Cardiac muscle is a structurally complex tissue optimized for life-long pumping function, and so it is challenging to replicate in vitro1. The ECM is nature’s template of a biomaterial and, through decellularization, such a scaffold could be produced to reliably engineer bioartificial cardiac tissue. Here, we aimed at evaluating several methods to efficiently decellularize human cardiac tissue, and test if the tissue scaffold could be used for tissue engineering.
Human cardiac tissue can be harvested during interventional biopsies or through segmental resection of postmortem hearts unsuitable for transplantation purposes. However, the vascular network of these tissue samples is often inaccessible, prohibiting perfusion of decellularization agents. In these cases, decellularization is often accomplished via tissue immersion in decellularization agents under gentle agitation to promote efficient diffusion. Among the factors that need to be considered when using this approach is tissue thickness. This provides reasoning why Method A, that was implemented on micrometer thick myocardium slices14, was ineffective to decellularize ~ 1 mm thick tissue pieces, that understandably have a higher cell density. An almost complete removal of nucleic material for all human cardiac tissues tested was only achieved in method D, in which the exposure times to detergents is practically equal to those used in whole-human heart perfusion-decellularization8. Lack of homogeneous exposure to decellularization solutions in methods using immersion is a clear limitation compared to perfusion-decellularization, and consequently prolonged exposure times are mandatory to eliminate dsDNA and other genetic material present in millimeter thick tissues. Even though mechanical damage to the ECM associated with pressure flow is minimal in immersion methods compared to perfusion-decellularization20, the need to extend detergent action in these methods undeniably influenced ECM biochemical composition. Quantitative analyses of some of the most prevalent ECM components, such as collagens and s-GAGs, revealed extended exposure to detergents significantly affected soluble collagen and s-GAG presence in decellularized tissue through method D. Others have also found out the retention of soluble components, such as s-GAGs21, soluble collagen8, elastin8 and growth factors22, can be severely affected using SDS/Triton-X-100 combinations, especially with extended exposure times. Conversely, insoluble collagen, that is more mature and mechanically resilient owed to cross-linking, is better retained.
Interestingly, we observed high variability when assessing the efficiency of methods B and C. Method B was sufficient to substantially reduce tissue cellularity and dsDNA content in donor 1, but the same was not verified for donors 2 and 3. Method C was an attempt to improve method B without extending detergent exposure, but was insufficient to significantly reduce cellular content in donors 2 and 3. The 48 h SDS period was based on successful protocols optimized to decellularize porcine cardiac tissue10. Inter-species differences may therefore prevent direct translation of protocols implemented in animals to human tissue. The reasons for such disparities may be manifold, in which tissue age may play a part. Human tissue samples are typically obtained from older donors, like the ones used here, while animal tissue can be harvested at earlier stages of life. Ageing induces structural changes in the myocardium, such as cardiomyocyte hypertrophic remodeling, increased collagen deposition and cross-linking, and increased proteolytic degradation via matrix metalloproteinase activity23, that could well impact the efficiency of decellularization agents. Even among animals, it has been shown SIS-ECM harvested from pigs that differ only in age presented dissimilar mechanical, structural and biological properties24. In addition, porcine and human cardiac ECM can simply have a distinct biological composition, and so be differentially susceptible to the type, concentration and exposure time of decellularization agents. For example, the higher lipidic content of decellularized human cardiac ECM when compared with its porcine analogue compromised in vitro gelation after solubilization25.
We could have further evaluated a range of ionic detergent treatment between 2 and 8/9 days, aiming at identifying an optimal protocol less harsh than method D; however, such an alternative may not exist since dsDNA was still present (in low amounts) in donors 2 and 3 after method D. Instead, combining detergents with an endonuclease may be preferable to reduce detergent exposure, even though nucleases can persist in tissue after the rinsing step19. Despite prolonged SDS/Triton-X-100 exposure being useful to reduced dsDNA to acceptable levels, we were unable to completely eliminate cellular components in donors 2 and 3 after method D (and to a lesser extent in donor 4; view Fig. 4b). The denaturant properties of these detergents seem to be efficient in solubilizing cell membranes, and consequently wash away nucleic material, but SDS action, in particular, was insufficient to effectively disrupt intracellular protein–protein interactions and aid their solubilization, regardless of cardiac tissue incubation with a lysis buffer solution beforehand (i.e., method C). In postmortem muscle tissue, as ATP is depleted, myosin heads remain bound to actin, indefinitely creating a rigor actomyosin complex that causes muscle tension and is challenging to properly solubilize26. This could explain why cellular remnants persisted in donors 2 and 3 after method D, which are possibly derived from cardiomyocytes’ contractile apparatus proteins and not from non-myocytes. In fact, attempting to decellularize mice left ventricle tissue with SDS alone created protein aggregates, that included macroproteins such as titin, and these were difficult to solubilize without enzymatic digestion27. Replacing SDS/Triton-X-100 with a trypsin/Triton-X-100 combination showed to better eliminate cellular remnants during perfusion-decellularization of thick porcine cardiac tissue slabs28. Nevertheless, others have reported trypsin has nonspecific degradation effects on heart ECM, particularly affecting preservation of collagen IV, laminin and elastin29. To overcome these issues, it has been proposed sarcolemma permeabilization followed by incubation with non-enzymatic solutions that allow sarcomere relaxation and disassembly can effectively fully decellularize rat myocardium patches, outperforming SDS-based methods30. Still, it remains elusive if the same protocol can be consistently applied to human tissue or whether it requires tuning.
Besides failure in using protocols established in animal tissue, human donor-to-donor variability is still apparent in our study. For donor 1, method B or C are suitable decellularization methods, showing successful removal of nucleic material and preservation of collagen content. However, when attempting to come up with a “one-size fits all” protocol (i.e., method D), the anisotropy of donor 1 was irreversibly lost alongside with noteworthy changes in its composition. For donors 2 and 3 tissue architecture was maintained and dsDNA significantly removed, yet tissue biochemical composition is thoroughly affected. Therefore, obtaining an acellular scaffold from human cardiac tissue that is both sufficiently complex and presents minimal variability might require optimization of current methods in a personalized way. The interindividual heterogeneity across human cardiac tissue is surprising but not unexpected, as a previous study already confirmed, through a global proteomics analysis, human myocardial ECM preparations show significant donor-to-donor variability31. The reasons behind this are challenging to discern and may be a combination of multiple factors like age, gender, and differential tissue cellular and molecular profiles, among others that warrant further investigation. Despite the low number of biological samples analyzed, our study reinforces the idea tissue-to-tissue variability should be considered when evaluating a decellularization protocol, which is often overlooked in the field. Furthermore, a compromise between cell removal and ECM preservation must be established, since for most strategies complete elimination of cells is incompatible with full retention of ECM proteins29, and an acellular scaffold may not even be essential if the ECM is not intended for in vivo implantation.
We further confirmed the suitability of the decellularized scaffold for human myocardium bioengineering. Previous attempts at recellularizing cardiac ECM have either used animal-derived neonatal proliferative cardiomyocytes6,7, non-cardiac cells14,32 or human uncommitted cardiac progenitor cells33. Fewer studies have used hiPSC-CM, and those that have attempted to recellularize human cardiac ECM with these cells have done it with low cell densities8,14,34, impairing efficient cell retention and robust tissue formation. In this work, we were able to fully repopulate decellularized cardiac tissue with hiPSC-CM, and the scaffold ensured cell retention, alignment and sustained spontaneous contractions up to 2 weeks in culture. In two-dimensional planar culture conditions hiPSC-CM lack proper organization and are unable to replicate the bundle-like appearance of cardiac muscle1. Immature neonatal cardiomyocytes are responsive to substrate stiffness, that influences twitch force and myofibril structure35. Substrate stiffness equally tunes the mechanical output of hiPSC-CM, which is improved when it matches the physiological stiffness of the myocardium36. Although further studies are needed, our decellularized human cardiac scaffold may be exerting a similar effect, thus improving the mechano-microenvironment of hiPSC-CM, and possibly features of structural maturation16.
Taken all together, producing an ECM preparation from human cardiac tissue requires a protocol tailored to the donor, with specific criteria outlining how biomimetically faithful the scaffold needs to be as to find a balance between acellularity and ECM compositional and architectural preservation. This perspective is in line with the goals of personalized medicine, ensuring product quality when designing an ECM biomaterial that is adequate for each experimental or clinical scenario. For tissue bioengineering applications we envision method D suffices; however, to study the cardiac ECM’s role during differentiation4 or how an aged37 or diseased38 ECM contributes to alteration in cardiac phenotype, the harshness of method D might compromise product quality attributes and other decellularization strategies should be pursued. Despite how tempting the notion of protocol standardization is, for human tissue it is likely unfeasible.
