ELABELA Is an Endogenous Growth Factor that Sustains hESC Self-Renewal via the PI3K/AKT Pathway

http://www.cell.com/cell-stem-cell/abstract/S1934-5909(15)00365-3?rss=yes by


To view the full text, please login as a subscribed user or purchase a subscription. Click here to view the full text on ScienceDirect.

Figure 1

ELA Is Associated with Human Embryonic Stemness

(A) NANOG, POU5F1, and PRDM14 syn-expression groups define a common list of 33 transcripts which are at the core of the human pluripotency circuitry. ELA is one of these genes.

(B) Luciferase reporter assay demonstrating that ELA is under direct transcriptional control by an upstream POU5F1 enhancer.

(C) ELA mRNA levels were measured by qPCR in control and POU5F1-knockdown hESCs.

(D) ELA mRNA expression in undifferentiated hESCs and during embroid body differentiation. Left axis: ELA and POU5F1; right axis: PAX6, SOX17, and NKX2.5, expressed as percentages relative to Day 0.

(E) Secreted ELA is detected in the supernatant of hESCs by ELISA. shRNA-mediated stable knockdown of ELA (shELA) reduces levels of secreted ELA by approximately 85%.

(F) Inducible CRISPR/Cas9 vector for derivation of ELAiCRISPR and AAVS1iCRISPR hESCs.

(G) FACS analysis of ELAiCRISPR and AAVS1iCRISPR hESCs serially passaged in the presence of DOX to track the persistence of GFP-positive genome-edited cells over four passages (P0 to P3).

(H) The percentage of GFP-positive ELAiCRISPR hESCs rapidly declines over four passages compared to GFP-positive AAVS1iCRISPR hESCs. Data are represented by the mean of six wells ± SEM.

(I) Immunofluorescence of ELA in control and shELA hESCs.

See also Figure S1 .

Figure 2

ELA Is Necessary and Sufficient for hESC Growth and Viability

(A) Cell index measurements (xCELLigence) of shELA and shβ2M (inset) hESCs seeded as single cells over 5 days. Cell Index is an approximation of cell numbers.

(B) FACS analysis of SSEA3 and TRA-1-60 3 days after induction of shRNA by DOX.

(C) Immunofluorescence for SSEA3, POU5F1, and TRA-1-60 in control and shELA hESC colonies after four passages of knockdown.

(D) Control and shELA hESCs were injected subcutaneously into NOD-SCID-GAMMA mice. Teratoma formation was visualized after 1 month.

(E) ELA and mutant ELARR > GG (R31G, R32G), with an intramolecular cystine bond between conserved C39 and C44 residues, were synthetically produced to 98% purity.

(F) By immunofluorescence, recombinant ELA, but not ELARR > GG, labeled with N-terminal FITC is rapidly taken up by hESCs.

(G) Brightfield images of shRNA hESCs and wild-type hESCs cultured with exogenous ELA or ELARR > GG after 4 days of shRNA or peptide treatment.

(H) Real-time cell index measurements of hESCs cultured with exogenous ELA or ELARR > GG over 4 days.

(I) Real-time cell index measurements over 5 days of shELA hESCs rescued with exogenous ELA, but not ELARR > GG.

(J) hESCs were cultured with affinity purified α C antibody, which inhibited their growth. This neutralizing activity can be outcompeted by the mutant non-signaling ELARR > GG peptide which competes for the α C antibody.

(K) Real time cell index measurements of multipotent human embryonal carcinoma cells (ECs) or unipotent human chondrosarcoma and primary fibroblast cells cultured with exogenous ELA over 5 days, with no apparent effect.

(L) H1, H9, and SHEF4 hESC lines were grown with exogenous ELA or α C antibody. Cell numbers were measured after 4 days. Data are represented by the mean of six wells ± SEM.

See also Figure S2 .

Figure 3

APLNR Is Not the ELA Receptor in hESCs

(A) FACS analysis of surface-expressed APLNR in undifferentiated (Day 0) versus Day 3 hESC-derived mesoendoderm (Day 3).

(B) AP-ELA binding assay on a variety of hESCs and also differentiated human cell lines. Data represent intensity of colorimetric readout and are means of three wells ± SEM.

(C) AP-ELA binding assay on shControl and shAPLNR undifferentiated hESCs (Day 0) versus Day 3 hESC-derived mesoendoderm (Day 3).

(D) Mutation of the indicated residues to glycine affects binding of AP-ELA to SHEF4 and HES3 hESCs. Input supernatants were normalized by their AP activity to ensure that equal amounts of each AP-ELA mutant were used for binding.

(E) Biotinylated ELA peptide applied to hESCs was detected using streptavidin with or without prior permeabilization of the cell surface membrane by digitonin.

(F) FITC-labeled ELA was bound to hESCs with or without a 5 min pretreatment of cells with a low concentration of trypsin, followed by FACs analysis.

(G) FITC-labeled ELA was bound to hESCs in the presence or absence of methyl-β-cyclodextrin (MβCD) or Chlopromazine, followed by FACs analysis.

See also Figure S3 .

Figure 4

ELA Is the Endogenous Signal for Activation of PI3K/AKT in hESCs

(A) Schematic of SILAC-based phospho-proteomic analysis to elucidate immediate signal transduction of ELA in hESCs.

(B) Mass spectra of a PRAS40-derived peptide showing phosphorylation on T246 by ELA, but not ELARR > GG, suggesting activation of the AKT pathway.

(C) hESCs were pulsed with ELA and lysed at the indicated time points. Western blots show immediate activation of the PI3K/AKT and mTORC1 pathways. Lysates for the top and bottom panels were derived from separate technical replicates.

(D) Activation of AKT by ELA is dependent on PI3K and is abrogated by pan-PI3K inhibitor LY294002 (LY).

(E) Activation of AKT by ELA in shAPLNR hESCs is not impaired.

(F)Western blots of pAKT in Control and shELA hESCs grown in decreasing INSULIN concentrations for 24 hr reveal the requirement for ELA-mediated AKT activation.

(G) By real-time cell index analysis over 5 days, ELA, but not ELARR > GG, can partially rescue the requirement for INSULIN in hESCs growth medium.

See also Figure S4 .

Figure 5

ELA and INSULIN Are Functionally Distinct

(A) PCA analysis of microarray data using probesets that showed at least greater than 1.5-fold change (one-way ANOVA, p < 0.05) between at least one pair of conditions.

(B) Venn diagrams depicting the overlap of probesets changed by more than 1.5-fold in ELA-treated (representing 42 upregulated and 16 downregulated genes) and INSULIN-treated (1 upregulated and 5 downregulated genes) hESCs.

(C) Self-organizing map of all probesets in the dataset with greater than 1.5-fold change between at least one pair of conditions (one-way ANOVA, p < 0.05). Selected clusters of ELA-dependent genes are highlighted to depict their variable dependence on PI3K/AKT activity.

See also Figure S5 .

Figure 6

ELA Promotes Translation and Proliferation and Protects against Stress-Induced Apoptosis

(A) GSEA profile plots depicting negative enrichment of ribosomal genes and genes involved in translation in shELA compared to control hESCs. NES, normalized enrichment score.

(B) Pulse-chase analysis by metabolic labeling to measure the rate of newly synthesized proteins in shControl versus shELA hESCs. Cells were harvested for FACs analysis of incorporated fluorescent amino acid 15 and 75 min after pulsing.

(C) Metabolic labeling to measure the rate of newly synthesized proteins in Control versus ELA-treated hESCs in the absence (top) or presence (bottom) of Rapamycin and LY. Cells were harvested for microscopic measurement of label incorporation 15 min after the addition of the amino acid label.

(D) 23 hours following release from a double thymidine block, shELA hESCs show an accumulation of cells in the G1 phase as measured by DNA content.

(E) Cell-cycle analysis using FUCCI-H9 hESCs synchronized by a double thymidine block following treatment with ELA, LY, or both 19 hr post-release.

(F) Quantitation of data in (E) at the indicated time points following thymidine block release.

(G) Cell numbers 48 hr following thymidine block release.

(H) Western blot analysis of CYCLIN D1 levels following an ELA pulse.

(I) By FACS analysis, a larger proportion of shELA hESCs are positive for ANNEXIN V, which marks apoptotic cells.

(J) Activated CASPASE 3 can be detected in shELA hESCs by immunofluorescence, but not in control hESCs.

(K) ELA can partially replace the ROCK inhibitor to prevent anoikis following single-cell dissociation.

(L) ELA-treated hESCs are more resistant to γ-irradiation compared to control hESCs. shELA hESCs are sensitized to γ-irradiation, which can be rescued by the addition of ELA. Data are representative of three independent experiments.

(M) ELA-treated hESCs are more resistant to low levels of Actinomycin D treatment, which induces transcriptional stress and p53-dependent cell death. shELA hESCs are more sensitive but can be rescued by addition of ELA. Data are representative of six independent experiments.

See also Figure S6 .

Figure 7

ELA Poises hESCs toward the Mesendodermal Lineage

(A) ELA, but not INSULIN, activates pSMAD3, as shown by western blot.

(B) ELA-mediated activation of pSMAD3 is insensitive to PI3K inhibition by LY.

(C) Immunofluorescence of pSMAD3 in hESCs pulsed with ELA.

(D) qPCR analysis of mesendoderm lineage genes in hESCs grown in the presence of ELA or depleted of ELA for 72 hr.

(E) FACs analysis of SSEA3 and TRA-1-60 levels on hESCs grown in the presence of ELA for 72 hr.

(F) qPCR analysis of POU5F1 and NANOG in hESCs grown in the presence of ELA for 72 hr.

(G) qPCR analysis of germ layer markers during embroid body formation from hESCs grown in the presence (pink) or absence (black) of exogenous ELA prior to differentiation.

(H) SOX17 immunofluorescence of hESC-derived definitive endoderm at Day 3 and Day 5 of differentiation.

(I) Quantitation of SOX17-positive cells in (H).

See also Figure S7 .


  • ELA is a peptide hormone secreted by hESCs that activates the PI3K/AKT pathway
  • ELA promotes self-renewal via cell-cycle progression and protein translation
  • ELA potentiates the TGFβ pathway, priming hESCs toward the endoderm lineage
  • hESCs do not express APLNR, so ELA may have an alternate unknown receptor


ELABELA (ELA) is a peptide hormone required for heart development that signals via the Apelin Receptor (APLNR, APJ). ELA is also abundantly secreted by human embryonic stem cells (hESCs), which do not express APLNR. Here we show that ELA signals in a paracrine fashion in hESCs to maintain self-renewal. ELA inhibition by CRISPR/Cas9-mediated deletion, shRNA, or neutralizing antibodies causes reduced hESC growth, cell death, and loss of pluripotency. Global phosphoproteomic and transcriptomic analyses of ELA-pulsed hESCs show that it activates PI3K/AKT/mTORC1 signaling required for cell survival. ELA promotes hESC cell-cycle progression and protein translation and blocks stress-induced apoptosis. INSULIN and ELA have partially overlapping functions in hESC medium, but only ELA can potentiate the TGFβ pathway to prime hESCs toward the endoderm lineage. We propose that ELA, acting through an alternate cell-surface receptor, is an endogenous secreted growth factor in human embryos and hESCs that promotes growth and pluripotency.

To access this article, please choose from the options below

Purchase access to this article

You must be logged in to purchase this article.

Claim Access

If you are a current subscriber with Society Membership or an Account Number, claim your access now.

Subscribe to this title

Purchase a subscription to gain access to this and all other articles in this journal.

Institutional Access

Visit ScienceDirect to see if you have access via your institution.