Cardiogenic programming of human pluripotent stem cells by dose-controlled activation of EOMES

Abstract

Master cell fate determinants are thought to induce specific cell lineages in gastrulation by orchestrating entire gene programs. The T-box transcription factor EOMES (eomesodermin) is crucially required for the development of the heart-yet it is equally important for endoderm specification suggesting that it may act in a context-dependent manner. Here, we define an unrecognized interplay between EOMES and the WNT signaling pathway in controlling cardiac induction by using loss and gain-of-function approaches in human embryonic stem cells. Dose-dependent EOMES induction alone can fully replace a cocktail of signaling molecules otherwise essential for the specification of cardiogenic mesoderm. Highly efficient cardiomyocyte programming by EOMES mechanistically involves autocrine activation of canonical WNT signaling via the WNT3 ligand, which necessitates a shutdown of this axis at a subsequent stage. Our findings provide insights into human germ layer induction and bear biotechnological potential for the robust production of cardiomyocytes from engineered stem cells.

Fig. 1

EOMES knockout hESCs fail to differentiate into cardiomyocytes. a Immunoblot confirming EOMES expression and its absence in WT and KO cells, respectively, at the cardiac mesoderm stage of directed differentiation. b EOMES KO cells fail to express the early cardiomyocyte marker NKX2.5 following exposure to a directed differentiation protocol. Scale bar: 50 µm. c EOMES KO cells show a general failure in markedly upregulating essential pan-cardiac genes (qPCR data, n = 3; error bars: s.e.m.). For each gene, the expression level in the highest expressing sample (“max” = WT in all three cases) is set to 1. d Differentiation of EOMES KO hESCs under signaling factor-assisted, non-cardiac mesoderm/endoderm-permissive culture conditions. Top: EOMES KO cells fail to express α-fetoprotein following Activin A-assisted endodermal induction. Scale bar: 50 µm. Bottom: Non-cardiac mesodermal differentiation competence of EOMES KO hESCs. KO cells were differentiated using varied meso and endoderm-permissive induction conditions. The data denote transformed p-values of annotation terms enriched in gene sets upregulated in EOMES KO samples compared to undifferentiated hESCs or differentiated WT controls (cutoffs: 10 or 3-fold, respectively). Underlying data: Supplementary Data 1

Fig. 2

EOMES programs hESCs into functional CMs at high efficiency. a Illustration of growth factor-mediated and optimized EOMES induction-based cardiac differentiation protocols. Bottom right: Immunoblot validating doxycycline-dependent EOMES expression in a transgenic EOMES^KO/E.TET-ON hESC line. b Typical yields of hESC-CMs (left, flow cytometry) and NKX2.5 expression (right, immunoblot) obtained with the two protocols (day 10). c Immunostainings 21 days after the initiation of EOMES induction. Weak perinuclear ANP staining is typical in overall MLC2v-positive hESC-CMs. Scale bars: 25 (top) and 50 µm (bottom). d Acceleration and slowdown of spontaneous beat rates in pCMs following exposure to 10 µM isoprenaline and 10 µM propranolol, respectively, on multielectrode arrays. e Microarray-based time course analysis comparing the indicated protocols and cell lines. RESCUE cells carry an inducible EOMES transgene on EOMES^KO HuES6 background. Underlying data are from Supplementary Data 2

Fig. 3

Contextual requirements of EOMES-mediated CM programming. a DOX dose dependency of the EOMES TET-ON protocol using the WT^E.TET-ON hESC line. Top panel: Immunostains at 1.5 wk. Scale bar: 100 µm. Bottom: Flow cytometry analysis. Numbers indicate average percentages of cTnT-positive CMs from 3–6 experiments per condition. b CM programming by EOMES necessitates suppression of autocrine WNT signaling from the third day of transgene induction (qPCR data, n = 2). c DOX dose-dependent CM programming using an independent WT^E.TET-ON hiPSC line. The data shows immunostaining (scale bar: 100 µm), RT-qPCR (n = 4), and FACS analysis (n = 3) performed at 1 wk of differentiation. The asymmetrical shape of the DOX titration data with this line is in part due to the incomplete repression of SOX2 at 0.05 µg/ml, which caused overgrowth of the cultures with neural precursors (also see Supplementary Fig. 3f). Error bars: s.e.m.

Fig. 4

Cardiac programming by EOMES is based on the downstream induction of WNT3. a Short-term induction of canonical WNT but not BMP target genes during the time of DOX treatment (array data from Supplementary Data 2; error bars: bead s.d.). b Signaling inhibitor experiments with constant DOX at days 0–2 using indicated treatments with WNT (top, 200 nM C-59) and BMP (bottom, 50 nM dorsomorphin) antagonists (qPCR data, n = 3). C-59 had to be administered to all samples of the bottom experiments to provide overall CM-permissive differentiation conditions as a baseline (see Fig. 3b). Data are normalized to the top samples in each of the two panels. c Canonical WNT ligand gene induction in the TET-ON protocol (from Supplementary Data 2; error bars: bead s.d.). d EOMES ChIP-qPCR using DOX-induced hESCs differentiated until the day 2 cardiac mesoderm stage (n = 3; error bars: s.e.m.). Amplicons overlap with hit regions identified by ChIP-seq. No-DOX samples served as specificity controls. e RT-PCR validation of frameshift-causing splice mutations induced by CRISPR/Cas9n in the WNT3 and WNT3A genes (also see Supplementary Fig. 4b). All cell lines were differentiated into cardiac mesoderm prior to RNA isolation. f TET-ON differentiation of the indicated cell lines (n = 3; error bars: s.e.m.). g WNT3 but not WNT3A is required for EOMES-mediated cardiac induction (top), and its disruption can be compensated by extrinsic WNT activation using CHIR99021 (8 µM, bottom panel, representative immunostains at 2 wk). Scale bar: 100 µm. h General model illustrating the interplay between EOMES and WNT signaling in cardiac mesoderm induction. Subsequent WNT inhibition silences the module to allow cardiac differentiation to proceed. A too weak or a too strong activation of the induction loop promotes non-cardiac fates as indicated. The regulatory link between canonical regulatory link between canonical WNT signaling and EOMES has previously been established