Association of Human iPSC Gene Signatures and X Chromosome Dosage with Two Distinct Cardiac Differentiation Trajectories - PubMed (original) (raw)

. 2019 Nov 12;13(5):924-938.

doi: 10.1016/j.stemcr.2019.09.011. Epub 2019 Oct 24.

Margaret K R Donovan 2, William W Young Greenwald 2, Jennifer Phuong Nguyen 2, Kyohei Fujita 1, Sherin Hashem 3, Hiroko Matsui 1, Francesca Soncin 4, Mana Parast 4, Michelle C Ward 5, Florence Coulet 1, Erin N Smith 1, Eric Adler 3, Matteo D'Antonio 6, Kelly A Frazer 7

Affiliations

Association of Human iPSC Gene Signatures and X Chromosome Dosage with Two Distinct Cardiac Differentiation Trajectories

Agnieszka D'Antonio-Chronowska et al. Stem Cell Reports. 2019.

Abstract

Despite the importance of understanding how variability across induced pluripotent stem cell (iPSC) lines due to non-genetic factors (clone and passage) influences their differentiation outcome, large-scale studies capable of addressing this question have not yet been conducted. Here, we differentiated 191 iPSC lines to generate iPSC-derived cardiovascular progenitor cells (iPSC-CVPCs). We observed cellular heterogeneity across the iPSC-CVPC samples due to varying fractions of two cell types: cardiomyocytes (CMs) and epicardium-derived cells (EPDCs). Comparing the transcriptomes of CM-fated and EPDC-fated iPSCs, we discovered that 91 signature genes and X chromosome dosage differences are associated with these two distinct cardiac developmental trajectories. In an independent set of 39 iPSCs differentiated into CMs, we confirmed that sex and transcriptional differences affect cardiac-fate outcome. Our study provides novel insights into how iPSC transcriptional and X chromosome gene dosage differences influence their response to differentiation stimuli and, hence, cardiac cell fate.

Keywords: X chromosome erosion; X chromosome inactivation; iPSC; iPSC differentiation; iPSC-derived cardiomyocytes; iPSC-derived cardiovascular progenitor cells; iPSC-derived epicardium; scRNA-seq; single-cell transcriptomics.

Copyright © 2019 The Authors. Published by Elsevier Inc. All rights reserved.

PubMed Disclaimer

Figures

None

Graphical abstract

Figure 1

Figure 1

Characterization of Cellular Heterogeneity in iPSC-CVPC Samples (A) Overview of the study design. Skin fibroblasts from 181 subjects were reprogrammed to iPSCs and differentiated to iPSC-CVPCs (191 lines, 232 differentiations). After WNT pathway activation at day 0 and its inactivation by IWP-2 at days 3–5, cells differentiate to CMs if WNT signaling is successfully inhibited. If WNT signaling is not sufficiently inhibited, cells differentiate to EPDCs. Of the 232 differentiations, 193 were completed (day 25), and we observed that different CVPC samples had different proportions of CMs and EPDCs. (B) Distribution of %cTnT. Dashed red line represents the median value. (C–E) Immunofluorescence staining of (C) iPSC-CVPCs, (D) human atrium, and (E) ventricle with markers DAPI, ACTN1, and CX43. (F–H) Immunofluorescence staining of iPSC-CVPCs with markers DAPI, MLC2a+ and MLC2v+, and MLC2v+MLC2a+ (F). scRNA-seq UMAP plots showing (G) the presence of three populations: CMs (orange), EPDCs (blue), and ESCs (green), and (H) the distribution of the nine analyzed samples (eight iPSC-CVPC lines and one ESC line) across the three clusters. (I) Scatterplot showing the correlation between the %cTnT and the fraction of cells in population 1 (CMs) for each of the nine samples. (J) Heatmap showing across all 34,905 single cells the expression markers for stem cells, CMs, EMT, fibroblasts, and smooth muscle. See also Figures S1 and S2.

Figure 2

Figure 2

Transcriptomic Features of 180 iPSC-CVPC Samples (A) Relative distributions of cell populations estimated using CIBERSORT across 180 iPSC-CVPC samples. (B) Scatterplot showing the correlation between %cTnT (x axis) and the fraction of population 1 in the iPSC-CVPCs calculated using CIBERSORT (y axis). (C) Heatmap showing the expression levels of CM and EPDC marker genes (Figure 1J) in 180 iPSC-CVPC samples. Samples are colored based on their fraction of population 1. (D) PCA of the 1,000 genes with highest variability from 184 iPSC samples, 180 iPSC-CVPCs (triangles colored according to their percentage of population 1), and samples from GTEx (squares—left ventricle, right ventricle, coronary artery, and aorta).

Figure 3

Figure 3

iPSC Gene Signatures Associated with Cardiac Differentiation Fate (A) Testing of ten CM/EPDC ratios (0:100 to 90:10, with 10% increments) to determine the optimal threshold for defining an iPSC as CM-fated or EPDC-fated. For each threshold, the number of iPSC lines defined as CM-fated (orange) or EPDC-fated (blue) is shown. (B) At the same thresholds indicated in (A), shown are the numbers of differentially expressed autosomal genes between the iPSC lines defined as CM-fated and EPDC-fated. The 30:70 threshold has the maximum number of differentially expressed genes. (C) Volcano plot showing mean difference in expression levels for all autosomal genes between CM-fated iPSC lines and EPDC-fated iPSC lines (x axis) and p value (y axis, t test). A positive difference indicates overexpression in CM-fated iPSCs, whereas a negative difference indicates overexpression in EPDC-fated iPSCs. Significant genes are indicated in red. (D) Expression levels of the 91 signature genes in iPSCs as a function of the %CM population in their corresponding iPSC-CVPC samples. Thick lines represent the average for 36 genes overexpressed in CM-fated iPSCs (orange) and for 55 genes overexpressed in EPDC-fated iPSCs (blue). (E) WNT/β-catenin pathway, muscle/cardiac related, or EMT/mesenchymal development signature genes (those differentially expressed with nominal p values [p < 0.0015] indicated with an asterisk). (F) GLM estimate (%CM population/expression) calculated for each signature gene. Mean and 95% confidence interval are shown. (G) Bar plot showing the percentage of variability in iPSC fate that is explained by each of the 91 signature genes. Bars highlighted in red show the 35 signature genes identified by L1 normalization that independently contributed to variance. Due to the fact that the 91 genes do not have independent expression, the total sum of the percent variance explained is >1. See also Figures S3–S5.

Figure 4

Figure 4

X Chromosome Gene Dosage Plays a Role in Cardiac Differentiation Fate (A) GSEA results. For each gene set, −log10(q value) is shown. Positive values correspond to gene sets enriched in CM-fated iPSCs, whereas negative values correspond to EPDC-fated iPSCs. For autosomes all iPSCs were included (top), for the chromosome X only the 113 female iPSCs were analyzed (bottom). Storey q value was used to adjust for multiple testing hypothesis; q values <0.05 were considered significant. (B) Cartoon showing the positions of differentially expressed loci on chromosome X and of _ELK1_ and _PORCN_. (C–F) Bar plot (C) showing the associations between sex and differentiation outcome (orange: iPSC-CVPC samples with CM fraction >30%; blue: with EPDC fraction >70%). p values were calculated using Z test. Density plots showing the differences in allelic imbalance fraction between: (D) autosomal genes (pink) and chrX genes outside of the pseudoautosomal region (maroon) in female iPSCs; (E) chrX genes in female CM-fated (light orange) and EPDC-fated (light blue) iPSCs; (F) chrX genes in female day 25 iPSC-CVPC samples with CM fraction >30% (orange) and EPDC fraction >70% (blue). p values in (D) to (F) were calculated using the Mann-Whitney U test. See also Figure S6.

Figure 5

Figure 5

Validation of Association between iPSC Gene Signatures, Sex, and Differentiation Outcome (A) Schematic depicting differences between the iPSCORE and Yoruba iPSC samples. (B) Estimated fractions of CMs and EPDCs for 13 Yoruba iPSC-CM samples from RNA-seq using CIBERSORT (two iPSC-CMs did not have RNA-seq). (C) Scatterplot showing the correlation between %cTnT and the fraction of cells in population 1 for 13 Yoruba iPSC-CM samples. (D) Box plots showing the distribution of estimated fraction of cells in population 1 in females and males. (E) Box plots showing correlation of gene expression in all 184 iPSCORE iPSCs with RNA-seq (purple), 34 Yoruba iPSCs with RNA-seq used for differentiation, and the pairwise comparison of the Yoruba iPSCs against the iPSCORE iPSCs (gray). (F) Volcano plot showing mean difference in expression levels for all autosomal genes between 14 Yoruba iPSC lines that were successfully differentiated and 125 iPSCORE iPSC-CM-fated lines and p value (y axis, t test). Significant genes are indicated in red. (G) Smooth color density scatterplot showing gene-expression differences between iPSCs with different fates in 184 iPSCORE iPSCs to the expression differences between iPSCs with different outcomes in Yoruba iPSCs (14 successful versus 20 terminated) (y axis). A positive difference indicates shared overexpression of genes between CM-fated iPSC in iPSCORE and successfully differentiated iPSC in the Yoruba set, whereas a negative difference indicates shared overexpression of genes between EPDC-fated iPSC in iPSCORE and terminated iPSC in the Yoruba set. Of the 91 signature genes that were differentially expressed in the iPSCORE iPSCs based on cell fate, eight had nominally significant expression differences in the same direction in the Yoruba iPSC set (shown in red). (H) Bar plot showing that the eight iPSCORE differentially expressed genes in (G) with nominal significant expression differences in the same direction (e.g., overexpressed or down regulated) in the Yoruba iPSCs are greater than random expectation. See also Figure S7.

Figure 6

Figure 6

iPSC Characteristics that Influence Their Cardiac-Fate Determination Cartoon showing iPSC characteristics that influence their cardiac-fate determination, including: (1) the expression levels of 91 genes grouped into three gene signature classes (WNT/B-catenin pathway, cardiac development genes, and genes involved in EMT); (2) sex: female iPSCs are more likely to differentiate to CMs than males; and (3) X chromosome activation state: female iPSCs that have activated both X chromosomes (XaXa) are more likely to differentiate to CMs.

References

    1. Banovich N.E., Li Y.I., Raj A., Ward M.C., Greenside P., Calderon D., Tung P.Y., Burnett J.E., Myrthil M., Thomas S.M. Impact of regulatory variation across human iPSCs and differentiated cells. Genome Res. 2018;28:122–131. -PMC -PubMed
    1. Bao X., Lian X., Hacker T.A., Schmuck E.G., Qian T., Bhute V.J., Han T., Shi M., Drowley L., Plowright A. Long-term self-renewing human epicardial cells generated from pluripotent stem cells under defined xeno-free conditions. Nat. Biomed. Eng. 2016;1 -PMC -PubMed
    1. Barakat T.S., Ghazvini M., de Hoon B., Li T., Eussen B., Douben H., van der Linden R., van der Stap N., Boter M., Laven J.S. Stable X chromosome reactivation in female human induced pluripotent stem cells. Stem Cell Reports. 2015;4:199–208. -PMC -PubMed
    1. Bargehr J., Ong L.P., Colzani M., Davaapil H., Hofsteen P., Bhandari S., Gambardella L., Le Novere N., Iyer D., Sampaziotis F. Epicardial cells derived from human embryonic stem cells augment cardiomyocyte-driven heart regeneration. Nat. Biotechnol. 2019;37:895–906. -PMC -PubMed
    1. Bruck T., Yanuka O., Benvenisty N. Human pluripotent stem cells with distinct X inactivation status show molecular and cellular differences controlled by the X-linked ELK-1 gene. Cell Rep. 2013;4:262–270. -PubMed

Publication types

MeSH terms

Grants and funding

LinkOut - more resources