Controlling expansion and cardiomyogenic differentiation of human pluripotent stem cells in scalable suspension culture

Abstract

To harness the potential of human pluripotent stem cells (hPSCs), an abundant supply of their progenies is required. Here, hPSC expansion as matrix-independent aggregates in suspension culture was combined with cardiomyogenic differentiation using chemical Wnt pathway modulators. A multiwell screen was scaled up to stirred Erlenmeyer flasks and subsequently to tank bioreactors, applying controlled feeding strategies (batch and cyclic perfusion). Cardiomyogenesis was sensitive to the GSK3 inhibitor CHIR99021 concentration, whereas the aggregate size was no prevailing factor across culture platforms. However, in bioreactors, the pattern of aggregate formation in the expansion phase dominated subsequent differentiation. Global profiling revealed a culture-dependent expression of BMP agonists/antagonists, suggesting their decisive role in cell-fate determination. Furthermore, metallothionein was discovered as a potentially stress-related marker in hPSCs. In 100 ml bioreactors, the production of 40 million predominantly ventricular-like cardiomyocytes (up to 85% purity) was enabled that were directly applicable to bioartificial cardiac tissue formation.

Graphical abstract

Figure 1

Efficient Cardiomyogenic Differentiation in Static Suspension Culture Works Only in a Tight Range of CHIR Concentration (A) Scheme of the expansion/differentiation protocol in a 12-well format. (B) Microscopic assessment of NKX2.5-GFP transgenic HES3 EBs on day 10 in response to the respective CHIR concentration supplemented for 24 hr at day 0. (C) Flow cytometry on day 10 revealed the highest content of GFP⁺ cells at 7.5 μM CHIR (n = 3 independent experiments, mean ± SEM). (D) Bars represent the total cell count per well and respective GFP content (green), confirming 7.5 μM CHIR as the most efficient concentration for CM induction (n = 3 experiments, mean ± SEM). (E and F) Differentiation of the hiPSC lines hHSC_F1285T_iPS2 and hCBiPS2 at 7.5 μM CHIR confirmed the robustness of the established protocol, as shown by microscopic images of plated cells derived from day 10–13 EBs (E; NKX2.5 in yellow, α-actinin and cTNT in red, DAPI in blue; scale bars represent 100 μm) and flow cytometry of d10 EB-derived CBiPS2 cells specific to respective sarcomeric proteins. (F) Isotype controls in gray. See also Figure S1.

Figure 2

Protocol Scale Up to Rotated Erlenmeyer Flasks (A) Culture scheme in flasks rotated at 75 rpm. (B) Microscopic assessment of NKX2.5-GFP expression in EBs on day 13 of differentiation in response to 7.5 μM CHIR. (C) Immunofluorescent staining specific to α-actinin and cTNT on EB sections (top) and dissociated/plated cells derived thereof (bottom). (D) Flow cytometry on day 7–10 revealed ∼55% of GFP⁺ (n = 5 of four independent experiments; mean ± SEM). (E) Representative flow cytometry histograms of day 10 EB-derived cells. See also Figure S1.

Figure 3

Feeding Strategy Determines the Differentiation Outcome in Stirred Bioreactors (A) Representative aggregate images generated from batch (top) and C-perfusion (bottom). The scale bar represents 1 mm. (B) Significantly larger spheres were observed at C-perfusion 24 hr postinoculation (n = 3 bioreactor runs, mean ± SEM). (C) Upregulated genes in C-perfusion-derived CPAs (top) and batch-derived Bas (bottom) detected by microarray analysis. Gene lists represent the intersecting set of >2-fold upregulated and significantly regulated genes identified by RCUTAS and Qlucore Omics Explorer, respectively. (D) Scheme of the defined expansion/differentiation protocol in bioreactors. (E) Microscopic assessment of differentiating aggregates/EBs from days 1 to 5 (left) and GFP fluorescence on day 7, which was discovered only upon differentiation of C-perfusion-derived cells (right). The scale bar represents 1 mm. (F) Gene expression analysis by qRT-PCR for markers of pluripotency (NANOG, OCT4), primitive streak (T-brachyury and MIXL1), mesoderm (MESP1, GATA4, TBX3), and early cardiomyogenesis (ISL1, NKX2.5) comparing differentiation of BAs (gray columns) and CPAs (green) (n = 3 independent bioreactor runs each). See also Figures S1–S3 and Movie S1.

Figure 4

Characterization of Cardiomyogenesis in Cyclic Perfusion-Initiated Differentiations Revealed an Average Production of 40 × 10⁶ CMs per Bioreactor Run, with >60% Average and up to ∼85% CM Content (A) Gene expression analysis by qRT-PCR for cardiac specific markers over time (n = 3 independent bioreactor runs). (B) Microscopic assessment of NKX2.5-GFP transgenic HES3 on day 10 of differentiation shows homogenous GFP expression in nearly all EBs. (C) Relative (top) and total (bottom) numbers of NKX2.5-GFP- or cTNT-positive cells on day 10 of differentiation (n = 3 independent bioreactor runs). (D) Immunofluorescent staining specific to sarcomeric proteins on HES-derived EB sections (top; the scale bar represents 100 μm) and iPS-derived, plated cells from EBs dissociated on day 10 (bottom; the scale bar represents 50 μm). (E) Representative flow cytometry data of EB-derived cells. See also Figure S4.

Figure 5

Electrophysiological and Pharmacological Characterization and Generation of BCT from Bioreactor-Derived EBs (A–C) MEA-derived field potentials (FPs) revealed a positive chronotropic response by the beta-adrenergic agonist isoproterenol at 2 μM (A), prolonged FP duration at >1 μM quinidine (Q), and reduced spike amplitude at 10 μM (B), and a negative chronotropic response accompanied by a shortening of the FP duration in response to 300 nM verapamil (V) treatment (C). Effects were reversible by washout (W) (baseline [B]). n = 2 independent experiments. (D) Representative recordings of a spontaneously active CM displaying a ventricular-like AP (top) and evoked APs representing ventricular- and atrial-like cells (bottom). Arrows in the bottom panels denote initial voltage responses to intracellular stimulation by short depolarizing current steps (left: 600 pA, 1 ms; right: 900 pA, 1 ms). (E) Distribution of atrial- and ventricular-like cells derived from two differentiations. (F) BCTs generated from EBs without (w/o, black lines/columns) or with (w, red) HFF on day 21. The scale bar represents 1 mm. (G) Force measurements of BCTs generated with HFF showed significantly higher active forces (L_max; top right) and a more physiological Frank Starling mechanism (top left) compared with BCTs without HFF, but no significant difference in passive forces (bottom left and right). n = 4–5 of independent BCTs. See also Figure S5.