Mechanisms of Development 91 (2000) 61±68

www.elsevier.com/locate/modo

The onset of germ cell migration in the mouse embryo

Robert Anderson a, Trevor K. Copeland b, Hans SchoÈler c, Janet Heasman a, Christopher Wylie d,*
a
Department of Genetics, Cell Biology, and Development, University of Minnesota Medical School, Minneapolis, MN 55455, USA
b
Molecular, Cellular, Developmental Biology & Genetics Program, University of Minnesota Medical School, Minneapolis, MN 55455, USA
c
EMBL, Gene Expression Programme, Meyerhofstrasse 1, D69012, Heidelberg, Germany
d
Department of Pediatrics, University of Minnesota Medical School, Minneapolis, MN 55455, USA
Received 27 August 1999; received in revised form 12 October 1999; accepted 14 October 1999

Abstract
Mouse primordial germ cells (PGCs) are speci®ed between embryonic day 6.5 (E6.5) and E7.5, when they have been visualized as an
alkaline phosphatase-positive (AP1) cell population in the developing allantois. By E8.5, they are embedded in the hind-gut epithelium.
Previous experiments have suggested different sites for PGCs' origin, and it is unclear how they reach the gut epithelium. We have used
transgenic mice expressing GFP under a truncated Oct4 promoter to visualize living PGCs. We ®nd GFP1/AP1 cells in the posterior end of
the primitive streak as a dispersed population of cells actively migrating into the allantois, and directly into the adjacent embryonic
endoderm. Time-lapse analysis shows these cells to be actively migratory from the time they exit the primitive streak. q 2000 Elsevier
Science Ireland Ltd. All rights reserved.
Keywords: Alkaline phosphatase; Allantois; Cell migration; Filopodia; FM4-64; Germline; Green ¯uorescent protein; Hind-gut; Mouse/murine development;
Oct4; Primitive streak; Primordial germ cell; SSEA-1; Time-lapse confocal microscopy

1. Introduction antigens cannot be used to look at earlier stages, since

The primordial germ cells (PGCs) in mice are thought to
be set aside from other lineages between E6.5 and E7.5 from
a pluripotential population of cells in the proximal epiblast
(Lawson and Hage, 1994). They arise from cells that also
give rise to the proximal allantois, and this step requires that
BMP4 be released from the extraembryonic ectoderm
(Lawson et al., 1999). Once formed, the PGCs enter the
embryonic endoderm that will give rise to the hind-gut
(Tam and Snow, 1981), where they display motile charac-
teristics both in vivo and in vitro (Clark and Eddy, 1975;
Godin et al., 1990). From E9.5±11.5, PGCs migrate from the
hind-gut to the genital ridges, and are migratory in vitro
(Clark and Eddy, 1975; Donovan et al., 1986). Immunos-
taining of the PGC cell surface molecules SSEA1 and
EpCAM during this stage shows that PGCs are extensively
in contact via dendritic processes during migration, a
complex morphology that was not anticipated by staining
for the germ cell marker alkaline phosphatase (Gomperts et
al., 1994; Anderson et al., 1999b). However, these surface
* Corresponding author. Box 206 Mayo, University of Minnesota Medi-
cal School, 420 Delaware St. SE, Minneapolis, MN 55455, USA. Tel.: 11-
612-624-3110; fax: 11-612-626-7031.
E-mail address: wylie@gene.med.umn.edu (C. Wylie)
0925-4773/00/$ - see front matter q 2000 Elsevier Science Ireland Ltd. All rights reserved.
PII: S 0925-477 3(99)00271-3
SSEA1 is not expressed on PGCs before E9.5 (Wylie et
al., 1986), and EpCAM is expressed by all epithelia (Litvi-
nov et al., 1994). Therefore, the only marker used to study
PGC origins and behavior prior to their entry into the hind-
gut has been alkaline phosphatase (Chiquoine, 1954).
Because of this, the exact origin, and route by which
PGCs enter the embryonic endoderm, have been uncertain.
Based upon alkaline phosphatase staining alone, different
studies have suggested different sites of origin for the PGCs,
including the posterior primitive streak (Chiquoine, 1954;
Snow, 1981; Copp et al., 1986), the yolk sac endoderm
(Chiquoine, 1954), and the extraembryonic allantoic meso-
derm (Ginsburg et al., 1990). In particular, a small group of
highly expressing cells, found in the allantoic diverticulum
has been reported as the initial PGC population (Ginsburg et
al., 1990). However, grafting experiments suggest that
allantoic tissue at this time cannot give rise to tissues of
the embryonic body, including the germ cells (Downs and
Harmann, 1997). Furthermore, the ¯ow of allantoic meso-
derm cells out of the posterior primitive streak outwards into
the allantois creates a logistical problem for the subsequent
movement of cells at this site into the embryonic endoderm.
In this study we have used a novel living marker of early
germ cells. We have previously described a transgenic
mouse (Oct-4D PE:GFP 1) in which green ¯uorescent
62 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68

protein (GFP) is expressed in PGCs under control of a trun- 1990; Yeom et al., 1996; Anderson et al., 1999a) in conjunc-
cated Oct-4 promoter (Yeom et al., 1996; Anderson et al., tion with a distinctive cell morphology, we conclude that the
1999a). Using transgenic embryos, we identi®ed a popula- extreme ventral-posterior GFP 1/AP 1 cells of the primitive
tion of cells that are both GFP 1 and AP 1 cells at the poster- streak and all GFP 1 cells in the de®nitive endoderm are
ior end of the primitive streak. These cells move out of the PGCs.
primitive streak, both directly into the embryonic endoderm, On embryonic day 7, visceral endoderm (AFP 1) is
and into the allantois. Using both time-lapse movies of replaced by de®nitive endoderm (AFP 2) which arises
living embryos at this stage, and the shapes and positions from the anterior primitive streak (Dziadek and Adamson,
of PGCs in ®xed tissues, we conclude that PGCs arise in the 1978; reviewed in Hogan et al., 1994). In E7.5±E8 embryos,
posterior primitive streak and move by active locomotion we found that the endoderm adjacent to the posterior primi-
directly into the subjacent endoderm, where they begin their tive streak expressed intermediate or low levels of AFP;
migration to the gonad. more proximal endoderm was clearly AFP 1, more distal
endoderm was AFP 2 (Fig. 2G and data not shown, respec-
tively). This result suggests that PGCs occupy a junctional
2. Results zone between visceral and de®nitive endoderm, and is in
agreement with ultrastructural observations made by Clark
We have previously shown that, in Oct4D PE:GFP 1 and Eddy (1975).
embryos, GFP expression is limited to the germline after
E10.5 (Anderson et al., 1999a). To test the speci®city of
this marker at earlier stages, we analyzed GFP expression
in living and ®xed Oct4D PE:GFP 1 embryos from the 1 cell
stage to E8.5. As shown in Fig. 1A,B, GFP was expressed in
all cells at the morula stage (E2.5) and then became
restricted to the inner cell mass in blastocyts (E3.5) (Fig.
1C). Since the Oct4D PE:GFP construct in these embryos is
paternally derived (see Section 4), all GFP expression is
exclusively zygotic. Embryos E6±E8 were staged by the
system of Downs and Davies (1993). In early streak/mid
streak (ES/MS, ,E6.75) embryos, GFP expression was
found throughout the epiblast and in the vicinity of the
primitive streak, but not in other tissues (Fig. 1D). In no
bud (OB, E7.25±E7.75) stage Oct-4D PE:GFP 1 embryos,
the entire epiblast and primitive streak expressed GFP,
although the level of expression in the epiblast was clearly
higher (Fig. 1E,F), even if minor differences in cell density
are considered.
The ®rst indications of GFP 1 PGCs were found in slightly
older embryos (early bud (EB) stage, Figs. 2D and 3H; late
bud (LB) stage, Fig. 2E) approximately E7.5±E8. In these
embryos, we identi®ed a group of GFP 1 cells in the poster-
ior-most region of the primitive streak. In the endodermal
epithelium immediately ventral to the posterior primitive
streak, we detected a small population of individual GFP 1
cells (Fig. 2E). The ®rst distinguishing feature of early
PGCs is the expression of AP (Chiquoine, 1954; Ginsburg
et al., 1990; MacGregor et al., 1995). Therefore, we stained
Oct4D PE:GFP 1 EB and LB embryos (,E7.5±E8) for AP Fig. 1. GFP expression in early Oct4D PE:GFP 1 embryos. (A,B) Zygotic
GFP expression initiates in all cells of an E2.5 morula. (C) In E3.5 blas-
activity. As shown in Fig. 2A±D, the posterior-most cells of
tocysts, the trophoectoderm begins to lose GFP expression, whereas the
the primitive streak were AP 1 and GFP 1. All endoderm- inner cell mass remains GFP 1. (D) In an ES/MS stage embryo (,E6.75),
embedded GFP 1 cells were also found to be AP 1. In addi- GFP is expressed by the ectoderm and in the early primitive streak. Label-
tion, endoderm-embedded GFP 1 cells were found to be ing of regions based on orientation of ®xed section. (E,F) Anatomy and GFP
negative for a -fetoprotein (AFP), a visceral endoderm expression in ,E7.25 embryo. (E) Propidium iodide staining of OB stage
(,E7.25) embryo. (F) GFP is expressed in the ectoderm and the primitive
marker, and negative for SSEA-1, a marker for PGCs
streak of the same embryo. AM, amnion; B, blastocoel; C, chorion; EC,
after E9.5 (Fig. 2G and data not shown, respectively). ectoderm; ICM, inner cell mass; PS (and arrow in E), primitive streak; T,
Based on this pro®le of markers (Chiquoine, 1954; Dziadek trophoblast. (D±F) Proximal is up; anterior is to the left. Scale bar for D 100
and Adamson, 1978; Wylie et al., 1986; Ginsburg et al., mm; scale bar for E/F in E 100 mm.
 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68 63

Fig. 2. PGCs are GFP 1 in Oct4D PE:GFP 1 E7.5 embryos. (A±C,D) Two ventral views of the posterior primitive streak of a ®xed EB stage (,E7.5) embryo
stained for AP (red), GFP is in green. (A±C) AP staining is localized to the membrane and GFP expression is cytoplasmic, as expected. Arrow points to the
same cell in A±C, illustrating this GFP/AP co-expression. (D) Note that the posterior-most GFP 1 cells are also AP 1 (arrows) and appear to be leaving the
primitive streak. The allantois is behind the plane of focus of this image and also demonstrates GFP/AP co-expression. (E) A live LB stage (,E7.5±E8) embryo
viewed laterally in the vicinity of the posterior primitive streak. Distinct GFP 1 cells (arrows) are found interspersed within the GFP 2 endoderm (asterisks). The
ectoderm is out of focus in this image. (F±I) A lateral view of an EHF stage (,E7.5±E8) embryo expressing GFP and stained for laminin (F), AFP (G), and
propidium iodide (H). (F) There is a discontinuous basement membrane between the endoderm (asterisks) and the terminal primitive streak. Note the PGC
(arrow) embedded within the endoderm proximal to the level of the allantois. (G) A PGC (arrow) is near to the AFP 1 visceral endoderm (dotted arrow) and
AFP - de®nitive endoderm (arrowhead). (H/I) The endoderm (asterisks) contains two isolated GFP 1 cells (presumed PGCs) (arrows). (I) There is a mass of
GFP 1 cells in the terminal primitive streak adjacent to the endoderm (arrowhead), and the allantois has a variegated appearance due to a mixed population of
GFP 1 and GFP 2 cells. The junction between the posterior primitive streak and the extraembryonic mesoderm is indicated by the dotted line. (J,K) Schematic
diagram of E7.5±E8 embryo. (K) Detail of schematic showing region pictured in panels F±I. Dotted line indicates plane shown in panels A±D. AM, amnion;
AL, allantois; EC, ectoderm; PS, primitive streak; SQ, squamous endoderm. (A±D) Posterior is up; right is to the right. (E±I) Posterior is up; dorsal is to the left.
Scale bar for A±C in A, H/I in I, all scale bars: 10 mm.

To determine the source of the endoderm-embedded ventral movement of GFP 1 cells from the posterior primi-
GFP 1 cells, we used time-lapse confocal microscopy to tive streak into the underlying endoderm (Fig. 3A±E,H±I).
®lm the posterior region of an EB embryo over a 160-min The route of this initial movement of GFP 1 cells is demon-
period. Using this technique, we documented the dorsal-to- strated by comparison of the details in Fig. 3 with the posi-
64 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68
Fig. 3. PGCs undergo active migration directly from the posterior primitive
streak into the endodermal epithelium. (A±C) Time-lapse cinematography
of an Oct4D PE:GFP 1 EHF stage (E7.5-E8) embryo in the vicinity of the
posterior primitive streak, ventral view; three Z-sections at 4 mm apart
projected for clarity and to show movement is both lateral and into the
plane of the page. (C) After 160 min, a PGC (arrow) has emerged, moving
from the posterior primitive streak into the GFP 2 endoderm. (D±F) Three-
dimensional reconstruction of same time points as A±C from side perspec-
tive: A±C rotated clockwise 908 out of the page to show separation of the
PGC from the PS along the dorso-ventral axis. (G) Schematic diagram of
E7.5±E8 embryo posterior region. Dotted line represents plane shown in A±
C, square represents plane shown in D±F. Asterisks are in the endoderm;
(H±I) live, single Z-section from Oct4D PE:GFP 1 EB (E7.5±E8) embryo.
(H,I) Red is FM4-64 (Molecular Probes) stain of all cell membranes includ-
ing those in the GFP 2 endoderm. (I) Green is GFP 1 PGCs at the edge of the
PS and in the endoderm; AL, allantois; AM, amnion; A, anterior; D, dorsal;
EC, ectoderm; EN, endoderm; L, left; P, posterior; PS, primitive streak; R,
right; V, ventral. The scale bar for A±F in C; the scale bar for H,I in I; all
scale bars: 10 mm.
tions of GFP 1 PGCs in the z-series shown in Fig. 4, in which
panel F is in the plane of the endoderm, panel E is imme-
diately ventral to the posterior primitive streak, and panel D
is through the posterior end of the primitive streak. We
interpret these data to indicate that GFP 1 cells have passed
directly from the posterior primitive streak into the endo-
derm.
It was previously impossible to identify E7.5 PGCs in the
un®xed/unstained embryo, and as a result, little is known of
their phenotype. Therefore, we examined the morphology
and behavior of live EB stage and LB stage PGCs (,E7.5±
E8) by confocal microscopy. These early PGCs displayed a
motile phenotype, with some projecting ®lopodia up to
several cell diameters in length (Fig. 4B). Time-lapse analy-
sis allowed us to capture the morphological changes as an
E7.5 PGC extends a ®lopodium (Fig. 4G±J). This is char-
acteristic of most E7.5 PGCs which were typically highly
polarized and teardrop shaped, very similar to the reported
shapes of E10.5 PGCs and motile Drosophila PGCs
(Gomperts et al., 1994; Jaglarz and Howard, 1995). In
contrast, although migratory, GFP 1 mesodermal cells of
the primitive streak were completely devoid of long ®lopo-
dia and were typically trapezoidal or rectangular in shape
(not shown). A further difference was that PGCs appeared to
disperse and migrate as individuals, whereas mesodermal
cells migrated en masse. In EB stage embryos, PGCs were
found in a relatively discreet cluster in the posterior primi-
tive streak (Fig. 2D); in LB-EHF stage embryos, PGCs were
in a crescent-like distribution: ventral, lateral and posterior
to the primitive streak (Fig. 4B,C).
At slightly later stages (E8, two to four somites), most
PGCs had translocated posteriorly and/or laterally from the
primitive streak, and were located in the endoderm of the
invaginating hind-gut (Fig. 5A,B). Due to the dynamic
changes in the morphology we could not track in detail
the path of PGCs in the invaginating hindgut. However,
we did observe in older E8.5±E9 embryos with 12±14
somites, that PGCs were located in a broad area of hind-
gut endoderm deep inside the caudal region of the embryo
(Fig. 5C). PGCs of this stage had the appearance of highly
motile cells (Fig. 5D).
The allantoides of LB stage (Fig. 6A) and stage EHF (Fig.
2I) embryos contained a population of GFP 1 cells mixed
with GFP 2 cells, giving the allantoides a variegated appear-
ance. In agreement with previous reports (Ginsburg et al.,
1990; MacGregor et al., 1995), we noted that the cells of the
proximal allantois also expressed AP (Fig. 2G). Unlike the
population of PGCs that migrate directly into the endoderm,
the GFP 1 allantois cells were spherical in shape and
displayed no ®lopodia or lamellipodia (Fig. 6A). Further,
GFP 1 cells in the allantois were frequently closely adjacent
to each other (Figs. 2H,I and 6A), in sharp contrast to early
PGCs which were scattered (compare with Fig. 4C). In E8.5
embryos, GFP 1 cells of the allantois were frequently found
clustered together in tight aggregates (Fig. 6B), whereas the
PGCs were widely dispersed in the hind-gut (Fig. 5C).
3. Discussion
Previous studies have focused on the location of PGCs in
E7 and E8 embryos using AP as a marker (Tam and Snow,
1981; Ginsburg et al., 1990). Here we show that in Oct4D
PE:GFP 1 embryos, GFP can be used to identify PGCs, and
is a de®nitive marker for PGCs as soon as they enter the
endodermal epithelium of the future hind-gut. We have
taken advantage of the fact that GFP expression can readily
be detected in vivo in order to examine the source and
behavior of the earliest PGCs.
Ginsburg et al. (1990) identi®ed a cluster of AP 1 cells in
the extraembryonic mesoderm at the base of the allantois
(above the level of the amnion) in ,E7.5 embryos (approxi-
 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68 65

Fig. 4. PGCs appear to migrate actively out of the posterior primitive streak. (A) Schematic detail of E7.5±E8 embryo. Letters labeling dotted lines correspond
to D±F. (B) Projection of serial 2 mm Z-sections over a total of 40 mm. Note the long ®lopodia on several PGCs (arrows). (C) Low power ventral view of a
living Oct4D PE:GFP 1 LB stage (E7.5±E8) embryo showing the posterior primitive streak (GFP 1) surrounded by endoderm (GFP ±). Note the dispersed PGCs
(GFP 1) in the endoderm. (D±F) Higher power optical sections of the embryo shown in (C). The plane of (E) is 20 mm ventral to (D); the plane of (F) is 40 mm
ventral to (D). (G±J) GFP 1 PGC extends ®lopodia. Projection of 12 mm serial Z-sections at 4 mm intervals shows GFP 1 cell in PS extending ®lopodia (arrows).
AL, allantois; AM, amnion; EC, ectoderm; Asterisk (*), endoderm. Axis key for A±C,E,F in C. Scale bar for B,D±F in B; G±J in J, all scale bars: 10 mm.

mately stage EB or OB) as PGCs. From this area, they
suggested that the AP 1 PGCs moved into the adjacent endo-
derm during slightly later stages (approximately OB or LB
stage). In this study, we identi®ed a GFP 1/AP 1 cluster of
cells in the ventral/posterior primitive streak of Oct4D
PE:GFP 1 stage EB embryos. Further, we demonstrated
that these cells migrated outward from the ventral/posterior
primitive streak into the adjacent allantois as well as both
extraembryonic and embryonic endoderm (Figs. 2I, 3 and
6). Based on this evidence, we propose that PGCs that will
eventually colonize the genital ridges are the ones that
migrate directly into the endodermal epithelium. This
conclusion is consistent with a recent report by Downs
and Harmann (1997), who found that, in grafting experi-
ments, the allantois does not contribute to any structures
of the embryo proper. Our results are also consistent with
an observation by Chiquoine (1954), who described an ante-
rior-to-posterior gradient of increasing AP activity in the
posterior primitive streak (see Fig. 2A,C,D,G). We interpret
this correlation of AP activity, GFP expression and a
distinctive morphology in the ventral-posterior-most cells
of the primitive streak as indicating that they are PGCs.
It has been known for some time that E8.5±E11.5 PGCs
migrate extensively in vivo, including movement within the
hind-gut endoderm, splanchnic mesoderm of the gut,
mesentery, and urogenital ridges (Clark and Eddy, 1975).
However, the mechanism by which E7.5 PGCs enter the
epithelium of the future hind-gut was not known, because
66 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68

extraembryonic endoderm immediately posterior to the

embryonic endoderm. Since ®xation and staining for AFP
is required to identify the embryonic/extraembryonic endo-
derm boundary, it has not been possible to follow the fates
of these in living embryos.
We identi®ed a large number of GFP 1/AP 1 cells in the
proximal allantoides of stage EB (E7.5±E8) and older
Oct4D PE:GFP 1 embryos. It is possible that these cells
were extraembryonic PGCs (or PGC precursors) that
would have eventually reached the endodermal epithelium,
as has been previously proposed (Ginsburg et al., 1990).
However, cell movement in the allantois is proximal to
distal (Downs and Harmann, 1997), and extraembryonic
PGCs in the base of the allantois would need to migrate
against this morphogenetic movement in order to reach
the endodermal epithelium. In addition, we know that

Fig. 5. PGCs in E8 and E8.5 embryos, ventral view. (A,B) PGCs (arrow-
heads) are located in the invaginating hind-gut of living Oct4D PE:GFP 1
E8 (three somite) embryos. (A) Filming of this embryo (not shown)
revealed that PGCs were swept into the forming hind-gut diverticulum
(gross motion indicated by the curved arrow). (B) In this plane, PGCs
did not appear highly motile. However, several PGCs (arrows) were slightly
angular. (C) PGCs are dispersed in the hind-gut of a living 14 somite
Oct4D PE:GFP 1 (E8.5±E9) embryo. The endoderm, mesoderm, and tail-
bud ectoderm can be distinguished by their relative levels of GFP expres-
sion. (D) Higher power image of the embryo shown in (C). PGCs at this
stage show obvious signs of motility, including the extension of ®lopodia
and lamellipodia (arrows). EN, hind-gut endoderm; M, mesoderm; PS,
primitive streak; T, tailbud ectoderm. (A±D) Posterior of embryo is up.
Scale bars, A,C: 100 mm; scale bars, B,D: 10 mm.

of an absence of suitable markers. Using Oct4D PE:GFP 1

embryos, we found that E7.5 PGCs were highly motile.
Indeed, E7.5 PGCs were morphologically indistinguishable
from migratory PGCs at E10.5 (Gomperts et al., 1994).
Although we could not stain for AP activity in living
embryos, our analysis of ®xed tissue suggests that nascent
PGCs acquire their distinct migratory phenotype and AP
activity at about the same time (see Fig. 2A±D).
Our observations of E8 PGCs are in agreement with
previous reports, where E8 PGCs were described as round
cells with occasional small pseudopodial projections (Spie-
gelman and Bennett, 1973; Clark and Eddy, 1975). In early
E8 (two to four somite) embryos, we found that PGCs had
fewer features of motile cells than at slightly earlier or later
stages (see Fig. 5B). However, the invagination of the hind-
gut at this stage signi®cantly changes the orientation of the
endoderm, making it extremely dif®cult to visualize the
morphology of PGCs. In older E8.5±E9 embryos the Fig. 6. GFP 1 cells in the allantoides of Oct4D PE:GFP 1 E7.5 and E8.5
embryos. (A) Projection of serial 2 mm Z-sections of a LB stage (E7.5±E8)
PGCs had the appearance of motile cells, which is in agree-
allantois over a total of 40 mm. This allantois is from the embryo shown in
ment with their behavior in culture and previously described Fig. 2D. (B) GFP 1 cells in the allantois/connecting stalk of an E8.5 (14
morphology. somite) embryo. This allantois from the embryo shown in Fig. 5C/D. (A,B)
We also found a small number of GFP 1/AP 1 cells in the Distal allantois is up. Scale bars: 10 mm.
 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68
Fig. 7. A model of PGC formation and behavior. (A) PGCs arise in mouse
embryos on the 7th day of development, (B) in the region of the posterior
primitive streak. (1) PGCs and cells of the proximal allantois arise from a
common precursor (Lawson and Hage, 1994). (2) In the ventral posterior
primitive streak, PGCs differentiate and migrate actively into the endoder-
mal epithelium. (3) Cells of the proximal allantois remain outside of
embryo proper, and do not contribute to the germline. Gray, ectoderm;
green, mesoderm; red, extraembryonic mesoderm; dark purple, de®nitive
endoderm; light purple, visceral endoderm.
many GFP 1/AP 1 cells remain in the allantois, since we see
them there on E8.5, when they are aggregated together and
appear non-motile. We assume that GFP 1 cells in the allan-
toides of Oct4D PE:GFP 1 E8.5 embryos are descendants of
GFP 1 cells found there a day earlier at E7.5.An alternative
route to the genital ridges exists for these cells, since the
allantois is later incorporated into the umbilicus. However,
the allantoic grafting experiments of Downs and Harmann
(1997) included portions of the allantois corresponding to
those we found to contain large numbers of GFP 1 cells (Fig.
6A), and their results indicated that none of the allantoic
derivatives are incorporated into the embryo proper.
It is well established that the PGCs are only loosely direc-
ted toward the somatic gonads during development, and that
many of them become ectopic (Bendel-Stenzel et al., 1998).
This is seen most obviously when PGCs migrate from the
gut to the genital ridges, during which many PGCs die in
ectopic locations. From the data presented here, it seems
likely that this principle also applies earlier. `Potential
PGCs' (de®ned by their expression of GFP and AP) migrate
ventrally and posteriorly out of the posterior primitive streak
into surrounding structures, including the allantois, as well
as the underlying extraemabryonic and embryonic endo-
derm. Those that enter the endoderm become incorporated
into the hind gut and migrate to the gonad, whilst those out
of range of the endoderm that becomes the hindgut remain
in ectopic locations and eventually die there. We do not
know the `range' here, because we cannot track germ
cells during the E7.5±E8.5 period due to morphogenetic
movements. So some of the potential PGCs in the proximal
allantois and extraembryonic endoderm may be within
range. However, we ®nd many non-migratory cells the
next day still in the allantois, and so assume these were
out of range.
In summary, we present a simple model of the onset of
67
PGC migration, in which cells that can enter the germ line
emerge as a dispersed population of cells migrating out of
the posterior primitive streak into adjacent tissue, which
includes a direct route into the embryonic endoderm
where they become incorporated into the hind gut (Fig. 7).
We would therefore de®ne the posterior primitive streak as
the site of origin of their migration in the mouse embryo.
4. Experimental procedures
4.1. Mice
Establishment of Oct4D PE:GFP 1/2 mice has been
described previously (Anderson et al., 1999a). All embryos
analyzed were (50% CD1, 50% FVB)-Oct4D PE:GFP/0
(Charles River, JAX Mice). This hemizygous genotype
was generated by mating FVB-Oct4D PE:GFP 1/1 males
to CD1 females and is referred to throughout this work by
reference to the expression of GFP under the control of
Oct4D PE (as Oct4D PE:GFP 1). Conception was estimated
to occur at midnight prior to the morning of the vaginal
plug. The approximate stages of E6±E8 mouse embryos
was determined by the system of Downs and Davies
(1993). All experiments involving mice were carried out
under the guidelines of the University of Minnesota's
Research Animal Resources department, which conform
to all pertinent government regulations.
4.2. Antibodies
Polyclonal rabbit anti-mouse laminin-1 was used at a
dilution of 1:500 (Sigma). Polyclonal rabbit anti-mouse
AFP, a kind gift from Eileen Adamson, was used at a dilu-
tion of 1:100. Cy-3-conjugated anti-rabbit Ig was purchased
from Jackson Immunoresearch, and was used at a dilution of
1:400.
4.3. Histology and wholemount embryo analysis
For paraf®n sections, embryos were ®xed in 4% parafor-
maldehyde (PF) for 30 min, dehydrated in a series of etha-
nol, cleared in xylene, and embedded in paraplast (Sigma).
Sections (10 mm) were cut, mounted on silanized slides
(Sigma), and dried overnight at 408C. Next, slides were
dewaxed in xylene, rehydrated in a graded ethanol series,
washed brie¯y in phosphate-buffered saline (PBS), and
blocked with 10% goat serum 1 0.02% sodium azide in
PBS (PSA).
For immuno¯uorescent staining of paraf®n sections, the
primary antibody was diluted in PSA, and slides were incu-
bated with this solution overnight at room temperature in a
humidi®ed chamber. Propidium iodide (Sigma) was used at
a ®nal concentration of 5 mg/ml. After several washes in
PBS, the secondary antibody (diluted in PSA) was incubated
1±4 h at room temperature in a humidi®ed chamber. After
several washes in PBS, slides were mounted in an aqueous
68 R. Anderson et al. / Mechanisms of Development 91 (2000) 61±68
mounting solution (90% glycerol, 10% H2O with 100 mg/ml
DABCO) (Sigma). Specimens were visualized using a 1024
laser confocal microscope (BioRad).
For wholemount analysis with GFP as a PGC marker, live
Oct4D PE:GFP 1/2 tissues were immersed in M2 media
(Sigma), placed in cavity slides and directly visualized by
confocal microscopy. For time-lapse analysis, tissues were
incubated at 378C in an environmental chamber throughout
the duration of the experiment. To visualize endoderm in the
living tissue, we stained the tissue with the vital cell
membrane dye FM4-64 (Molecular Probes) at 25 mM in
M2 media for 15 minutes before transferring to fresh M2
on a cavity slide for time-lapse analysis. For whole mount
AP staining, Oct4D PE:GFP 1/2 embryos were ®xed in 4%
paraformaldehyde for 5 min, washed extensively in PBS,
and stained with Fast Red as reported previously (Garcia-
Castro et al., 1997). After extensive washing in PBS, AP-
stained tissues were directly visualized by confocal micro-
scopy. The Fast Red precipitate was readily detected using
Cy-3/rhodamine ®lter set. In all imaging (of both ®xed and
living tissue), GFP was present in suf®cient quantity to be
visualized by its own ¯uorescence with the appropriate ®lter
set. Processing of images was done with Adobe Illustrator
and Adobe Photoshop, NIH Image (http://rsb.info.nih.gov/
nih-image), and Confocal Assistant (ftp://ftp.genetics.bio-
rad.com/Public/confocal/cas/).
Acknowledgements
We are very grateful to Eileen Adamson for the gift of the
anti-AFP antibody, Young Il Yeom and Karin HuÈbner for
generating the plasmid pOct-4DPE:GFP, Electra Coucouva-
nis and Michael Bendel-Stenzel for helpful discussions, and
Kyle Schaible for technical assistance. Financial support for
this work came from the National Life and Health Insurance
Medical Research Fund, the Harrison Fund, the Institute of
Human Genetics, and the National Institutes of Health
(HD33440-01).
References
Anderson, R., FaÈssler, R., Georges-Labouesse, E., Hynes, R.O., Bader,
B.L., Kreidberg, J.A., Schaible, K., Heasman, J., Wylie, C., 1999a.
Mouse primordial germ cells lacking beta1 integrins enter the germline
but fail to migrate normally to the gonads. Development 126, 1655±
1664.
Anderson, R., Schaible, K., Heasman, J., Wylie, C., 1999b. Expression of
the homophilic adhesion molecule Ep-CAM in the mammalian germ-
line. J. Reprod. Fertil. 116, 379±384.
Bendel-Stenzel, M., Anderson, R., Heasman, J., Wylie, C., 1998. The origin
and migration of primordial germ cells in the mouse. Semin. Cell Dev.
Biol. 9, 393±400.
Chiquoine, A., 1954. The identi®cation, origin and migration of the primor-
dial germ cells of the mouse embryo. Anat. Rec. 118, 135±146.
Clark, J.M., Eddy, E.M., 1975. Fine structural observations on the origin
and associations of primordial germ cells of the mouse. Dev. Biol. 47,
136±155.
Copp, A.J., Roberts, H.M., Polani, P.E., 1986. Chimaerism of primordial
germ cells in the early post-implantation mouse embryo following
microsurgical grafting of posterior primitive streak cells in vitro. J.
Embryol. Exp. Morphol. 95, 95±115.
Donovan, P., Stott, D., Cairns, L., Heasman, J., Wylie, C., 1986. Migratory
and post-migratory mouse primordial germ cells behave differently in
culture. Cell 44, 831±838.
Downs, K.M., Davies, T., 1993. Staging of gastrulating mouse embryos by
morphological landmarks in the dissecting microscope. Development
118, 1255±1266.
Downs, K.M., Harmann, C., 1997. Developmental potency of the murine
allantois. Development 124, 2769±2780.
Dziadek, M., Adamson, E., 1978. Localization and synthesis of alpha-
foetoprotein in post-implantation mouse embryos. J. Embryol. Exp.
Morphol. 43, 289±313.
Garcia-Castro, M., Anderson, R., Heasman, J., Wylie, C., 1997. Interac-
tions between germ cells and extracellular matrix glycoproteins during
migration and gonad assembly in the mouse embryo. J. Cell Biol. 138,
471±480.
Ginsburg, M., Snow, M., McLaren, A., 1990. Primordial germ cells in the
mouse during gastrulation. Development 110, 521±528.
Godin, I., Wylie, C., Heasman, J., 1990. Genital ridges exert long-range
effects on mouse primordial germ cell numbers and direction of migra-
tion in culture. Development 108, 357±363.
Gomperts, M., Garcia-Castro, M., Wylie, C., Heasman, J., 1994. Interac-
tions between primordial germ cells play a role in their migration in
mouse embryos. Development 120, 135±141.
Hogan, B., Beddington, R., Constantini, F., Lacy, E., 1994. Manipulating
the Mouse Embryo, 2nd ed. Cold Spring Harbor Press, Cold Spring
Harbor.
Jaglarz, M., Howard, K., 1995. The active migration of Drosophila primor-
dial germ cells. Development 121, 3495±3503.
Lawson, K.A., Hage, W.J., 1994. Germline development. In: Goode, J.
(Ed.). CIBA Foundation Symposium, vol. 182, Wiley, Chichester, pp.
68±91.
Lawson, K.A., Dunn, N.R., Roelen, B.A., Zeinstra, L.M., Davis, A.M.,
Wright, C.V., Korving, J.P., Hogan, B.L., 1999. Bmp4 is required for
the generation of primordial germ cells in the mouse embryo. Genes
Dev. 13, 424±436.
Litvinov, S.V., Bakker, H.A., Gourevitch, M.M., Velders, M.P., Warnaar,
S.O., 1994. Evidence for a role of the epithelial glycoprotein 40 (Ep-
CAM) in epithelial cell-cell adhesion. Cell Adhes Commun. 2, 417±
428.
MacGregor, G.R., Zambrowicz, B.P., Soriano, P., 1995. Tissue non-speci®c
alkaline phosphatase is expressed in both embryonic and extraembryo-
nic lineages during mouse embryogenesis but is not required for migra-
tion of primordial germ cells. Development 121, 1487±1496.
Snow, M.H., 1981. Autonomous development of parts isolated from primi-
tive-streak-stage mouse embryos. Is development clonal? J. Embryol.
Exp. Morphol. 65, 269±287.
Spiegelman, M., Bennett, D., 1973. A light- and electron-microscopic study
of primordial germ cells in the early mouse embryo. J. Embryol. Exp.
Morphol. 30, 97±118.
Tam, P., Snow, M., 1981. Proliferation and migration of primordial germ
cells during compensatory growth in mouse embryos. J. Embryol. Exp.
Morphol. 64, 133±147.
Wylie, C., Stott, D., Donovan, P., 1986. Primordial germ cell migration. In:
Browder, L. (Ed.). The Cellular Basis of Morphogenesis, Plenum Press,
New York, pp. 433±450.
Yeom, Y., Fuhrmann, G., Ovitt, C., Brehm, A., Ohbo, K., Gross, M.,
HuÈbner, K., SchoÈler, H., 1996. Germline regulatory element of Oct-4
speci®c for the totipotent cycle of embryonal cells. Development 122,
881±894.