Protein-free media for cardiac differentiation of hPSCs in 2000 mL suspension culture

Abstract

Background: Commonly used media for the differentiation of human pluripotent stem cells into cardiomyocytes (hPSC-CMs) contain high concentrations of proteins, in particular albumin, which is prone to quality variations and presents a substantial cost factor, hampering the clinical translation of in vitro-generated cardiomyocytes for heart repair. To overcome these limitations, we have developed chemically defined, entirely protein-free media based on RPMI, supplemented with L-ascorbic acid 2-phosphate (AA-2P) and either the non-ionic surfactant Pluronic F-68 or a specific polyvinyl alcohol (PVA).

Methods and results: Both media compositions enable the efficient, directed differentiation of embryonic and induced hPSCs, matching the cell yields and cardiomyocyte purity ranging from 85 to 99% achieved with the widely used protein-based CDM3 medium. The protein-free differentiation approach was readily up-scaled to a 2000 mL process scale in a fully controlled stirred tank bioreactor in suspension culture, producing > 1.3 × 10⁹ cardiomyocytes in a single process run. Transcriptome analysis, flow cytometry, electrophysiology, and contractile force measurements revealed that the mass-produced cardiomyocytes differentiated in protein-free medium exhibit the expected ventricular-like properties equivalent to the well-established characteristics of CDM3-control cells.

Conclusions: This study promotes the robustness and upscaling of the cardiomyogenic differentiation process, substantially reduces media costs, and provides an important step toward the clinical translation of hPSC-CMs for heart regeneration.

Keywords: Bioreactor; Cardiomyocytes; Protein-free differentiation media; hPSC.

Fig. 1

BA + PF-68 and BA + PVA1 achieve CM yield and purities comparable to CDM3 in shaken Erlenmeyer flasks. A Schematic of differentiation. hPSC aggregates are formed over three days of expansion in STBRs and transferred to Erlenemeyer flasks, where different differentiation media were used. B Expression of MIXL1-GFP on dd1 for the investigated differentiation media assessed by flow cytometry compared to CDM3 (dashed line; n = 3; results are mean ± s.d.). C Cell yields on dd10 for respective media formulations compared to CDM3 cell yields (dashed line; n = 3; results are mean ± s.d.). D Representative flow cytometry plots revealing the expression of investigated CM markers cardiac troponin T (cTnT), pan-myosin heavy chain (MHC), and sarcomeric actinin (SA) for CDM3, BA + PF-68, and BA + PVA1

Fig. 2

Protein-free differentiation media closely resemble typical patterns observed for CDM3 at 150 mL process scale. A Schematic of differentiation. hPSC aggregates were formed over two or three days (depending on cell line). The viable cell density was adjusted to 75 × 10⁶ hPSCs per 150 mL bioreactor, and the differentiation was performed. B Viable cell number and viability for the hESC line HES3 MIXL1-GFP for CDM3 (black), BA + PVA1 (blue), BA + PF-68 (green) throughout the 10-day lasting differentiation process (n = 3 for each medium; mean ± s.d.) and C for the hiPSC line Phoenix HSC_ADCF_SeV_iPS2 for CDM3 (black) and BA + PVA1 (blue; n = 3 for each medium; mean ± s.d.). D CM-specific markers cTnT, pan-MHC, and SA for the hESC line MIXL1-GFP for CDM3 (black), BA + PVA1 (blue), BA + PF-68 (green; n = 3 for each medium; mean ± s.d.) and E for the hiPSC line Phoenix for CDM3 (black) and BA + PVA1 (blue; n = 3 for each medium; mean ± s.d.). F Aggregate size development during differentiation for exemplary processes in CDM3 or G BA + PVA1 (cell line Phoenix; each dot represents a single analyzed aggregate, red lines indicate mean ± s.d.)

Fig. 3

Contraction and electrophysiological analysis revealed no substantial differences between CMs from protein-free differentiation and CDM3. A Bioartificial cardiac tissues (BCTs) were produced from three independent differentiation runs in either BA + PVA1 or CDM3 medium (cell line Phoenix, 3 BCTs per biological replicate for 3 biological replicates; mean ± SEM). The paced and spontaneous contraction forces increase to a similar level (mN/mm²) with increasing preload. The values were normalized to the diameter of the individual BCT. B Individual BCT forces (mN/mm.²) at the maximum stretch (1200 µm; mean ± SEM). C Spontaneous beating frequency (Hz) for BCTs from CMs differentiated in BA + PVA1 or CDM3 at 0 µm and 1200 µm preload (mean ± SEM). D Representative ventricular-like action potentials (APs) for BA + PVA1-derived and E CDM3-derived, seeded CMs were revealed by patch clamp analysis. F Duration of the action potential at 50% of the amplitude (APD50) for CMs derived from BA + PVA1 and CDM3 conditions. G Frequency of APs/min for CMs derived from BA + PVA1 and CDM3 conditions H AP amplitude (in mV) for CMs derived from BA + PVA1 and CDM3 conditions. I Resting membrane potential (in mV) for CMs derived from BA + PVA1 and CDM3 conditions (for BA + PVA1 n = 3 biological replicates, for CDM3 control n = 1 in F-I) J-M Comparison of APD50, AP frequency, AP amplitude, and RMP in three biologically independent CM batches produced in BA + PVA1 (all mean ± SEM)

Fig. 4

Transcriptomic analysis of CMs from BA + PVA1 and CDM3 revealed significant differences. A Principal component analysis of three biologically independent replicates of CMs from BA + PVA1 (red) and CDM3 (blue) conditions according to K-means clustering. B Volcano plot of transcripts identified in all samples with ≥ 10 counts. Significantly up/downregulated transcripts (p ≥ 0.05) with a fold change ≥ 1/ ≤ − 1 between the two groups are marked in green or red, respectively. Three independent biological samples with comparable cTnT purity (according to flow cytometry analysis) were analyzed for both conditions. C + D Gene Ontology analysis of the most significant results between CMs differentiated in BA + PVA1 or CDM3 to be upregulated in BA + PVA1 (C) or downregulated in BA + PVA1 (D). Only genes of transcripts with a fold change ≥ 1/ ≤ − 1 and p ≥ 0.05 were considered

Fig. 5

Successful upscaling of differentiation to 2000 mL bioreactor (Bioflo 320). A Viable cell number and viability for the hiPSC line hHSC_1285T_iPS2 in a 2000 mL bioreactor differentiated in BA + PVA1 throughout the 10-day differentiation process (n = 3; mean ± s.d.). B CM-specific markers cTnT, pan-MHC, and SA for the hiPSC line for differentiation runs in A (n = 3; mean ± s.d.). C Microscopic analysis of aggregate diameters during the 10-day differentiation process are comparable between an exemplary 150 mL process and a 2000 mL process when utilizing the same cell line hHSC_1285T_iPS2 (each dot represents a single aggregate; mean ± s.d.) D Exemplary depiction of the respective markers (cTnT, pan-MHC, SA) for a 2000 mL differentiation process. E Total biomass generated in one 2000 mL differentiation process