The PERK Branch of the Unfolded Protein Response Safeguards Protein Homeostasis and Mesendoderm Specification of Human Pluripotent Stem Cells

Abstract Cardiac development involves large‐scale rearrangements of the proteome. How the developing cardiac cells maintain the integrity of the proteome during the rapid lineage transition remains unclear. Here it is shown that proteotoxic stress visualized by the misfolded and/or aggregated proteins appears during early cardiac differentiation of human pluripotent stem cells and is resolved by activation of the PERK branch of unfolded protein response (UPR). PERK depletion increases misfolded and/or aggregated protein accumulation, leading to pluripotency exit defect and impaired mesendoderm specification of human pluripotent stem cells. Mechanistically, it is found that PERK safeguards mesendoderm specification through its conserved downstream effector ATF4, which subsequently activates a novel transcriptional target WARS1, to cope with the differentiation‐induced proteotoxic stress. The results indicate that protein quality control represents a previously unrecognized core component of the cardiogenic regulatory network. Broadly, these findings provide a framework for understanding how UPR is integrated into the developmental program by activating the PERK‐ATF4‐WARS1 axis.

Relative protein expression
Figure S4.Generation of PERK knockout hESCs A, Schematic of CRISPR/Cas9-mediated knockout (KO) of PERK and genotyping of two PERK KO hESC clones by Sanger sequencing.B, Sanger sequencing of the potential off-target sites of sgRNA1 or sgRNA2 predicated by Cas-OFFinder in WT and PERK KO hESCs.C, Alkaline phosphatase (AP) staining analysis of WT and PERK KO hESCs.Scale bar, 5 mm.D, Immunofluorescence analysis of the pluripotency markers NANOG, OCT4, and SOX2 in WT and PERK KO hESCs.Scale bars, 100 μm.E and F, Representative (left) and quantitative (right) flow cytometry analysis of pluripotency markers TRA-1-81 (E) and SSEA4 (F) in WT and PERK KO hESCs.n = 4 biologically independent experiments.G, Immunofluorescence analysis of the proliferation marker Ki67 in WT and PERK KO hESCs.Scale bars, 100 μm.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a post-hoc Tukey test.

Figure
Figure S5.PERK is essential for cardiogenesis of H1 hESCs A, Representative flow cytometry analysis of cTNT expression in D10 cultures differentiated from WT, PERK KO, or PERK re-expressed H1 hESCs.B and C, Representative (B) and quantitative (C) immunoblot analysis of the CM markers α-actinin and cTNT in D10 cultures differentiated from WT, PERK KO, or PERK re-expressed H1 hESCs.β-actin was used as a loading control.n = 6 biologically independent experiments.***P<0.001vs. ESC; ### P < 0.001 vs. the corresponding PERK KO clone.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a posthoc Tukey test.
Figure S6.PERK is essential for cardiogenesis of WTC hiPSCs A, Immunoblot analysis of PERK in WT, PERK KO, and PERK re-expressed WTC hiPSC clones.B and C, Representative (B) and quantitative (C) flow cytometric analysis of cTNT expression in D10 cultures differentiated from WT, PERK KO, or PERK re-expressed WTC hiPSCs.n = 6 biologically independent experiments.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.D, Immunofluorescence analysis of CM markers in D10 cultures differentiated from WT, PERK KO, and PERK re-expressed WTC hiPSCs.Scale bars, 200 μm.E and F, Representative (E) and quantitative (F) immunofluorescence analysis of the protein aggregates in D1 cultures differentiated from WT, PERK KO, and PERK re-expressed WTC hiPSCs.n = 3 biologically independent experiments, 10 fields of view per experiment.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.Scale bars, 50 μm.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a posthoc Tukey test.
Figure S7.PERK is essential for cardiogenesis in an alternative RPMI-B27 cardiac differentiation protocol A, Schematic of the RPMI-B27 cardiac differentiation protocol.B, Immunofluorescence analysis of CM markers in D10 cultures differentiated from WT, PERK KO, and PERK re-expressed H1 hESCs using the RPMI-B27 protocol.Scale bars, 200 μm.C and D, Representative (C) and quantitative (D) flow cytometric analysis of cTNT expression in D10 cultures differentiated from WT, PERK KO, or PERK re-expressed H1 hESCs using the RPMI-B27 protocol.n = 6 biologically independent experiments.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a post-hoc Tukey test.
Figure S8.PERK is essential for atrial CM differentiation A, Schematic of the atrial CM differentiation protocol.B, Immunofluorescence analysis of the general CM marker cTNT, ventricular CM marker MLC2v, and atrial CM marker COUP-TFI in D10 cultures differentiated from H1 hESCs using the protocol descripted in A. Scale bars, 100 μm.C and D, Representative (C) and quantitative (D) flow cytometric analysis of cTNT expression in D10 cultures differentiated from WT, PERK KO, or PERK re-expressed H1 hESCs using the protocol shown in A. n = 6 biologically independent experiments.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a post-hoc Tukey test.

FigureFigure
Figure S10.PERK is essential for lateral mesoderm differentiation A, Schematic of the lateral mesoderm differentiation protocol.LM, lateral mesoderm.B, Immunofluorescence analysis of LM markers NKX2-5 and ISL1 in D2 cultures differentiated from H1 hESCs using the protocol shown in A. Scale bars, 100 μm.C and D, Representative (C) and quantitative (D) immunofluorescence analysis of the protein aggregates at each stage of LM differentiation of H1 hESCs.n = 3 biologically independent experiments, 10 fields of view per experiment.***P<0.001vs. ESC; ### P<0.001 vs. D1.Scale bars, 50 μm.E and F, Representative (E) and quantitative (F) immunofluorescence analysis of the protein aggregates in D1 ME cells differentiated from WT, PERK KO, and PERK re-expressed H1 hESCs.n = 3 biologically independent experiments, 10 fields of view per experiment.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.Scale bars, 50 μm.G and H, Representative (G) and quantitative (H) immunofluorescence analysis of the protein aggregates in D2 LM cells differentiated from WT, PERK KO, and PERK re-expressed H1 hESCs.n = 3 biologically independent experiments, 10 fields of view per experiment.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.Scale bars, 50 μm.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a post-hoc Tukey test.

Figure S13 .Figure S14 .Figure S15 .
Figure S13.Stage specific effect of PERK depletion on CM differentiation A, Representative flow cytometry analysis of cTNT expression in H1 hESCs-derived D10 cultures treated with GSK2606414 at various time windows during differentiation.B, RT-qPCR analysis of PERK in H1 hESCs-derived D2 cultures 6 hours after transfected with the Scramble or PERK siRNA at D1. n = 6 biologically independent experiments.***P<0.001vs. Scramble siRNA.C and D, Representative (C) and quantitative (D) flow cytometric analysis of cTNT + cells in H1 hESCs-derived D10 cultures transfected with Scramble or PERK siRNA at D2. n = 6 biologically independent experiments.***P<0.001vs. Scramble siRNA.Data represent mean ± SD.Statistical significance was determined by unpaired two-tailed t-tests.

Figure S16 .Figure
Figure S16.PERK KO hESCs fail to activate the mesoderm transcriptional program during cardiac specificationA, Heatmap showing up-and down-regulated genes in WT and PERK KO mesoderm cells (D2), as compared to WT ESCs determined by RNA-seq.GO analysis of genes deregulated by at least 2-fold in PERK KO mesoderm cells as compared to WT mesoderm cells are presented in the right panel.B-D, Gene set enrichment analysis of the RNA-seq data from WT and PERK KO mesoderm cells (D2).Gene sets from the GO term 'mesoderm development' and 'mesoderm morphogenesis', the ESC-enriched genes, as well as the Reactome Pathways term 'unfolded protein response' and Hallmark Gene Sets term 'unfolded protein response' are used.

FigureWARS1Figure S23 .Figure S24 .
Figure S22.WARS1 activation rescues PERK KO-induced CM differentiation defect A, Representative flow cytometry analysis of cTNT expression in D10 cultures differentiated from WT and PERK KO, as well as PERK KO H1 hESCs that receive WARS1-, HERPUD1-, or PSAT1overexpression, respectively.B, Immunofluorescence analysis of CM markers in D10 cultures differentiated from WT, PERK KO, and WARS1-overexpressed PERK KO WTC hiPSCs.Scale bars, 200 μm.C and D, Representative (C) and quantitative (D) flow cytometric analysis of cTNT expression in D10 cultures differentiated from WT, PERK KO, or WARS1-overexpressed PERK KO WTC hiPSCs.n = 6 biologically independent experiments.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.E and F, Representative (E) and quantitative (F) immunofluorescence analysis of the protein aggregates in D1 cultures differentiated from WT, PERK KO, and WARS1-overexpressed PERK KO WTC hiPSCs.n = 3 biologically independent experiments, 10 fields of view per experiment.Scale bars, 50 μm.***P<0.001vs. WT; ### P<0.001 vs. the corresponding PERK KO clone.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a posthoc Tukey test. 50 Immunofluorescence analysis of CM markers in D10 cultures differentiated from the shScram control and ATF4 KD WTC hiPSCs.Scale bars, 200 μm.I and J, Representative (I) and quantitative (J) immunofluorescence analysis of the protein aggregates in D1 cultures differentiated from the shScram control and ATF4 KD WTC hiPSCs.n = 3 biologically independent experiments, 10 fields of view per experiment.Scale bars, 50 μm.***P<0.001vs. shScram.K and L, Representative (K) and quantitative (L) immunoblot analysis of the ubiquitin in D1 ME cells differentiated from shScram control and ATF4 KD H1 hESCs.β-actin was used as a loading control.n = 6 biologically independent experiments.***P<0.001vs. shScram.Data represent mean ± SD.Statistical significance was determined by one-way ANOVA with a post-hoc Tukey test. W