Totipotent embryonic stem cells arise in ground-state culture conditions

Abstract

Embryonic stem cells (ESCs) are derived from mammalian embryos during the transition from totipotency, when individual blastomeres can make all lineages, to pluripotency, when they are competent to make only embryonic lineages. ESCs maintained with inhibitors of MEK and GSK3 (2i) are thought to represent an embryonically restricted ground state. However, we observed heterogeneous expression of the extraembryonic endoderm marker Hex in 2i-cultured embryos, suggesting that 2i blocked development prior to epiblast commitment. Similarly, 2i ESC cultures were heterogeneous and contained a Hex-positive fraction primed to differentiate into trophoblast and extraembryonic endoderm. Single Hex-positive ESCs coexpressed epiblast and extraembryonic genes and contributed to all lineages in chimeras. The cytokine LIF, necessary for ESC self-renewal, supported the expansion of this population but did not directly support Nanog-positive epiblast-like ESCs. Thus, 2i and LIF support a totipotent state comparable to early embryonic cells that coexpress embryonic and extraembryonic determinants.

Graphical abstract

Figure 1

Culture in 2i Does Not Eliminate Endoderm Marker Expression in the Early Embryo (A) HV transgenic embryos were collected at E2.5 and cultured in vitro for different time periods before live confocal microscopy imaging. Representative images are shown for each stage. At E2.5, an extended focus image is shown. E3.5 and E4.5 images show confocal optical sections. (B) Scoring of the HV expression pattern in embryos collected at E2.5 and cultured in either control or 2i medium. Day 0 = E2.5, day 1 = E3.5 (early blastocyst), day 2 = E4.5 (midblastocyst), and day 3 = E5.5 (late blastocyst, equivalent to E4.5 in vivo due to culture induced delay). Only HV-expressing transgenic embryos were scored. TB, trophoblast; ICM, inner cell mass. (C) Confocal optical section of an E2.5 HV embryo cultured for 3 days immunostained for the trophoblast marker CDX2. Red arrowheads indicate trophoblast cells coexpressing HV and CDX2. Although in vivo PE is specified by E4.5, this segregation is delayed in vitro, and, hence, embryos were cultured until E5.5 to observe differentiated PE. (D) HV embryos were flushed at E2.5 and cultured in 2i media for 3 days before live confocal microscopy imaging. Red arrowheads indicate HV-expressing trophoblast cells, whereas white arrowheads indicate HV⁺ PE cells. (E) Confocal optical sections of immunostained transgenic HV embryos after flushing at E2.5 and 3 days in culture either in control or 2i medium.

Figure 2

ESCs Cultured in 2i Are Heterogeneous (A) Flow cytometry of HV and ESC marker PECAM-1 in cells cultured in serum/LIF, 2i, and 2i/LIF. Gates set using unstained E14 ESCs (Figure S1B). Black boxes indicate sorting gates for the upper and lower 25% of HV expression used to separate HV⁻ and HV⁺ populations. (B) Flow cytometry of clones (represented by individual bars) after expansion from single HV⁻ or HV⁺ sorted cells from 2i or 2i/LIF. Red circles indicate mean. (C) Heat map of sorted HV⁻ and HV⁺ populations from serum/LIF and 2i/LIF culture, based on gene expression data from RNA-seq. Heat map shows differentially expressed genes identified by pairwise comparison of all sorted fractions. Data were normalized by subtracting the average log expression from all samples. Genes are hierarchically clustered by average Euclidean distance. Two biological replicates are shown per sample. Red represents upregulation and green represents downregulation of expression. (D) Immunostaining displaying representative images of CDX2-positive cells in trophoblast differentiation. High-magnification images are shown as insets. (E and F) Quantification of in vitro differentiation of sorted HV⁻ and HV⁺ ESCs after differentiation in trophoblast stem cell conditions for 7 days (E), LIF withdrawal (F), or neural differentiation (G). (E) Trophoblast differentiation was quantified by counting CDX2-positive cells, which were negative for GATA6 and BRACHYURY (n = 3). Values are shown relative to HV⁺. (F) Endoderm differentiation was quantified by counting the number of GATA6-positive endodermal clusters (see also Figures S3A and S3C; n = 3). Values are shown relative to HV⁻. (G) Neural differentiation was quantified by counting elongated TUJ1-positive neurons (n = 5). Values are shown relative to HV⁻. Error bars indicate mean ± SD of biological replicates. See also Figures S1, S2, and S3.

Figure 3

Subpopulations of ESCs Cultured in 2i Demonstrate Extraembryonic Potential In Vivo (A and B) HV cells, expressing a constitutive LACZ marker, cultured in 2i (A) or 2i/LIF (B), were sorted via FACS into HV⁻ (A: n = 18; B: n = 60) and HV⁺ (A: n = 26; B: n = 55) PECAM-1⁺ cells and aggregated with wild-type F1 morulae. Embryos were collected at E6.5 and X-gal stained. Black bars indicate the plane of section. Arrowheads mark extraembryonic endoderm (black) and trophoblast (red). (C–F) Sorted HV⁺ cells expressing a constitutive H2B-Tomato fusion protein, cultured in 2i/LIF, were used in morula aggregations. Embryos were cultured until E4.5 (C) or dissected at E9.5 (D). (C) Confocal optical section of an immunostained chimera after 3 days of culture (equivalent to E4.5 in vivo). Red arrowheads indicate H2B-Tomato cells within the trophoblast. Extended view throughout the embryo also shown for orientation. (D) E9.5 chimera. White line indicates plane of the low-magnification section in (E). White box indicates area of placenta shown in the high-magnification image marked with an asterisk (^∗). (E) Wax sections of an E9.5 chimeric embryo showing HV H2B-Tomato cells in the placenta, yolk sac, and embryo. The uppermost panels show a low-magnification section of the whole embryo (shown in D) within the yolk sac and attached to the placenta. ys, yolk sac; yse, yolk sac endoderm; ysm, yolk sac mesoderm; em, embryo; n, notochord; p, placenta; um, umbilical cord. (F) High-magnification image of placental section. See also Figure S4.

Figure 4

Single HV⁺ Cells in 2i Can Contribute to Both Epiblast and Extraembryonic Lineages Single HV⁻ or HV⁺ cells from serum/LIF or 2i were sorted by FACS and analyzed by qRT-PCR using the Biomark HD system (Fluidigm). (A) Heat map for each condition and cell type showing the expression of genes with reliable primer melting curves. Adjacent squares show technical replicates for each cell. (B) Single cells were sorted by flow cytometry into LIF withdrawal conditions. After 7 days, differentiated colonies were immunostained for GATA6, CDX2, and BRACHYURY. (C) Single-sorted HV⁺ cells (sorted from the top 10% of HV expression), previously cultured in 2i/LIF, constitutively expressing H2B-Tomato were injected into morulae and dissected at E6.5 and chimeras (n = 23/64) were assessed for lineage contribution. Examples of different contribution patterns of single cells are shown. Some cells showed even contribution across the whole epiblast while others showed a contribution that was biased toward a particular area of the epiblast. Red arrowheads indicate trophoblast contribution, green arrowheads indicate visceral endoderm contribution, and black arrowheads indicate parietal endoderm contribution. Epiblast contribution alone was observed in nine chimeras, extraembryonic contribution alone in one chimera, and 13 chimeras showed both epiblast and extraembryonic contribution. See also Figure S5.

Figure 5

LIF Supports an Extraembryonically Primed Population of Cells (A) Flow cytometry of HV or TNGB Nanog-GFP cells cultured in serum or 2i medium, with increasing LIF concentrations (n = 3). Raw data and controls are in Figures S6A–S6C. Error bars indicate mean ± SD of two biological replicates. (B) qRT-PCR showing relative gene expression of extraembryonic markers in ESCs cultured in 2i or 2i/LIF. Data were normalized to the housekeeping gene TBP and are shown relative to 2i samples. Error bars indicate mean ± SD of biological replicates. (C) Confocal optical section showing immunostaining of HV cells for NANOG after culture of cells in serum/LIF, 2i, or 2i/LIF. (D) qRT-PCR showing relative gene expression in Stat3^+/+ or ^−/− ESCs. Data are shown relative to wild-type expression levels. All qRT-PCR transcript levels in this figure were normalized to TBP and are shown as mean ± SD. (E) Statistical significance of correlation of gene expression differences between ESCs cultured in the presence and absence of LIF and tissue-specific gene expression in the GNF Ver.3 database. Correlation was estimated from data on 5,085 genes, which represent the overlap between 10,000 most informative genes in our RNA-seq data and 10,000 most informative genes in the GNF database. Organs/tissues sorted by decreasing correlation. (F) Confocal optical section of HV cells immunostained for the proliferation marker KI67. White arrowheads indicate KI67-negative cells. (G) Quantification of KI67 immunostaining in HV⁺ and HV⁻ ESCs. The absolute number of proliferating cells (expressing KI67, red bar) and quiescent cells (not expressing KI67, black bar) were counted. Error bars indicate SD of the mean from five colonies in each condition, analyzed by confocal microscopy. See also Figure S5.

Figure 6

Model Comparing Standard Serum/LIF ESC Culture to 2i ESC Culture The model relates serum/LIF and 2i ESC cultures to embryonic development. 2i/LIF cultures contain totipotent single cells that are comparable to early preimplantation stages when embryonic cells are still totipotent. As Fgf is needed for later lineage segregation, inhibiting Fgf signaling in 2i seems to block embryos and ESCs at a point prior to this commitment event. Serum/LIF ESC cultures are reminiscent of later embryonic stages when cells are pluripotent and contain more restricted cell types. Although partially differentiated mesoderm and ectoderm appear to be lost in 2i/LIF, endoderm is maintained. Based on our data, totipotent ESCs are able to dynamically interconvert to a pluripotent epiblast precursor state and may also exist in equilibrium with a similar primitive endoderm precursor, although we have not been able to distinguish this population in this study. LIF supports the expansion of the totipotent ESC state and primitive endoderm precursors in serum.

Figure S1

HV Heterogeneity Can Be Observed after Clonal Rederivation of ESC Lines in 2i, Related to Figure 2 (A) Phospho ERK western blots of cells cultured in serum/LIF, 2i or 2i/LIF. Total ERK and Tubulin are shown as loading controls. (B) Plots showing gating for all flow cytometry in this study. Gating for HV and PECAM-1 expression was based on a control E14 cell line without the HV reporter and unstained for PECAM-1. Cells were separated from debris by gating on forward (FSC) and side scatter (SSC). Dead cells, showing high DAPI levels, were eliminated. (C) Purity checks after sorting by FACS of upper and lower 25% of HV expressing ESCs in 2i. (D–F) Flow cytometry of individual clones in serum/LIF (D), in 2i (E) and 2i/LIF (F).

Figure S2

Transcriptomics Suggest the Presence of a Population Primed toward Trophoblast as well as Endoderm, Related to Figure 2 (A–D) Common RNA-seq signatures in specific gene classes. The behavior of the individual genes belonging to the classes described in the text is shown. Plots compare mean log intensity values for individual genes among four sorted populations; serum/LIF HV-, serum/LIF HV+, 2i/LIF HV-, 2i/LIF HV+. Error bars represent s.d. between 2 biological replicates. (A) Markers of undifferentiated ESCs, (B) imprinted genes (^∗ = member of Dlk1-Dio3 cluster), (C) trophoblast markers, (D) 2-cell stage embryo retroviral genes. (E) High-magnification image of GATA6 and CDX2 immunostaining of ESCs, after 7 days differentiation in TSC conditions, demonstrating that there was no coexpression of these 2 markers. (F) qRT-PCR of HV- and HV+ populations cultured in serum/LIF or 2i after LIF withdrawal for 4 days. Values were normalized to the housekeeping gene TBP and are shown relative to serum/LIF HV+. Error bars indicate mean ± SD.

Figure S3

Quantification of Lineage Priming In Vitro, Related to Figure 2 (A) A typical GATA6+ endodermal colony, also expressing HV, as scored in differentiation assays. (B) Quantification of number of GATA6+ cells per endodermal cluster. (C and D) Immunostaining of ESCs differentiated by LIF withdrawal (C) or neural differentiation (D) after previous culture in serum/LIF or 2i and then sorting into HV- and HV+, PECAM-1+ ESCs. Cells were plated immediately into differentiation conditions after sorting. Endoderm levels were quantified by GATA6 immunostaining (n = 3) (values relative to serum/LIF HV+), (D) and neural differentiation by TUJ1 immunostaining (n = 5) (Values relative to serum/LIF HV-) (E and F) qRT-PCR showing relative gene expression of sorted cells differentiated by LIF withdrawal (E) or neural differentiation (F). Data were normalized to the housekeeping gene TBP and is shown relative to serum/LIF HV sample. Values indicate mean ± SD of biological replicates. (G) Flow cytometry of sorted HV- and HV+ populations following LIF withdrawal and staining for the endoderm marker PDGFR-alpha.

Figure S4

HV⁺ ESCs Demonstrate Extraembryonic Potential in Morula Aggregations, Related to Figure 3 (A) Table containing a summary of all chimeras produced by morula aggregation in this study. (B–D) HV ESCs were cultured in 2i/LIF. HV+ cells, expressing a constitutive H2B-Tomato fusion protein, were sorted and used in morula aggregations. Embryos were dissected at E6.5 (B-D) and imaged by confocal microscopy. (B) Example chimeras, one showing only epiblast contribution, the other both epiblast and extraembryonic contribution. Red arrowheads indicate H2B-Tomato cells in the trophoblast, green arrowheads indicate H2B-Tomato cells in the extraembryonic endoderm. (C) Whole-mount immunostaining of chimeras for an endoderm (GATA6) and trophoblast marker (KRT7). (D) Optical sections through whole-mount immunostained embryos. Upper panels show both visceral endoderm and trophoblast contribution at the epiblast/extraembryonic border. Lower panels indicate trophoblast giant cell contribution outside of the distal epiblast. White box marked with ^∗ indicates area of high magnification in lower panel. (E) Immunostaining of E3.5 and E4.5 wild-type blastocysts. Images show confocal optical sections.

Figure S5

Single HV⁺ Cells Are Totipotent in the Ground State, Related to Figure 4 Single cell qRT-PCR was carried out on a BioMark HD System (Fluidigm) using a 96 × 96 chip. DELTAgene Assay custom designed primers (Fluidigm) were used for amplification. Technical replicates were generated for each single cell. (A) Histograms showing the distribution of expression of 4 genes in each population. Data shown unnormalized. (B) Single cells were sorted by flow cytometry into LIF withdrawal conditions. After 7 days differentiated colonies were immunostained for GATA6, CDX2 and BRACHYURY. Quantification of marker expression patterns. 2 and 3 marker exclusive expression indicates colonies that contain independent cells positive for different lineage markers. (C) Whole-mount immunostaining of an E6.5 chimeric embryo generated by injection of a single HV+ ES cell constitutively expressed H2B-Tomato, previously cultured in 2i/LIF, into a morula stage wild-type embryo. Upper whole-mount image shows an extended focus view of the entire embryo. Panels below show optical sections, generated by confocal microscopy, with ES cells contributing to trophoblast and visceral and parietal endoderm (VE and PE, respectively). (D) E6.5 chimeras generated by injection of a single HV- cell that had been cultured in 2i/LIF. The majority of single cells contributed only to epiblast (11/13), with only 2 showing extraembryonic contribution.

Figure S6

LIF Increases HV Expression within an Undifferentiated Population, Related to Figure 5 (A and B) Flow cytometry of HV cells (A) and Nanog-GFP TNG cells (B) in the presence and absence of LIF. Control E14 cells are also shown. Cells were cultured at high density to prevent differentiation. (C) While LIF stimulates the HV reporter (Figure S1) it does not enhance levels of autofluorescence in wild-type E14 ESCs. (D) Quantification of colocalization of HV and NANOG in serum/LIF, 2i and 2i/LIF. Data, based on confocal optical sections of 5 individual colonies of cells per condition, shown as mean ± SD. (E) Differentiation by LIF withdrawal from unsorted cultures in 2i or 2i/LIF. Cells were immunostained for GATA6 (n = 3). (F) Quantification of GATA6 immunostaining. Values relative to 2i. Error bars indicate mean ± SD of biological replicates (^∗∗∗p < 0.001). White dotted lines highlight GATA6+ endoderm clusters. See also Figure S3 and methods section. (G) Summary of RNA-seq expression data for endoderm and trophoblast (left), canonical ESC (middle) and mesoderm and neurectoderm markers (right) for unsorted samples cultured in 2i or 2i/LIF. Error bars indicate mean ± SD of two biological replicates. Values are relative to 2i sample.