The origin of mouse extraembryonic endoderm stem cell lines

Mouse extraembryonic endoderm stem (XEN) cell lines can be derived from preimplantation embryos (pre-XEN) and postimplantation embryos (post-XEN). XEN cells share a gene expression profile and cell lineage potential with primitive endoderm (PrE) blastocysts. However, the cellular origin of XEN cells in embryos remains unclear. Here, we report that post-XEN cell lines are derived both from the extraembryonic endoderm and epiblasts of postimplantation embryos and that pre-XEN cell lines are derived both from PrE and epiblasts of blastocysts. Our strategy consisted of deriving post-XEN cells from clumps of epiblasts, parietal endoderm (PE) and visceral endoderm (VE) and deriving pre-XEN cell lines from single PrE and single epiblasts of blastocysts. Thus, XEN cell lines in the mouse embryo originate not only from PrE and PrE-derived lineages but also from epiblast and epiblast-derived lineages of blastocysts and postimplantation embryos.


INTRODUCTION
After the 8-to 16-cell stage, mouse embryos start to compact and initiate the first cell fate decision to form two distinct cell types: the outer and inner cells. The outer cells will develop mostly into trophectoderm, which ultimately give rise to the fetal portion of the placenta. The inner cells will develop mostly into the inner cell mass (ICM) 1  of postimplantation embryos [8][9][10] . Extraembryonic endoderm stem (XEN) cell lines were first derived from blastocysts. 11 XEN cell lines can also be converted from ES cell lines by ectopic gene expression or by adding chemical factors to the culture medium [12][13][14][15][16][17] or be induced from fibroblasts 18 . We have previously reported that XEN cell lines can be derived efficiently from postimplantation embryos 19 and that the derivation and maintenance of XEN cell lines does not require PDGRFA 20 . We termed XEN cell lines derived from postimplantation embryos "post-XEN cell lines" to distinguish them operationally from XEN cell lines derived from blastocysts, which we termed "pre-XEN cell lines". The embryonic origin of pre-XEN and post- 4 XEN cell lines remains unclear. Mouse fibroblasts pass through an XEN-like state before forming induced pluripotent stem cells by chemical reprogramming 21 .
Here, we derived post-XEN cell lines from clumps of epiblasts, PE and VE and derived pre-XEN cell lines from single PrE and single epiblasts of blastocysts. 5

embryos.
We set up a natural mating between a homozygous female of the gene-targeted Cre reporter strain R26-tauGFP41 22 and a heterozygous male of the gene-targeted driver strain Sox17-Cre 23 . In (R26-tauGFP41 x Sox17-Cre) F1 embryos, cells become permanently labeled with GFP upon expression of Sox17 (which occurs in PrE but not in epiblasts), and the descendants of these cells, including XEN cells, are also labeled permanently.
We have reported the derivation of post-XEN cell lines from postimplantation embryos at E6.5 19,20 . We sectioned 14 (R26-tauGFP41 x Sox17-Cre) F1 decidua at the E6.5 stage and found that the overwhelming majority of GFP+ cells resided within the ExEn (Fig. 1A). Then, we collected 5 GFP+ and 4 GFP-E6.5 (R26-tauGFP41 x Sox17-Cre) F1 embryos from a natural mating between a homozygous R26-tauGFP41 female and a heterozygous Sox17-Cre male. Previously, we derived post-XEN cell lines from postimplantation embryos at E6.5 with whole embryos (termed the whole embryo method) 19,20 . Here, we incubated the embryos in 2 mg/ml collagenase in Ca 2+ /Mg 2+ -free PBS for 20 min at room temperature and disaggregated them into small pieces and single cells using a 120-200 µm glass pipette (termed the disaggregated embryo method) (Fig. 1B). We placed each disaggregated embryo into a well of a 4-well dish in TS cell medium with F4H. The cultures of the GFP+ embryos were intended to establish post-XEN cell lines, and the cultures of the GFPembryos were used for immunofluorescence analysis. On day 4, we observed two 6 types of colonies: GFP+ XEN-like colonies and GFP-flat colonies (Fig. 1C). We picked colonies of the two types using two hypodermic 30G needles and a 20-µl plastic pipette tip and pooled each colony type. We disaggregated each pool into a suspension of small pieces and single cells and transferred the suspension into a well of a 4-well dish in TS cell medium with F4H. Cultures from pooled GFP+ XEN-like colonies grew quickly, and we established 5 post-XEN cell lines from 5 pools (Fig.   1D). Cultures from pooled GFP-flat colonies initially maintained a flat morphology, with GFP+ cells surrounding on day 10; then, on day 35, only GFP+ XEN-like cells remained (Fig. 1E). It appears that cells of GFP-flat colonies convert into GFP+ XENlike cells in culture. We established 3 post-XEN cell lines from 5 pools of GFP-flat colonies.
We performed immunofluorescence on outgrowths from disaggregated GFP-E6.5 embryos on day 7 and on cultures on days 18 and 36. The outgrowths expressed GATA6, SOX7 and OCT4 and weakly or no NANOG on day 7 (Fig. 1F). On day 18, flat colonies displayed GATA6 and SOX7 expression, strong OCT4 expression, and weak or no NANOG expression (Fig. 1G). Marker expression in flat colonies distinguishes them from epiblast stem cells (which express NANOG and GATA6) and ES cells (which express NANOG but not GATA6) [24][25][26] . On day 36, most cells from flat colonies had the typical appearance of XEN-like cells, with strong expression of GATA6 and SOX7 and no OCT4 or NANOG expression (Fig. 1H). Another reason could be that single ExEn and single epiblast cells could not grow well in the single-cell stage. A previous study reported that epiblast stem cells would be induced widespread cell death after being disaggregated to single cells by trypsin or other single-cell dissociation methods 24 . However, clumps from disaggregated methods could survive and establish the post-XEN cell line 19 . We traced the origin of XEN cells in E6.5 (R26-tauGFP41 x Sox17-Cre) F1 embryos in clumps from ExEn, 9 epiblasts and EXE. In (R26-tauGFP41 x Sox17-Cre) F1 embryos, cells become permanently labeled with GFP upon expression of Sox17 (which occurs in PrE but not in epiblasts), and the descendants of these cells, including XEN cells, are also labeled permanently 23 . GFP+ cells were located in ExEn, and epiblasts were located in the middle of the embryo piece and consisted of GFP-cells. We produced 4 GFP+ E6.5 (R26-tauGFP41 x Sox17-Cre) F1 embryos by natural mating between a homozygous R26-tauGFP41 female and a heterozygous Sox17-Cre male ( Fig. 2A). We used two hypodermic 30G needles to cut each embryo into three pieces: an extraembryonic ectoderm piece (EXE piece) consisting of EXE and ExEn surrounding EXE; an epiblast piece (EPI piece) consisting of epiblast and ExEn surrounding epiblast; and a middle piece consisting of the transition between EXE and epiblast ( Fig. 2B). We discarded the middle piece. We incubated the EXE and EPI pieces separately in 2 mg/ml collagenase in Ca 2+ /Mg 2+ -free PBS for 20 min at room temperature. First, we used a glass needle of 200 µm inner diameter to separate EXE, EPI, and ExEn pieces. Using a glass needle of 60-80 µm inner diameter, we disaggregated the EXE, EPI and ExEn pieces into small clumps. We then picked several clumps of ExEn (combining ExEn from EXE and EPI pieces), EPI, and EXE.
The ExEn clumps expressed GFP and had a dark background; in contrast, the EPI and EXE clumps did not express GFP and had a white background (Fig. 2C) because any GFP+ cells in the EPI and EXE clumps were detected by fluorescence on day 0.
We pooled and transferred each of the three types of clumps into a well of a 4-well dish in TS cell medium with F4H. On day 2, the ExEn clumps showed strong GFP fluorescence; the EPI clumps showed weak GFP fluorescence (on day 0, no GFP 10 fluorescence); and the EXE clumps had no GFP fluorescence. On days 4 and 6, the three types of clumps formed outgrowths. On day 16, the ExEn outgrowths abounded with GFP+ XEN-like cells; the EPI outgrowths were a mixture of GFP+ XEN-like cells and GFP-cells; and the EXE outgrowths were devoid of GFP+ cells.
Thus, GFP+ cells from E6.5 (R26-tauGFP41 x Sox17-Cre) F1 embryos could originate from ExEn and can also be converted in culture from epiblasts but not from EXE. ExEn; the middle piece, which is discarded, contains some EXE and some EPI; and the EPI piece contains EPI and ExEn. (C) Culture of clumps of ExEn, EPI, and EXE on various days.

embryos.
We have already derived post-XEN cell lines from E5.5 and E6.5 embryos 19 . Here, we derived post-XEN cell lines from E7.5 embryos. We produced 4 GFP+ E7.5 (R26-tauGFP41 x Sox17-Cre) F1 embryos by natural mating between a homozygous R26-tauGFP41 female and a heterozygous Sox17-Cre male. By the disaggregation method, we picked several clumps of ExEn and as few epiblast cells as possible from each embryo into a well of 4-well dishes coated with gelatin, covered them with MEF, and cultured them in TS medium with F4H. On day 7, we picked XEN-like colonies, and on day 30, we established 4 post-XEN cell lines from 4 E7.5 embryos (Fig. 3A). To confirm that post-XEN cells from E7.5 can contribute to ExEn, we injected two post-XEN cell lines (X-E7.5-AC1563-1 and X-E7.5-AC1563-2) into 17 C57BL/6J blastocysts and transferred 17 blastocysts into two foster mothers. We obtained 13 deciduae of E7.5 stage, sectioned the deciduae, and then stained the sections with an antibody against PDGFRA and E-cadherin. We found that 6 chimeras from 13 deciduae with GFP cells contributed to ExEn (Fig. 3B).
The transgenic driver strain Sox2-Cre is epiblast-specific 27 and has been widely used to mark epiblast-derived cell lineages. Sox2-Cre strain is heterozygous 27 . Finally, to determine whether post-XEN cell lines are derived from PE and/or VE, we separated 11 GFP+ E7.5 (R26-tauGFP41 x Sox2-Cre) F1 embryos (Fig. 3C) into two pieces: Reichert's membrane (RM) (Fig. 3D) and an embryo piece (epiblast with VE and EXE) (Fig. 3E). RM is formed by PE cells together with trophoblast giant cells 28 . GFP+ cells were located in the epiblast region, which is in the middle of the 13 embryo piece. VE surrounds epiblasts and EXEs and consists of GFP-cells. We cut the embryo piece into two pieces (EXE piece and EPI piece) with two hypodermic 30G needles and discarded the EXE piece, which was GFP-. We transferred a piece of each RM and each EPI piece (epiblast with VE) separately into a well of a 4-well dish in TS cell medium with F4H. On days 3 and 10, an outgrowth of XEN-like cells surrounded the RM, and 7 of 11 did not contain GFP+ cells, but 4 of 11 contained a few GFP+ cells (Fig. 3F). On day 12, we disaggregated the outgrowths and passaged the cells. On day 38, we established 9 post-XEN cell lines from 11 RM pieces (called PE-XEN). We treated 6 of the 11 EPI pieces with 2 mg/ml collagenase in Ca 2+ /Mg 2+free PBS for 10-20 min at room temperature and disaggregated them into small clumps using a 200-300 µm glass pipette (Fig. 3G). We picked the VE clumps and  all single cells from ICMs were used as a group (Fig. 4A). If the group has any outgrowth cells that express GFP+, we term the ICMs R26-tauGFP41 x Sox2-Cre. If the group had no outgrowth cells that expressed GFP+, the ICMs were considered wild-type sox2. We placed 170 single cells from the 15 ICMs each into a well of a 4well dish (all single cells did not express GFP) and cultured them in ES cell medium with LIF.
We observed two different outgrowths from single ICM cells from days 3 to 8. The outgrowth with GFP-expression and a phenotype of XEN-like on day 3 was termed a single PrE, and the outgrowth with strong GFP+ expression and an ES-like phenotype on day 3 was termed a single epiblast (Fig. 4B-4C). However, unexpectedly, on days 5 and 8, we observed that a few GFP+ cells appeared in the outgrowths that were previously termed single PrE cells, and on day 30, a few GFP+ cells were still in the pre-XEN cell line, which was established from a single PrE (Fig.   4B). This result indicated that the Sox2-cre driver was activated in a few PrE cells during the XEN cell line derivation procedure. The outgrowths from single epiblast cells showed typical ES-like cells on day 3, day 5, and day 8. After disaggregating the outgrowth to single cells and clumps and passaging the cells, both ES-like and XENlike colonies appeared in the plates (Fig. 4C). We got ES-like cells with some XENlike cells. We cultured the cell lines either in ES medium with LIF or in TS medium with F4H on day 30. We changed the medium every other day and passaged the cells every 1-2 weeks. The cells were cultured with crowded, and some of XEN-like cells 17 were in suspension medium, we collected the suspension medium and spun down, and passaged the cells into a new plate which coated with gelatin and covered with MEF, according to the protocol as we shown in previous report 20 , but we didn't treat the cells with retinoic acid (RA) and activin A. We got XEN cell lines by collected the cells in suspension medium, in TS medium with F4H or in ES medium with LIF.
To confirm outgrowth with a typical XEN-like phenotype and that most cells were GFP-, which was termed from a single PrE cell, or outgrowth with a typical ES-like phenotype, and that most cells were GFP+, which was termed from single EPI cells, we performed immunofluorescence for the outgrowths that were from the ICMs of R26-tauGFP41 x Sox2-Cre on day 7. Four of 8 outgrowths had GFP+ expression and an ES-like phenotype. Some cells that were in the middle of the outgrowths expressed OCT4+, but GATA6+ expression was very weak. Therefore, these outgrowths could be derived from single epiblast cells. Another 4 of 8 outgrowths that had little or no GFP cells and with an XEN-like phenotype had strong GATA6+ expression and no OCT4+ expression; therefore, we concluded that these outgrowths could be derived from single PrE cells. Figure 4D shows immunofluorescence staining of such epiblast-derived outgrowth and PrE-derived outgrowth on day 7. In epiblast-derived outgrowth, cells were GFP+, some cells that were in the middle of the outgrowth still expressed OCT4+ and surrounded by GATA6+ cells, and OCT4 cells coexpressed GATA6, while the expression was very weak. PrE-derived outgrowths had GATA6 expression but no OCT4 expression, and there were a few GFP+ cells that arose from single PrE cells (Fig. 4D). Thus, by performing 18 immunofluorescence, we confirmed that Pre-XEN cells could be derived both from a single PrE and from a single epiblast. Figure 4E shows immunofluorescence staining of ES-AC1558-2-8, which was from a single epiblast, cultured crowded in ES medium with LIF on day 15. Which shows two distinct cell types: one is ES-like cells that is with OCT4 and NANOG expression, some of the ES-like cells are with weak GATA6 and SOX7 expression; another one is XEN-like cells that are with GATA6 and SOX7 expression. Figure 4F   In the process of deriving pre-XEN cell lines from blastocysts, the success rate in ES medium with LIF (56%) is higher than in TS cell medium with F4H (21%) 31  Here, we revealed a mixed origin of XEN cell lines. How can epiblasts of blastocysts and postimplantation embryos contribute to XEN cell lines? ICM comprises three distinct cell types. The first type coexpresses NANOG and GATA6 and differentiates into epiblasts or PrE; the second type forms epiblasts; and the third type forms PrE 1 . ES cells go from an epiblast-like stage to a NANOG+ GATA6+ 23 coexpression stage, over to a PrE-like stage, and then convert into XEN cells 36 . We speculate that, likewise, cells from a blastocyst could go from the NANOG+ GATA6epiblast stage to the NANOG+ GATA6+ coexpression stage, over to a NANOG-