Efficient derivation of extraembryonic endoderm stem cell lines from mouse postimplantation embryos

Various types of stem cell lines have been derived from preimplantation or postimplantation mouse embryos: embryonic stem cell lines, epiblast stem cell lines, and trophoblast stem cell lines. It is not known if extraembryonic endoderm stem (XEN) cell lines can be derived from postimplantation mouse embryos. Here, we report the derivation of 77 XEN cell lines from 85 postimplantation embryos at embryonic day E5.5 or E6.5, in parallel to the derivation of 41 XEN lines from 69 preimplantation embryos at the blastocyst stage. We attain a success rate of 100% of XEN cell line derivation with our E5.5 whole-embryo and E6.5 disaggregated-embryo methods. Immunofluorescence and NanoString gene expression analyses indicate that the XEN cell lines that we derived from postimplantation embryos (post-XEN) are very similar to the XEN cell lines that we derived from preimplantation embryos (pre-XEN) using a conventional method. After injection into blastocysts, post-XEN cells contribute to extraembryonic endoderm in chimeras at E6.5 and E7.5.

Sox17-Cre male 34 , and one hemizygous Gata6-mTomato female 36 mated with a homozygous Cdx2-GFP male 37 ( Table 1). Xist1loxGFP is a GFP-containing targeted mutation in the Xist locus on the X-chromosome; Sox17 and Gata6 are XEN-cell markers; and Cdx2 is a marker for trophoblast stem cells. We removed the ectoplacental cone of the embryos as much as possible, and transferred each embryo separately into a well of 4-well dish coated with 0.1% gelatin and covered with MEF in TS cell medium including 25 ng/ml FGF4 and 1 μ g/ml heparin (referred to as F4H). One day later, the embryos had attached to the surface and started to form an outgrowth. The embryos had formed a large outgrowth after 5 days. We used TrypLE Express to disaggregate the outgrowths and passaged cells into a well of a 4-well dish. After cells reached 70-80% confluency, they were passaged into a well of a 12-well dish. After they reached 70-80% confluency again, cells were passaged into a well of a 6-well dish, and we then obtained stable post-XEN cell lines. The intrinsic red fluorescence of mTomato produced from the Gata6 promoter in the transgene was sufficiently high to detect it in the whole embryo and outgrowth, but not in the established post-XEN cell line at day 60 (Fig. 2a). We thus derived, using the whole-embryo method, a total of 30 post-XEN cell lines from 38 E6.5 embryos, at a 79% success rate (Table 1). We found that mTomato expressed from the Gata6 transgenic promoter in these cells can be detected by immunofluorescence with antibodies against RFP, together with GATA6 expressed from the endogenous Gata6  locus with antibodies against GATA6 (Fig. 2b). As was the case for our pre-XEN cell lines, our post-XEN cell lines are positive for XEN cell markers GATA4, GATA6, SOX7, SOX17, and DAB2 but negative for ES cell markers OCT4 and NANOG, and negative for TS cell marker CDX2.
Derivation of post-XEN cell lines from disaggregated E6.5 embryos. During the process of post-XEN cell line derivation from whole E6.5 embryos, we observed that some embryos had difficulty forming large outgrowths. We reasoned that some unidentified cell types in these embryos may inhibit XEN cell growth. We therefore proceeded to derive post-XEN cell lines from disaggregates of these embryos. We isolated 30 E6.5 embryos from three types of natural matings: a homozygous R26-tauGFP41 female with a heterozygous Sox17-Cre male 36 , a heterozygous ROSA26-STOP-taulacZ female with a heterozygous Sox17-Cre male, and a wild-type CD1 female with a heterozygous PDGFRa-GFP male 28 (Table 1). We prepared disaggregates (Fig. 3a) by exposing the embryos to collagenase and deoxyribonuclease followed by TrypLE Express, and then gently pipetting the embryos in a glass pipette with a diameter of 50-60 μ m. We plated the cell suspension, which consisted of a mixture of single cells and clumps of cells, from one disaggregated embryo separately into a well of 4-well dish coated with 0.1% gelatin and covered with MEF in standard TS medium including F4H. On day 3, XEN-like cells colonies appeared. We picked these colonies, disaggregated them by a glass pipette or by TrypLE Express for 5 min at 37 °C, and passaged them into a well of a 4-well dish. (We did not succeed in deriving post-XEN cell lines from single GFP+ cells that we placed in wells of a 96-well dish.) We thus derived, using the disaggregation method, a total of 30 post-XEN cell lines from 30 E6.5 embryos, at a 100% success rate ( Table 1).
As was the case with our pre-XEN cell lines and post-XEN cell lines derived from E6.5 whole embryos, cells are immunoreactive for XEN markers GATA4, GATA6, SOX7, SOX17, and DAB2, but negative for ES cell markers OCT4 and NANOG, and negative for TS cell marker CDX2 (Fig. 3b). Thus, we increased the success rate of post-XEN cell derivation from 79% for whole E6.5 embryos to 100% for disaggregated E6.5 embryos.
Derivation of post-XEN cell lines from E5.5 whole embryos. In our final set of experiments, we derived post-XEN cell lines from 5.5 day-old embryos. We isolated 17 embryos from 33 E5.5 implantation sites from a mating of three homozygous R26-tauGFP41 females with three heterozygous Sox17-Cre males, and a mating of a homozygous ROSA-STOP-taulacZ female with a heterozygous Sox17-Cre male ( Table 1). We removed as much of the ectoplacental cone as possible, and transferred one whole embryo separately into a well of a 4-well dish coated with 0.1% gelatin and covered with MEF, in TS medium including F4H (Fig. 4a). In our experience, the less ectoplacental cone and extraembryonic ectoderm remains in a dissected embryo, the easier it is to derive a post-XEN cell line. We surmise that ectoplacental cone cells and extraembryonic ectoderm cells (trophoblast-derived cells) compete with or inhibit XEN cells in culture. For 14/17 embryos, large outgrowths had appeared by day 5. We disaggregated these outgrowths with TrypLE Express on day 7, and passaged the cells into a well of a 4-well dish. On day 11, XEN-like colonies had accumulated. We picked these colonies, combined them, disaggregated them with a glass pipette or with TrpLE Express for 5 min at 37 °C, and passaged them into a well of a 4-well dish. We thus established 14 post-XEN cell lines after ~21 days in culture. For the remaining 3/17 embryos, trophoblast-derived cells surrounded the outgrowth. We performed a disaggregation by pipette on day 3, and switched to ES medium with LIF. When XEN cells became abundant, we passaged the cells onto gelatin-coated dishes without MEF, and switched back to TS medium including F4H. Three post-XEN cell lines were established after ~50 days culture. We thus derived a total of 17 post-XEN cell lines from 17 E5.5 embryos, at a 100% success rate (Table 1).
These post-XEN cell lines are immunoreactive for XEN cell markers GATA4, GATA6, SOX7, SOX17, and DAB2, but negative for ES cell markers OCT4 and NANOG, and negative for TS cell marker CDX2 (Fig. 4b).
Post-XEN cell morphology, population characteristics, and in vitro differentiation. Like pre-XEN cell lines 10 , our cultures of post-XEN cell lines contain at least two cell morphologies: a rounded, highly refractile cell type and a more stellate epithelial-like cell type (Fig. 5a). At higher densities, post-XEN cells can form epithelial sheets (Fig. 5b) and often a lattice-type structure (Fig. 5c). To determine if these two cell morphologies represent two cell types in the cultures, we FACS-sorted single GFP+ post-XEN cells (X-E6.5-Z0617-5) directly into wells of a 96-well dish, and derived two subclones. Cells of these subclones continued to exhibit either round or epithelial-like cell morphologies (Fig. 5d).
To evaluate the population characteristics of post-XEN cell lines, we analyzed three GFP-expressing cell lines (X-E6.5-Z0617-5, X-E6.5-Z0617-2 and X-E5.5-10) by immunofluorescence for GATA4 and counterstaining with DAPI. Sets of fluorescence images were captured for each line visualizing the intrinsic fluorescence of GFP, DAPI, and GATA4 immunoreactivity (Fig. 5e). The images were evaluated with a custom CellProfiler pipeline that segmented cells using the DAPI fluorescent signal. A cutoff for size and nuclear roundness was used to exclude the MEF population as much as possible. We then evaluated each cell for levels of GFP and GATA4 signal. In these three post-XEN cell lines, > 94% of cells evaluated exhibited GFP fluorescence and are GATA4+ (Fig. 5f). The populations of cells that expressed either GFP or GATA4 but not both represented 0-3% of the cells evaluated. The remaining population of cells, which expressed neither GFP nor GATA4, may reflect another cell type, or MEFs that escaped exclusion during the cell identification step in the custom CellProfiler pipeline.
We asked if post-XEN cells can differentiate into a VE identity by incubation with BMP4 38,39 . We cultured four post-XEN cell lines (X-E5.5-9, X-E6.5-Z0617-2, X-E6.5-Z0617-5 and X-E6.5-78097-7) and three pre-XEN cell lines (X42, X47, X-ICM-4) in gelatin-coated dishes in TS medium with F4H, plus 10 ng/ml BMP4. In parallel, we cultured the same cell lines in TS medium with F4H without BMP4. Four days later, we performed immunofluorescence for E-cadherin, a VE marker. We found that culture with BMP4 induces expression of E-cadherin in post-XEN cells, as well as in pre-XEN cells 10 (Fig. 5g- NanoString gene expression analyses of XEN and ES cell lines. We applied the NanoString multiplex platform for gene expression [40][41][42] and agglomerative clustering, in order to compare the patterns of expression of selected genes in four pre-XEN cell lines, three post-XEN cell lines, and three ES cell lines that we had derived in other experiments (Fig. 6). We find that all our XEN cell lines have high levels of expression of XEN-specific genes, such as Dab2, Gata4, Gata6, Pdgfra, Sox7, and Sox17, versus low or no expression of ES cell-specific genes such as Nanog, Pou5f1/Oct4, Sox2, and Nr0b1. There is no expression of EpiSC-specific genes First column: intrinsic (indicated with an asterisk after GFP) green fluorescence of GFP expressed from the ROSA26 locus after activation by Cre recombinase that is expressed from the gene-targeted Sox17 locus. Second column: cells are immunoreactive (magenta) for XEN markers GATA4, GATA6, SOX7, and SOX17. Third column: cells are immunoreactive for XEN marker DAB2, but negative for ES cell markers OCT4 and NANOG, and negative for TS cell marker CDX2. Fourth column: DAPI (blue).
such as Cer1 and Fgf5 (data not shown). Thus, the NanoString gene expression analysis confirms and extends the immunofluorescence profiles.

Discussion
We report here that XEN cell lines can be derived efficiently from postimplantation embryos, and from a wide variety of strains and crosses. We believe that it is prudent to continue to refer to these two types of cell lines with regard to the embryonic stage from which they were derived. Additional work will be necessary to identify and characterize possible differences, subtle or substantial, between pre-XEN and post-XEN cell lines.
With our E5.5 whole-embryo method and E6.5 disaggregated-embryo method, we attained a success rate of 100%: we derived 17 post-XEN cell lines from 17 embryos and 30 post-XEN cell lines from 30 embryos, respectively. There are only ~11 PrE cells at the late blastocyst stage 45 . As E5.5 and E6.5 embryos contain ~95 and ~250 extraembryonic endoderm cells respectively 46 , these higher numbers of potential source cells may explain the 100% efficiency of XEN cell line derivation from postimplantation embryos. Other, experimental, reasons for the highly efficient derivation may be the reduction of the negative effects of trophoblast-derived cells on XEN cells in the disaggregated E.6 embryo method, and the promotion of XEN cells over trophoblast-derived cells by culturing in ES medium with LIF in the whole-embryo E5.5 embryo method. Regardless of these numerical and experimental explanations, a more exciting, and biological, explanation is that extraembryonic tissues preserve developmental plasticity through implantation.
Since the first report of XEN cell derivation from blastocysts in 2005 (ref. 10), there are three types of methods in place to derive XEN cell lines, either directly from blastocysts 10,20 or by conversion from ES cells [20][21][22][23][24][25][26] or, recently, by reprogramming from fibroblasts 27 . The success rate for derivation of XEN cell lines from blastocysts was 21% in TS medium including F4H and 56% in ES medium supplemented with LIF 20 , which is identical to our success rate of 57% in ES medium supplemented with LIF. Recently, LIF has been reported to support PrE expansion during pre-implantation embryo development 47 , and LIF could also be supporting XEN cell expansion. The conventional protocol to derive ES cell lines from blastocysts involves also ES medium supplemented with LIF 48,49 ; this medium can be used to derive XEN cell lines from blastocysts, by picking XEN cell colonies or by removing ES cell colonies, but there is always the risk of deriving ES cell lines 50 . The indirect generation of XEN cell lines from ES cell lines requires, obviously, the prior generation or availability of such lines.
TS cell lines can be derived from the extraembryonic ectoderm of postimplantation embryos in TS medium including FGF4 and heparin 9,11 . Interestingly, we did not derive TS cell lines from either whole or disaggregated embryos using this medium. We did observe transiently cells and colonies with a morphology that is consistent with TS cells or EpiSCs, but they disappeared from the culture with time. In our immunofluorescence analyses, our XEN cell lines do not contain cells that express the trophectoderm marker Cdx2, and our Nanostring analyses do not reveal expression of TS-specific genes either. The cell lines that we derived from postimplantation   and E-cadherin (blue), fluorescence of DAPI (white) and GFP (green), and merge. The single GFP+ cell in (k) apposes but does not belong to the VE, as it is not immunoreactive for E-cadherin. (a-g) and (j,k) were imaged using a Zeiss LSM 710 confocal microscope, and (h-i) were imaged using a Nikon SMZ25 stereofluorescence microscope. In (a) and (e-g) the red counterstain is propidium iodide. embryos in TS medium are thus not TS cell lines, not ES cell lines, and not EpiSC lines, but represent XEN cell lines. Why then are XEN cell lines obtained rather than TS cell lines with our method, which cultures the cells in the same TS medium? The protocol for deriving TS cell lines from postimplantation embryos prescribes that the extraembryonic ectoderm be dissected out and that the embryo and the VE and PE be discarded 12 . By contrast, we remove the ectoplacental cone from the embryo. We speculate that the VE and PE surrounding the extraembryonic ectoderm inhibit TS cell growth, and/or that VE-and PE-derived cells divide at a faster rate than TS cells and dominate with time the cell culture. Chemical reprogramming of mouse fibroblasts to the iPS cells involves a XEN-like intermediary stage as a bridge between somatic and pluripotent cells 19 . In another study, iXEN clones arise in parallel to iPS clones during OSKM-mediated reprogramming of mouse fibroblasts 27 . More knowledge about the derivation, biology, culture, and conversion of XEN cell lines may be beneficial to develop robust and efficient protocols for the derivation of iPS cell lines by chemical reprogramming with small molecules.
In future experiments, it will be interesting to determine the cell type(s) that are at the origin of the post-XEN cell lines.

Derivation of ES cell lines.
Embryos were collected at the 2-8 cell stage by flushing oviducts using M2 medium (Sigma #M7167), and cultured in KSOM medium (Millipore #MR-106-D) to the blastocyst stage. The zona pellucida of blastocysts was then removed using acid Tyrode solution (Sigma #T1788). Blastocysts were transferred separately into a well of a 96-well dish (Falcon) coated with 0.1% gelatin (Specialty Media #ES-006-B) and covered with MEF in ES medium supplemented with LIF and 1 μ M PD0325901 (Axon #1408) and 3 μ M CHIR99021 (Axon #1386), a combination of chemicals that is typically referred to as "2i". After 4 days TrypLE Express (Gibco #12604-013) was used to disaggregate the embryonic outgrowths, and cells were passaged to a well of 24-well dish to derive ES cell lines.

Derivation of pre-XEN cell lines from blastocysts.
Embryos were collected at the 2-8 cell stage embryos and cultured in in KSOM medium to the blastocyst stage. The zona pellucida of blastocysts was removed using acid Tyrode solution. Blastocysts were transferred separately into a well of a 4 well-dish (Nunc #176740) coated with 0.1% gelatin and covered with MEF in ES medium supplemented with LIF. The XEN lines were derived as described 20 .
Derivation of post-XEN cell lines from whole E5.5 or E6.5 embryos. The ectoplacental cone was removed with forceps or needles. A whole embryo was placed in a well of a 4-well dish (Nunc #176740) coated with 0.1% gelatin and covered with MEF in TS medium including F4H. After the embryos formed a large outgrowth, TrypLE Express was used to disaggregate the outgrowths and passage cells into a 4-well dish. When cells reached 70-80% confluency, they were passaged into a well of a 12-well dish until XEN cell lines were obtained, which were then passaged into a well of a 6-well dish. If the outgrowth of an E5.5 embryo grew well and XEN cells thrived, we continued to culture cells in TS medium including F4H. But if the outgrowth grew slowly and XEN cells were surrounded by trophoblast-derived cells, we cultured cells in ES medium supplemented with LIF, culture conditions that would inhibit trophoblast-derived cells; when XEN cells started to become abundant, we switched to TS medium including F4H. To prepare XEN cells for RNA extraction, cells were cultured in dishes coated with 0.1% gelatin but without MEF in TS medium including F4H.
Scientific RepoRts | 6:39457 | DOI: 10.1038/srep39457 Derivation of post-XEN cell lines at E6.5 from disaggregated embryos. The ectoplacental cone was removed with forceps or needles. A whole embryo was treated with 0.1 mg/ml collagenase (Gibco #17104-019) and 0.01 mg/ml deoxyribonuclease (Gibco #D5025) for 20-30 min at room temperature, followed by 0.2 mg/ ml TrypLE for 5 min at room temperature. The embryo was disaggregated into single cells using a glass pipette with a diameter of 50-60 μ m. Cells were transferred into a well of a 4-well dish (Nunc #176740) coated with 0.1% gelatin and covered with MEF in TS medium including F4H. Three days later XEN colonies appeared. We picked these colonies, disaggregated them with a glass pipette or by TrypLE Express for 5 min at 37 °C, and passaged them into a well of a 4-well dish. When cells reached 70-80% confluency, they were passaged into a well of a 12-well dish until XEN cell lines were obtained, which were then passaged into a well of a 6-well dish.
Population characteristics. Images representing Fields of View of post-XEN cell lines were imaged for intrinsic fluorescence of GFP, fluorescence of DAPI, and GATA4 immunoreactivity. Automated cell population characteristics were determined with CellProfiler http://cellprofiler.org/ (Broad Institute, Cambridge, MA, USA). The DAPI signal was used to segment individual cells by thresholding, de-clumping, and applying size and roundness filter in an effort to evaluated single cells that are not MEFs. After filtering, the GFP and GATA4 signals were evaluated in each cell, and the population characteristics for these two markers were quantified.
In vitro differentiation. Gelatin-coated plates were prepared by coating with 0.1% gelatin overnight at room temperature. XEN cells were cultured in TS medium with F4H and with or without 10 ng/mL BMP4 (Peprotech, 120-05) on gelatin-coated plates for four days.
Immunofluorescence and imaging. Cell lines X42, X44, X47 (PDGFRa-GFP × CAG::mRFP1); NanoString multiplex gene expression analysis. Cells were cultured in 12-well plates with TS medium including F4H or ES medium supplemented with LIF. Dissociated cells were collected by trypsinization and centrifugation. Cell pellets were dispensed directly in RNAlater Stabilization Solution (Qiagen) and stored in − 80 °C for later use. Cell pellets were lysed in RLT Lysis Plus Buffer using a TissueLyser LT (Qiagen) at 40 Hz for 2 min. Total RNA extraction was performed using RNeasy Plus Micro kit (Qiagen) according to manufacturer's protocol. The custom NanoString CodeSet "Extra" was used. 100 ng of total RNA samples were hybridized at 65 °C for 18 hr and processed with the nCounter Analysis System GEN1 (NanoString Technologies). The reporter counts were processed using nSolver Analysis Software v2.5 (NanoString). Two normalizations were performed to the counts, the first normalization to the generic positive controls, followed by normalization to the reference genes, Actb and Gapdh. Normalized counts are displayed in a heatmap generated by the nSolver Analysis Software v2.5, using agglomerative clustering.