Systematic identification of culture conditions for induction and maintenance of naive human pluripotency

Abstract

Embryonic stem cells (ESCs) of mice and humans have distinct molecular and biological characteristics, raising the question of whether an earlier, "naive" state of pluripotency may exist in humans. Here we took a systematic approach to identify small molecules that support self-renewal of naive human ESCs based on maintenance of endogenous OCT4 distal enhancer activity, a molecular signature of ground state pluripotency. Iterative chemical screening identified a combination of five kinase inhibitors that induces and maintains OCT4 distal enhancer activity when applied directly to conventional human ESCs. These inhibitors generate human pluripotent cells in which transcription factors associated with the ground state of pluripotency are highly upregulated and bivalent chromatin domains are depleted. Comparison with previously reported naive human ESCs indicates that our conditions capture a distinct pluripotent state in humans that closely resembles that of mouse ESCs. This study presents a framework for defining the culture requirements of naive human pluripotent cells.

Graphical abstract

Figure 1

A Reporter System for Naive Human Pluripotency Based on Endogenous OCT4 Distal Enhancer Activity (A) Proximal enhancer (PE) targeting strategy in human ESCs containing a 2A-GFP sequence in frame with the 3′ UTR of OCT4. (B) Southern blot analysis confirming disruption of PE in OCT4-2A-GFP ESCs. NdeI-digested genomic DNA was hybridized with 5′ and 3′ external probes. Expected fragment size: WT (wild-type) = 5.6 kb, T (targeted) = 6.4 kb. (C) Images of OCT4-2A-GFP human ESCs before (left) and after (right) TALEN-mediated deletion of the PE. 40× magnification. (D) Single-molecule RNA FISH analysis for OCT4 and GFP transcripts in OCT4-2A-GFP human ESCs before and after TALEN-mediated disruption of the PE. (E) Flow cytometric analysis of the proportion of OCT4-ΔPE-GFP+ cells obtained after DOX induction of lentiviral KLF2, NANOG, or KLF2+NANOG. After primary infection WIBR3 human ESCs containing the OCT4-ΔPE-GFP reporter allele were trypsinized and treated with primed human ESC medium (hESM), 2i/L, or 2i/L/DOX for 1 week. R1, total proportion of GFP+ cells, includes weak GFP activity observed in primed human ESCs; R2, subset of cells with high GFP activity observed only upon combined overexpression of KLF2+NANOG. (F) Phase and fluorescence images and flow cytometric analysis of a clonal line of WIBR3 OCT4-ΔPE-GFP+ cells derived in 2i/L/DOX (left). Phase and fluorescence images and flow cytometric analysis after replating in the absence of DOX for 1 week (right) are also shown. 40× magnification. (G) Quantitative gene expression analysis for lentiviral FUW-KLF2, lentiviral FUW-NANOG, endogenous OCT4, and endogenous KLF4 in WIBR3 hESCs cultured in hESM and clonal OCT4-ΔPE-GFP+ derivatives generated in 2i/L/DOX. Error bars indicate ± 1 SD of technical replicates. (H) Phase and fluorescence images of primitive neural stem cells (pNSCs) derived by treating WIBR3 hESCs containing the OCT4-ΔPE-GFP allele with 2i/L for three passages. 100× magnification. (I) Immunofluorescence staining for OCT4 and NESTIN in a clonal line of OCT4-ΔPE-GFP+ cells derived in 2i/L/DOX and a clonal line of OCT4-ΔPE-GFP− pNSCs derived in 2i/L. 100× magnification. (J) Model representing the distinct phenotypic responses of hESCs to treatment with 2i/L and 2i/L/DOX. OCT4-ΔPE-GFP+ cells generated in 2i/L/DOX do not maintain reporter activity upon transgene withdrawal. OCT4-ΔPE-GFP+ cells can revert back to the conventional “primed” hESC state by re-exposure to serum and FGF.

Figure 2

Identification of Small Molecules that Maintain OCT4-ΔPE-GFP Activity after Transgene Withdrawal (A) Strategy for screening a kinase inhibitor library to identify compounds that maintain OCT4-ΔPE-GFP reporter activity upon withdrawal of DOX-dependent KLF2 and NANOG expression. (B) Raw data obtained from high-throughput flow cytometric analysis of the proportion of OCT4-ΔPE-GFP+ cells in 96-well plates supplemented with a kinase inhibitor library (n = 2). (C) Hit compounds from maintenance screen using a clonal line of WIBR3 OCT4-ΔPE-GFP+ ESCs established in 2i/L/DOX. (D) Phase images (top) and flow cytometric analysis of the proportion of OCT4-ΔPE-GFP+ cells (bottom) in a clonal line of OCT4-ΔPE-GFP+ cells derived in 2i/L/DOX and maintained for 10 passages without DOX in the presence of each candidate compound. 40× magnification. (E) Quantitative gene expression analysis for lentiviral FUW-KLF2, lentiviral FUW-NANOG, endogenous OCT4, and GFP in a clonal line of OCT4-ΔPE-GFP+ cells maintained in 2i/L/DOX or for five passages without DOX in the presence of each candidate compound. Error bars indicate ± 1 SD of technical replicates. (F) Chemical structure of the BRAF inhibitor SB590885. (G) Phase images of a clonal line of WIBR3 human ESCs established in 2i/L upon DOX-mediated induction of KLF2 and NANOG (top), and the same line maintained for eight passages without DOX in 2i/L/SB590885 (1 μM) (bottom). 40× magnification. (H) Quantitative gene expression analysis for lentiviral FUW-KLF2, lentiviral FUW-NANOG, and endogenous OCT4 in two clonal lines of WIBR3 human ESCs maintained for eight passages without DOX in 2i/L/SB590885 (1 μM). Error bars indicate ± 1 SD of technical replicates.

Figure 3

Optimization of Medium for Maintaining Viable OCT4-ΔPE-GFP+ Cells (A) Flow cytometric analyses of the proportion of OCT4-ΔPE-GFP+ cells in a 96-well plate 1 week after culture in 2i/L/DOX, 2i/L alone, or 2i/L/SB590885 (1 μM). Top panel shows quantification of OCT4-ΔPE-GFP+ cells without including live/dead discrimination. Bottom panel shows quantification of OCT4-ΔPE-GFP+ cells after gating out DAPI+ cells. (B) Strategy for screening a kinase inhibitor library to identify compounds that improve the fraction of viable (DAPI−) OCT4-ΔPE-GFP+ cells maintained without DOX for two passages in 2i/L/SB590995 (1 μM). (C) Raw data obtained from high-throughput flow cytometric analysis of the proportion of DAPI−/OCT4-ΔPE-GFP+ cells in 96-well plates supplemented with one plate of a kinase inhibitor library (n = 2). Hit compound WH-4-023 is indicated with chemical structure (Martin et al., 2006). (D and E) High-throughput flow cytometric quantification of the proportion of DAPI−/OCT4-ΔPE-GFP+ cells in 96 wells cultured for one passage (D) or two passages (E) in 64 different concentrations of PD0325901, CHIR99021, and SB590885. Asterisk denotes score of the standard concentration of the three inhibitors used in the preceding experiments (1 μM PD0325901, 3 μM CHIR99021, and 1 μM SB590885). Arrows indicate concentrations producing the highest score at P2-DOX. (F) Phase and fluorescence images (top) and flow cytometric analysis of the proportion of OCT4-ΔPE-GFP+ cells (bottom) in a clonal line of OCT4-ΔPE-GFP+ cells derived in 2i/L/DOX and maintained for two passages without DOX in 2i/L/SB590885^opt/Y-27632 or 2i/L/SB590885^opt/Y-27632/WH-4-023. Opt, optimized concentrations of PD0325901, CHIR99021 and SB590885 (see Figure 3E). 100× magnification. (G) Phase and fluorescence images of a clonal line of WIBR3 OCT4-ΔPE-GFP+ cells (left) and a clonal line of wild-type WIBR3 human ESCs generated in 2i/L/DOX (right) and maintained for three passages in PD0325901/IM12/SB590885/Y-27632/WH-4-023 (5i) and hLIF. 40× magnification. (H) Teratoma generated from wild-type WIBR3 human ESCs maintained in PD0325901/IM12/SB590885/Y-27632/WH-4-023 (5i) and hLIF after transgene withdrawal. Representative tissues of the three germ layers are indicated. 200× magnification.

Figure 4

Direct Conversion of Conventional Human ESCs to Naive Pluripotency in 5i/L (A) Strategy for assessing direct conversion of primed human ESCs into OCT4-ΔPE-GFP+ cells under optimized chemical conditions. (B) Phase and fluorescence images of emerging naive colony and expanded cells from WIBR3 OCT4-ΔPE-GFP human ESCs treated with 5i/L for 10 days. Left and right panels are 40× magnification, and middle panel is 100×. (C) Phase and fluorescence images and flow cytometric analyses of the proportion of GFP+ cells during conversion experiments in 5i/L supplemented with FGF and/or Activin A (FA). 40× magnification. (D) Phase images of wild-type naive WIBR2 human ESCs converted in 5i/L supplemented with FGF and/or Activin A (FA). 40× magnification. (E) Phase image of a primary human ESC line derived in 5i/L/FA from an explanted human blastocyst. Cell line is designated as Whitehead Institute Naive Human ESC line 1 (WIN1). 100× magnification. (F) Top, green: strategy for generating secondary naive human iPSCs from secondary derived fibroblasts (Hockemeyer et al., 2008) harboring inducible OCT4, SOX2 and KLF4 transgenes and OCT4-ΔPE-GFP allele. Top, blue: cell culture media conditions used at representative stages of reprogramming experiment. Bottom: phase and fluorescence images of primary primed iPSCs and secondary derived fibroblasts and the reactivation of GFP in naive reprogrammed secondary iPSCs. 40× magnification. (G) Flow cytometric analysis of the proportion of OCT4-ΔPE-GFP+ cells three passages after withdrawal of individual inhibitors and growth factors. (H) Quantitative gene expression analysis for NANOG and KLF4 three passages after withdrawal of individual inhibitors and growth factors from 5i/L/FA control cells. Error bars indicate ± 1 SD of technical replicates.

Figure 5

Evaluation of Alternative Culture Conditions for Naive Human Pluripotency (A) Table comparing the components of four recent protocols for capturing naive-like human ESCs with 5i/L/A medium. Note that the protocol for naive conversion from Ware et al. (2014) involves preculture with the HDAC inhibitors butyrate and suberoylanilide hydroxamic acid. (B) Phase and fluorescence images and flow cytometric analyses showing the response of OCT4-ΔPE-GFP− primed cells to recently reported protocols for naive human pluripotency (see Figure 5A) and 5i/L/A. 40× magnification. (C) Quantification of the proportion of GFP+ cells in WIBR3 OCT4-GFP and OCT4-ΔPE-GFP human ESCs upon removal of DOX-inducible KLF2 and NANOG expression in primed human ES medium (hESM) and four alternative conditions for naive human pluripotency. (D) Flow cytometric analysis of the proportion of OCT4-ΔPE-GFP+ cells in 5i/L/A and the JNK inhibitor SP600125 (6i/L/A) in serum-free N2B27 basal medium versus 20% KSR basal medium. (E) Quantitative gene expression analysis for OCT4, SOX2, KLF2, and NANOG in human ESCs cultured in 6i/L/A and supplemented with 1%, 5%, 7.5%, or 10% FBS or KSR. Error bars indicate ± 1 SD of technical replicates. (F) Phase and fluorescence images of induction of OCT4-ΔPE-GFP activity from the primed state in 6i/L/A, and 6i/L/A supplemented with 1%, 5%, 7.5%, or 10% KSR. 40× magnification.

Figure 6

Transcriptional Profiling of Naive Human ESCs in 5i/L/A (A) Cross-species hierarchical clustering of naive and primed pluripotent cells from mice and humans, as performed previously by Gafni et al. (2013). Affymetrix expression data were normalized using RNA spike-in. Two groups of human ESC samples are included: WIBR2 (P12 and P14), WIBR3 (P9), and WIN1 (P10) human ESCs derived in our optimized naive medium (5i/L/A or 6i/L/A, as indicated) and parental WIBR2 and WIBR3 human ESCs in primed human ESC medium (hESM). Correlation matrix of gene expression was clustered using Pearson correlation coefficients (PCCs). The average linkage hierarchical clustering of the Pearson correlation is shown in the heatmap. mEpiSCs, mouse EpiSCs; mESC, mouse ESC; miPSC, mouse iPSC. Note that NOD miPSC samples described in Hanna et al. (2009) were metastable and acquired an EpiSC-like identity after undergoing removal of exogenous transcription factors. (B) Volcano plots showing fold change (x axis) between naive human ESC samples and primed human ESCs on all genes, as reported in this study, Gafni et al. (2013), Chan et al. (2013), and Ware et al. (2014). The light blue dots represent those genes that exhibit the most significant gene expression changes (defined as those that have a log2 fold change > 1 and < −1 and wherein p < 0.05). Highlighted in red or green are genes of interest that are upregulated in the naive state or downregulated in the naive state, respectively. (C) Fold changes in expression of naive pluripotency-associated transcripts in our naive human ESC samples versus primed human ESCs (blue), and the naive human samples published by Gafni et al. (2013) versus primed human ESCs (red). (D) For comparison with (C), fold changes in expression of naive pluripotency-associated transcripts in naive mouse ESCs versus primed mouse EpiSCs were curated from a previously published study (Najm et al., 2011). (E) Quantitative gene expression analysis for NANOG and STELLA in human ESCs cultured in parallel in primed medium, the medium of Gafni et al. (2013), and 5i/L/FA (P4). Error bars indicate ± 1 SD of technical replicates. (F) Moving average plots of expressed genes along the X chromosome in female primed human ESCs (WIBR2 and WIBR3 hESM) compared to the average of three male control iPSCs in hESM (top) and converted naive human ESCs in 6i/L/A (bottom). Shown are representative genes on the X chromosome and the location of the centromere (Cen). (G) Single-molecule (sm) RNA FISH analysis using OCT4, NANOG, KLF4, and REX1 probes in human ESCs cultured in primed hESM, the medium of Gafni et al. (2013), or 5i/L/A.

Figure 7

Chromatin Landscape of Naive Human Pluripotency (A–E) ChIP-Seq tracks for H3K4me3 and H3K27me3 in WIBR2 human ESCs cultured in primed hESM (red) or naive 6i/L/A medium (P18) (blue) at five classes of genes: (A) developmental genes that are bivalent in the primed state and exhibit loss of H3K27me3 in the naive state; (B) naive pluripotency genes that are bivalent in the primed state and exhibit loss of H3K27me3 in the naive state; (C) naive pluripotency genes that acquire H3K4me3 in the naive state; (D) master transcription factors that have a signal for H3K4me3, but not H3K27me3, in both naive and primed states; and (E) genes that are strongly downregulated in the naive state and acquire increased H3K27me3 signal. (F) ChIP-Seq analysis for H3K4me3 and H3K27me3 at Polycomb target genes in WIBR2 human ESCs cultured in primed medium (left) or naive 6i/L/A medium (right). (G) Average H3K4me3 and H3K27me3 signal at Polycomb target genes in WIBR2 human ESCs cultured in primed medium (red) or naive 6i/L/A medium (blue).