM) plus BSA (5%) [7-day n=16, 28-day n=28], and glyoxal (50mM) plus BSA (5%) [7-day n=36, 28-day n=15]. Data plotted are means with 95% confidence intervals. Significance was determined by Dunnett’s method comparing to PBS.

By 28 days, there was stronger separation in the failure curves of the groups with glucose or glyoxal from the PBS when compared to the curves at 7 days (Figure 6A vs 6F). Significant changes occurred in the baseline mechanical properties with a 12% decrease in degree of relaxation, a 12% relative increase in failure strain, and an 18% decrease in elastic modulus when comparing 7-day and 28-day PBS incubations. This change resulted in no treatments showing significant changes to viscoelastic relaxation relative nor ultimate tensile strength relative to the 28-day PBS control (Figure 7G). Glucose alone caused a significant increase in elastic modulus (Figure 6I). Glyoxal, by itself, caused significant decrease in failure strain (Figure 6H) and increase in elastic modulus (Figure 6I), which both alterations twice the change observed at 7 days. BSA by itself caused no significant changes in mechanical properties at 28 days. The glucose-BSA co-incubation induced both decreased failure strain (Figure 6H) and increased stiffness (Figure 6I). The glyoxal-BSA co-incubation caused significant decrease in failure strain (Figure 6H) and increase in elastic modulus (Figure 6I)

4. DISCUSSION

The presented studies provided multifaceted information concerning the susceptibility of BHV to glycation-related pathophysiology. Our results documented BHV glycation kinetics, processes by which glycation contributes to BHV SVD, and the deleterious effects of serum albumin that contribute to BHV glycation. Glucose effects modeled in vitro in the present studies were not examined in our previous report[23]. However, glucose incorporation into BP occurred at comparable nanomolar levels incorporation as glyoxal (Figure 1A versus Figure 1E), and produced IHC results in BP in vitro that are comparable to failed clinical explants (Figure 4) [23]. These in vitro BP studies also demonstrated disrupted collagen structure (Figure 5) of BP with a morphology comparable to clinical explants[23], and caused significant increase in stiffness (Figure 6I).

These model system studies demonstrated that BHV are highly susceptible to rapid and mostly irreversible glycation (Figure 1) by both glucose and glyoxal, with or without serum albumin. The present results concerning the reaction kinetics of glucose and glyoxal with BP (Figure 1A&E), while not previously reported in BHV studies, are comparable to previous model system results by others using radiolabeled glycation reagents [30–34]. For example, carboxymethyl lysine formation from ¹⁴C-glyoxal in bovine serum albumin has been shown in studies by others to have comparable reaction kinetics to those observed in our BHV investigations[31]. In addition, prior studies of ¹⁴C-glucose incorporation into retinal basement membranes[34] demonstrated reaction rate results that were also comparable to those for BP in our experiments. These comparisons indicate that the susceptibility of BHV to glycation occurs through mechanistic pathways that are operative in other pathophysiologies.

Non-crosslinked BP demonstrated greater nanomolar levels of glucose incorporation and comparable glyoxal incorporation as glutaraldehyde pretreated BP (Figure 2). The expectation was that prior glutaraldehyde fixation by reacting with lysine amino residues would decrease the number of reactive sites and result in diminished incorporation of both glucose and glyoxal compared to non-fixed BP. The fact glucose incorporation was significantly reduced in glutaraldehyde fixed BP indicates that AGE formation was occurring at sites other lysines that had reacted with glutaraldehyde. The glyoxal results were surprising, since glyoxal is a dialdehyde, and the data showed that glyoxal incorporation into either non-crosslinked or glutaraldehyde-fixed tissues were comparable. This result suggests that glyoxal mediated glycation reactivity and complexity beyond that occurring with glutaraldehyde fixation. AGE reaction products are complex and involve interactions with other amino acids, such as arginine and histidine[15,16,35], besides free amino groups, such as lysine amino groups, which are the only substrate for glutaraldehyde. Glycation of proteins in living systems, unlike devitalized heterograft tissue, is dependent on protein turnover and is not comparable to the present findings.

The dry mass change of BP showed a significant baseline loss in mass, BSA caused significant increase in mass, and glyoxal presence significantly enhanced the mass increase while glucose showed no similar enhancement (Figure 3). This baseline mass loss could be attributed to residual proteases in the BP which have been shown to remain active through glutaraldehyde-fixation[36]. The enhancement by glyoxal may be the result of the glycated BSA being crosslinked by glyoxal, a dialdehyde, within the collagen substrate. The amount of glyoxal bound to BP at 28 days in the glyoxal-BSA co-incubation would only cause a 0.06% increase in mass, making it reasonable to assume all the mass change was from BSA addition. Similarly, if all the glucose molecules were fully intact, the added glucose from the glucose-BSA co-incubation would result in only a 0.1% increase in BP mass. It is likely that the glucose reaction products resulting from Maillard and Amadori reactions are low molecular weight, and thus do not reflect hexose binding, thus resulting in even smaller mass contribution. While glucose should promote BSA binding via similar mechanisms as glyoxal and there were comparable amounts of glucose and glyoxal bound to BSA by day 28 (Figure 1). The IHC results show that glucose and glyoxal preferentially form different AGE species (Figure 4). Thus, the AGE involved in crosslinks to bind BSA to BP may be less likely to form from glucose than glyoxal. By adding mass to BHV due to BSA uptake, hypothetically the mechanical load over time will increase, resulting in compromised hemodynamic function.

There was significantly reduced glucose and glyoxal incorporation into BP and CS with BSA co-incubations (Figures 1&2). The mechanisms responsible for this are complex, and the diminished glucose and glyoxal uptake were most likely due to BSA reactively binding to sites that would otherwise have reacted with glucose or glyoxal. BSA has a multitude of amino acids capable of engaging in the formation of AGE, including lysines, arginines, and histidines. This is supported by our results showing the glycation kinetics of BSA (Figure 1D&H). This kinetic information (Figure 1D&H) also allowed us to rule out BSA sequestering enough glucose and glyoxal to affect the glycation kinetics of BP as the fraction of glucose and glyoxal bound to BSA was significantly smaller than the change in glycation capacity of BP.

The IHC micrographs (Figure 4) provided insight into the formation of CML, general AGE, and glucosepane as well as confirming BSA addition to the BP in the co-incubations with BSA. CML stained most strongly in the glyoxal-BSA co-incubation despite a smaller amount of glyoxal binding to the substrate relative to BP was less than with glyoxal by itself. This effect may be due to BSA itself preferentially forming CML as the glyoxal-BSA co-incubation does have the largest amount of BSA addition (Figure 3). The general AGE antibody only showed some weak staining from glucose exposure (Figure 4I) and the staining was much fainter than our previous in vivo explants[23]. This is likely a result of the AGE antibodies being produced through immunization of rabbits using BSA that had been oxidized by hydrogen peroxide. Similar oxidative stresses would occur commonly in in vivo models but less so in our in vitro system, producing weaker staining. The most intense BSA staining in the glyoxal-BSA co-incubation (Figure 4X) is consistent with the observation of the greatest mass increase in the glyoxal-BSA co-incubation (Figure 3) and supported the idea that glyoxal enhanced BSA addition to BP.

SHG analyses detailed how collagen structural alters in BP due to glycation and serum albumin uptake; these alterations included collagen malalignment and loss of collagen crimp (Figure 5). These results highlighted the synergistic effects of serum exposure and glycation as the co-incubations of glucose with BSA or glyoxal with BSA showed more severe degradation than glucose or glyoxal alone, respectively. Collagen crimp, that structurally represents collagen’s folding during its unloaded state, is important to BHV function and its loss causes is indicative of diminished compliance and increased susceptibility to mechanical injury[37]. This study’s collagen modifications were comparable to the changes observed using SHG in our clinical study evaluating failed clinical BHV tissue, in vivo rat subdermal explanted BP, and in vitro pulse duplicator experiments[23]. Taken together these observations validate our current in vitro models as reproducing representative alterations in collagen structure, comparable to clinical explant BHV leaflets with SVD[23].

CS, as a model for the BP collagen network, made possible uniaxial mechanical testing that documented alteration of the viscoelastic and linear elastic properties in CS due to AGE formation and serum protein incorporation (Figure 6). The results showed progressive increase of stiffness and accelerated loss of viscoelasticity resulting from glycation. This is most likely the result of new crosslinks, particularly with BSA exposure, being introduced into the substrate changing how the collagen is loaded. These crosslinks also impede the collagen fibers from realigning which attenuates the relaxation behavior. The stiffness change for glyoxal doubled from day-7 to day-28 (Figure 6) which highlights how this is a time-dependent degenerative process as this is much greater than the change in the glyoxal incorporation (Figure 1). Despite BSA presence significantly attenuating glucose and glyoxal incorporation (Figure 1), BSA provided no protective effects against structural degeneration (Figure 5) nor mechanical degeneration (Figure 6). This underscores the importance of the glycation state of BSA to BHV SVD. Increased stiffness and loss of viscoelasticity means that through a cardiac cycle, the instantaneous load on glycated BHV tissue would be greater and the material would have less capacity to disperse the stress. These mechanical results provide a mechanistic explanation for our previous in vitro pulse duplicator findings that demonstrated AGE formation caused hydrodynamic dysfunction of clinical grade tricuspid BHV that resulted in decreased effective orifice area and increased pressure gradient[23]. These dysfunctional effects are directly related to clinical SVD. The present results suggest that the hydrodynamic changes observed in the pulse duplicator studies are likely due to increased stiffness and loss of viscoelasticity making the valve leaflets more resistant to cyclic opening.

There are some caveats and limitations of this study to be noted. The concentrations of reagents used in the glycation studies were adjusted to provide an accelerated model system. An isotonic 5% BSA solution was used. However, glyoxal was utilized at 50mM, based on prior model studies using this reagent [23,31,38,39], rather than the physiologic concentration (~154nM)[40]. Glucose, at 100mM, was also used at diabetic level, hyperglycemic concentrations rather than physiologic (~5.6mM)[41] levels. Nevertheless, this glucose concentration was representative of concentrations used in previous accelerated in vitro glycation studies by others[32–34,42–44]. Furthermore, glucose was not investigated in our prior AGE-BHV study of clinical explants[23], and is of strong relevance because of the more rapid SVD BHV failure process in diabetics[14]. Thus, the glucose results in the present results addressed the important contributions of glucose mediated glycation of BHV to SVD. The IHC assays of the study focused on well known, representative immunostaining targets reported in other glycation studies and these were: AGE, carboxymethyl lysine and glucosepane. These same IHC markers were endpoints in our clinical study[23] as well as other clinical studies[25,45–50], and thus have mechanistic and clinical relevance. Uniaxial testing of CS showed that glycation results in accelerated loss of viscoelasticity and increased stiffness (Figure 6). BP was not studied because of the well-known technical challenges of its anisotropic nature of BP[51–53]. This property was shown in our SHG scans of BP and the challenge of orienting the tissue would be compounded by loss of collagen alignment observed in the glycated tissues (Figure 5). As discussed earlier, the present uniaxial testing observations account for the degeneration of the hydrodynamic properties of BHV following glycation and serum protein modifications observed in our pulse duplicator study[23].

The results of this study have several implications addressing SVD. AGE-mitigating strategies for use with BHV have not been previously investigated; the presents results indicate anti-AGE agents may be of interest. Currently available anti-AGE treatments include AGE inhibitors and AGE breakers. AGE inhibitors are pharmacologic agents, e.g. aminoguanidine and pyridoxamine, that prevent the formation of AGE by mechanisms such as scavenging the reactive carbonyl intermediates and blocking the oxidation of the Amadori intermediate, respectively[54]. AGE breakers, e.g. phenacyl-thiazolium bromide and alagebrium, function by cleaving glycation crosslinks and have been demonstrated to have the capacity to prevent or reverse glycation modifications in vascular disease[55,56]. The present study’s model systems are capable of evaluating in vitro those treatments’ efficacy at preventing or reversing the glycation modifications of BHV. Furthermore, our studies demonstrated the impact of serum protein infiltration in vitro on BHV structure and mechanical properties, and provided insights to explain clinical and experimental observations. Lastly, diabetic patients are at higher risk for earlier SVD[14,57,58]. Since the present studies showed significant effects of glucose on structure and mechanical properties of BP and CS, future in vivo model studies using diabetic animals represent an important direction. A notable example of such a study investigated both collagen and elastin scaffolds implanted subdermally in control and diabetic rats[59]. The samples explanted from diabetic rats demonstrated significantly increased biaxial stiffness compared to controls.

5. CONCLUSIONS

The results of the present study support the hypothesis that glycation and serum protein infiltration can contribute to SVD pathophysiology and lead to the following conclusions: 1) BP and CS are susceptible to rapidly progressive glycation and serum albumin incorporation; 2) BSA, with our without glyoxal or glucose, disrupts collagen structure; 3) Glucose alone, while not dramatically altering collagen structure after 28 days compared to PBS, does alter viscoelastic properties; 4) Similarly, collagen’s uniaxial properties are significantly altered by both glycation and serum albumin incorporation. Taken together glycation and serum protein infiltration contribute to SVD via the structural changes in collagen that were observed in these studies and the related disruption of viscoelastic and linear elastic properties.

Rock Acta Significance Statement 5 6 2020b

Bioprosthetic heart valves (BHV) derived from glutaraldehyde-fixed heterografts, typically bovine pericardium or porcine valves, are widely used for heart valve surgery. BHV have desirable hemodynamic properties and catheter-deployed BHV shifted the treatment paradigm, becoming an increasing fraction of valve replacement procedures. However, BHV functional lifespan is limited due to leaflet structural degeneration, a process that is not fully understood. This study evaluated advanced glycation end products’ (AGE) modification of BHV leaflets as a mechanism of BHV degeneration. While AGE have been linked with other soft-tissue pathologies, this is the first study to show that glucose, the most physiologically relevant AGE precursor, can induce pathological changes in glutaraldehyde-fixed collagen’s microarchitecture and biomechanics. Therefore, anti-glycation strategies merit investigating to mitigate BHV degeneration.

Table 1.

Significance tests performed on changes in dry mass of of glutaraldehyde-crosslinked bovine pericardial samples (BP) using glycation conditions in the presence of bovine serum albumin (BSA) to simulate serum protein exposure (Figure 3).

Mass Uptake Comparison	Statistical test	p-value
Initial vs. Day-28: PBS	Paired t-test	0.028
Initial vs. Day-28: BSA Only	Paired t-test	0.021
Initial vs. Day-28: BSA + Glucose	Paired t-test	0.020
Initial vs. Day-28: BSA + Glyoxal	Paired t-test	0.009
Day-28 Change: PBS vs. BSA Only	Dunnet’s method	<0.001
Day-28 Change: PBS vs. BSA Only	Dunnet’s method	<0.001
Day-28 Change: PBS vs. BSA Only	Dunnet’s method	<0.001
Day-28 Change: BSA only vs. BSA + Glucose	Tukey’s HSD	0.77
Day-28 Change: BSA only vs. BSA + Glyoxal	Tukey’s HSD	<0.001
Day-28 Change: BSA + glucose vs. BSA + Glyoxal	Tukey’s HSD	<0.001

TABLE OF ABBREVIATIONS

AGE

advanced glycation endproducts

BHV

bioprosthetic heart valves

BP

bovine pericardium

BSA

bovine serum albumin

CML

carboxy-methyl-lysine

CS

collagen sponge

IHC

immunohistochemistry

PBS

phosphate buffered saline

SHG

second harmonic generation

SVD

structural valve degeneration