A role for intermediate filaments in the establishment of the primitive epithelia during mammalian embryogenesis

ELSEVIER

Mechanisms of Development 53 (1995) 305-321

A role for intermediate filaments in the establishment of the primitive epithelia during mammalian embryogenesis Stella M. Schwarza, G. Ian Gallicanob, Robert W. McGaugheyc, David G. Capcoc3* aBecker College, 955 Main Street, Leicester, MA, USA bVniversity of Chicago, Howard Hughes Medical Institute, 5841 S. Maryland Avenue, Room N-314, Chicago, IL 60637, USA ‘Molecular and Cellular Biology Program/zoology, Arizona State University, Tempe. AZ, 85287-1501. USA

Received 14 July 1995; accepted25 July 1995

Abstract

Investigations of the cytoskeleton in mammalian eggs and embryos have revealed the existence of an unusual array of crosslinked intermediate filaments composed of cytokeratins 5, 6, 16, and ‘Z’ that are referred to as cytoskeletal sheets. We have been investigating the function of these cytoskeletal sheets during embryogenesis. In this investigation we report the rapid appearance of extensive arrays of tonofilaments extending across blastomeres and in association with intercellular desmosomal junctions appearing at the time the embryo hatches from its zona pellucida, through the time of implantation of the embryo into the uterine wall. Just prior to the time of gastrulation these tonofilaments disappear. Electron microscopy and immunoconfocal microscopy demonstrate that the tonofilaments are composed of cytokeratins characteristic of the type found earlier in development, that is types 5 and 6; whereas, cytokeratin type 8 which has been shown to be synthesized in blastocysts is localized primarily at perinuclear regions. Cytokeratins 8 and 18 are synthesized to about the same extent as actin at the time the tonofilaments appear whereas the synthesis of cytokeratins 5 an 6 is greatly reduced. Our results suggest that cytokeratins 5 and 6 in the tonofilaments may arise from the stored form of cytokeratins in the cytoskeletal sheets. Consequently, our results suggest that the sheets may serve as a maternal reserve of cytokeratin employed by the embryo at the time of implantation to form extensive arrays of tonofilaments in the embryo that likely provide structural integrity to the embryo as it is subjected to mechanical stress during invasion and implantation into the uterine wall. Keywords:

Intermediate filaments; Cytoskeletal sheets; Cytokeratin; Embryogenesis

1. Introduction The cytoskeleton provides structure to cells and mediates many cellular functions (Koonce and Schliwa, 1986; Vale et al., 1986; Schroeder, 1990; Theriot and Mitchison, 1991; Navara et al., 1994). In the typical cell, three distinct filament systems exist, microtubules, actin filaments, and intermediate filaments, and together with numerous’ filament system-associated proteins, they comprise the cytoskeleton. Several years ago we proposed that the cytoskeleton within oocytes, eggs, and embryos would exhibit cytoskeletal specializations to accommodate the developmental challenges facing these special cells. Moreover, we predicted that unique cytoskeletal elements might exist to mediate some of the major devel* Corresponding author. Tel.: +l 602 9657100; Fax: +l 602 9652519; E-mail: [email protected].

opmental challenges faced by these special cells. In support of our assertions, several investigators have reported special functions for the cytoskeleton during development (Jeffery and Meier, 1983; Shimizu, 1984, 1988; Jeffery, 1985; Klymkowsky et al., 1987; Hill and Strome, 1988; Navara et al., 1994; Schatten, 1994). Further, we identified an unusual cytoskeletal element in mammalian eggs and embryos which appeared to be involved in each of the major developmental transitions of mammalian embryogenesis (Capco and McGaughey, 1986; Gallicano et al., 1991, 1994a, 1995; Capco et al., 1993). These unique elements, which we refer to as sheets, are composed of a highly crosslinked array of assembled intermediate filaments coated with particulate components (McGaughey and Capco, 1989; Gallicano et al., 1991, 1994a, 1995; Capco et al., 1993). We speculate that these sheets appear only in mammalian eggs and embryos and not eggs and embryos of other organisms because they participate in

0925-4773/95/$09.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0925r4773(95)00440-C

306

S.M. Schwarz et al. I Mechanisms

the developmental transitions which exist uniquely during mammalian embryogenesis, that is embryonic compaction, blastocyst formation, and formation of the primitive ectoderm and primitive endoderm of the egg cylinder (Snell and Stevens, 1966; Tam and Beddington, 1987, 1992; Tam, 1989). The unique developmental transitions noted above are specializations which lead to placental development, a process of embryogenesis in all eutherian mammalian embryos that allows the embryo to acquire nutrients from the maternal organism (Renfree, 1983; Fisher et al., 1989; Wallace et al., 1991; Tam et al., 1993). In forming the placenta, the trophectoderm develops as the outer layer of cells of the early embryo and differentiates into two cell layers each with specific organization, the cytotrophoblast as an inner cell layer and the syncitiotrophoblast as the outermost layer (Ramsey et al., 1976; Tuttle et al., 1985; Fisher et al., 1989; Ohlsson, 1989). The remainder of the embryo, in contrast, develops from the inner cell mass cells that subsequently form the primitive ectoderm and primitive endoderm, which together form the egg cylinder. The placental cells invade, and drag the embryo into the uterine wall, a process termed implantation. We have shown that the sheets undergo major changes in structural organization at all developmental transitions up to the blastocyst stage and in this report we show another major change at the time the embryo implants into the uterine wall. Elucidation of the function of the sheets required both in vitro and in vivo biochemical and ultrastructural analyses (Capco and McGaughey, 1986; McGaughey and Capco, 1989; Gallicano et al., 1991, 1992, 1995; Capco et al., 1993). We show here that promptly after the sheets disassemble at the blastocyst stage, tonofilaments form first in trophectodermal cells, the cells in which sheets first disassemble. An extensive number of tonofilaments appear throughout all cell types hours before the embryo begins to invade the uterine wall followed by the disappearance of the tonofilaments just as rapidly as they arose immediately before gastrulation. Because of the rapid appearance of large numbers of tonofilaments, we predicted that they would be formed from intermediate filaments of the sheets (i.e., cytokeratins 5, 6, 16 and ‘Z’; Gallicano et al., 1994). As corollaries to this prediction, we anticipated that there would be relatively little new synthesis of these sheet-derived cytokeratins and that cytokeratins 8 and 18 which are synthesized extensively by the blastocyst stage (Jackson et al., 1980; Oshima et al.,

of Developmenf 53 (1995) 305-321 1983) would not contribute to the tonofilaments. In this report we provide data to support all of the predications noted above. Our results indicate that the sheets serve as a maternal store of preassembled intermediate filaments that are used at least at the time of implantation to contribute to the formation of tonofilaments. Our results suggest that the tonofilaments form to provide structural integrity to the developing embryo as it faces mechanical stress during its implantation into the uterine wall. Our results also suggest that cytokeratin 15 may be another sheet protein. 2. Results 2.1. Culture of embryos In order to assess the developmental fate of the cytoskeletal sheets and to determine if any other cytoskeletal structures derived from the sheets arose during postblastocyst development it was necessary to employ in vitro conditions that allowed development beyond the blastocyst stage. Culture was essential because it is virtually impossible to detect and dissect embryos from the wall of the uterine horn at early stages of implantation. Fig. 1A demonstrates the appearance of 11 blastocysts flushed from uterine horns, while Fig. 1B illustrates a population of blastocysts 12 h after culture, in which two were undergoing release from the zona pellucida (i.e. hatching), and two which had already hatched. Twelve hours after these embryos hatch, they anchor to the culture plate at approximately the same time as implantation would occur in vivo (Fig. 1C). As attachment of the embryo to the culture plate continues, the trophectodermal cells begin to spread out on to the culture plastic from the point of blastocyst attachment (Fig. 1D). While virtually all blastocysts anchor on to the culture plastic, only those that anchor with the ICM located near the culture plastic develop normally (Wu et. al., 1981). As shown in Fig. lE, the ICM eventually formed an egg cylinder, which then elongated. If the ICM is near the surface of the culture plastic, elongation will take place laterally or upwards in the culture plate and elongation is not impeded. However, if the ICM resides near the top of the embryo, away from the culture plate at the time of anchorage, then as the egg cylinder elongates it encounters the culture plastic and the shape of the egg cylinder and embryo proper becomes distorted, causing a premature death of the embryo. About (25%) of the embryos anchored with their ICM near the culture plate resulting in normal development.

Fig. I. Mouse embryos were viewed in culture a: varying time intervals. (A) Eleven blastocysts flushed from the uterus. The arrowheads indicate ICM cells and the arrows indicate the trophectoderm cells. The fluid-filled blastocoelic cavity is indicated by a ‘C’. (B). Blastocysts cultured for 12-24 h. The bottom two embryos have hatched (arrow) from their zona pellucida (arrowheads) while the top two blastocysts are midway in the process of hatching. (C) An embryo that has hatched and just attached to the culture plate. (D) An embryo that has been attached to the culture plate for 12-24 h. The arrows point to trophectoderm cells surrounding the embryo and spreading on the culture plate. (E) An embryo approximately 72 h after attaching to the culture plate. The embryo in the center (arrow) has attached with ICM near the culture plate, whereas the embryo on the left (arrowheads) has not and is severely distorted.

S.M. Schwarz et al. I Mechanisms of Development 53 (1995) 305-321

307

308

S.M. Schwarz et al. I Mechanisms of Development 53 (1995) 305-321

Fig. 2. Thick sections of toluidine blue stained embryos at various stages viewed using light microscopy. (A) A blastocyst stage embryo. Note the ICM cells in the process of mitosis (arrow). (B) A blastocyst in the process of hatching. The zona pellucida on the left is indicated by a ‘2’ and the traphectoderm cells ‘T’ have already hatched (arrows). The arrowheads indicate ICM cells still inside the zona pellucida. (C) An embryo 12-24 h after attachment to the culture plate. Note egg cylinder (arrows) which has begun to form. (D) An embryo 72 h after attachment to the culture plate. ECT indicates the primitive ectoderm and arrowheads indicate the primitive endoderm. Arrows indicate the trophectoderm cells. (E) A vertical cross-section of an 84-100 h embryo. Note the extraembryonic endoderrn cells (arrowheads) surrounding the primitive ectoderm (arrows). (F) A longitudinal section of an embryo at the beginning of gastrulation. Arrowheads represent the endoderm cells, and arrows represent the ectoderm cells. The embryo is distorted by the ectoplacental cone ‘EPC’ region.

SM. Schwarz et al. I Mechanisms of Development 53 (I 995) 305-321

Plastic embedded sections viewed in the light microscope after staining with toluidine blue demonstrated normal development in the culture conditions (Fig. 2AF). In Fig. 2A-D the plane of the section runs parallel to the long axis of the developing egg cylinder, thus showing a representation of cells both near the interior and near the exterior of the forming egg cylinder. Fig. 2A represents a normal blastocyst prior to culture, while Fig. 2B illustrates the hatching blastocyst during culture. Fig. 2C shows a section through an embryo 24 h after hatching and Fig. 2D illustrates an embryo 72 h post-hatching. Note that at this time the cells within the egg cylinder begin to differ from each other forming the primitive endoderm and the primitive ectoderm (Gardner and Rossant, 1979; Stern, 1992; Tam and Beddington, 1992). The point of anchorage to the culture plate is at the bottom of the micrograph. The cells at the outer edge of the egg cylinder represent the primitive endoderm and are vesiculated, whereas cells in the interior are not. In addition, some of the outer cells have a high number of filopodia suggesting that they may have been migrating over the more internal cells. At the egg cylinder stage, Fig. 2E shows a vertical cross section and Fig. 2F shows a horizontal cross section of an embryo at 100 h after hatching. Notice that the outer cells are highly vesiculated and at higher magnification have a fuzzy surface appearance. This fuzzy surface is indicative of numerous microvilli (see below, this is confirmed by ultrastructural analysis) indicating this primitive endoderm developed in good health. The interior cells resemble primitive ectoderm cells in that they do not have a brush border and are not highly vesiculated. This embryo was likely in the process of gastrulating as distinct tissue layers could be seen on the right side of the embryo as shown in the micrograph (Fig. 2E) (Tam et al., 1993). 2.2. Analysis of the sheet and tonojihment distribution prior to blastocyst hatching The number and distribution of sheets in a two-cell embryo is shown in Fig. 3. Fig. 3A is a low magnification view showing sheets as small linear objects throughout the blastomeres (see arrows). At higher magnification, the sheets appear as a crosslinked network of 10 nm filaments (Fig. 3B). At the late blastocyst stage, prior to hatching, individual 10 nm filaments are seen in the outer trophectodermal cells and a few sheets are detected remaining in the ICM cells in agreement with previously published studies (Gallicano et al., 1991; Capco et al., 1993). However, in addition to the previously published study of Gallicano et al. (1991) unusual clusters of 10 nm filaments in trophectodermal cells were detected (see Fig. 3C). These clusters did not resemble the crosslinked structures of typical sheets but instead resembled bundles of intermediate filaments that could have been in the process of packing more tightly to form tonofilaments. Note that in trophectoderm cells it appears that there were

309

areas where sheets were disassembling into individual intermediate filaments. This process was observed only in the outer cells of the embryo (trophectoderm) and not in ICM cells of these well developed blastocysts. Approximately 12-24 h after blastocysts were in culture, a rupture formed in the zona pellucida. The blastocysts squeezed through this rupture until free of their zona pellucida, a process referred to as hatching, shown at the light microscope level in Figs. 1B and 2B. At this time, more structures resembling tonofilaments appeared in the trophectoderm of the hatching blastocyst (Fig. 4A). In the ICM cells of the hatching blastocyst, tonofilaments were found only in very specific ICM cells; that is those that lined the blastocoelic cavity (i.e. primitive endoderm cells) and had proceeded through the hatching site of the zona pellucida (Fig. 4B). ICM cells still inside the zona did not contain detectable tonofilaments. The embryonic ectoderm cells of the ICM cells which had hatched also were devoid of distinct tonofilaments but did have areas where sheets appeared to be disassembling and associating with individual intermediate filaments and desmosoma1 junctions (Fig. 4C). Quantification for number of sheets per pm2 area of cytoplasm demonstrated a progressive reduction in number of sheets from two-cell embryos, through postcompaction stages, the blastocyst stage, the hatchingblastocyst stage, and the hatched-blastocysts stage; the results are shown the histogram of Fig. 5A. Quantification of the number of tonofilaments per pm* throughout this same time period during early development showed that tonofilaments were not detected in the embryo until very late in blastocyst development and appeared in cells of the embryo only at the time that sheets were disassembling (Fig. 5B). 2.3. Analysis of the sheet and tonofilament distribution after blastocyst hatching Blastocysts 12-24 h after hatching began showing increased numbers of tonofilaments in both the ICM cells and in the trophectodermal cells. These numbers increased up to 48 h after hatching, as shown in the histogram (Fig. 5B). Electron micrographs of a typical trophectoderm cell and a typical inner cell mass cell in a blastocyst 12-24 h after hatching and attaching to the culture plate are shown in Fig. 6A. Fig. 6A shows a trophectoderm cell with tonofilaments extending parallel to the long axis of the cell. Fig. 6B illustrates tonofilaments seen in inner cell mass cells at this developmental stage. Approximately 72 h post attachment of the embryo to the culture plate, tonofilaments were not detected in any part of the embryo although numerous junctional complexes existed (data not shown). Embryos observed 84-100 h after hatching were devoid of tonofilaments (Fig. 7). It was determined by the following developmental criteria that the developing embryos were healthy. At this stage of development a nor-

310

SM. Schwurz et al. I Mechunisms of Development 53 (1995) 305-321

SM. Schwarz et al. I Mechanisms

ofDevelopment 53 (1995) 305-321

311

Fig. 4. Electron microscopic views of a hatching blastocyst. (A) Note tonofilaments (arrows) in the trophectoderm cell. (B) An ICM cells facing the blastocoelic cavity that has already hatched from the zona pellucida. Note tonofilaments which appear to be at an early stage of formation. (C) In embryonic ectoderm cells within the KM in a hatching blastocyst, sheets appear to he splaying apart (arrowheads) and reorganizing into tonofilamentlike structures which are associating with desmosomes (arrows) at cell contacts.

ma1 embryo would contain two distinct epithelia, the primitive endoderm and the primitive ectoderm, which are morphologically distinct and posses different developmental functions. In addition, at this stage gastrulation begins during in vivo development. Embryos processed 84-100 h after hatching met all of these developmental criteria. The low magnification views illustrate a typical region of an 84-100 h embryo (Fig. 7). Note that two clearly distinct cell layers exist (Fig. 7A). The outer layer

Fig. 3. Electron microscopic views of two cell and blastocysts stage pellucida and arrows point to the sheets. (B) A high magnification and associated vesicles. (C) A high magnification view of a portion apart into intermediate filaments (arrowheads). Tonofilament-like ‘MT’ is a microtubule.

indicated with (EN) is the primitive endoderm and the inner layer indicated with a (EC) is the primitive ectoderm. These cell layers were clearly distinct as the primitive endoderm exhibited numerous microvilli at its apical surface and also contained numerous vesicles near the apical surface. The primitive ectoderm appeared as columnar cells which contained very few microvilli (Fig. 7B). The primitive endoderm cells (data not shown) contained well developed microvilli, healthy mitochondria

embryos. (A) Low magnification view of a two cell embryo. ‘Z’ indicates the zona view of a two-cell embryo. Arrows indicate sheets and arrowheads indicate Golgi of a trophectoderm cell shows sheets (large arrow) which appears to have splayed structures (small arrow) also are shown. The ‘A’ represents an actin filament and

312

SM. Schwarz et al. I Mechanisms of Development 53 (I 995) 305-32 I

BIASTOCYST

HATCHING

HATCHED

12-24 hr.

48 hr.

k

k

8 CI

i

-72 hr.

96 hr.

---BLASTOCYST

HATCHING

-72 hr.

96 hr.

Fig. 5. Quantification of sheets and tonofilaments during mammalian development. (A) A histogram of sheets perpm2 from the two-cell stage through to the hatched blastocyst stage. (B) A histogram of tonofilaments per pm2 which are observed first in trophectoderm cells in late blastocyst stage and persist through to the 84-100 h embryo. * represents the density of tonotilaments in both primitive endoderm and ectoderm cells.

and endoplasmic reticulum, junctional complexes, and vesicular material typical of these cells (Tam et al., 1993); however, no tonofilaments were detected in any of the

images observed at high magnification. This latter observation is in agreement with Jackson et al. (1981), Franke et al. (1982), and Tam et al., (1993) who have shown

SM. Schwarz et al. I Mechanisms of Development 53 (I 995) 305-321

313

Fig. 6. Electron microscopic views of cultured embryos. (A) High magnification view of an ICM cell in a 48 h attached embryo shows the typical arrangement of tonofilaments (arrows) in these cells. (B) View of a 12-24 h attached embryo. Note tonofilaments (arrows) in both the trophectoderm ‘T’ cell and the KM ‘I’ cell.

junctional complexes connecting the cell borders of each cell to the next without associated cytoplasmic tonofilaments. As noted previously, embryos at this stage of development would typically undergo gastrulation. Gastrulation is the stage at which cells of the primitive ectoderm differentiate into mesoderm cells which then migrate through the primitive streak. Once migration between the primitive tctoderm and primitive endoderm is complete, the embryo contains three primary germ layers, the endoderm, ectoderm and mesoderm. In some embryos grown in vitro, a layer of cells was observed apparently migrating between the primitive endoderm and primitive ectoderm as is shown in Fig. 7C. In this micrograph the endodermal cell labeled (EN) and the primitive ectoderm label (EC) contained another cell layer between their surfaces. These cells contain numerous larnellapodia and appeared to be migrating as would occur at the time of gastrulation. This was not observed in all embryos due to the presence

of the ectoplacental cone. During in vivo development ectoplacental cells would continue to grow and occupy space within the uterine wall, however in vitro, these cells could not expand into the plastic of the cell culture plate and consequently they would expand abnormally into the embryo, often distorting the embryo’s shape. Such a cell mass derived from an ectoplacental cone is shown in Fig. 7D. Note that these cells do not resemble primitive ectoderm or primitive endoderm. These cells are not columnar, as would be the case for cells of the primitive ectoderm, and these cells do not contain microvilli or vesiculated areas on their apical surface as would be the case for the primitive endoderm. These cells also contained no detectable tonofilarnents or cytoskeletal sheets. 2.4. Tonofilaments are composed of the same cytokeratins as the sheets To determine the type of intermediate filament proteins within tonofilaments found in these embryos,

314

S.M. Schwarz et al. I Mechanisms

of Development 53 (1995) 305-321

Fig. 7. Electron micrographs of 84-100 h embryos grown in culture. (A) Low magnification view of the endoderm cells ‘EN’ and the ectoderm cells ‘ECT’. (B) A high magnification view of the ectoderm cells. Note the columnar appearance of these cells. (C) View illustrating embryo in process of gastrulating. Note presumable mesoderm ‘ME’ cells migrating between endoderm ‘EN’ and ectoderm cells ‘ECT’, without the presence of tonofilaments. (D) Ectoplacental cone cells contain no tonofilaments. These are the cells responsible for distorting the embryo as it grows beyond this stage under the culture conditions described in Section 2.

hatched blastocysts were subjected to immunocytochemistry employing antibodies directed against specific types of keratin intermediate filaments, types 5, 6, and 8 (i.e.

K5, K6, and K8). Antibodies to K5 and K6 were used because the sheets have been shown to contain cytokeratins type 5, 6, 16, and ‘2’ (Gallicano et al., 1994) and

Fig. 8. Hatched blastocysts were subjected to antibodies directed against KY6 (A and B) or K8 (D and E) followed by fluorescently tagged second antibodies and analyzed using laser scanning confocal microscopy. (A) Stereoscopic view of a stack of twenty 1 Sprn optical sections from the trophectoderm cells of a hatched embryo demonstrates staining of KY6 at intercellular junctions (arrowheads) and in distinct tracts extending between perinuclear regions and the intercellular junctions. (B) A stack of ten 1.5 ym sections through the middle of an embryo demonstrates KY6 in a diffuse pattern in ICM cells and at intercellular junctions (arrowheads) in trophectoderm cells. (C) Control embryo stained only with rhodamine-conjugated goat-antimouse antibodies shows no staining. (D) Stereoscopic view of a stack of 20 I .5,~m optical sections from a hatched embryo demonstrates staining of K8 at perinuclear regions (arrowheads) in trophectoderm cells. No staining was observed at intercellular junctions or on tracts. (E) A stack of 10 1.5pm sections through the middle of an embryo shows K8 surrounding nuclei in both ICM cells and trophectoderm cells. (F) Control embryo stained with FITC-conjugated goat-antirat antibodies shows no staining.

S.M. Schwurz et ul. I Mechanisms

of Development 53 (1995) 305-321

315

316

SM. Schwarz et al. I Mechanisms of Development 53 (1995) 305-321

possibly K15 (see Section 3). Antibodies to K8 were used because it has been reported that K8 (and its pair K18) are the first types of intermediate filament proteins expressed in the mammalian embryo (Oshima et al., 1983; Houliston, 1987; Thorey et al., 1993). Confocal stereo-images of hatched blastocysts demonstrated an enrichment of K5 and K6 both at the junctions between blastomeres of trophectodermal cells and in tracts extending across trophectodermal cells (Fig. 8A). Fig. 8A demonstrates these tracts in images composed of 20 optical sections (each at 1.5 pm thick) in trophectodermal cells. Ten optical sections (each at 1.5 pm thick) through the middle of the embryo demonstrates a diffuse staining pattern of K5/6 in the ICM cells as well as a staining pattern at intercellular junctions (Fig. 8B). Hatched blastocysts stained for K8 revealed a staining pattern very different from that found for K5 and K6. In both trophectodermal cells and ICM cells, K8 is localized in the perinuclear area and largely excluded from the region of blastomere-blastomere contact (Fig. 8D; 20 optical sections). Moreover, K8 is not present in tracts extending across entire blastomeres (Fig. 8D). Ten optical sections (each at 1.5 pm thick) through the center of the embryo showed a diffuse perinuclear staining of KS in both ICM cells and trophectoderm cells (Fig. 8E). Hatched embryos challenged to rhodamine-labelled anti-mouse (Fig. 8C) or FITC-labelled anti-rat (Fig. 8F) antibodies both show virtually no staining.

addition, anti-actin antibodies were used to immunopurify actin, which served as a control for the amount of protein synthesis. Actin synthesis is relatively high at this stage of mammalian embryogenesis and has been used as a control for the amount of protein synthesis by other investigators (Oshima et al., 1983; Overstrom et al., 1989). Samples were subjected to SDS-polyacrylamide gel electrophoresis and autoradiography. Hatched embryos immunopurified with anti-K8 and K18 antibodies resulted in relatively strong signals at 55 kDa and 53 kDa, the proper molecular weight for KS and K18, respectively (lane 1,

2.5. (35S]Methionine defection ofnewfy synthesized keratins Newly synthesized K5 and K6 or KS and K18 were detected in hatched blastocysts by incubating equal numbers of blastocysts (a stage during development when K8 and K18 have been shown to be synthesized (Oshima et al., 1983; Thorey et al., 1993) in H-l medium containing [35S]methionine (5OO,~Ci/ml) for 20 h at 37°C. Once the blastocysts hatched, the appropriate cytokeratins were immunopurified with anti K5/K6 or anti K8 and K18. In Fig. 9. Newly synthesized KS and K6 or KS and K18 and actin were. detected in hatched blastocysts. (A) Autoradiographic analysis of immunopurified KS and K18 from 250 hatched blastocysts is shown in lane 1. The polypeptide at 55 kDa represents K8 and the polypeptide at 53 kJIa represents K18. Immunopurified K5 and K6 from 250 hatched blastocysts are. shown in lane 2. Immunopurified actin is in lane 3. The polypeptide at 59 kDa represents K5 and the 56 kDa polypeptide represents K6. the third polypeptide may he a keratin pair to K5 (i.e. K14) or K6 (i.e. K16) which co-immunopurified with anti-K5/6. (B) Lane 1 demonstrates Western blot analysis of 150 hatched blastocysts using anti-K8 and K18 antibodies followed by chemiluminescence. Two bands are detected at 55 kDa and 52 kDa representing K8 and K18, respectively. The staining pattern of 150 hatched blastocysts stained with anti-K5/6 is in lane 2. Two bands are detected at 59 kDa and 56 kDa representing KS and K6, respectively. (C) The synthesis of KS, K6, K8, K18 was compared densitometrically to each other as well as to the synthesis of actin from hatched embryos. Histogram shows the results from two experiments. The gray bars represent the bands from the autoradiograph in (A).

C

KS

K6

KS

K18 ACTIN

IMMUNOBWRIHED PROTEIN (EC-kentintype)

S.M. Schwarz et al. I Mechanisms of Development 53 (1995) 305-321

Fig. 9). However, hatched embryos immunopurified with antibodies to W and K6 resulted in relatively weak signals at 59, 56, and 51 kDa (lane 2, Fig. 9). Based on the M, of each polypeptide, the 59 kDa polypeptide represented K5 and the 56 kDa polypeptide represented K6. The third polypeptide may represent a co-immunopurified keratin pair of K5 or K6 (K5 pairs with K14 or K15 and K6 pairs with K16). Western blot analysis of 150 blastocysts with these antibodies revealed the presence of K5, K6, K8, and K18 in these embryos (Fig. 9B). The amounts of incorporation of [35S]methionine into K5, K6, K8, K18 were compared densitometrically to each other as well as to the synthesis of actin (lane 3, Fig. 9A), a prominent cytoskeletal protein synthesized in blastocysts (Jackson et al., 1980; Overstrom et al., 1989). Fig. 9C shows the results from two separate experiments using identical incubation conditions and the same batch of [35S]methionine. The density of each band was measured and expressed as arbitrary units reflecting the optical density of the bands. The ratio of K8 synthesis to K18 synthesis was approximately 1: 1 (see lane 1, Fig. 9A for example) which would be expected since both must pair to make an intermediate filament. Comparison of synthesis of K5 and K6 which do not pair to make a filament (K5 pairs with K14 or K15 and K6 pairs with K16) also resulted in a ratio of approximately 1: 1 (see lane 2, Fig. 9A for example). However, when K8 and K18 synthesis was compared to that of K5 and K6, the ratio in each experiment was approximately 3:l (K8 and K18:K5 and K6; see lanes 1 and 2, Fig. 9A for example). As a control for protein synthesis in these embryos, actin synthesis also was measured, resulting in a ratio of approximately 1:l when compared to K8 and K18 (Fig. 9C). 3. Discussion The structure of any cell is mediated by its cytoskeleton, and moreover, the cytoskeleton plays a central role in many activities of cells. Thus, it is likely that the cytoskeleton mediates many cellular functions in cells progressing through early development, such as embryonic blastomeres. Previous work (Capco and McGaughey, 1986; Capco et al., 1993; Gallicano et al., 1993, 1994, 1995) identified an unusual cytoskeletal element that appears to be involved with major developmental transitions in early mammalian embryos. These elements were termed ‘sheets’ by Capco and McGaughey (1986). The sheets in mammalian eggs and embryos represent a highly ordered array of intermediate filaments held in register by crossbridges ‘and coated with a particulate component detectable when viewed at the ultrastructural levels (Capco et al., 1993; Gallicano et al., 1991, 1994). In this study we show that there appears to be a rapid appearance and then disappearance of tonofilaments between the time of hatching from the zona and immediately prior to gastrulation. The tonofilaments appear first

317

in the cell exhibiting epithelial characteristics, the trophectodermal cells, and subsequently in the epithelial-like cells derived from the ICM cells, that is the primitive ectoderm and primitive endoderm. Other investigators have identified tonofilaments in cells of the post hatching blastocyst (Jackson et al., 1980, 1981; Franke et al., 1982). However, here we report the transient, pulse-like appearance, and then subsequent disappearance of tonofilaments in cells of the early embryo immediately prior to gastrulation. Moreover, we are the first to identify a possible correlation between the timing of the disappearance of sheets and the subsequent initial appearance of tonofilaments. Tonofilaments appear as the sheets disappear both in trophectodermal cells and in ICM cells. In both trophectodermal cells and ICM-derived cells (i.e. primitive ectoderm and primitive endoderm), the tonofilaments increase in number eightfold after their initial appearance. Then 72 h after the initial appearance of tonofilaments in the cells of the primitive endoderm, primitive ectoderm and trophectoderm tonofilaments seam to disappear. Reports have clearly shown that embryonic ectodermal cells are connected to one another by gap junctions and in some areas, also by tight junctions, however, junctional complexes associated with tonofilaments (desmosomes) have not been identified in these cells (Tam et al., 1993). Why would tonofilaments and desmosomes be absent at the time of gastrulation in these cell types? Based on our results as well as a number of other observations (Croft and Tarin, 1970; Klymkowsky et al., 1989; Page, 1989; Tam et al., 1993), we propose that the absence of tonofilaments and desmosomes may promote gastrulation in mammalian embryos. To date, little information exists concerning the mechanics of gastrulation in mammalian embryos; however gastrulation involves migration of cells and comparable types of mobility of epithelial cells have been demonstrated in a number of in vitro studies. These studies have shown that the disappearance of intermediate filaments, and consequently tonofilaments, consistently correlates with increased mobility of individual cells within an epithelial sheet (Klymkowsky et al., 1989; Page, 1989) especially in the movement of epidermal cells during wound healing (Croft and Tarin, 1970). In vivo studies provide further evidence that the lack of tonofilamentidesmosomal complexes may promote gastrulation. These studies in species other than mammals such as those in the urodelian family have demonstrated that tonofilamentidesmosome complexes are not present in cells undergoing gastrulation (Perry, 1975). Thus, it is possible that the absence of tonofilaments at the time of gastrulation in mammals may help to promote both subtle and dramatic ingressive movements of embryonic ectodermal cells through the primitive streak. What is the functional significance of the transient presence of tonofilaments? The current theory of tonofilament function is that they provide stability across sin-

318

S.M. Schwarz et al. I Mechanisms

gle cells and their association with intercellular junctions known as desmosomes provide stability across groups of cells (Fuchs, 1994). Recent evidence supporting this assumption has been provided using transgenic mice in which human truncated K5 and/or K14 genes which lack the carboxyl tail and about one third of the a-helical rod segment result in rupturing of basal epithelial cells of the epidermis in both the Fl and F2 generations (Vasser et al., 1991; Fuchs, 1994). This rupturing of epithelial cells due to mutations in specific types of keratin intermediate filaments has been shown to be directly responsible for the blistering in patients with specific skin diseases (Coulombe et al., 1991; Cheng et al., 1992). Based on these and related observations (Croft and Tarin, 1970; Perry, 1975; Coulombe et al., 1991; Fuchs, 1994) we propose that the tonofilaments may provide structural integrity to the embryo as it implants into the uterine wall. The disappearance of tonofilaments just prior to gastrulation could subsequently permit cell mobility that is essential for gastrulation. What types of keratins are present in the embryo and tonofilaments prior to, during, and after implantation? Our data show that the tonofilaments which appear in these embryos are composed of K5 and K6 and very little K8 and K18. This was shown by correlating the intracellular location of tonofilaments in desmosomes at intercellular junctions and tracts of tonofilaments across blastomeres with the distribution of K5 and K6 (detected by immunocytochemical analysis). K5 and K6 were found associated with intercellular junctions and in distinct tracts across blastomeres whereas K8 was found in neither of these locations. Instead K8 was identified in a diffuse perinuclear area. Our studies show that only small amounts of K5 and K6 are synthesized (when compared to K8, K18, and actin) during the time that the tonofilaments are formed, suggesting that the tonofilaments are assembled from preexisting K5 and K6. Since the sheets serve as an extensive supply of assembled K5 and K6 protein (Gallicano et al., 1994), it is likely that the tonofilaments recruit intermediate filament proteins from the sheets. Further support for this assertion comes from the timing of the disappearance of sheets and the appearance of tonofilaments. In both trophectoderm cells and ICM cells, tonofilaments appear within 12 h after the sheets have disassembled into individual intermediate filaments. This temporal correlation also suggests that the sheets serve as a reserve of assembled intermediate filaments, which in turn may contribute to the tonofilaments. The sheets form and are stored in the female gamete during oogenesis under the direction of the maternal genome (Gallicano et al., 1994b and references therein). Storage of materials in the oocyte to be used after fertilization is a common developmental theme (Jeffery et al., 1983; Hensen and Begg, 1988; Larabell and Capco, 1988; Spudich et al., 1988; Bonder et al., 1989; Bement et al., 1992) and appears to be applicable for the sheets.

of Development 53 (1995) 305-321 There has been some confusion in the literature as to when intermediate filament proteins are present in mammalian embryos. Oshima et al. (1983) reported the synthesis of K8 and K18 beginning at the 8-16 cell stage, but other investigators employing broad spectrum antibodies detected intermediate filament proteins in mouse eggs (Lehtonen, 1985, 1987). We believe our results have helped clarify some of this confusion. Identification of KS and K18 synthesis in 8-16 cell mouse embryos was accomplished by employing specific antibodies produced by Oshima and co-workers (1983). These antibodies did not cross react with K5, K6, K16 or ‘Z’ (Gallicano et al., 1994, 1995) and thus did not recognize the intermediate filaments in the sheets, whereas other investigators using broad spectrum antibodies likely detected the intermediate filament proteins in the sheets (Lehtonen, 1985, 1987; Gallicano et al., 1994). The sheets are a product of the maternal genome and thus represent a maternal store of this protein. Since the sheets are composed of K5, K6, K16, and ‘Z’, the above cited studies did not detect them with monospecific antibodies to K8 or K18. Furthermore, the investigation by Emerson, (1988) which injected antibodies to K8 into early embryos, did not test the biological function of the intermediate filaments comprising the sheets and the K8 and K18 knockout experiments of Baribault et al. (1993) likely had no affect on K5, K6, K16, and ‘Z’. In a recent publication, we demonstrated that the sheets were composed of K5, K6, K16, and ‘Z’ (Gallicano et al., 1994). It is known that K6 pairs with K16 to make a filament and it has been generally accepted that K5 pairs with K14 to make a filament; however, we did not detect any K14 with our primary antibodies (i.e. AEl should recognize K14; Moll et al., 1982; Weiss et al., 1984) in our two-dimensional Western blots of 250 mouse eggs (see Fig. 8 in Gallicano et al., 1994). In recent work from Elaine Fuchs’ laboratory (Lloyd et al., 1995), strong evidence is provided showing that K5 also pairs with K15 in basal epithelial cells. Based on this new evidence we obtained a polyclonal antibody specific for K15 from the Fuchs laboratory. Western blot analysis of mouse eggs and hatched blastocysts using this polyclonal antibody revealed that K15 also is present in these eggs and embryos (data not shown). Consequently, the filaments within sheets may be composed of K5 and K15 in addition to K6, K16 and ‘Z’. Further evidence that K15 may be the keratin type that pairs with K5 in sheets comes from gene ablation experiments where K14 is knocked out in transgenic mice. The Fl and F2 progeny of these mice engineered with ablated K14 gene are able to develop to term normally (Vassar et al., 1991) suggesting that K14 may not be present in sheets and, thus, not utilized at implantation. Alternatively, if our antibody detection system (from Gallicano et al., 1994) did not recognize K14 and K5 does pair with K14 in sheets it is possible that the mutation introduced

S.M.

Schwurz

et al. I Mechanisms

into K14 may not have been severe enough to completely alter its function in the sheets. Whichever the case, molecular approaches such as these may be a valuable tool for future investigations in the functional significance of the sheets. It is not surprising that K5, K6, Kl5, and K16 are the intermediate filament types found in tonofilament/desmosome complexes in embryos undergoing implantation. K6 and its pair K16 have been found in many types of highly proliferative, somatic epidermal cells (Weiss et al., 1984) and K5 and its pair Kl5 are found in proliferative, basal stratified squamous epithelial cells (Lloyd et al., 1995). The implanting embryo is comprised of numerous cells most of which are undergoing rapid differentiation and proliferation into primitive ectodermal cell types. In summary, our data show that sheets appear to represent a storage form of assembled intermediate filaments. The intermediate filaments appear to be recruited from the sheets as individual intermediate filaments as embryogenesis proceeds in those cells which form epithelial layers. Later, beginning at the time of hatching from the zona pellucida, the intermediate filaments which descend from the sheets (i.e. K5, K6, K15, K16, ‘Z’) appear to form tonofilaments that associate with desmosomes and form tracts across the cells. These tonofilaments probably provide mechanical strength to junctions between cells and the entire epithelial layer during the time of implantation. Finally, the tonofilaments in both the trophectoderm and inner cell mass cells disappear when the embryo nears gastrulation. 4. Experimental

procedures

4.1. Embryo procurement and culture Female Swiss Albino mice (Mus musculus) were injected with 8 international units (IU) of pregnant mares’ serum gonadotropin (PMSG) between 1400 h and 1500 h and again 48 h later with 8 IU of human chorionic gonadotropin (hCG). Each superovulated female was placed in a cage with one male to allow copulation and subsequent fertilization to occur. Blastocysts stage embryos were obtained 3.5 days post coitum (p.c.) after cervical dislocation of the female upon necropsy. Blastocysts were flushed from the uterus using a 3 cm3 syringe, filled with 2A-HEPES (McGaughey, 1977), connected to a 26 gauge needle. The injection was made into the cervix allowing the uterine horns to balloon outward. The ovaries and oviducts then were cut off and the remaining medium was flushed through the cervix allowing the blastocysts to wash out from the distal end of the uterine horns. The blastocysts were immediately placed in organ-well culture dishes containing pre-gassed (Trimix; 5% CO,, 5% 0,, 90% Nz), prewarmed (incubated for 1 h at 37’C) Ham’s F-12 medium supplemented with 15% fetal bovine serum (FBS) and 1% (w/v) penicillin/streptomycin (Pen-Strep)

of Development

53 (1995)

305-321

319

covered with mineral oil (H-l medium; Wiley et al., 1978; Gonda and Hsu, 1980). The embryos were incubated in their culture dishes in a closed chamber warmed at 37°C and gassed with Trimix (defined above). Hatching of these embryos from their zonae pellucidae typically occurred within 12-24 h after which they anchored to the organ culture dish. In agreement with the studies by Gonda and Hsu (1980) and Wu et al. (1981), virtually all of the blastocysts attached and a large proportion (about 80%) attached with the inner cell mass (ICM) near the culture plate allowing an upward growing cylinder. Usually at least 25% (of the 80%) continued to develop to pregastrulation stage embryos, which were in an egg cylinder formation (Wu et al., 1981). Embryos that did not continue normal development were usually hindered because the ICM cells were anchored in a sideways or downward position, which inhibits egg cylinder elongation, instead of the normal upward slant. Embryos cultured under the above conditions were observed only when necessary for fixation of a certain stage or when the medium needed to be changed. Medium was changed every 3 days. Embryos were fixed at increasing time intervals beginning at the late blastocyst stage prior to hatching through to gastrulation stage (approximately 100+ h post hatching). Embryos were fixed with 2.0% glutaraldehyde made 1.0% (w/v) with tannic acid for 2-4 h, followed by postfixation in 1.O% 0~0~ overnight. Subsequently, specimens were dehydrated in alcohol and embedded in Spurrs resin. In the case of late stage embryos which were attached to the culture plate upon fixation and dehydration, prior to their placement in Spurrs resin, the embryos were very gently dislodged from the bottom of the culture dish with a rubber policeman, while in 100% ethanol. Extreme care had to be taken or subsequent handling would damage the very friable embryo. Any physical trauma imposed on an embryo during this procedure would render it unsuitable for any microscopic observation. Sections were taken at a thickness of 1 pm and stained with toluidine blue to observe the specific location in the embryo. This enabled the proper location to be identified for preparation of subsequent thin sections for observation in the transmission electron microscope. Thin sections were taken at a thickness of 90 nm and stained with 2% uranyl acetate for 12 min and then Reynold’s lead citrate for 12 min and viewed using a Philips CM12 Electron Microscope at 80 kilovolts (kV). 4.2. Confocal microscopy Embryos used for confocal microscopy were fixed with methanol at -20°C for 10 min. They then were washed 3X in intracellular buffer (ICB; 100 mM KCI, 5 mM MgCl2, 3 mM EGTA, 20 mM Hepes, pH 6.8; Aggeler, et al., 1983; Webster and McGaughey, 1990) and incubated for 1 h at 37°C in ICB + 1.0% BSA which

320

S.M. Schwarz et al. / Mechanisms of Development 53 (1995) 305-321

contained primary antibodies directed against either keratin 5 and 6 (mouse monoclonal; Boehringer, Indianapolis, IN; catalogue no. 1273 396) or keratin 8 (Troma-1; rat monoclonal, a generous gift from Dr. Robert Oshima, La Jolla Cancer Foundation, La Jolla, CA). Following three 20 min washes in ICB, embryos were placed into ICB + 1.0% BSA containing rhodamine-conjugated anti-mouse IgG (Sigma) or FITC-conjugated anti-rat IgG (Sigma) for 1 h at 37°C. Embryos then were washed 3 times. for 30 min each in ICB and mounted onto clean glass slides. Using a Molecular Dynamics confocal microscope equipped with a 60X Plan Apo lens, each embryo was scanned through the Z axis by an argon laser to establish the embryo center. Sections then were scanned at 0.5 pm intervals through the X/Y axis beginning at the top of the embryo and ending at the bottom of the embryo. The images were digitized and stored on an erasable optical disc. 4.3. Gel electrophoresis and Western blotting Embryos used for gel electrophoresis were solubilized in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer (Laemmli, 1970) in a 1.5 ml microcentrifuge tube. Electrophoresis was performed using the method described previously (McGaughey and Capco, 1989). After electrophoresis, proteins in gels either were stained with Coomassie blue and Silver stain (Gallicano et al., 1991) or blotted onto PolyscreenTM (NEN, Wilmington, DE) for immunodetection. Detection of antigen/antibody complexes were conducted using a chemiluminescence antibody detection system as described by (Gallicano et al., 1994). Contamination of Westerns by finger proteins was controlled for by the procedure described in Gallicano et al. (1994a). No finger proteins were detected 4.4. [35S]Methionine labeling and immunopurifcation of intermediate filament proteins To identify newly synthesized intermediate filament proteins in hatched embryos, blastocysts were incubated in pregassed, prewarmed H-l media containing SOOyCi/ml [35S]methionine for 24 h. Labelled proteins were immunopurified by one of four different antibodies. One antibody was anti-cytokeratin type 5/6. These two types of cytokeratins were recently shown to be major components of cytoskeletal sheets in both mouse (Gallicano et al., 1994) and hamster eggs (Capco et al., 1993). The second antibody used to immunopurify labelled proteins was Troma-1 (see above), which recognizes K8. The third antibody used was Endo B antiserum, a polyclonal antibody which recognizes K18. These latter antibodies were a kind gift from Dr. Robert Oshima. The fourth antibody used was anti-actin (Boehringer, catalogue no. 1378 996). Immunopurification of [35S]methionine labelled intermediate filament proteins was accomplished by a modified procedure from Sefton et al.,

(1993). Briefly, 250 radiolabelled hatched blastocysts were washed in phosphate buffered saline (PBS) 2x for 15 min each and placed into 5 ~1 of immunopurification buffer (IP; 0.15 M NaCl, 1.0 mM EDTA, and 10 mM Tris-base pH 7.5) containing 2.0% SDS and 6 M urea. Urea then was removed via dialysis. Dialysis medium was composed of IP buffer + 2.0% SDS. IP/SDS buffer containing the embryos then was diluted 1:20 with IP buffer containing 0.65% NP40, followed by addition of primary antibody (1:lOO dilution for all antibodies). This slurry was continuously rocked for 1 h at room temperature. IgGs were removed from solution with the addition of 50 ~1 washed protein G (Sigma) and continuously rocked for 1 h at room temperature. The Protein G was pelleted and washed 5 times using a microcentrifuge. The subsequent pellet was dissolved in Hoefer sample buffer. Sample buffer containing immunopurified protein then was electrophoresed in a 10% SDS polyacrylamide gel which was dried, and exposed to autoradiography film between two enhancer plates for 2 days. Immunopurified proteins were quantified using a densitometer. Three controls were employed to determine the efficiency of the immunopurification. Firstly, the ratio of antibody to antigen was increased from 1:lOO in the first two experiments to 1:50 in the third. After exposing all radiographs for the same time, no significant difference in density of the bands was observed. Secondly, all immunopurifications were performed twice on each sample. No detectable signal was observed in the second immunopurification from any of the three experiments indicating that in the first round of immunopurification, most of the protein was removed from the extract. Thirdly, actin synthesis was used as a positive control for protein synthesis. Actin synthesis has been shown to be relatively high at this stage of mammalian embryogenesis and has been used as a control for the amount of protein synthesis by other investigators (Oshima et al., 1983; Overstrom et al., 1989) 4.5. Morphometric analysis Measurements were made from negatives of sectioned embryos. At least three different embryos were employed for quantification of sheets or tonofilaments per pm2 in each cell type. Sheets and tonofilaments were quantified in 50 random locations at each developmental stage within the cytoplasmic regions of the cell to produce a mean and a standard error of the mean (SEM) at each stage. Morphometric data are presented as mean f SEM.

Acknowledgments We thank Dr. Elaine Fuchs and Dr. Catriona Lloyd for the K15 antiserum. This work was supported by NIH grant HD27 15 1.

S.M. Schwarz et al. I Mechanisms of Development 53 (1995) 305-321

References Aggeler, .I., Takemura, R. and Werb, 2. (1983) J. Cell Biol. 97, 14521458. Bement, W.M., Gallicano, G.I. and Capco, D.G. (1992) Microsc. Res. Tech. 22,23-48. Baribault, H., Price, J., Miyai, K. and Oshima, R.G. (1993) Genes Dev. 7.1191-1202. Bonder, E.M., Fishkind, D.J., Cotran, N.M. and Begg, D.A. (1989) Dev. Biol. 134, 327-341. Capco, D.G. and McGaughey, R.W. (1986) Dev. Biol. 115,446-458. Capco, D.G., Gallicano, G.I., McGaughey, R.W., Downing, K.H. and Larabell, C.A. (1993) Cell Motil. Cytoskel. 24.85-99. Cheng, J., Syder, A.J., Yu, Q.-C., Letai, A., Paller, A.S. and Fuchs, E. (1992) Cell 70, 811-819. Coulombe, P.A., Hutton, M.E., Letai, A., Hebert, A., Paller, A.S. and Fuchs, E. (1991) Cell 66, 1301-1311. Croft, B. and Tarin, D. (1970) J. Anat. 106.6377-6383. Emerson, J.A. (1988) Development 104, 219-234. Fisher, S.J., Cui, T.-Y., Zhang, L., Hartman, L., Grahl, K., Guo-Yang, Z., Tarpey, J. and Damsky, C.H. (1989) J. Cell Biol. 109,891-902. Fleming, T.P., Garrod, D.R. and Elsmore, A.J. (1991) Development 112.527-539. Franke, W.W., Schmid, E. and Grund, C. (1982) Cell 30, 103-l 13. Fuchs, E. (1994) J. Cell Biol. 125,511-516. Gallicano, G.I., McGaughey, R.W. and Capco, D.G. (1991) Cell Mot& Cytoskel. 18, 143-145. Gallicano, G.I., McGaughey, R.W. and Capco, D.G. (1992) J. Exp. Zool. 263, 194-203. Gallicano, G.I., McGaughey, R.W. and Capco, D.G. (1994a) Mech. Dev. 45.211-226. Gallicano, G.I., McGaughey, R.W. and Capco, D.G. (1994b) Microsc. Res. Tech. 27, 134-144. Gallicano, G.I., McGaughey, R.W. and Capco, D.G. (1995) Dev. Biol. 167,482-501. Gardner, R.L. and Rossant, J. (1979) J. Embryol. Exp. Morph. 52, 141152. Gonda, M.A. and Hsu, Y.-C. (1980) J. Embryol. Exp. Morphol. 56,2339. Hensen, J.H. and Begg, D.A. (1988). Dev. Biol. 127.338-348. Hill, D.P. and Strome, S. (1988) Dev. Biol. 125.75-84. Houliston, E., Pickering, S.J. and Mare, B. (1987) J. Cell Biol. 104, 1299-1308. Howe, J.G. and Hershey, J.W.B. (1984) Cell 37,85-93. Jackson, B.W., Grund, C., Schmid, E., Burki, K., Franke, V.W. and Illmensee, K. (1980) Differentiation 17, 161-179. Jackson, B.W., Grund, C., Winter, S., Franke, W.W. and Illmensee, K. (1981) Differentiation 20, 203-216. Jeffery, W.R. (1985) Dev. Biol. 110,217-229. Jeffery, W.R. and Meier, S. (1983) Dev. Biol. 96, 125-143. Jeffery, W.R., Tomlinson, CR. and Broduer, R.D. (1983) Dev. Biol. 99,408-417. Klymkowsky, M.P., Maynell, L.A. and Polson, A.G. (1987) Development.‘lOO, 543-557. Klymkowsky, M.W., Bachant, J.B. and Domingo, A. (1989) Cell Motil. Cytoskel. 14, 309-331. Koonce, M.P. and Schliwa, M. (1986) J. Cell Biol. 103.695-612. Laemmli, U.K. (1970) Nature (London) 227.680-685. Larabell, C.A. and Capco, D.G. (1988) Roux’s Arch. Dev. Biol. 197, 175-183.

321

Lawrence, J.B. and Singer, R.H. (1986) Cell 45,407-415. Lehtonen, E. (1985) J. Embryol. Exp. Morphol. 90, 197-209. Lehtonen, E., (1987) Cum Top. Dev. Biol. 22, 153-173. Lloyd C., Yu, Q-C., Cheng, J., Turksen, K., Degenstein, L., Hutton, E. and Fuchs, E. J. (1995) Cell Biol. 129, in press. McGaughey, R.W. (1977) Exp. Cell Res. 109,25-30. McGaughey, R.W. and Capco, D.G. (1989) Cell Motil. Cytoskel. 13, 104-111. Navara, C.S., First, N.L. and Schatten, G. (1994) Dev. Biol. 162, 2940. Ohlsson, R. (1989) Cell Differ. Dev. 28, 1016-1022. Oshima, R.G., Howe, W.E., Klier, F.G., Adamson, E.D. and Shevinsky, L.H. (1983) Dev. Biol. 99,447-l55. Overstrom, E.W., Benos, D.J. and Biggers, J.D. (1989) J. Reprod. Fertil. 85,283-295. Page, M. (1989) Development 105,97-107. Perry, M. M. (1975) J. Embryol. Exp. Morphol. 33, 127-146. Ramsey, E.M., Houston, M.L. and Harris, J.W.S. (1976) Am. J. Obstet. Gynecol. 124,647-652. Renfree, M.B. (1983) Oxf. Rev. Reprod. Biol. 5.272-289. Schatten, G. (1994) Dev. Biol. 165, 299-335. Schroeder, T.E. (1990). In Conrad, G.W. and Schroeder, T.E. (eds.), Cytokinesis: Mechanisms of Furrow Formation During Cell Division, Ann. N.Y. Acad. Sci., New York Academy of Science, New York. Vol. 582, pp. 78-87. Sefton, M., Johnson, M.H. and Clayton, L. (1993) Development 115, 313-318. Shimizu, T. (1984) Dev. Biol. 106.414426 Shimizu, T. (1988) Dev. Biol. 125,321-331. Snell, G.D. and Stevens, L.C. (1966) In Green, E.L. (ed.), Biology of the Laboratory Mouse, McGraw-Hill, New York. 2nd edn., pp. 205-245. Spudich, A., Wrenn, J.T. and Wessells, N.K. (1988) Cell Motil. Cytoskel. 9.85-96. Stem, CD. (1992) Trends Genet. 8, 158-163. Tam, P.P.L. (1989) Development 101, 55-67. Tam, P.P.L. and Beddington, R.S.P. (1987) Development 99, 109-126. Tam P.P.L. and Beddington, R.S.P. (1992) In Postimplantation Development of the Mouse, CIBA Foundation Symposium 165, Wiley, Chichester. pp. 27-49. Tam, P.P.L, Williams, E.A. and Ghan, W.Y. (1993) Microsc. Res. Tech. 26,301-328. Theriot, J.A. and Mitchison, T.J. (1991) Nature 352, 126-131. Thorey, I.S., Meneses, J.J., Neznanov, N., Kulesh, D.A., Pederson, R.A. and Oshima, R.G. (1993) Dev. Biol. 160.519-534. Tuttle, S.E., O’Toole, R.V., O’Shaughnessy, R.W. and Zuspan, F.P. (1985) Am. J. Obstet. Gynecol. 153,239-244. Vale, R.D., Scholey, J.M. and Sheetz, M.P. (1986) Trends Biochem. Sci. 11, 46ti68. Vasser, R., Coulombe, P.A., Degenstein, L.. Albers, K. and Fuchs, E. (1991) Cell 64.365-360. Wallace, R.A., Sanders, G.P. and Ferl, R.J. (1991) In Davies, G. (ed.), Biology: the Science of Life. Harper Collins, New York. Wiley, L.M., Spindle, A.I. and Pederson, R.A. (1978) Dev. Biol. 63, l10. Webster, S.D. and McGaughey, R.W. (1990) Dev. Biol. 142.61-74. Weiss, R.A., Eichner, R. and Sun, T.-T. (1984) J. Cell Biol. 98, 13971406. Wu, T.C., Wan, Y.J. and Damjanov, I. (1981) J. Embryol. Exp. Morphol. 248.45-54.