Combining genome and tissue engineering for next-generation human biomimetics

Abstract

Tissue engineering for cardiovascular implants has largely utilized primary human cells to generate human tissue-engineered matrices (hTEMs). However, due to donor-to-donor variability and limited passage numbers, a more robust alternative to primary cells would be beneficial. To overcome these limitations, we have defined a new differentiation protocol for human-induced pluripotent stem cells (hiPSCs) into isogeneic cardiac fibroblast-like cells (iCFs) using animal sera-free and chemically defined methods. Morphology, extracellular matrix (ECM) deposition, and global transcriptomics revealed similarity between iCFs and primary human cardiac fibroblasts. Additionally, by overexpressing specific ECM and ECM-related proteins through gene-editing approaches, the ECM composition can be modulated as a building block to create “designer” next-generation hTEMs. Proteomics of gene-edited iCF-derived hTEMs demonstrated an increase in proteins involved in collagen and elastic fiber assembly. Furthermore, analysis of gene-edited iCF-derived hTEM mechanical functionality through biaxial mechanical testing exhibited increased collagen function, attributed to increased crosslinking and maturation. In sum, we have combined hiPSC technology with genome engineering to lay the foundation for next-generation tissue engineering applications by generating a novel cell source, gene-edited iCFs, that are able to modulate the composition as well as the functional mechanics of hTEMs.

Graphical abstract

Graphical Abstract Depicts the building blocks of next-generation human tissue-engineered matrices (hTEMs). Human-induced pluripotent stem cells (hiPSCs) as the starting cell source were gene edited, using the TALEN-based genome engineering approach, to induce a stable upregulation of genes involved in the biosynthesis and regulation of extracellular matrix (ECM) proteins (ie, lysyl oxidase [LOX], lysyl oxidase like 1 [LOXL1], elastin [ELN], and fibulin 5 [FBLN5]). These isogeneic hiPSCs were then differentiated into isogeneic cardiac fibroblast-like cells (iCFs) and used as building blocks to manufacture hTEMs for cardiovascular applications with modulated ECM composition and tissue mechanical properties.

Significance statement.

Through the integration of hiPSC technology with genome engineering, this study has established a platform for advancing cardiovascular tissue engineering applications. We have created a novel cell source, iPSC-derived isogeneic cardiac fibroblast-like cells (iCFs), that closely resemble the primary human cardiac fibroblast genotype and phenotype. Our findings highlight iCFs as a valuable alternative to human primary cells used in manufacturing human tissue-engineered matrices (hTEMs). Moreover, through targeted gene-editing iCFs to overexpress key extracellular matrix (ECM) and ECM-related proteins involved in collagen and elastin organization and maturation, we have generated next-generation hTEMs with improved structural and mechanical properties.

Introduction

Cardiovascular disease remains a major health concern, with many degenerative or congenital heart diseases requiring surgical intervention for vascular or valvular replacement^1,2 Autologous vascular grafts are preferred; however, synthetic grafts such as Teflon and Dacron are commonly used.³ The gold standard for heart valve replacements is mechanical or bioprosthetic valves generated from porcine or bovine pericardium.^4,5 Importantly, neither of the synthetic or bioprosthetic materials carry growth, regenerative, or remodeling capabilities, which emphasizes the necessity to develop alternative treatment options, especially for young patients.⁶

The ideal cardiovascular replacement requires the capacity to recruit the host’s endogenous cells to initiate adaptive remodeling and regeneration of the implanted tissue.^7–10 In situ tissue engineering (TE), which harnesses the host’s body regenerative capacities to remodel a cell-free implant, has been suggested as an alternative to overcome the hurdles of the current treatment options.^8–10 Human tissue-engineered matrices (hTEMs) are typically produced with primary human cells, with the final composition being decellularized human extracellular matrix (ECM) to obtain an off-the-shelf available implant with in situ remodeling capabilities.^11–19

Because ECM production directly correlates to the cell type used, an essential component of TE is determining the cell source. The cell source must synthesize the required ECM components as well as facilitate the structure and maturation of these proteins, which directly impacts the mechanical properties of the final tissue.²⁰ TE approaches have used primary human dermal fibroblasts, myofibroblasts, valvular interstitial cells, and bone marrow-derived stem cells, among others, as ECM-producing cell sources.^21,22 However, due to donor-to-donor variability in primary cell sources, there is inconsistent ECM production.^20,23,24 Furthermore, cell proliferation and the ability to secrete ECM proteins are known to be affected by the passage number and the age of the donor, thus introducing more variability.^16,25–28

To overcome the limitations displayed by primary cell sources, human induced pluripotent stem cells (hiPSCs) represent a promising alternative for hTEM-based constructs because of their reproducibility, differentiation potential, and high proliferative capacities.^29,30 hiPSCs represent a promising cell source not only due to their high proliferation rate, but also the ability to differentiate into reproducible cells each time to reduce donor-to-donor variability in the ECM synthesis. Furthermore, hiPSCs are able to differentiate into theoretically any cell type, allowing us to generate specific cells to produce specific tissues, such as cardiac cell types for cardiovascular implants. In addition, a variety of studies have utilized hiPSC-derived cardiac cell types in TE.^31–33 However, a disadvantage of using hiPSCs is the lack of standardized differentiation methods, with several protocols available for nearly every cell type, as extensively reviewed elsewhere.^34–37

In the present study, we define a hiPSC differentiation protocol for isogeneic cardiac fibroblast-like cells (iCFs), using chemically defined methods toward cGMP compliant manufacturing requirements. The newly defined iCFs demonstrate a genotype and phenotype comparable to primary human CFs. Furthermore, iCF-derived hTEMs exhibited ECM production that was reproducible and with increased functional integrity when compared to primary human dermal fibroblast (hDFB)-derived hTEMs. However, remaining challenges are still present in the composition and maturation of the ECM produced in today’s hTEMs. We therefore utilized transcription activator-like effector nuclease (TALEN) to implement gene-editing techniques to upregulate key ECM and ECM-related proteins necessary for collagen and elastin (ELN) synthesis and maturation. Thus, fine-tuning the biological and mechanical behavior of our hTEMs.

Through bridging gene-editing and hiPSC technology in TE, we designed and produced gene-edited iCFs, which were used as building blocks to create customizable hTEM-based constructs. Our objective was to create a stable and reliable cell source that is able to modulate the composition and function of the ECM, thus creating a “designer” matrix for next-generation hTEMs.

Materials and methods

Extended materials and methods can be found in the online data supplement.

hiPSC differentiation into cardiac fibroblast-like cells

hiPSCs were cultured until 60%-70% confluency. On Day 0, cells were cultured in Rosswell Park Memorial Institute (RPMI-1640) medium (Gibco) supplemented with B27-minus insulin (Gibco) and 12 µM CHIR99021 (STEMCELL Technologies) for 24 h. On Day 1, cells were cultured in RPMI + B27 minus insulin for 24 h. On Day 2, RPMI + B27 minus insulin was supplemented with 5 µM IWR-1 (STEMCELL Technologies) for 24 h. On Day 3, medium was changed to cardiac fibroblast medium (CFBM), which contains Advanced Dulbecco's Modified Eagle's Medium (DMEM; Gibco) supplemented with FibroLife Serum-Free Fibroblast LifeFactors Kit (Lifeline Cell Technology) and 75 ng/mL basic fibroblast growth factor (bFGF) (Miltenyi Biotec) until Day 18 (key supplements listed on Table S10). On Day 18, cells were plated directly on plastic non-coated 6-well plates at a ratio of 1:3 until Day 22 to remove any undifferentiated cells and purify the iCF cell populations. On Day 22, cells were plated on collagen type 1 (COL1) (Merck) coated 6-well plates and maintained in CFBM + bFGF for subsequent passages.

Plasmid production and amplification

All plasmids were purchased from Addgene, backbone plasmid (AAVS1-mEGFP; #114404); TALEN Left (AAVS1-TALEN-L; #59025); and TALEN Right (AAVS1-TALEN-R; #59026). Benchling was used to design the plasmid library. Primers for Gibson assembly (listed in Table S2) were designed to add the appropriate overhang for subsequent restriction digestion for cloning into the vector. The target genes (ELN, fibulin-5 [FBLN5], lysyl oxidase [LOX], and lysyl oxidase-like 1 [LOXL1]) were amplified by Polymerase Chain Reaction (PCR) from complementary DNA (cDNA) using primers to introduce the required overhangs for subsequent cloning into the vector. Amplified cDNAs were then cloned into the linearized vector using Gibson assembly following the chemically competent cells transformation protocol (E5510; New England BioLabs). Clones of transformed bacteria were then inoculated in autoclaved Lysogeny Broth (LB) media (composed of 5 g NaCl, 5 g Typtone, 2.5 g yeast extract, 100 µg/mL ampicillin, and MiliQ water to 500 mL) and cultured overnight at 37 °C on an agitator. Vectors were isolated using the Mini plasmid purification kit (PureLink HiPure Plasmid Mini plasmid purification Kit). Sanger sequencing of the plasmids was performed to confirm integration of the cDNA into the plasmid using alignment software (Benchling). Colonies containing the desired plasmids were then expanded in a large LB culture (AAVS1-ELN, AAVS1-FBLN5, AAVS1-LOX, and AAVS1-LOXL1) using the Maxi plasmid purification kit (PureLink Fast Low-Endotoxin Maxi Plasmid Purification Kit).

hiPSC transfection and clonal selection

hiPSCs were cultured to 70% confluency and incubated for 2 hours in 10 µM Rho kinase inhibitor (Miltenyi Biotec) at 37 °C, before detachment. hiPSCs were split into a concentration of 200 000 cells in 20 µL of mTeSR1. AAVS1-TALEN-L, AAVS1-TALEN-R, and the vector plasmid were added to the cell suspension in a ratio of 1:1:2, respectively. hiPSCs were electroporated using 1100 V, 30 milliseconds, 1 pulse (Neon). Following electroporation, hiPSCs were cultured for 3-4 days before starting the clonal selection by supplementing mTeSR1 medium with puromycin at 1 µg/mL concentration for an additional 3 days. Puromycin-resistant colonies were then selected and expanded. To validate integration of the transgene, PCR analysis of the insertion site was performed using specific primers (Forward: TCATGGCATCTTCCAGGGGT and Reverse: TCTAACCCCCACCTCCTGTT), as well as by gene expression and sequencing using specific primers for the PCR amplification of the homology arms left and right, followed by Sanger sequencing. Primers can be found in Tables S3 and S4.

RNA library preparation and sequencing

Total mRNA was isolated from cells using the RNeasy mini kit (Qiagen) with additional DNAse I treatment. Isolated RNA quality was determined using a Qubit Fluorometer (Life Technologies) and a Fragment Analyzer (Agilent). Samples with 260/280 nm ratio between 1.8-2.1 and a 28S/18S ratio within 1.5-2 were further processed. The Functional Genomics Center Zurich (FGCZ) then performed the library preparation and sequencing using Stranded mRNA Prep Ligation (Illumina). Briefly, 100-1000 ng total RNA were polyA enriched and reverse transcribed to double-stranded cDNA. cDNA samples multiplexed with unique dual indices TruSeq adapters. Libraries were prepared according to the NovaSeq6000 Reagent Kit (Illumina) workflow. Cluster generation and sequencing were performed on a NovaSeq6000 System (Illumina) with a run configuration of single-end 100 bp. Subsequent enrichment analysis was made with the online tool Metascape.³⁸

hTEM manufacturing

Human tissue-engineered matrices were produced in triplicate (n = 3 per group) using non-woven polyglycolic acid (PGA) mesh (Confluent, Inc.) coated with 1% poly-4-hydroxybutyrate (P4HB) (TEPHA, Inc.), according to our previous protocols.³⁹ Briefly, scaffolds were sutured onto sterile stainless steel rings, and 1 × 10⁶ cells/cm² were seeded onto the scaffolds using fibrin (Sigma) as a cell carrier.⁴⁰ Following 24 hours, culture media was supplemented with 0.25 mg/mL L-ascorbic acid 2-phosphate (Sigma) and 5 ng/mL transforming growth factor beta 1 (TGFβ1; PeproTech), hTEMs were moved to an orbital shaker at 37°C to be cultured for 4 weeks. hTEMs were then decellularized using a detergent composed of 0.25% Triton X-100, sodium deoxycholate, and 0.02% ethylenediaminetetraacetic acid, followed by Benzonase (Millipore) treatment to remove residual dsDNA. Decellularized hTEMs were then stored in Dulbecco's Phosphate-Buffered Saline (DPBS) containing 1% penicillin/streptomycin at 4°C.

Proteomics analysis

Label-free mass spectrometry of hTEMs was performed at the FGCZ. Based on initial quality control results, the sample size was set to n  = 4 per group. Briefly, proteins were isolated, purified, and digested using KingFisher SP3 digestion (Thermo Fisher). Carboxylated magnetic beads were added using a KingFisher Flex system (Thermo Fisher), according to the manufacturer’s instructions. Beads were digested overnight at 37°C in 50 mM Triethylammonium Bicarbonate (TEAB) with trypsin before drying the eluted samples. Dried samples were dissolved in 3% Acetonitrile with 0.1% formic acid prior to being loaded on Evotip (Evosep) and analyzed on an Evosep One liquid chromatography coupled to a TMIS TOF Pro mass spectrometer. Data analysis was performed with DIA-NN (version 1.8.2) using a library-free approach with the Homo sapiens database and the following parameters; variable modifications: Acetyl (Protein-N-term), Oxidation (M), Met-loss (protein N-terminal), Fixed modifications: Carbamidomethyl (C). The R-Package prolfqua was used to filter, normalize, generate visualization, and compute differential expression of the proteomics raw data.⁴¹ To further improve the differential expression, the protein variances are updated using a variance prior estimated from all the proteins in the experiment. Differential expression analysis (DEA) was made by comparing hTEMs produced by non-edited vs edited BION-iCFs.

Biaxial mechanical testing

Decellularized hTEMs were cut into 5 × 5 mm squares, and the thickness was measured between 2 thin metal plates using a digital caliper. hTEM squares were then mounted and attached in a planar biaxial mechanical tester (BioTester 5000, CellScale) using the BioRake sample mounting system (CellScale, tine diameter 254 µm, tine spacing 0.7 mm, puncture depth 1.4 mm). Samples were stretched to 30% equibiaxially. Prestretching cycles were repeated 5 times, consisting of 15 seconds of loading and unloading. Final measurements were derived from the average of the subsequent 2 cycles.

Statistical analyses

Reverse Transcription-Quantitative Polymerase Chain Reaction (RT-qPCR) measurements, with primers listed in Table S1, were analyzed with the unpaired t-test in GraphPad Prism. Data are represented as mean ± SEM, *P < .05, **P < .005, ***P < .0005, ****P < .0001, n = 3 per group. Mechanical testing measurements were analyzed using an unpaired t-test in GraphPad Prism and represented as mean ± SEM, *P < .05, **P < .005, ***P < .0005, ****P < 0.0001, n = 3 per group.

Results

Generation of hiPSC-derived iCFs

We have developed a novel hiPSC differentiation protocol that uses fetal bovine serum-free media and a cocktail of growth factors, all of which are chemically defined to produce iCFs (Figure 1A). Briefly, hiPSCs were first differentiated into mesoderm with CHIR99021, followed by induction of cardiac mesoderm using a Wnt . The resulting cardiac lineage cells were then differentiated into fibroblasts using bFGF. Previous iCF protocols have used similar methods; however, we have included a purification step to remove any unwanted, undifferentiated cells from the pool of iCFs.^33,42 The resulting iCFs have a purity of 93.2% when compared to the unpurified population of 73.9% (Figure 1B). Two different hiPSC lines were used to generate iCFs, WTB6 and BION-10C, respectively, in order to demonstrate the proficiency of the differentiation protocol among multiple cell sources. iCFs derived from both WTB6 and BION-10C, described here as WTB6-iCFs and BION-iCFs, displayed similar morphology to primary human cardiac fibroblasts (hCFs) (Figure 1C).

Figure 1.

Differentiation of hiPSCs into isogenic cardiac fibroblast-like cells (iCFs). (A) Differentiation protocol schematic of hiPSCs into iCFs. (B) Flow cytometry analysis of pre-selection and post-selection on plastic of differentiated iCFs, using fibroblast-associated markers. (C) Brightfield images of primary human cardiac fibroblasts (hCFs), WTB6-iCFs, and BION-iCFs (scale: 200 µm). (D and E) Relative mRNA expression using RT-qPCR of fibroblast-associated genes for BION-iCFs and WTB6-iCFs, compared to hCFs (n = 3 per group). Data were analyzed with unpaired t-test and represented as mean ± SEM, ***P < .0005, ****P < .0001. (F) Immunostaining of hCFs, WTB6-iCFs, and BION-iCFs for fibroblast markers αSMA (green) and VIM (green), as well as ECM proteins ELN (red) and FN (red) (scale: 25 µm). αSMA, alpha-smooth muscle actin; CD90, cluster differentiation 90; CFBM, cardiac fibroblast basic medium; CHIR, CHIR99021; COL1, collagen type 1; ELN, elastin; FGF, basic fibroblast growth factor; FN, fibronectin; IWR-1, Wnt pathway inhibitor; PDGFRα, platelet-derived growth factor receptor alpha; PDGFRβ, platelet-derived growth factor receptor beta; VIM, vimentin.

Further characterization of WTB6-iCFs and BION-iCFs showed complete downregulation of pluripotency markers when compared to hiPSCs and similar expression to that of hCFs, indicating terminal differentiation (Figure S1). Fibroblast-associated markers, platelet-derived growth factor receptor alpha (PDGFR-α), platelet-derived growth factor receptor beta (PDGFR-β), vimentin (Vim), fibronectin (FN), and COL1, showed a significant upregulation in their gene expression when compared to hiPSCs (Figure 1D and E). Overall, iCFs demonstrated similar gene expression of COL1 when compared to primary hCFs. However, iCFs had increased expression in the case of PDGFR-α, PDGFR-β, and Vim, while the gene expression of FN was reduced when compared to hCFs (Figure 1D and E). Immunostaining of intracellular proteins α-smooth muscle actin and Vim, as well as ECM proteins ELN and FN indicated similar protein levels and organization between primary hCFs, WTB6-iCFs, and BION-iCFs (Figure 1F). In summary, iCFs demonstrated terminal differentiation, while fibroblast-associated markers were upregulated. However, these fibroblast-associated markers are known to not be specific to fibroblasts alone, especially not specific to cardiac fibroblasts. We therefore employed bulk RNA sequencing in order to compare WTB6- and BION-iCFs to primary hCFs and hDFBs.

Global transcriptomic profiles (21,057 genes) were assessed and analyzed for total mRNA transcripts, and global hierarchical clustering and principal component analysis (PCA) plots can be found in Figure S2. Of those, 2600 genes showed similarity in the gene expression between iCFs and hCFs, but were not expressed in hDFBs (Figure 2A). Gene ontology analysis of the similarly expressed genes (a total of 2600 genes) between iCFs and hCFs indicated enrichment of genes heavily involved in cardiovascular system development and processes (Figure 2B and Table S5). Among the identified genes, cardiac-related factors, including HEY1, HAND2, PDGFβ, HOPX, NOTCH1, TEK, TGFβ2, ANKRD1, JUP, and BMP2, showed an upregulation expression in both iCFs and hCFs (Figure 2C). This suggests a successful cardiac lineage differentiation which is maintained in our iCFs, similar to primary hCFs. Furthermore, these results highlight that the iCFs are biologically distinct from primary hDFBs, and may therefore be a better cell type for creating cardiac-specific hTEMs.

Figure 2.

Transcriptomics of iCFs compared to primary hCFs. (A) Heatmap of similarly expressed genes (2600 genes) in primary hCFs, BION-iCFs, and WTB6-iCFs, which are not expressed in primary human dermal fibroblasts (hDFBs) (n = 2 per group). (B) Gene ontology analysis indicating the top 5 biological processes of the 2600 similarly expressed genes between hCFs, BION-iCFs, and WTB6-iCFs using the online tool Metascape. (C) Comparison of the expression levels (values in fragments per kilobase million, FKPM) of the genes involved in cardiovascular developmental processes. ANKRD1, ankyrin receptor domain 1; BMP2, bone morphogenetic protein 2; HAND2, heart and neural crest derivatives expressed 2; HEY1, Hes-related family BHLH transcription factor with YRPW motif 1; HOPX, homeodomain-only protein homeobox; JUP, junction plakoglobin; NOTCH1, notch receptor 1; PDGFβ, platelet-derived growth factor beta; TEK, tyrosine-protein kinase receptor; TGFβ2, transforming growth factor beta 2.

Isogeneic hiPSCs with upregulation of ECM-specific genes

In order to create the building blocks for designer hTEMs, we next created isogenic hiPSC lines with specific upregulation of ECM or ECM-related genes. The selected genes for targeted stable transfection were LOX, LOXL1, FBLN5, and ELN, all of which are involved in the deposition and maturation of vessel or heart valve ECM. Specifically, collagen and ELN fibrils are crosslinked by LOX and LOXL1, while FBLN5 and ELN are key for elastinogenesis.^43,44 Utilizing the TALEN approach, we integrated specific ECM genes successfully into the adeno-associated virus integration site 1 (AAVS1), a known safe harbor locus that does not disrupt cell function.^45–47 The plasmid DNA template structure consisted of left and right homology arms specific to the AAVS1 locus, an SA sequence, indicating where mRNA needs to be spliced, followed by T2A a sequence, who mediates polypeptide cleavage during translation, a puromycin resistance gene for selection, a human phosphoglycerate kinase promoter for ubiquitous expression, and the target DNA sequence (ie, ELN, FBLN5, LOX or LOXL1) (Figure 3A and Figure S3). The AAVS1 locus is inside intron 1; therefore, the splice acceptor (SA) site is important for the splicing (removing the remaining intron sequence before the puromycin sequence to then link it to exon 1), which results in an mRNA sequence containing AAVS1 exon 1 followed by the puromycin resistance gene. The T2A sequence allows for the separation of the puromycin resistance gene from the exon 1 during translation by the ribosome, resulting in the activation of the puromycin resistance. To confirm proper integration of the transgenic cassette at the AAV1 locus, we performed Sanger sequencing at the integration site of edited hiPSCs (Figure 3B). Pluripotency and differentiation capacity of the non-edited vs edited hiPSCs indicated no reduction (Figures S4 and S5). qPCR analysis of isogenic hiPSCs compared to non-edited hiPSCs showed a significant increase in the expression of the target mRNAs (Figure 3C). Isogenic hiPSC lines were then differentiated into iCFs using our optimized protocol described in Figure 1. qPCR analysis demonstrated increased ECM-related gene expression in the iCFs, similar to their edited hiPSC counterparts (Figure 3D). Furthermore, immunostaining of each upregulated ECM gene of interest showed protein localization and increased deposition (Figure 3E). Taken together, we have successfully used gene-editing to upregulate the expression of key proteins needed for collagen and ELN production and maturation in isogenic hiPSC lines. Furthermore, the subsequent edited iCFs can potentially be used as building blocks for the production of designer next-generation hTEMs.

Figure 3.

Design and characterization of ECM-related gene-edited hiPSCs and iCFs. (A) Schematic of the donor DNA insert designed for the TALEN gene-editing approach, which includes the homology arm left, followed by a splice acceptor site (SA), thosea asigna virus 2A-like peptide (T2A) sequence, and puromycin resistance gene, followed by a PGK promoter and the gene of interest (target DNA), ending with the homology arm right. (B) Sanger sequencing of the homology arm left and right indicating proper insertion at the AAVS1 safe harbor locus for elastin (ELN), fibulin 5 (FBLN5), lysyl oxidase (LOX), and lysyl oxidase-like 1 (LOXL1) in the edited BION-iCFs. (C) RT-qPCR quantification of mRNA expression levels after TALEN knock-in of ELN, FBLN5, LOX, or LOXL1 compared to unedited BION-iPSCs (n = 3 per group). (D) RT-qPCR quantification of mRNA expression levels after TALEN knock-in of ELN, FBLN5, LOX, or LOXL1 with subsequent differentiation into BION-iCFs, compared to unedited BION-iCFs (n = 3 per group). (E) Immunostaining comparing unedited- to edited-BION-iCFs with the ECM-related genes of interest (ie, ELN, FBLN5, LOX, and LOXL1) (scale: 25 µm). Data were analyzed using unpaired t-test and represented as mean ± SEM, *P < .05, **P < .005, ***P < .0005, ****P < .0001. AAVS1, adeno-associated virus integration site 1; iCF, isogeneic cardiac fibroblast-like cell; ECM, extracellular matrix; TALEN, transcription activator-like effector nuclease.

Generation and characterization of iCF-based hTEMs

To investigate hTEM ECM production, non-edited WTB6- and BION-iCFs were cultured onto a PGA scaffold coated with 1% P4HB (Figure 4A). The iCF-derived hTEMs were compared to hTEMs produced with primary hDFBs, used in TE because of their ability to generate large amounts of ECM.^48,49 After 4 weeks of tissue culture, scaffolds were decellularized to obtain a cell-free tissue construct composed of a hybrid material of human ECM and remaining scaffold material. Initial evaluation of the iCF-derived hTEMs indicated ECM deposition that can be visualized by the shiny layer of tissue surrounding the PGA/P4HB scaffold, similar to the hDFB-derived hTEMs (Figure 4B-D). Histology and immunohistochemistry (IHC) were used to assess the overall ECM deposition and maturation between hDFB- and iCF-derived hTEMs. H&E staining demonstrated similar ECM deposition between all groups (Figure 4B-D). Immunohistochemistry of COL1 and collagen type 3 (COL3) also indicated similar deposition between all groups. Elastic van Gieson (VGEL) staining did not reveal the presence of mature elastic fibers in any group (Figure 4B-D). Overall, hTEMs derived from iCFs have similar ECM deposition and maturation when compared to hDFBs.

Figure 4.

Manufacturing and characterization of human tissue-engineered matrices (hTEMs) derived from hDFBs, non-edited iCFs, and edited iCFs. (A) Illustration of hTEM production protocol. (B-D) Representative images (n = 3 per group) of hTEMs derived from hDFBs, unedited WTB6-, and BION-iCFs. (E-H) BION-iCFs edited with either ELN, FBLN5, LOX, or LOXL1. Following 4 weeks of tissue culture, representative images (n = 3 per group) of (I) macroscopic appearance of hTEMs post-decellularization; (II) hematoxylin and eosin (H&E) staining, (III) collagen type 1 (COL1) staining, (IV) collagen type 3 (COL3) staining, and (V) Verhoeff’s van Gieson Elastin (VGEL) staining. Scale: 100 µm. ELN, elastin; FBLN5, fibulin 5; hDFBs, human dermal fibroblasts; iCF, isogeneic cardiac fibroblast-like cell; LOX, lysyl oxidase; LOXL1, lysyl oxidase-like 1.

To assess the effect of gene-editing on ECM biosynthesis and maturation, hTEMs were generated using BION-iCFs edited to overexpress either ELN, FBLN5, LOX, or LOXL1. H&E staining of FBLN5 edited BION-iCFs showed similarities in the deposition of ECM with non-edited iCFs and hDFBs, as well as similar COL1 and COL3 staining (Figure 4E). However, histology analysis of ELN, LOX, and LOXL1 edited BION-iCFs was speculated to have denser ECM deposition in H&E staining (Figure 4F-H). Furthermore, COL1 and COL3 staining showed increased compaction of mature collagen fibers (Figure 4F-H). However, mature ELN fibers were not present in any of the edited iCF-derived hTEMs (Figure 4E-H). To determine if the change in ECM was a function of increased cell proliferation, we performed growth curves of non-edited and edited BION-iCFs compared to hDFBs, which indicated no change in growth over 72 hours for most cell types except ELN and LOX-edited iCFs showing increased proliferation (Figure S6). Increased iCF proliferation cannot be excluded as the cause of denser ECM in the histology data. Quantification of overall collagen content using hydroxyproline as well as glycosaminoglycan content using biochemical assays indicated downregulation of both HYP and GAG content in WTB6-iCF-derived hTEMs compared to hDFB-derived hTEMs, while HYP content being similar and GAGs even upregulated in BION-iCF-derived hTEMs (Figure S7A and B). Measurements of edited iCF-derived hTEMs indicated an upregulation of GAGs, while HYP demonstrated no change in edited iCF-derived hTEMs, except for FBLN5-edited iCF-derived hTEMs having slight downregulation. Additionally, Picrosirius Red staining was performed as a semi-quantitative measurement of overall collagen deposition between each group. These results indicate an overall increased trend of collagen deposition for each of the edited iCF cell types (Figure S8). Taken together, these results indicate that the addition of gene editing for the upregulation of select ECM and ECM-related proteins can augment ECM deposition in iCF-derived hTEMs.

hTEM ECM composition

To further elucidate the global composition of the edited vs non-edited iCF-derived hTEMs, proteomics was performed using liquid chromatography coupled to mass spectrometry. Differential expression analysis identified 6522 proteins with a minimum of 2 non-redundant peptides. Matrisome analysis of the DEA results, using the MatrisomeDB 2.0, classified the proteome into 3 main categories.⁵⁰ Among the 6522 proteins identified, 6109 were categorized as non-matrisome (93.67%), 251 as matrisome-associated (3.85%), and 162 as core matrisome proteins (2.48%) (Figure 5A). Further analysis of the 251 matrisome-associated proteins indicated that 48.21% were categorized as ECM regulators, 32.67% as secreted factors, and 19.12% as ECM-affiliated proteins (Figure 5B). Finally, among the 162 core matrisome proteins, 70.37% were identified as glycosaminoglycans, 18.52% as collagens and 11.11% as proteoglycans (Figure 5C).

Figure 5.

ECM composition of edited vs non-edited BION-iCFs using proteomics. (A) Total number of identified proteins (6522), categorized into non-matrisome, core matrisome, and matrisome-associated proteins. Representation of each category as a percentage of the total number of identified proteins (6522). (B) Categorization of the matrisome-associated proteins (251) into ECM regulators, secreted factors, and ECM-affiliated proteins. Representation of each category as a percentage of the total number of proteins in the matrisome associated (251). (C) Categorization of the core matrisome proteins (162) into glycosaminoglycans (GAGs), collagens, and proteoglycans. Representation of each category as a percentage of the total number of proteins of the core matrisome (162). (D) Heat map of the differential expression analysis comparing ELN-, FBLN5-, LOX-, and LOXL1-edited hTEM proteome to the unedited hTEMs. (E-H) Gene ontology analysis of the upregulated proteins, for hTEMs derived from BION-iCF-ELN, BION-iCF-FBLN5, BION-iCF-LOX, and BION-iCF-LOXL1 compared with hTEM produced by non-edited BION-iCFs. Gene ontology analyses were made using the online tool Metascape. ECM, extracellular matrix; ELN, elastin; FBLN5, fibulin 5; hDFBs, human dermal fibroblasts; hTEMs, human tissue-engineered matrices; iCF, isogeneic cardiac fibroblast-like cell; LOX, lysyl oxidase; LOXL1, lysyl oxidase-like 1.

Although the most abundant fibrillar collagens in human tissues, COL1A1 and COL1A2, showed similar expression in ELN-, FBLN5-, and LOX-edited hTEMs when compared to unedited hTEMs, LOXL1-edited hTEMs showed significant upregulation of COL1A2 (Figure 5D). However, FBLN5-, LOX-, and LOXL1-edited hTEMs demonstrated a marked increase in specific collagens of interest such as COL8A1 (Figure 5D), which is expressed in blood vessel and adult heart.⁵¹ Furthermore, COL18A1, known to be expressed in early developmental stages of myocardium and atrioventricular valves (AVs), showed a significant increase in ELN-, FBLN5-, LOX-, and LOXL1-edited hTEMs.^52,53 Additionally, COL12A1, known to participate in the stabilization of COL1 fibrils and maintain mechanical function, was upregulated in LOX-edited hTEMs (Figure 5D).^54,55

Proteoglycans also demonstrated an increase in the edited compared to the unedited hTEMs. Specifically, aggrecan, involved in vascular plasticity and remodeling, was upregulated in all edited hTEM groups (Figure 5D).^56,57 Hyaluronan and proteoglycan link protein 1, involved in compression resistance of tissues and tissue regeneration, increased in FBLN5-, LOX-, and LOXL1-edited hTEMs.^58,59 Lumican, involved in collagen fibrillogenesis, was increased in ELN-, LOX-, and LOXL1-edited hTEMs (Figure 5D).⁶⁰ Furthermore, biglycan, known to have multiple functions such as muscle development and collagen fibril assembly, was increased in hTEMs edited with FBLN5, LOX, or LOXL1 (Figure 5D).

Glycosaminoglycans, holding the majority of the core matrisome proteins identified in our proteomics, had significantly increased in all edited hTEMs. Namely, fibrillin-2, involved in elastic fiber assembly^61,62; matrilin-2, involved in the process of matrix assembly, was increased in ELN-, FBLN5, and LOX-edited hTEMs⁶³; latent transforming growth factor beta binding protein 1, which promotes fibrillin assembly⁶⁴; thrombospondin type 1 domain containing 4 , known to promote fibrillin assembly and participates in elastic fiber formation^65,66; and cellular communication network factor 2, a matricellular protein having various functions in ECM remodeling and angiogenesis⁶⁷ (Figure 5D).

To determine key ECM pathways enhanced by each of the edited BION-iCFs, gene ontology of the upregulated proteins was performed. The top 5 biological processes for the edited hTEMs were enriched for pathways, unsurprisingly, involved in ECM organization and processing, but also pathways that are cardiovascular specific. Key examples being ELN-edited hTEMs being enriched for blood vessel morphogenesis (GO: 0048514); FBLN5-editing showed upregulation of tube morphogenesis (GO: 0035239); LOX-edited hTEMs showed upregulation of angiogenesis (GO: 0001525); as well as LOXL1 promoting blood vessel development (GO: 0001568) (Figure 5E-H and Tables S6-S9). Overall, the in-depth proteomics analysis revealed that genetically edited iCFs that overexpress specific ECM or ECM-related proteins can augment the hTEM ECM composition. Specifically looking at the core matrisome proteins, editing a single ECM-related gene was able to upregulate pathways that are not only associated with ECM organization and fibril formation, but also cardiovascular-specific pathways. These results indicate that these gene-edited iCFs may be a building block for customizable hTEMs for cardiovascular implants. Importantly, these changes in ECM composition and structure may also impact the mechanical functionality of hTEMs.

hTEM mechanical properties

To test ECM functionality, we evaluated the mechanical properties of each hTEM group using biaxial mechanical tests. Samples demonstrated isotropic behavior across all the tissues, with no significant variance between the 2 axes (Figure S9). Each stress and stretch curve starts with a toe region, which represents the contribution of the soft matrix and translates to the elastic properties. The transition region, the curvature in the graph, indicates the point where the tissue’s response shifts from being influenced by soft matrix to when collagen fibers start resisting the stress stimuli. The heel region, the final exponential line in the graph, translates to the properties of the collagen fibers in the tissue. Initial assessment of the stress and stretch curves revealed similar behavior to a collagen-rich tissue.^68–72

The stress and stretch curves comparing hTEMs derived from hDFBs and non-edited BION-iCFs showed similar behavior between all samples (Figure 6A). However, closer analysis indicated an extended toe region trend in the iCF-derived hTEMs when compared to hDFB-derived hTEMs (Figure 6F). The heel region had an increasing trend, although not significant, for the iCF-derived hTEMs (Figure 6K). Furthermore, while the transition strain did not show a significant difference, a trend was observed where a higher strain was required to activate the collagen fibers in the iCF-derived hTEMs (Figure 6P). In summary, the hTEM mechanical properties produced by non-edited iCFs exhibited similar behaviors to the hDFB-derived hTEMs. However, detailed analysis of the toe, heel, and transition region indicated that non-edited iCF-derived hTEMs tend to have slightly more elastic properties and require higher strain before inducing the contribution of the collagen fibers to the mechanical properties of the tissue. This can be linked to an increase in the waviness of the collagen fibers.^73,74

Figure 6.

Mechanical functionality of non-edited and edited hTEMs. (A-E) Biaxial mechanical testing depicting stretch vs stress curves in the Y direction of (A) hTEMs derived from hDFBs (black), non-edited BION-iCFs (blue), and non-edited WTB6-iCFs (red); (B) hTEMs derived from hDFBs (black), non-edited BION-iCFs (blue), and edited BION-iCFs with ELN (orange); (C) hTEMs derived from hDFBs (black), non-edited BION-iCFs (blue), and edited BION-iCFs with FBLN5 (green); (D) hTEMs derived from hDFBs (black), non-edited BION-iCFs (blue), and edited BION-iCFs with LOX (pink); (E) hTEMs derived from hDFBs (black), non-edited BION-iCFs (blue), and edited BION-iCFs with LOXL1 (purple). (F-J) Toe region as a quantification of the elasticity of the hTEM. (K-O) Heel region as a quantification of the stiffness of the hTEM. (P-T) Transition region depicting the phase where mechanical properties shift from elastic to collagen-driven mechanical behavior. n = 3 per group. Data were analyzed with unpaired t-test and represented as mean ± SEM, *P < .05. ELN, elastin; FBLN5, fibulin 5; hDFBs, human dermal fibroblasts; hTEMs, human tissue-engineered matrices; iCF, isogeneic cardiac fibroblast-like cell; LOX, lysyl oxidase; LOXL1, lysyl oxidase-like 1.

Stretch and strain analyses of the gene-edited BION-iCF-derived hTEMs indicated differences in the mechanical behavior in ELN (Figure 6B, G, L, and Q), LOX (Figure 6D, I, N, and S), and LOXL1 (Figure 6E, J, O, and T) edited hTEMs, while FBLN5 was similar to the unedited iCFs and hDFBs (Figure 6C, H, M, and R). Specifically, ELN-edited hTEMs have a significant increase in the toe region, indicating increased elastic properties in the tissue, when compared to hDFB-derived hTEMs (Figure 6G). Furthermore, ELN-, LOX-, and LOXL1-edited hTEMs exhibited significantly steeper curves in the heel region (Figure 6L, N, and O), indicating increased collagen functionality and therefore increased stiffness of the tissue.

To address if the residual polymer contributes to the mechanical strength of the hTEMs, we must look at the typical degradation of PGA and P4HB. However, the composition and mechanical strength of degrading PGA/P4HB are difficult to quantify. PGA degrades much quicker than P4HB, whereas P4HB is more elastic.^75–77 Furthermore, degradation may not be consistent throughout the matrix, as demonstrated by 3D grating interferometry in vitro.⁷⁸ Consequently, the tissue-engineered ECM becomes the primary contributor to the construct’s mechanical strength. To fully address this question, a systematic investigation is required, which is beyond the scope of this manuscript.

Overall, the results indicate that the ECM produced by iCFs has an increase in elastic properties, measured in the toe region, and a trend toward an increase in collagen function, measured in the heel region, when compared to hDFB-derived hTEMs. Additionally, gene-edited ELN-iCFs also increased the elastic properties of the hTEMs. Furthermore, gene-editing of ELN, LOX, and LOXL1 significantly increased the heel region, representing an increase in collagen fiber function and stiffness of the material. The stiffening of the material can be attributed to an increase in the crosslinking of the collagen fibers, resulting in a denser and more mature collagen network. This correlates with the histology data of ELN-, LOX-, and LOXL1-edited hTEMs seen in Figure 4.

Discussion

Extracellular matrix composition and organization represent a major aspect of in situ TE implants, as the ECM plays a role in the functionality, adaptability, and remodeling capacity. The ECM’s role not only provides the mechanical strength to withstand the environment, but also influences host immune reaction toward adaptive or maladaptive remodeling.⁷⁹ Thus, the cell source used to generate TE implants is a key component to establish ECM composition and organization. Human primary cells, such as hDFBs, have been used in the manufacturing process of hTEMs due to their ability to synthesize great amounts of ECM.^48,80 However, donor-to-donor variability and limited passage number can affect the reproducibility of ECM production.^{16,20,23–28} To address these limitations, the reproducibility, differentiation potential, and proliferative capacities of hiPSCs make them an interesting alternative cell source for hTEM-based constructs. The reasoning behind the use of iPSC-derived cardiac fibroblast cells for the production of hTEMs is that by using cardiac-specific fibroblasts, we can generate a more cardiac-specific ECM, containing cardiovascular-specific proteins and growth factors important to support infiltrating cells upon implantation. By implanting an ECM that is more similar to native cardiovascular tissue, we may be able to promote an adaptive tissue remodeling response. Specifically, utilizing iCFs in hTEM production may produce a matrix that more closely mimics the native cardiac microenvironment. Thus, iCFs may provide a more suitable microenvironment for infiltrating host cells to promote better tissue remodeling and a better outcome for the patient. Another challenge in today’s hTEMs is the lack of mature ELN in the final tissue-engineered product, which may affect its integrity and durability once implanted.^81–83 Therefore, combining hiPSC technology with genome engineering to target specific genes involved in the ECM biosynthesis process represents an innovative approach for next-generation hTEMs. By using gene-editing, we can fine-tune the ECM composition and potentially trigger elastinogenesis in vitro.

This study introduces the combination of genome and TE, as a first step toward the modulation and optimization of the ECM composition of hTEM-based tissue constructs. We have successfully generated a novel method of differentiating hiPSCs to iCFs from animal sera-free and chemically defined materials, of which only one supplement uses BSA as a stabilizer that would need to be replaced to reach cGMP compliance. Yet, our protocol possesses the purity, reproducibility, and has the growth potential for large-scale manufacturing. FACS analysis comparing iCFs pre- and post-purification revealed an increase in purity up to 93.2% (Figure 1C), when compared to previously published differentiation methods.^33,42,84 Analysis of the gene expression profile showed complete downregulation of pluripotency markers, while fibroblast-associated markers indicated significant upregulation (Figure 1D and E), indicating complete differentiation and cardiac lineage of the cells, respectively. Additionally, the global cardiac specificity of our iCFs was demonstrated using transcriptomics. In-depth RNA-seq characterization showed similarities in primary hCFs compared to iCFs in expression profiles and gene pathways involved in vasculature development, angiogenesis, and heart development (Figure 2B), highlighting cardiac-specificity of our iCFs. This suggests that iCFs may be an interesting cell source to produce cardiac-specific matrix and cardiovascular-related tissue-engineered implants.

Using these newly defined iCFs to produce hTEMs resulted in histologically similar matrices when compared to hDFB-derived hTEMs. However, an important distinction in utilizing hTEMs for cardiovascular implants is their functional mechanics. Analysis of the mechanical characteristics indicated an increasing trend in the toe and heel region of iCF-derived hTEMs in contrast to the hDFB-derived hTEMs. This suggests that hTEMs generated by iCFs exhibit increased elasticity, while also displaying characteristics of a stiffer matrix, which aligns with the higher collagen content identified in the proteomics. Furthermore, analysis of the transition region demonstrated an increasing trend in the iCF-derived hTEMs. The transition region is the point where the mechanical properties of the tissue shift from being governed by elastic characteristics to being influenced by the collagen fibers. Thus, this region gives an indication of the structure of the collagen,^73,74 suggesting an increased waviness in the collagen fibers in iCF-derived hTEMs. Taken together, iCFs were able to generate hTEMs with equal gross morphology and slightly better mechanical properties than hDFB-derived hTEMs. Therefore, iCFs represent an interesting alternative to hDFBs for the manufacturing of hTEMs, which circumvents limitations imposed by hDFBs such as donor-to-donor variability and limited human primary cell proliferation.

Using the TALEN approach, we targeted our ECM-related genes of interest to the AAVS1 locus to ensure that our genetic modification would not disrupt normal cell behavior and function.⁴⁷ hTEMs produced by edited iCFs displayed differences in the deposition of ECM and the resulting mechanical properties in comparison to both hDFB- and unedited iCF-derived hTEMs. Histological analysis of ELN-, LOX-, and LOXL1-edited hTEMs indicated increased ECM density and compaction of mature collagen fibers (Figure 4H-K). However, our edited hTEMs had no indication of mature ELN fibers in the histology analysis. Mass spectrometry analysis of the hTEMs produced by the edited BION-iCFs provided us with a global overview of the ECM components contributing to the mechanical properties of the hTEMs. The proteomics showed the presence of various collagens, many of which contributing to ECM organization and fibril formation as well as cardiovascular-related pathways (Figure 5). Furthermore, the data showed an increase in the relative abundance of proteoglycans and glycosaminoglycans involved in various processes ranging from matrix assembly to collagen fibers assembly and elastic fibers assembly (Figure 5). This correlates with the histological results of ELN-, LOX-, and LOXL1-edited hTEMs showing increased compaction of the deposited collagens. Histology staining did not identify mature ELN fibers; however, the production of tropoelastin is increased, and mechanical testing shows an increase in the hTEMs elasticity. A potential reason that might explain the absence of mature ELN fibers can be the absence of load on the hTEMs, which does not trigger the assembly of tropoelastin molecules to form ELN fibers. However, the increased deposition of tropoelastin molecules in the hTEMs might provide the necessary building blocks to allow faster assembly once implanted and subject to mechanical load. To investigate this hypothesis, the first step is to produce the hTEMs in a pulsatile bioreactor, in order to modulate the pressure and assess the effect of mechanical load on the formation of ELN fibers. Secondly, the produced hTEM heart valves should be implanted to study their functionality, growth potential, and de novo ECM formation.

Mechanically, edited iCF-derived hTEMs showed an increased trend in the elastic properties when compared to those derived from hDFBs, which could be attributed to an increase in immature ELN and GAGs in the tissue. Furthermore, the gene-edited hTEMs displayed increased stiffness in the heel region and increased strain in the transition region, suggesting higher collagen fiber crosslinking, waviness, and maturation, which is supported by the histological and mass spectrometry analyses. In sum, the upregulation of ELN, LOX, and LOXL1 in iCF-derived hTEMs increased production of mature collagen fibers, which in return enhanced the mechanical properties of the tissue. This indicates that application of genome engineering to TE can specifically modulate the composition and functional properties of hTEMs. This in turn may positively influence the in vivo host response and promote adaptive remodeling of the tissue over time.

Although direct comparisons with other studies are challenging due to differences in culture methods and mechanical testing procedures, our iCF-derived hTEM achieved mechanical properties in the same order of magnitude as previously developed hTEM-based TE heart valve replacements.⁶⁸ In the first study, hDFBs were used in a rotation-based culture method to produce TE heart valves and showed a Young’s modulus between 4 and 6 MPa in the heel region using uniaxial mechanical tests, while our edited iCF-derived hTEMs were between 2 and 3 MPa.⁸⁵ In the second study, TE heart valves were produced by culturing ovine cells in a sophisticated bioreactor system.⁸⁶ These were successfully implanted in the pulmonary position of an ovine model in a long-term study.⁸⁶ Analysis of the mechanical properties showed a range of stress between 0.4 and 0.6 MPa at 15% stretch in the toe region, whereas our edited iCF-derived hTEMs were between 0.1 and 0.2 MPa at 30% stretch.⁸⁶ Furthermore, bioreactor culture settings are known to improve the stiffness and strength of hTEM constructs. Therefore, the implementation of the respective bioreactor system using the edited iCFs from this study may improve the hTEM mechanical properties even further.^39,87,88 However, in order to perform direct comparisons, edited iCF-derived hTEMs must be produced in similar bioreactor systems and assessed using the same mechanical testing method.

Overall, we have comprehensively demonstrated the potential benefits of using a combination of hiPSC biology and genome engineering to create building blocks for “designer” next-generation hTEMs. However, future studies are needed to apply the edited and non-edited iCFs in manufacturing vascular grafts or heart valve replacements for pre-clinical in vivo testing. Although outside of the scope of this study, in situ tissue remodeling and host immune response can only be assessed when implanted in vivo. Further investigation into additional components needed for elastinogenesis and elastin maturation for in vitro production would also prove beneficial.

Conclusion

In this study, we have combined hiPSC technology with genome engineering to lay the foundation for next-generation TE applications by generating a novel cell source, iCFs, that are able to modulate the composition as well as the functional mechanics of hTEMs. We established an animal sera-free and chemically defined differentiation protocol of hiPSCs towards iCFs that demonstrate genotypic and phenotypic similarity to primary hCFs. Furthermore, we stably upregulated ECM-related genes key in collagen and ELN production and maturation in these iCFs to be used as building blocks for the manufacturing of “designer” hTEMs. The data from this study suggest that iCFs represent a valuable cell source and reliable alternative to primary human cells in the manufacturing of hTEMs. iCF-derived hTEMs not only demonstrated their ability to produce tissue with similar structure, but also enhanced ECM composition and mechanical properties. With the addition of iCF gene-editing, we were able to target specific ECM-related proteins to improve ECM composition and function. This study paves the way for implementing hiPSC biology and genome engineering in the field of cardiovascular TE.

Author contributions

Eric K.N. Gähwiler (Conceptualization [lead], Data curation [lead], Formal analysis [lead], Funding acquisition [supporting], Methodology [lead], Visualization [lead], Writing—original draft [lead]), Valery L. Visser (Data curation [supporting], Formal analysis [supporting], Writing—review & editing [supporting]), Melanie Generali (Data curation [supporting], Formal analysis [supporting], Project administration [supporting], Writing—review & editing [supporting]), Dennis Zorndt (Data curation [supporting]), Darcie R. Jackson (Data curation [supporting]), Maximilian Y. Emmert (Project administration [supporting], Supervision [supporting], Writing—review & editing [supporting]), Simon P. Hoerstrup (Conceptualization [equal], Funding acquisition [lead], Project administration [equal], Supervision [equal], Validation [equal], Writing—review & editing [equal]), and Marcy Martin (Conceptualization [equal], Data curation [supporting], Project administration [lead], Supervision [lead], Validation [lead], Visualization [equal], Writing—original draft [lead], Writing—review & editing [lead])

Supplementary material

Supplementary material is available at Stem Cells Translational Medicine online.

Funding

This manuscript was funded through the Mäxi Foundation (S.P.H), which supported E.K.N.G and M.M. M.Y.E. received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program, grant agreement no. 852814 (TAVI4Life), which supported V.L.V. E.K.N.G received funding from the Theodor und Ida Herzog-Egli Foundation.

Conflicts of interest

S.P.H. is a shareholder at Xeltis BV and LifeMatrix AG. M.Y.E. is a shareholder at LifeMatrix. AG. All other authors declare no competing interests.

Data availability

Raw data from RNA-sequencing and proteomics were generated at the Functional Genomics Center Zurich (FGCZ). Data used to derive the reported findings of this study are available from the corresponding author M.Y.E. upon request.