Novel xeno-free human heart matrix-derived three-dimensional scaffolds

Ischemic heart disease is currently the leading cause of death in the United States. Yearly, ~525,000 Americans will experience their first myocardial infarction (MI) and 190,000 will have a recurrent attack [1]. During MI, blood flow to a section of the heart is cut off, subsequently contributing to local acute tissue ischemia and the death of approximately one billion cardiomyocytes [2]. Furthermore, the resulting necrotic tissue acute zone is typically replaced by dense fibrous scar, impeding cardiac tissue regeneration and restoration of normal heart function. Due to limited intrinsic repair potential exhibited by the heart, much research has been directed towards improved methods for cardiac tissue regeneration. Towards this end, cell therapy strategies—including stem cell delivery—are now actively developed, with some showing promising results for restoring cardiac function [3, 4]. Clinical trial data has shown a reduction in myocardial scarring, a degree of healthy myocardial regeneration, or an increase in left ventricular ejection fraction in response to post-infarction cell therapy [5‐7]. However, cell survival, engraftment, and control of cell differentiation state are major challenges associated with the local injection of cell populations for heart repair [8]. Additionally, cells injected after MI encounter a harsh environment that promotes apoptosis, and the majority of injected cells die within 4 days [9]. Furthermore, cells delivered through systemic circulation or coronary arteries have low rates of localization to the infarcted myocardium [9, 10], and it has been reported that only 10–15% of transepicardially injected cells are retained in the myocardium [11]. The majority of cells are lost to filter organs such as the liver and spleen [12]. These limitations must be addressed to increase treatment efficacy.

Autologous human myoskeletal cell suspensions injected into human cardiac MI sites have produced spontaneous arrhythmias [13, 14]. This pathology plus the long-standing inability to retain sufficient amounts of injected cells in MI tissue sites has prompted increasing use of biomaterials-based approaches to such cardiac tissue cellular therapeutic strategies. Notably, cell delivery modalities include gel-based injectable cell-containing carriers [15] and living cell sheets to retain cells at the injured site. The Okano group pioneered beating contiguous cardiomyocyte cell sheets for use as heart patches [16] and demonstrated their utility in porcine [17] and rat [18] models. Bel et al. also used cell sheets, but found a multilayer construct with increased ECM content provides better efficacy [19]. Chein’s group isolated progenitors that could ultimately be used to create 3D viable cell constructs that could regenerate a single portion of a damaged heart [20]. However, these groups have focused on cellular delivery, rather than on exploiting autologous extracellular matrix (ECM) either as a bulking agent or a therapeutic cell carrier. ECM proteins have also been used to regenerate damaged myocardium post-MI. Cardiac ECM has significant influence in cardiac development, and undergoes several changes during cardiac disease progression [21, 22]. ECM proteins provide essential biochemical and mechanical cues to guide cell differentiation and phenotype [23]. Although the influence of extracellular matrix in cardiac regeneration is not well understood, many studies now begin to examine this relationship. For example, ECM isolated from decellularized skeletal and cardiac muscle has been shown to improve cardiomyocyte differentiation and maturation [24], and periostin was found to increase myocyte proliferation and enhance cardiac repair [25]. Another study showed that isolated fetal cardiac ECM promoted myocyte adhesion and expansion better than adult cardiac ECM [26]. Significantly, extracellular matrix materials have been used in a variety of cardiac tissue engineering approaches. For instance, decellularized cardiac ECM in the form of an injectable hydrogel has been shown to preserve post-MI cardiac function in preclinical animal studies [27, 28]. Furthermore, decellularized cardiac biomaterials have been used as acellular patches for repair of ventricular defects in animal models [29, 30], and several groups are investigating use of cardiac ECM to promote cardiac phenotypes in 3-dimensional (3D) constructs in vitro [31‐33]. These studies indicate that cardiac-derived extracellular matrix materials have a significant impact on heart regeneration. However, these 3D scaffolds are made of decellularized cardiac tissue or hydrogels, both of which have limited geometric and mechanical properties.

In addition to biochemical stimuli from ECM materials, mechanical properties and scaffold geometry may be equally important for promoting mature cardiac tissue. For instance, materials of appropriate stiffness (~10 kPa) [34] and materials patterned to promote cardiomyocyte orientation [35] have both been reported to improve cardiomyocyte differentiation and maturation. Though cardiac ECM hydrogels have been researched, they inherently lack sufficient mechanical properties [36]. Furthermore, for 3D tissue engineering constructs, high scaffold porosity enables cell infiltration and transport of nutrients and wastes [34, 37]. However, decellularized ECM biomaterials have inherently fixed porosity (constituting the absent cells), density and geometry, dependent entirely upon the organ in which it was retrieved.

Although many decellularized extracellular matrix materials are clinically available, they are relatively limited in that their mechanical properties and porosity are dependent on the tissue of origin and these properties are not easily modified. Alloderm^® and Matristem^® are two such decellularized clinical biomaterials made from human dermis and porcine urinary bladder tissue, respectively [38]. The pores of these fibrous ECM materials arise from the removal of the original cells, and the small size of these pores, on the order of 10 microns or less [39], limits cell infiltration [38]. CorMatrix ECM™, decellularized porcine small intestinal submucosa, used for pericardial and cardiac tissue repair [40], is also limited by its small pore size. Other approaches for 3D tissue engineering suffer from similar issues. Lyophilization of ECM materials, a process used to produce porous protein foams, leads to a wide distribution of pore sizes, ~10–250 μm with an average around 70 μm [41]. Additionally, electrospun scaffolds have been shown to have small pore sizes that hinder cell infiltration [42]. Taken together, the limited porosity control of tissue engineering scaffolds, especially for decellularized biomaterials, limits recellularization into their interior and eventual homogeneous growth. Furthermore, many of the clinically available decellularized materials are not of human origin, and, importantly, none are composed of tissue-specific, cardiac-derived ECM, intuitively the most relevant compilation of proteins for cardiac-specific regeneration and repair.

Based on these limitations of tissue engineering, fabrication of allogeneic cardiac-derived extracellular matrix scaffolds with well-defined structure and large pores that allow for greater cell density should be both unique and highly advantageous for cardiac tissue engineering strategies. Such scaffolds may be used to deliver cells directly to an injury site, which has shown efficacy [19], and to provide cells with a favorable microenvironment and biochemical cues, improve cell survival and retention, and to ultimately promote improved myocardial regeneration. Importantly, efficacious scaffolds must be able to support viability in the presence of diffusion alone until host vascularization can occur. Otherwise, all implanted scaffolds will necrose prior to engraftment. Towards this aim, we have decellularized human myocardial tissue using a gentle decellularization procedure to maximally preserve endogenous protein content, created ECM scaffolds with a large, controlled, interconnected pore structure, and recellularized these scaffolds with human primary cardiac cells and human iPSC-derived cardiomyocytes. This complete, human heart-derived ECM carrier (1) utilizes the intrinsic regenerative capacity of both ECM and cardiac cells in concert, (2) uniquely retains mechanical and biochemical features of native cardiac ECM, (3) can deliver many millions of viable synchronously beating cells directly to cardiac tissue sites, and (4) can support viability for time periods sufficient to become vascularized in vivo.

It is commonly believed progressive heart failure following MI is associated with adverse cardiac remodeling. Several approaches exist to promote healthy cardiac remodeling and bulk up the infarcted wall. One approach is the use of hydrogels made of synthetic polymers [44] or animal derived ECM [27, 28] that are injected into infarcted areas. This treatment modality shown to help prevent scar expansion and wall thinning, increases left ventricular ejection fraction, and improves overall cardiac function [27, 28, 44]. To provide more relevant signaling cues and avoid undesired foreign body reactions, this study focused on ECM protein, rather than synthetic polymers. Specifically cardiac ECM, pursued by several groups [27, 28, 30], was utilized to provide the most relevant biochemical signaling cues and proteins specific for heart structural and physiological function, such as collagen and laminin (Figure 1a, Additional file 1: Figure S1). Moreover, to facilitate further clinical translation and reduce potential risks associated with xenogeneic proteins such as viral disease transmission and non- homologous signaling, we utilized purified human heart ECM protein, rather than animal-sourced ECM. There are over 3,000 hearts in the US that are not used for transplant, but are used for tissue and valves. A single heart can be utilized to create scaffolds that can patch dozens of hearts. Thus, human heart ECM scaffolds are a viable option for scaffold generation clinically. Importantly, this study not only uses human ECM, but also utilizes a novel method to form human heart ECM into perhaps unparalleled highly porous, voluminous scaffolds with strikingly interconnected, uniform pores. The human heart ECM purified in this study was solubilized and coated onto glass surfaces and supported primary heart cell phenotype (Figure 1b–d), showing promise as a biomaterial.

However, ECM alone as a biomaterial has so far been shown to be insufficient in addressing degenerative cardiac tissue or restoring damaged cardiac tissue to functional form. More effective treatment strategies for cardiac repair will likely comprise tissue-specific ECM and incorporated, viable cells capable of maintaining and sustaining tissue-context-dependent phenotypes and functions. Cell therapies of diverse forms have shown limited efficacy in regenerating damaged heart tissue, in large part by providing regenerative signaling cues [4]; however, only a small portion of these delivered cells make it to or remain within the compromised tissue zone in vivo [11]. A cellular delivery vehicle that better ensures tissue site-, context-, and function-specific localization of cells, their viability, phenotype and effective integration is required. The heart-derived, porous 3D-templated ECM matrix described here shows initial promise in cardiac cellular support, exploiting xenogeneic-free purified human heart ECM with cardiac cells.

Cell sheets have been utilized to deliver functional cells to defects in the heart, but the viability of these monolayers limits sheet thickness and subsequently cell number and accompanying ECM [45]. Other groups have also pursued fibroblast-grown ECM, however these constructs are still limited to 70 μm-thick, and increasing their thickness results in poor cell viability due to poor diffusion [46]. Porous, volumetric, 3D constructs, like those pursued in this study, are an option to maximize the number of cells and ECM protein that can be delivered to a heart defect, thereby overcoming limitations encountered with current cardiac repair approaches.

Three primary methods for generating analogous 3D porous scaffolds from ECM materials are commonly employed including electrospinning, lyophilization, and decellularization. However, all three methods yield limited pore sizes, thereby decreasing the number of cells that are capable of being delivered using the scaffold. Electrospinning generally provides matrix pore sizes of 25–100 µm [47], lyophilization yields pore sizes of 10–250 µm [41] that can be highly heterogeneous [48], and decellularized constructs only have pore sizes of the absent cells (e.g., 10’s of microns). Cells entomb themselves in extracellular matrix [49], so retroactively adding back cells to such decellularized, tight matrix voids (Figure 2) is highly inefficient and nearly impossible to control well. Cell infiltration of these scaffolds can be severely limited with small matrix pore sizes. Brown et al. reported poor cell infiltration into both decellularized porcine urinary bladder matrix (UBM) and small intestinal submucosa (SIS) [38]. Commercially available decellularized UBM, Matristem^® compared in this study with our ECM porous scaffolds showed both poor cell infiltration (limited to ~100 µm penetration distances from the outer surfaces) and resulting viability (Figure 3). Phipps et al. seeded cells onto electrospun PCL, collagen I, and hydroxyapatite composite matrices and only were able to achieve ~100 µm of cell penetration through the thickness of their scaffolds [42]. Another group using poly polyglycolic acid (PGA) woven mesh scaffolds had far fewer cells within the center of their scaffolds even with circulating media [50, 51]. This same group began using commercially available lyophilized collagen sponges that still only contained viable cells within 100 µm thick layer around the periphery of their scaffolds [51].

The novel human heart ECM-derived porous scaffolds described in this study are capable of breaking this apparent 100 µm viability barrier. Possessing large interconnected pores, they enable complete penetration throughout 750 µm thick constructs, 7.5 times greater than Matristem^® (Figure 3). The scaffolds in this study also demonstrate nearly twice as many viable cells as previously reported PGA scaffolds [52]. Moreover they support homogenous cell distribution that maintains high cell viability (62.15% ± SD 5.89%) out 7 days (Figure 3), even under static cell culture conditions—without need for circulating media or complex perfusion systems. This feat is of paramount importance as an implanted construct must be able to maintain viability until host vascularization can occur. 7 days has shown to be a sufficient time period for vascularization to occur [53]. The scaffolds developed in this study support a clinically relevant number of cells for a clinically relevant period of time. Importantly, the diameter of the scaffold has no affect on viability (data not shown), enabling larger constructs that cover larger surface areas of the heart to be employed. The ECM scaffolds developed in this study utilized porous polymer templates as sacrificial scaffolds with controlled geometry and density, a process commonly used in metallurgy, enabling precise control over final ECM protein structure and porosity. These sacrificial PCL scaffolds are endowed with fine control over pore sizes using porogens. Salt porogens were used to create polymer matrix pore sizes ranging from 250 to 500 µm, far larger than those commonly seen in known commercial ECM materials (Figure 2). After porogen removal, the porous polymeric scaffolds were coated with solubilized, purified human heart ECM, using air-drying processes to bond serially deposited ECM layers together. This process of drying creates strong, pure ECM protein scaffolds without the crosslinking utilized by many current ECM scaffold technologies [39, 41]. Exogenous chemical methods crosslink ECM proteins [54] in manners distinct from native lysyl oxidase crosslinking of native proteins [55]. Chemical crosslinking alters the mechanical properties of the scaffold, reducing elasticity and compliance, and importantly reducing degradative capacity [54]. Highly degradable ECM will better expedite tissue regeneration and replacement of scaffold ECM by native ECM, a feature readily afforded by ECM scaffolds lacking chemical crosslinking. Chemical crosslinking also increases the body’s innate foreign body response, resulting in a highly oxidative cellular environment often lethal to both exogenous and endogenous cells, thus inhibiting regeneration [56]. These known deleterious effects of the host foreign body response have limited some use of scaffolds [19]. Scaffolds absent of chemical crosslinking are afforded faster degradation and reduced foreign body reaction potential, and likely improved regenerative capacity compared to commercially available and commonly used chemically cross-linked ECM scaffolds.

This study utilized heterogeneous primary cardiac cells and homogenous iPSC human cardiomyocytes seeded in novel human heart derived ECM scaffolds. Our previous work has successfully cultured stem cells on similarly constructed adipose-derived scaffolds [57], which may also be pursued for cardiac regeneration therapy. Matrix pore sizes of 425–500 µm and monolith thickness of 0.75 mm for primary heterogeneous cardiac cells provided reliable cell viability and phenotypes (Figure 3). However, this heterogeneous population of cells possessed a large amount of cardiac fibroblasts and too few functional cardiomyocytes to enable a functional beating construct. iPSC derived cardiomyocytes were then utilized as a pure enough population to potentially enable synchronous beating within the scaffolds. To optimize sufficient cell densities for cardiac synchronization, scaffolds with pore sizes of 250 microns and thicknesses of 0.3 mm were employed. These matrix fabrication conditions enabled both cell seeding and maintenance of sufficient iPSC cardiomyocyte densities to mechanically contract the entire ECM scaffold in unison (Figure 4d, e; Additional files 3, 4: 2A, B). Though contracting occurred, subpopulations of cardiomyocytes were still present in the scaffold that likely affected the complete unity of the cells within the scaffold. Lack of complete confluency and the presence of multiple pacemaker cells driving the preparation could cause irregular beating. Future work will strive to achieve total confluence and uniformity. The iPSC cardiomyocytes seeded within these same ECM scaffolds displayed electrical competence and maintained long action potentials (Figure 4a, b) consistent with human cardiomyocyte action potentials [43, 58]. Though great variability with iPSCs in culture has been seen ranging from 250 to 1,500 ms [59].

These cells also responded to beta-adrenergic agonist and cardiac chronotrope, Isoproterenol, as expected by increasing the amplitude and frequency of their beats in response to this pharmacology (Figure 4f, g; Additional file 5, 6: 3A, B). This increase in amplitude could be due to an increase in recruitment of additional beating cells or the reduction in action potential duration dispersion due to faster beating. Similar tests utilizing isoproterenol have also been utilized by other groups to prove cardiomyocyte functionality [51, 60]. These metrics indicate that both pacemaker cells and cardiomyocytes maintain many of their responsive mechanisms within these ECM scaffolds. Future studies will analyze potential arrhythmia formation post isoproterenol and how syncytium structure and anisotropic force development affect functionality.

A microscopic analysis of seeded cells on these ECM scaffolds shows cells maintain several points of adhesion to the ECM scaffold (Figure 5a–c). Interestingly, when these cardiomyocyte-seeded scaffolds were placed on freshly explanted mouse hearts, the entire scaffold spontaneously adhered to the mouse ventricle tissue within minutes, making it impossible to remove the scaffold with forceps without tearing it (Figure 5d–g). The second harmonic optical image from a two-photon microscope highlights differences in collagen fibril orientation between the scaffold and murine heart tissue (Figure 5d). Scaffolds with aligned collagen fibrils will likely beat more efficiently. Future work will be to create scaffolds with a more linear pore structure to encourage linear collagen fibril and cardiomyocyte alignment. Electrical stimulation has also be used to increase both collagen fibril alignment and ultrastructure of tissue engineered heart scaffolds, a technique employed by Vunjak-Novakovic’s group that resulted in constructs with greater ultrastructure similarity to native heart tissue [61].

We have developed novel human derived scaffolds with sufficient pore size to enable the seeding and survival of a critical number of human cardiomyocytes to maintain cardiac synchrony. This construct acts as a cell delivery vehicle that can serve as a patch to treat ischemic heart tissue and directly target cells to damaged tissue sites. Furthermore these constructs are made entirely of human heart protein and human cells, and grown in a completely xeno-free system. Our process eliminates risk of xeno-disease transmission, optimizes regenerative capacity, and provides an expedited therapy for clinical translation.