ABSTRACT
We have characterized an induced mutation, called embryonic ectoderm development or eed, that disrupts A-P patterning of the mouse embryo during gastrulation. Positional cloning of this gene revealed it to be the highly conserved homologue of the Drosophila gene extra sex combs, which is required for maintenance of long-term transcriptional repression of homeotic gene expression. Mouse embryos homozygous for loss-of-function alleles of eed initiate gastrulation but display abnormal mesoderm production. Very little embryonic mesoderm is produced; in contrast, extraembryonic mesoderm is relatively abundant. These observations, along with mRNA in situ hybridization analyses, suggested a defect in the anterior primitive streak, from which much of the embryonic mesoderm of the wild-type embryo is derived. To analyse this defect, we initiated clonal analysis of the pre-streak epiblast in eed mutant embryos, using the lineage tracer horseradish peroxidase (HRP). The results of these studies indicate that epiblast cells ingress through the anterior streak, but the newly formed mesoderm does not migrate anteriorly and is mislocalized to the extraembryonic compartment. Abnormal localization of mesoderm to the extraembryonic region did not appear to be due to a restriction and alteration of distal epiblast cell fate, since the majority of clones produced from regions fated to ingress through the anterior streak were mixed, displaying descendants in both embryonic and extraembryonic derivatives. eed mutant embryos also fail to display proper epiblast expansion, particularly with respect to the A-P axis. Based on patterns of clonal spread and calculated clone doubling times for the epiblast, this does not appear to be due to decreased epiblast growth. Rather, epiblast, which is normally fated to make a substantial contribution to the axial midline, appears to make mesoderm preferentially. The data are discussed in terms of global morphogenetic movements in the mouse gastrula and a disruption of signalling activity in the anterior primitive streak.
INTRODUCTION
In the mouse, gastrulation begins approximately midway through embryonic day (E) 6 (for reviews, see Tam and Behringer, 1997; Tam et al., 1993). At this time, the embryo consists of a single layer of primitive ectoderm (epiblast) covered by a layer of primitive endoderm. On the posterior side of the embryo, a disruption in the epithelial basal lamina defines a transient structure called the primitive streak, through which epiblast cells ingress and differentiate into mesoderm and definitive endoderm. Mesodermal cells exiting the primitive streak migrate anteriorly around the surface of the epiblast, contributing to head and trunk mesoderm; mesoderm also moves proximally into the extraembryonic region, contributing to the formation of the amnion, chorion, yolk sac and allantois, tissues that are critical for the proper development of the maternal-fetal interface.
The rapidly growing number of mutations affecting various aspects of gastrulation is beginning to allow for elucidation of a molecular basis for this dynamic process (reviewed in Tam and Behringer, 1997). We have characterized an induced mutation, called embryonic ectoderm development or eed, that disrupts A-P patterning of the mouse embryo during gastrulation (Faust et al., 1995; Niswander et al., 1988). Positional cloning of this gene revealed it to be the highly conserved homologue of the Drosophila gene extra sex combs (esc), which is required for maintenance of long-term transcriptional repression of homeotic gene expression (Schumacher et al., 1996). Mouse embryos homozygous for loss-of-function alleles of eed (hereafter referred to as eed mutant embryos or eed mutants) initiate but fail to complete gastrulation and display abnormal mesoderm production (Faust et al., 1995; Niswander et al., 1988). Very little embryonic mesoderm is present, and it does not differentiate into the normal axial and paraxial derivatives, i.e., the notochord and somites. In contrast, abundant extraembryonic mesoderm is produced, forming an allantois and the mesodermal lining of the amnion, chorion and yolk sac. Accompanying reduced embryonic mesoderm production is a complete lack of anterior development: embryos display an underdeveloped embryonic ectoderm with no morphological or molecular evidence of neural induction (Schumacher et al., 1996). Because eed mRNA is normally expressed in all tissues of the gastrulating embryo, it is unclear whether lack of anterior development in eed mutant embryos is directly related to or independent of abnormal mesoderm production.
Extensive fate mapping of the wild-type mouse gastrula and primitive streak (Lawson et al., 1991; Tam and Beddington, 1987) suggests that the gross deficiency of embryonic mesoderm and absence of axial and paraxial derivatives in eed mutant embryos reflect a defect in the anterior primitive streak. This hypothesis is supported by lack of formation of a morphologically defined node, which is the mouse gastrula organizer (Beddington, 1994). In addition, Nodal mRNA fails to be localized to the streak and there is a loss of expression of Fgf4 and Gsc in the anterior streak (Faust et al., 1995). Evx1, which is normally expressed in the streak in a graded fashion with highest levels posteriorly (Dush and Martin, 1992), appears to be expressed at high levels throughout the eed mutant streak (Schumacher et al., 1996). Taken together, the data suggest a disturbance in the anterior primitive streak of eed mutant embryos. In contrast, the apparently unimpeded production of extraembryonic mesoderm and primordial germ cells (Faust et al., 1995) suggests that epiblast cells are able to traverse the posterior streak.
We wished to define further the anterior primitive streak defect in eed mutant embryos. The presence of sparse axial mesoderm (Faust et al., 1995) suggested that some initial ingression occurs, but might be subsequently blocked. Alternatively, continuous ingression of epiblast cells through the anterior streak might occur, but mesodermal cells leaving the streak move aberrantly into the extraembryonic region. In order to test these hypotheses, we initiated clonal analysis of the pre-streak epiblast in eed mutant embryos, using the lineage tracer horseradish peroxidase (HRP). Using this technique, the morphogenetic movements of eed epiblast cells through the streak and localization of their descendants in the embryo have been determined. Clonal analysis also enabled us to investigate if individual epiblast cell fate is altered, as well as to examine cell proliferation in the eed mutant embryos. The results of these studies indicate that there is no block in epiblast cell ingression through the anterior streak; instead, nascent mesoderm does not migrate anteriorly and is subsequently mislocalized to the extraembryonic region of eed mutant embryos.
MATERIALS AND METHODS
Embryos
The eed mutant mice used in this study (l(7)Rn53354SB) originated at the Oak Ridge National Laboratory (Rinchik and Carpenter, 1993) and were maintained as described (Faust et al., 1995). Heterozygous Tyrc-ch, +/Tyrc, l(7)Rn53354SB mice were crossed inter se to obtain embryos. Noon of the day of the appearance of a vaginal plug was designated E0.5. Embryos were dissected at approximately E6.7. At this time, embryos were classified as either pre-streak or early streak. Pre-streak embryos were either radially symmetrical or exhibited a thickened posterior epiblast, preceding the site of primitive streak formation. Early streak embryos displayed some mesoderm formation. Only pre-streak embryos were used for cell labelling.
Cell labelling
Embryos were labelled iontophoretically as described (Lawson et al., 1991), with minor modifications. The embryo injection assembly consisted of a single-chambered LabTek II slide (Nunc) in which the chamber had been removed. Embryos were injected in a large drop of Dulbecco’s modified minimal essential medium (DMEM) containing 10% fetal calf serum and 20 mM Hepes. Labelling was performed on a Nikon Diaphot microscope fitted with differential interference contrast (DIC) optics and epifluorescence. The microscope was attached to an image intensifier (Dage/MTI), standard CCD camera and video monitor to allow visualization of the embryos. One epiblast cell per embryo was injected with a mixture of 8% HRP (Boehringer Mannheim Biochemicals) and 2% lysinated-tetramethyl rhodamine dextran (LRDX, Mr 10×103; Molecular Probes) in 0.05 M KCl, at 3 nA discontinuous current (0.5 second duty cycle) for 15 seconds. Occasionally two adjacent epiblast sister cells were labelled due to leakage via cytoplasmic bridges (14% and 16% of the time for mutant and wild-type embryos, respectively). These embryos were included in the analysis and did not significantly affect interpretation of the data. An additional injection was made in one extraembryonic visceral endoderm cell at 3 nA discontinuous current for 12 seconds to identify retrospectively the ‘position of longitude’ of the injected epiblast cell in the egg cylinder (Lawson et al., 1991). The positions of the injected, fluorescently labelled cells were recorded with a Sony black and white videographic printer attached to the video monitor.
Embryo culture
Embryos were cultured in DMEM containing 50% rat serum as described (Lawson et al., 1986). Embryos were cultured for either 1 day (range 15-31.5 hours, average time 23.6 hours) or 2 days (range 34.5-45.5 hours, average time 39.6 hours). The percentage normal development was 83% and 80% for wild-type and eed mutant embryos, respectively, after 1 day in culture. The percentage normal development after 2 days in culture was 75% and 65% for wild-type and eed mutant embryos, respectively.
Detection of HRP-labelled cells and scoring of embryos
Detection of HRP-labelled cells was performed as described (Lawson et al., 1991). After staining, embryos were fixed in 2.5% glutaraldehyde in phosphate-buffered saline, dehydrated, cleared in benzyl alcohol: benzyl benzoate (1:2) and photographed under DIC optics. After clearing, those embryos displaying an abnormal phenotype and/or resembling homozygous eed mutants were subjected to genotype analysis. A portion of the yolk sac was removed, lysed and PCR was performed using primers flanking the l(7)Rn53354SB point mutation (Schumacher et al., 1996). eed homozygous embryos were identified by an AluI restriction polymorphism; the mutant allele disrupts this site. All embryos displaying normal development with respect to various morphological features (Table 1) were classified as ‘wild type.’
To confirm the number of descendants of the labelled cells and precisely determine their location, embryos were embedded in glycolmethacrylate (JB4; Polysciences), serially sectioned and stained with 0.2% methylene blue. Only those embryos wherein all of the descendants were evenly stained were used to calculate clone doubling times for labelled cells, using a previously established formula (Lawson and Pedersen, 1987; Lawson et al., 1991). Furthermore, clone doubling times were calculated using only those embryos cultured for 1 day, due to the potential loss of descendants from dilution of HRP after 2 days in culture (Lawson et al., 1986).
Statistical analysis
The two-sample t-test (Zar, 1984, pp.126-128) was used to determine whether the differences for mean dimensions (Table 1) and clone doubling times between wild-type and mutant clones were statistically significant. To determine if the percentage distribution of descendants to different embryonic regions differed significantly between wild-type and mutant embryos (Tables 3, 4, 6), Chi-square and the Wilcoxon rank-sum tests were performed using PROC FREQUENCY and PROCNPAR1WAY software (SAS Institute, Inc., Cary, NC). For all calculations, the level of significance employed was ≤5%.
RESULTS
Embryo injection and development in vitro
For the analyses reported here, single epiblast cells of wild-type and eed mutant embryos from the same litters were injected with HRP at the pre-streak stage and cultured for either 1 or 2 days before staining the descendants. Wild-type and eed mutant embryos yielded epiblast clones at similar efficiencies (Table 1).
At the time of injection, wild-type and eed mutant embryos displayed similar embryonic dimensions (Table 1) but, with careful observation, eed mutant embryos could be identified. In general, the diameter of the eed mutant extraembryonic region was narrower than wild type and the ratio of embryonic height to extraembryonic height was often greater than the 1:1 ratio evident in wild-type embryos (compare Fig. 1A,B). After 1 day in culture, most wild-type embryos developed to the late streak stage and had formed an amnion, chorion and allantoic bud (Fig. 1C; Table 1). In contrast, eed mutant embryos were often smaller and, although the primitive streak was elongated, it never extended to the distal tip of the embryo (Fig. 1D). Varying degrees of mesoderm production were observed, but no amnion or chorion were formed (compare Fig. 1C,D).
After 2 days in culture, most wild-type embryos possessed 3-6 somites (Fig. 1E; Table 1). While eed mutant embryos continued to grow between 1 and 2 days in culture and demonstrated some degree of A-P axial extension (Table 1), they were significantly smaller than their wild-type littermates. Mutant embryos displayed a well-developed extraembryonic region, with an enlarged yolk sac and formation of an amnion and chorion (Fig. 1F). A large mass of extraembryonic mesoderm was present that extended from the posterior streak into the exocoelomic cavity in a proximoanterior direction, reminiscent of an enlarged allantoic bud. Indeed, this mesoderm showed signs of allantoic differentiation in its distal-most region and often displayed fusion with the chorion (Fig. 1F). Blood islands were occasionally present in the yolk sac. In summary, the development of eed mutant embryos in culture was identical to their development in utero; this was confirmed molecularly by in situ hybridization analysis of genes expressed in the primitive streak and its derivatives (Faust et al., 1995; C. F. and T. M., unpublished data).
Experimental strategy
Wild-type fate maps constructed by clonal analysis were based on the division of the epiblast into 11 zones (Lawson et al., 1991). Considerable overlap in cell fate among different zones was observed. For this reason, and to simplify the presentation of the results reported in this study, we have focused our efforts on selected zones and pooled the data obtained from different zones, where applicable, into five distinct regions along the anterior-posterior axis (Fig. 2). The proximal two-fifths of the epiblast, which includes regions A and B, primarily contributes to embryonic ectoderm, proximal embryonic mesoderm and extraembryonic mesoderm after 1 day in culture, and to surface and amnion ectoderm as well as extraembryonic mesoderm after 2 days in culture (Fig. 2; Lawson et al., 1991; Lawson and Pedersen, 1992b). The middle two-fifths of the epiblast, which includes regions C and D, primarily contributes to embryonic ectoderm and mesoderm after 1 day in culture, and neural ectoderm and/or mesodermal derivatives after 2 days in culture (Fig. 2; Lawson et al., 1991; Lawson and Pedersen, 1992a,b). The distal-most portion of the epiblast, region E, contributes primarily to embryonic ectoderm after 1 day in culture and its derivative, neural ectoderm, after 2 days in culture (Fig. 2; Lawson et al., 1991; Lawson and Pedersen, 1992a).
Besides illuminating epiblast cell fate, clonal analysis yielded important information regarding the temporal and spatial incorporation of epiblast cells into the primitive streak, findings that are particularly relevant to this study (Lawson et al., 1991). At the time of primitive streak formation, the first cells to ingress lie at the posterior midline in regions B and D. As these cells leave the streak, they are replaced by progressively more lateral and anterior epiblast cells, including regions A and C. Thus, epiblast growth appears to be directed posteriorly towards the streak. Once epiblast cells reach the streak, they ingress, resulting in primitive streak extension and continuous mesoderm production. By the onset of neurulation at E7.5, the bulk of the epiblast initially located laterally and posteriorly (including regions B and D) will have been incorporated into mesoderm, while a substantial number of epiblast cells initially located more anteriorly (including regions A and C) will give rise to descendants that do not ingress through the streak. These cells will contribute to surface (regions A and C) and neural (region C) ectoderm.
Furthermore, clonal analysis revealed important information regarding growth of the epiblast and expansion of the A-P axis. In wild-type embryos, the A-P axis increases 3.5-fold in length during gastrulation, primarily due to elongation of the primitive streak and expansion of the area of the epiblast just anterior to the primitive streak (Lawson et al., 1991; Lawson and Pedersen, 1992a). As cells of regions A-D are progressively incorporated into the streak, the streak elongates. In addition, some descendants from region D remain in the ectoderm, inserting at the midline and contributing to the extension of the A-P axis (Lawson et al., 1991; Lawson and Pedersen, 1992a). As a result, descendants of the distal cap region E are propelled anteriorly, contributing to anterior expansion of the embryonic ectoderm.
(1)The results obtained from clonal analysis of wild-type embryos enabled us to propose several hypotheses and make the following predictions regarding the fate and behavior of cells in these different regions of the epiblast of eed mutant embryos.
(2) Sparse embryonic mesoderm production in eed mutant embryos is consistent with an abnormal anterior primitive streak. We propose two hypotheses to explain this defect in eed mutant embryos. One possibility is that epiblast cells ingress through the anterior streak, but as mesodermal cells exit, they move aberrantly into the extraembryonic region. If correct, clonal analysis of regions C and D in eed mutants would reveal a greater degree of contribution to extraembryonic mesoderm than observed in wild-type embryos. An alternative hypothesis is that ingression through the anterior primitive streak does not occur, or is greatly reduced. This was an attractive possibility because the small size of the epiblast of eed mutant embryos suggested a cell proliferation problem. If this were the case, we would predict seeing little posteriorly directed growth of the epiblast from anterior region C, and reduced or no contribution to mesoderm from this and posterior region D.
(3) The diminished epiblast expansion evident in eed mutant embryos might arise from either a general defect in epiblast proliferation, decreased primitive streak elongation, lack of incorporation of cells of region D into the midline ectoderm or a combination of these events. Clonal analysis was used to investigate these possibilities. To characterize epiblast expansion in eed mutant embryos further, we also examined the fate of cells in region E.
Cells traversing the posterior streak display normal fates
As predicted from our morphological and molecular analyses of eed mutant embryos, eed mutant epiblast cells from regions A and B displayed similar fates to those of their wild-type counterparts, with some minor differences. As in wild-type embryos, after 1 day in culture descendants of region A were spread posteriorly in the epiblast towards the streak (Fig. 1C,D; Table 2). In addition, contribution to embryonic ectoderm and mesoderm, primitive streak, amnion ectoderm and extraembryonic mesoderm was observed after 2 days in culture (Table 2 and data not shown).
In wild-type embryos, region B, which contains cells traversing the streak very early in gastrulation, contributed to proximal embryonic (3/5 embryos injected), as well as extraembryonic mesoderm (5/5) after 1 day in culture (Table 2 and data not shown). In contrast, 3/3 eed mutant embryos injected in this region contributed to the proximal posterior mesoderm, but only 1/3 contributed to the extraembryonic mesoderm after 1 day in culture (Table 2). This difference may be due to a 6-12 hour delay in the onset of gastrulation observed in eed mutant embryos. Nevertheless, after 2 days in culture, both eed mutant and wild-type embryos contributed extensively to extraembryonic mesoderm (Table 2 and data not shown).
Cells traversing the anterior streak display abnormal fates
(i)Region D
Epiblast cells from region D traverse the primitive streak early in gastrulation and give rise to the bulk of the paraxial and axial mesoderm and definitive endoderm of the embryo; their descendants also make a significant contribution to ventral midline neurectoderm (Lawson et al., 1991; Lawson and Pedersen, 1992a,b; K. A. L., unpublished results). Rarely are descendants of these cells found in extraembryonic mesoderm. Analysis of region D in eed mutant embryos revealed marked differences in colonization by descendants (Fig. 3). For example, after 1 day in culture, the majority of descendants of wild-type region D were found in embryonic mesoderm, with fewer contributing to embryonic ectoderm, primitive streak, definitive endoderm and extraembryonic mesoderm (Fig. 3A). Descendants of eed mutant region D were also found in the embryonic mesoderm, streak, embryonic ectoderm and definitive endoderm. However, the embryonic mesoderm descendants lay posteriorly in the embryo (Fig. 3B). Transverse histological sections revealed that, while mesoderm cells readily exited the streak, they accumulated posterior to it and demonstrated very little anterolateral migration away from it (data not shown). Cell mixing in eed mutant embryos was prevalent, in that descendants in the mesoderm were often dispersed widely among unstained cells. In some cases (2/16 embryos), clonal spread was extensive, spanning from the most distal embryonic mesoderm into the extraembryonic region. Other embryos displayed clones encompassing proximal embryonic and extraembryonic mesoderm (3/16 embryos), proximal embryonic ectoderm (2/16), distal embryonic mesoderm (6/16), or both (3/16).
The differences in contribution to mesoderm were amplified after 2 days in culture. In wild-type embryos, descendants from region D that after 1 day were found in the embryonic mesoderm, were now located in the cranial paraxial mesoderm and somites (Fig. 3A). Localization to the primitive streak, definitive endoderm, node and ventral midline neurectoderm was also observed. Descendants of eed mutant region D were still found in the embryonic mesoderm, primitive streak and definitive endoderm, but the bulk of the mesodermal descendants localized extraembryonically, in the allantoic bud-like mesoderm, as well as the differentiated allantois and yolk sac (Fig. 3B). 2/7 embryos contained clones wherein descendants were restricted to the extraembryonic region. In the remaining embryos examined, clones spanned either the proximal embryonic mesoderm and extraembryonic mesoderm (3/7) or extended from the distal into the proximal embryonic mesoderm (2/7). Again, transverse sections revealed very limited lateral migration away from the primitive streak (data not shown). Interestingly, once mesodermal cells reached the extraembryonic region, they were able to move more laterally and anteriorly, as descendants were found widely dispersed throughout the yolk sac and in the allantoic bud-like mesoderm and differentiated allantois (Fig. 3B).
The percentage distribution of descendants and incidence of embryos contributing to different embryonic and extraembryonic derivatives of region D are described in Table Chi-square analysis indicated that the overall distribution of descendants of region D to these different derivatives after culture was significantly different between wild-type and eed mutant embryos. After 1 day in culture, a small percentage of descendants localized to the embryonic ectoderm, primitive streak and axial mesoderm in both wild-type and eed mutant descendants in embryonic ectoderm and axial mesoderm was lower than wild type. Although the frequency of descendants contributing to embryonic mesoderm was similar for both wild-type and eed mutants, the incidence of eed mutant embryos contributing to embryonic mesoderm was much higher (14/16 compared to 6/10 embryos). The percentage distribution of descendants found in the endoderm was decreased in eed mutant embryos (7%, as compared to 17% for wild type). Both eed mutant and wild-type embryos displayed significant contribution to the extraembryonic mesoderm; however, although wild-type embryos exhibited a similar percentage of descendants in extraembryonic mesoderm, only 2/10 embryos contributed to this tissue, with one of these yielding a pure extraembryonic clone containing 24 descendants (accounting for 90% of the descendants in this region).
After 2 days in culture, 5/6 wild-type embryos contributed to embryonic ectoderm derivatives (ventral midline neurectoderm and/or the node; 21% of all descendants), while no contribution to embryonic ectoderm was seen in eed mutants (Table 3). The Wilcoxon rank-sum test (see Materials and Methods) indicated that this difference in contribution was highly significant. Both wild-type and eed mutant embryos continued to display a small contribution to the primitive streak. eed mutant embryos still contributed to embryonic mesoderm, although the percentage of descendants found here was decreased compared to both wild-type controls cultured 2 days and eed mutant embryos cultured for 1 day. Concomitantly, there was both an increase in the incidence of eed mutant embryos contributing to, as well as the percentage of descendants colonizing, the extraembryonic mesoderm in eed mutant embryos (5/7 versus 5/16 embryos; 49% of descendants versus 12% colonizing after 1 day in culture). In contrast, 0/6 wild-type embryos contributed to extraembryonic mesoderm after 2 days in culture; this difference in incidence was determined to be highly significant using the Wilcoxon rank-sum test. The percentage of descendants found in endoderm was still lower in eed mutant embryos and no contribution to axial mesoderm was seen. Whether the lack of eed mutant embryos displaying descendants in axial mesoderm after 2 days in culture is significant cannot be determined because of the low incidence of wild-type embryos contributing to this tissue (Table 3; K. A. L., unpublished results). Overall, the data indicate that epiblast cells from region D ingress through the anterior streak, but as their mesodermal descendants exit, they move extraembryonically.
(ii) Region C
Having established that epiblast cells from region D ingress through the streak and contribute to mesoderm formation in eed mutant embryos, we next analysed the fate of region C. As stated earlier, it was unclear whether cells from this anterior region would undergo proper growth toward the streak and subsequent ingression and, as a result, contribute to the loss of embryonic mesoderm. The localization of descendants from region C is depicted in Fig. 4; the percentage distribution of descendants and incidence of embryos contributing to different embryonic and extraembryonic derivatives are described Table 4. Chi-square analysis indicated that the overall distribution of descendants of region C to these different derivatives after culture was significantly different between wild-type and eed mutant embryos. In wild-type embryos after 1 day in culture, descendants of region C were found in the epiblast spread posteriorly towards the streak; epiblast cells initially lying more posteriorly in region C also gave rise to descendants in the primitive streak and mesoderm (Fig. 4A; Table 4). Although a similar incidence of eed mutant embryos displayed descendants in the embryonic ectoderm, a much higher percentage of descendants was found in this region as compared to wild type (Table 4). In addition, the bulk of the descendants were localized more anteriorly in eed mutants (compare Fig. 4A,B). Furthermore, both the incidence of eed mutant embryos contributing, and the percentage of descendants found in the primitive streak and mesoderm, were lower than wild type. These results could reflect the above-mentioned delay in the onset of gastrulation evident in eed mutant embryos.
After 2 days in culture, a significant percentage of descendants of wild-type region C did not ingress through the primitive streak; instead they contributed to embryonic ectodermal derivatives, the neural and surface ectoderm (Fig. 4A; Table 4). In eed mutant embryos injected in region C and cultured for 2 days, similar contribution to the embryonic ectoderm was observed (Table 4), and, like wild type, descendants spanned posteriorly toward the streak (Fig. 4B).
Descendants of wild-type epiblast cells initially lying more posteriorly in region C contributed substantially to lateral plate mesoderm, as well as paraxial and posterior (postnodal) mesoderm after 2 days in culture (Figs 1E, 4A; Table 4). Epiblast cells lying posteriorly in eed mutant region C also gave rise to descendants in the embryonic mesoderm, albeit at a lower percentage (30%, as compared to 58% for wild type). In contrast to descendants of wild-type region C, all embryonic mesoderm descendants were found proximally, near the embryonic/extraembryonic junction and either posterior or just lateral to the primitive streak. Furthermore, a higher incidence of eed mutant embryos displayed contribution to the extraembryonic mesoderm (3/5 embryos; 28% of descendants, compared to 2/7 embryos, 3% of descendants for wild type) (Figs 1F, 4B; Table 4). Thus, as in wild-type embryos, descendants of eed mutant region C are able to reach the streak and produce mesoderm; however, as was seen with mutant region D, very little anterolateral migration out of the streak occurs and the mesoderm tends to be displaced extraembryonically.
Descendants of eed mutant regions C and D that localized to the extraembryonic region of eed mutant embryos were usually found in more than one extraembryonic derivative. Descendants were found in either allantoic bud-like mesoderm (6/10 embryos, 28% of extraembryonic descendants), the mesoderm lining the yolk sac (7/10 embryos, 54% of extraembryonic descendants) and/or allantoic mesenchyme (2/10 embryos, 18% of extraembryonic descendants).
Mislocalization of mesoderm extraembryonically is not due to an inherent restriction of distal epiblast cell fate
It has been previously established that the epiblast of the pre-streak mouse embryo displays no regional restriction in cell fate (Lawson et al., 1991). In eed mutant embryos, mislocalization of descendants to the extraembryonic mesoderm could reflect an alteration and premature restriction of epiblast cell fate, such that distal epiblast cells now become restricted to an extraembryonic, rather than embryonic, fate. If this occurred, we would only expect to see clones wherein all of the descendants were localized either to embryonic or extraembryonic derivatives. This was not the case (Table 5). Although 25% of eed mutant embryos injected in regions C and D displayed clones wherein all of the descendants were localized strictly to the extraembryonic region, 42% displayed mixed clones, wherein descendants were found in extraembryonic mesoderm, as well as embryonic mesoderm and/ or endoderm, in a single clone. In contrast, only 15% of wild-type embryos injected in regions C and D displayed mixed clones containing both embryonic and extraembryonic derivatives after 2 days in culture. Thus, the propensity to form extraembryonic mesoderm in eed mutant embryos appears to be due to a tendency for most cells leaving the streak to move extraembryonically and not due to an inherent restriction of distal epiblast cell fate.
Epiblast cells at the distal cap display abnormal morphogenetic movements
Compared to wild-type embryos, the epiblast of eed mutant embryos does not expand well during gastrulation, particularly with respect to the A-P axis (Table 1). Reduced axial extension was associated with a lack of incorporation of descendants of eed mutant region D into the ectoderm, particularly at the ventral midline (compare Fig. 3A,B, 2 days). In addition, although we could not detect a significant difference in the ability of descendants of eed mutant regions C and D to colonize the primitive streak (see Tables 3, 4), decreased primitive streak elongation was apparent in eed mutant embryos. Normally the streak constitutes about 30% of the total axial length at the early streak stage, and by the onset of neurulation it accounts for approximately 50% of the entire axis (Lawson et al., 1991). The elongation of the primitive streak in eed mutants varied, but never constituted more that 40% of the embryonic axis, even after 2 days in culture (data not shown).
To characterize the lack of epiblast expansion further, we compared the fate of mutant and wild-type region E. Descendants of these cells contribute to anterior expansion of the embryonic ectoderm, presumably as a result of axial extension from the insertion of descendants of region D at the distal midline (Lawson et al., 1991; Lawson and Pedersen, 1992a). In wild-type embryos after 1 day in culture, the majority of descendants contributed to the epiblast anterior to the node (Fig. 5A). After 2 days in culture descendants were found spread widely throughout the head and trunk neural ectoderm. Rarely did descendants contribute to ectoderm posterior to the node or to embryonic mesoderm (Fig. 5A; Table 6). As with regions C and D, chi-square analysis indicated that the overall distribution of descendants of eed mutant region E was significantly different from wild type. After 1 day in culture, most of the descendants were found in ectoderm posterior to the distal tip and 2/5 embryos displayed contribution to embryonic mesoderm (Fig. 5B; Table 6). The posterior localization and contribution to mesoderm were striking after 2 days in culture (Fig. 5B; Table 6). Thus, anterior expansion of the epiblast, normally illustrated by the anterior spread of region E descendants, is very limited in eed mutant embryos.
Embryo growth
In a previous clonal analysis of epiblast cell fate, it was determined that the wild-type epiblast demonstrated a homogeneous proliferation rate, dividing about every 7.5 hours (Lawson et al., 1991). Consistent with these results, the clone doubling time obtained for labelled epiblast cells of wild-type embryos in this study was 7.4±1.5 hours (n=17 embryos; see Materials and Methods for further details). Mutant epiblast cells demonstrated a clone doubling time of 7.5±1.1 hours (n=17 embryos), not a statistically significant different value from wild-type controls. Clone-doubling times from region D alone (n=5 wild-type and n=9 mutant embryos) were similar to the whole embryo values. Because of the paucity of embryos suitable for proliferation analysis from other regions of the epiblast (see Materials and Methods), we were unable to determine if there were statistically significant differences in proliferation between wild-type and eed mutant embryos within these other regions.
DISCUSSION
We have used clonal analysis to determine that loss of embryonic mesoderm in eed mutant embryos is not due to lack of its formation. Rather, embryonic mesoderm is produced, but does not migrate anteriorly and accumulates in the extraembryonic compartment over time. Both the proximodistal clonal spread of descendants in mesoderm, as well as the gain of extraembryonic mesoderm at the expense of embryonic mesoderm between 1 and 2 days in culture support this notion.
Region D, in addition to the area of the epiblast just anterior to it and distal region B, gives rise to the bulk of the definitive endoderm in the gastrulating embryo (Lawson et al., 1991). In eed mutant embryos, contribution to this tissue was lower as compared to wild-type controls. Production of definitive endoderm ensues very early in gastrulation (Tam and Beddington, 1992), and it is possible that diminished contribution to this tissue in eed mutants simply reflects a delay in the onset of gastrulation. The fact that definitive endoderm is indeed produced in eed mutant embryos suggests that at least some of the appropriate conditions for the production of this tissue from the anterior primitive streak are present.
Abnormal localization of mesoderm to the extraembryonic region did not appear to be due to a restriction and alteration of distal epiblast cell fate, since the majority of clones produced from regions fated to ingress through the anterior streak displayed descendants in both embryonic and extraembryonic derivatives. Within these mixed clones, some descendants were able to adopt their normal fates (i.e., axial mesoderm and/or definitive endoderm) after having ingressed through the anterior streak. However, it is unknown if mesodermal cells moving into the extraembryonic region have truly adopted an extraembryonic mesodermal fate. If the mislocalized mesodermal cells exiting the anterior streak maintained their normal fates as they entered the extraembryonic region, we might expect to see misexpression of genes specific to embryonic mesoderm and anterior primitive streak derivatives. However, none of the mesoderm located in the extraembryonic region expresses Lim1, a marker of embryonic mesoderm (C. F. and T. M., unpublished data; Barnes et al., 1994; Shawlot and Behringer, 1995), or markers of the anterior streak and its derivatives, such as Fgf4, Mox1, Hnf3β or Shh (Faust et al., 1995; Schumacher et al., 1996, and references therein). Thus, while the gene expression data are suggestive of an alteration in fate as mesodermal cells move extraembryonically, they do not conclusively show the adoption of an extraembryonic cell fate.
eed mutant embryos fail to display proper epiblast expansion, particularly with respect to the A-P axis. Based on patterns of clonal spread from anterior regions A and C and calculated clone doubling times for the epiblast, this does not appear to be due to decreased epiblast growth. In wild-type embryos, extension of the A-P axis is primarily due to elongation of the primitive streak and expansion of the area of the epiblast just anterior to it, including region D (Lawson et al., 1991; Lawson and Pedersen, 1992a). A substantial portion of these cells remain in the epiblast, rather than ingress through the streak; moreover, they tend to colonize the ventral midline neurectoderm, resulting in axial extension (Lawson et al., 1991; Lawson and Pedersen, 1992a,b; K. A. L., unpublished results). Although after 1 day in culture a small number of descendants of eed mutant region D remained in the ectoderm just anterior to the streak, it is notable that by 2 days, no descendants were observed in embryonic ectoderm derivatives.
These data indicate that posterior epiblast cells tend to ingress through the streak, rather than contribute to embryonic ectodermal derivatives, particularly ventral midline neurectoderm. Thus, preferential mesoderm production from region D, perhaps combined with decreased elongation of the streak, might reduce axial extension in eed mutant embryos. Consequently, descendants of region E localize posteriorly and many ingress through the streak, probably exacerbating the defect in epiblast expansion.
In summary, eed mutant embryos display complex defects in gastrulation. First, it appears that mutant epiblast cells preferentially make mesoderm rather than contribute to expansion of the epiblast. Consistent with this observation, studies with eed mutant ES cells have revealed a tendency to differentiate prematurely into mesoderm in embryoid body differentiation assays (C. F. and T. M., unpublished results). The second major defect observed in eed mutant embryos is the inability of mesoderm to migrate anteriorly and its subsequent mislocalization to the extraembryonic region. Preliminary analysis of chimeric embryos produced by injection of eed mutant ES cells into wild-type host embryos suggests that the defect in anterior migration could be intrinsic to the mesoderm (C. F. and T. M., unpublished results).
Expansion of mesoderm in the gastrulating mouse embryo appears to be due to a variety of forces. Besides propulsion away from the streak by the rapidly dividing and ingressing epiblast cells, short-term time-lapse micrography of gastrulating mouse embryos has revealed that individual mesodermal cells actively migrate away from the streak in an anterodistal direction (Nakatsuji et al., 1986). Furthermore, grafting experiments have indicated that all but the most distal-third of the mesoderm present at the mid-streak stage is eventually displaced extraembryonically and that this displacement is greatly diminished by the onset of neurulation (Parameswaran and Tam, 1995).
An important question that remains is how preferential movement of mesoderm to the extraembryonic region in eed mutant embryos is implemented. Does mesoderm actively migrate into the extraembryonic region or is it passively displaced extraembryonically as new mesoderm propels it out of the streak? While the experiments presented here do not directly assess the ability of mutant mesodermal cells to migrate, the extensive cell mixing and proximodistal clonal spread of descendants seen in many embryos suggests that mesodermal cells might be able to migrate into the extraembryonic region, even though they are inhibited in their anterior migration. Alternatively, it is possible that a decreased ability of mesoderm to migrate anteriorly, coupled with limited epiblast expansion, results in excessive displacement of mesoderm proximally in eed mutant embryos, leading to preferential production of extraembryonic mesoderm.
When grafted to ectopic sites, the distal posterior epiblast of the early-streak-stage embryo, as well as the mid-streak-stage node, exhibit organizer activity, such that they can recruit host tissues into a secondary neural axis (Beddington, 1994; Tam et al., 1997). These findings, coupled with mutational analysis and regionalized gene expression in the primitive streak, support a model wherein the anterior tip of the streak emits a combination of morphogenetic signals that play a role in determination of neurectodermal/mesodermal cell fates, as well as pattern mesodermal cells as they leave the streak (Chapman and Papaioannou, 1998; Ciruna et al., 1997; Sasaki and Hogan, 1993; Yamaguchi et al., 1994; Yoshikawa et al., 1997). Based on their proximity to these signals, mesodermal cells leaving the streak are patterned accordingly: cells exiting the anterior streak differentiate into axial and paraxial derivatives, cells exiting the more medial regions of the streak differentiate into lateral and intermediate mesoderm, and cells exiting the posterior streak differentiate into extraembryonic mesoderm. If these proposed anterior streak-derived morphogenetic signals are disrupted, patterning of mesoderm leaving the streak, as well as proper determination of neurectodermal/mesodermal cell fate, might be disrupted.
Loss of the eed gene clearly affects transcriptional activity in the primitive streak (Faust et al., 1995; Schumacher et al., 1996). Furthermore, based on the function of its Drosophila homologue as a negative regulator of gene expression, it is possible that eed negatively regulates transcriptional activity in the posterior streak. Without eed, gene expression normally restricted to the posterior streak might be constitutively expressed throughout it; as a result, signalling in the anterior streak might be disrupted. Thus far, the strongest candidate for negative regulation by the eed gene is the homeobox-containing gene Evx1 (Schumacher et al., 1996). Direct evidence that Evx1 affects mesoderm patterning in the mouse awaits the production of conditional knockouts, since embryos homozygous for a null allele of Evx1 die prior to implantation (Spyropoulos and Capecchi, 1994). However, several studies strongly suggest that other vertebrate homologues of Evx1 are involved in posterior mesodermal patterning (Ahringer, 1996; Barro et al., 1995; Ruiz i Altaba et al., 1991; Ruiz i Altaba and Melton, 1989). Thus, it is possible that high levels of Evx1 in the anterior streak of eed mutant embryos might disrupt the normal signalling that occurs here, the results of which may be two-fold: the preferential production of mesoderm at the expense of ventral midline neurectoderm and/or the inability of cells leaving the anterior streak to respond to the appropriate morphogenetic signals required for the patterning of embryonic mesoderm.
ACKNOWLEDGEMENTS
We thank Dr Alfred Malouf for invaluable technical advice, members of our laboratory for many helpful discussions, and Drs Patrick Tam and Jaime Rivera-Perez for critical reading of the manuscript. This work was supported by a grant from the NIH (#HD24462) to T. M.