FGF Is an Essential Regulator of the Fifth Cell Division in Preimplantation Mouse Embryos

Home , TGF alpha

DEVELOPMENTAL BIOLOGY 198, 105–115 (1998) ARTICLE NO. DB978858 FGF Is an Essential Regulator of the Fifth Cell Division in Preimplantation Mouse Embryos

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Ning Chai,*,† Yogesh Patel,*,† Kristine Jacobson,*,†,1 Jill McMahon,‡ Andrew McMahon,‡ and Daniel A. Rappolee*,†,§,Ø *Department of Cell and Molecular Biology and †Department of Obstetrics and Gynecology, Northwestern University Medical School, Chicago, Illinois, 60611; ‡Biological Laboratories, Harvard University, 16 Divinity Avenue, Cambridge, Massachusetts, 02138; §Lurie Cancer Center and Feinberg Cardiovascular Research Institute, Northwestern University Medical School, Chicago, Illinois, 60611; and ØCenter for Reproductive Sciences, Northwestern University, Evanston, Illinois, 60620

Fibroblast growth factor (FGF) signaling is required prior to gastrulation in the mouse embryo. To test for the spatial and temporal requirements of FGF signaling, a dominant negative FGF receptor (dnFGFR) was used to make transgenic mouse embryos. In mosaic embryos, cell division ceased at the fifth cell division in all cells that expressed the mutant receptor, but cell death did not increase. After the fifth cell division, the progeny of unaltered cells and cells expressing lacZ continued to accumulate at the same rate, suggesting that the FGF requirement is cell autonomous. In mosaic embryos, lacZ, but not dnFGFR expression was detected in mitotic trophoblasts adjacent to the ICM. Conversely, dnFGFR-expressing extraembryonic ectoderm cells were detected at the abembryonic pole in postmitotic cells. In blastocysts expressing the dnFGFR in all cells, the morphology appeared normal and inner cell masses (ICMs) formed, but resultant embryos had only one-third the number of cells as control embryos. In these blastocysts, cell division had also ceased at the fifth cell division, but cavitation, a concurrent morphogenetic event, initiated and progressed normally. To test for the continuing requirement of FGF, FGFR-3 was overexpressed in all cells and resulted in an increase in cell numbers after the fifth cell cycle. In a model for postimplantation development, addition of FGF-4 to blastocyst outgrowths increased the number of extraembryonic ectoderm cells, suggesting a continuing role for FGF. Thus, FGF signaling induces the cell division of embryonic and extraembryonic cells in the preimplantation mouse embryo starting at the fifth cell division. The signal requirement for FGF is cell autonomous, but is not required to prevent cell death. This provides the first evidence for the necessity of a growth factor before implantation. ᭧ 1998 Academic Press

INTRODUCTION Fibroblast growth factor (FGF) signaling is essential for embryonic development in mice. Loss of a single FGF ligand In mammals, the peri-implantation period is the time of family member (FGF-4) in the mouse is lethal soon after greatest embryonic failure and loss (Cross et al., 1994). Em- implantation (Feldman et al., 1995). In the fgf4 null mutant, bryonic growth before implantation is density-dependent in the embryo implants into the uterus at 4.5 days after fertil- vitro, suggesting that cell division is regulated by diffusible ization (E4.5), but degenerates within 1–2 days. In an in factors. Growth factors including members of the insulin vitro model for early postimplantation events, the fgf4 null family and CSF-1 are sufficient to regulate cell division be- blastocyst grows out extraembryonic ectoderm cells, but fore implantation (Rappolee and Werb, 1994; Rappolee, the ICM dies within 2 days. Overexpression of FGF-4 in a 1998). However, no growth factor has been shown to be culture of isolated ICM cells leads to increased production necessary for cell division before implantation. of yolk sac endoderm, suggesting an FGF influence on ICM and/or early endoderm (Rappolee et al., 1994). Trophoblast and ICM cells divide sufficiently to produce a morphogenet- 1 Current address: Vysis Inc., 3100 Woodcreek Dr., Downer’s ically normal fgf4 null blastocyst (Feldman et al., 1995). Grove, IL 60515. However, since there is maternal FGF-4 mRNA in the oo-

0012-1606/98 $25.00 Copyright ᭧ 1998 by Academic Press All rights of reproduction in any form reserved. 105

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba 106 Chai et al. cyte (Rappolee et al., 1994), sufficient signal input in fgf4 Cross et al., 1995). The PGK promoter was based upon a sequence null mutant may continue for a number of zygotic cell divi- described previously (McBurney et al., 1991). To synthesize the sions. Other FGF ligands may also provide sufficient FGF dnFGFR, a BamHI–XbaI cDNA fragment containing the first 9 to sustain the embryo in early cell divisions. exons of FGFR-4 was excised from pKSFGFR-4 (Stark et al., 1991), resulting in a complete truncation of the tyrosine kinase domain. The FGF family consists of 15 FGF ligand (FGF-1 through This was ligated downstream of the myogenin promoter in a pSP72 FGF-15) genes of related sequence (Basilico and Moscatelli, vector (Yee and Rigby, 1993) to produce the pSP72MyodnFGFR 1992; Smallwood et al., 1996; Rappolee, 1997). The first 9 vector. A KpnI–XbaI genomic DNA fragment of FGFR-4 was used ligands are the best characterized. These heparin-binding to replace exons 5–9 of the FGFR-4 cDNA genes. The pSP72 vector factors are between 14 and 28 kDa. Some are secreted (FGF- contained a XbaI–HindIII fragment of bPA (Stewart et al., 1992). An 4 through FGF-9), while some lack the signal sequences oligonucleotide encoding a human myc epitope tag (EQKLISEEDL) required for secretion (FGF-1 through FGF-3). FGF-4 is ex- recognized by the 9E10 hybridoma (Evan et al., 1985) was cloned pressed in the ICM, but FGF-2 and FGF-3 have not been into the XbaI site between the 3؅ terminus of the truncated mutant detected in the blastocyst (Rappolee et al., 1988, 1994). of pSP72Myo-dnFGFR and the bPA site. This vector was named There are four high-affinity FGF receptors (FGFR-1 pSP72Myo-dnFGFRmycbPA and was used for another line of experiments. For synthesis of pPGKdnFGFRmycbPA, the myogenin pro- through FGFR-4; Kd Å20–480 pM). FGFR-1 through FGFR- 4 are related transmembrane receptors with split-tyrosine moter was digested with BamHI, blunted, and then excised with ClaI. The PGK promoter was digested in the PGKneo vector (Stew- kinase domains which require FGF ligand-dependent dimer- art et al., 1992) with PstI and then blunted and excised with ClaI ization for activation. Both tyrosine kinase activity and the and directionally cloned into the ClaI site in the pSP72 vector to split domain are required for mitogenic activity (Partanen create pSP72PGKdnFGFRmycbPA. The entire PGKdnFGFRmyc- et al., 1992; Johnson and Williams, 1993). An obligate low- bPA transgene was removed as a HindIII fragment and cloned into affinity FGF receptor (Kd Å 10 nM), such as syndecan (a the HindIII site in the PGKneobPA vector to create a vector with sulfated heparin), is believed to alter the FGF ligand and independent PGKneobPA and PGKdnFGFRmycbPA transgenes. allow it to bind high-affinity receptors. There is cross-reac- This construct is referred to as pPGKdnFGFRmyc. All references tivity between the four receptors and FGF-1 through FGF- in the text to expression of dnFGFR refer to this construct. 9 (Ornitz et al., 1996), although some alternately spliced A full-length FGF-4 was a gift of Dr. Martin (Hebert et al., 1990). forms of FGFR-1 through FGFR-3 have ligand specificity. For pPGK-FGF-4, the neo gene was removed from the pPGK-neo- Isoforms of all four FGF receptors bind FGF-4. FGFR-1, -3, bPA vector (Stewart et al., 1992; McBurney, 1991) and replaced with the XbaI–PstI fragment of the full-length coding sequence of and -4 mRNA have been detected in the blastocyst (Rap- mouse FGF-4 from FGF-K clone-12 (Hebert et al., 1990) by blunt- polee et al., 1994; Rappolee, 1998). FGFR-2 mRNA is de- end cloning. For the pPGKFGFR-4, a SalI–NotI fragment of the tected in the early postimplantation embryo (Orr-Urtreger full-length FGFR-4 gene was excised from pKSFGFR-4-Comp (Stark et al., 1991). et al., 1991) and replaced the excised neo gene by blunt-end cloning In this report, we have used a dnFGFR to test preimplan- (Stewart et al., 1992). For pPGKFGFR-3, a HindIII–XbaI fragment tation embryos for the time and location of their require- of the full-length wild-type gene was removed from MomFR3SV ment for FGF signal input. Surprisingly, ICM and adjacent (Ornitz et al., 1992) and directionally cloned into the pPGKneobPA trophoblast cells require FGF receptor signaling to divide after the neo gene was excised. beginning at the fifth cell division, which occurs before All the techniques used in transgene construction were standard implantation. techniques (Ausubel, 1993). All transgenes were transiently trans- fected into NIH3T3 fibroblasts or F9 teratocarcinoma stem cells or microinjected into one blastomere at the two-cell stage to produce mosaic embryos. The cells or mosaic embryos were then assayed for MATERIALS AND METHODS protein production by indirect immunocytochemistry (Rappolee et al., 1994). This test of expression for transgenesis was done for Materials. AmpliTaq DNA polymerase was purchased from pPGKdnFGFRmyc, pPGKFGFR-4, pPGKFGF-4, and pPGKFGFR-3. Perkin–Elmer Cetus (Norwalk, CT). Restriction enzymes and T4 Mouse eggs and embryos. Standard techniques were used for polynucleotide kinase were purchased from New England BioLabs obtaining eggs and zygotes. Female CD-1 (6–10 weeks old, Charles (Beverly, MA) and Gibco/BRL (Gaithersburg, MD). X-Phosphate (5- River, Wilmington, MA) or MF-1 (Harlan Sprague–Dawley, India- bromo-4-chloro-3-indolyl phosphate) was purchased from Boeh- napolis, IN) were mated with C57BL/6J X SJL/J F1 hybrid males ringer-Mannheim Biochemicals (Indianapolis, IN). Hoechst 33258 (Jackson Laboratories, Bar Harbor, ME). Noon of the day following was purchased from Eastman Kodak Co. (Rochester, NY). All other coitus was considered day E0.5. Embryos were obtained at the fol- chemicals were from Sigma Chemical (St. Louis, MO) or VWR lowing stages: E0.5 (fertilized zygote), E1.5 (two-cell stage), and E4.5 (Chicago, IL). Agarose was from FMC (Rockland, ME). Laminin (late cavitation blastocyst). antibody was purchased from ICN (Costa Mesa, CA), 9E10 hybrid- Embryo culture and microinjection. Preimplantation embryos oma was obtained from Dr. Bishop (Evan et al., 1985), and recombi- were cultured in serum-free microdrop culture under oil in CZB nant FGF-4 and anti-FGF-4 were gifts of Dr. Basilico (Rappolee et or KSOM media (Specialty Media, Lavallette, NJ). Postimplantation al., 1994). Polyclonal antisera to FGFR-3 and FGF-4 were purchased embryos were cultured in Ham’s F-12 with 4% FCS (Invitrogen from Santa Cruz Biotechnology (Santa Cruz, CA). low retinoic acid test) with or without 10 ng/ml recombinant FGF- DNA. pPGKneobPA (PGK, phosphoglycerate kinase; bPA, bo- 4. Approximately 0.5–1 pl of circularized DNA plasmids was mi- vine growth hormone polyadenylation site) was a gift from C. Stew- croinjected into the male pronucleus of single-cell zygotes or 0.5 art and b-actin lacZ was a gift from J. Cross (Stewart et al., 1992; pl into a nucleus of one blastomere in the two-cell zygotes by

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba FGF as an Essential Regulator of Mouse Embryo Fifth Cell Division 107

standard techniques (Hogan et al., 1994). In the circular form, ex- segments. Since the ICM is nearly always positive and we wanted pression is favored over integration (Brinster et al., 1985). The con- to score only trophectoderm cells, we scored segments one and centration of plasmid microinjected was 10 ng/ml of PGKdnFGFRmyc; eight as positive only if expression was detected in the outer extra- 20 ng/mlofb-actinlacZ; 10 ng/ml of PGKFGFR-4, PGKFGFR-3, and embryonic ectoderm cells. pPGKFGF-4; or a combination of 10 ng/ml PGKdnFGFRmyc and 20 Analysis of blastocyst outgrowths. To test the ability of em- ng/ml b-actin lacZ. Embryos were cultured as described above and bryos to form outgrowths, E3.5 embryos were cultured in Ham’s morphology was recorded daily at fixed times. Dead embryos were F-12 with 10% FCS with or without 10 ng/ml FGF-4 for 3 days. removed from the media and not included in the analysis. Approxi- Outgrowths were stained with Hoechst as above and photographed, mately 65% of all injected embryos progressed past microinjection and negatives were scanned using the Photoshop CD 3.0 program. through to compaction. During culture, some embryos microin- Some outgrowths were also analyzed for the number of laminin- jected with lacZ or dnFGFR did not progress beyond the two-cell positive cells. Sizes of outgrowths were measured with NIH Image stage. The transgenic embryos that did not progress beyond the 1.57. All statistics were done with the Biostatistics 3.0 program two-cell stage were not analyzed. After 1 day, embryos were exam- (Glantz, 1992) run on a Mac IIsi. ined under the microscope to detect the number of cells and then reincubated. After 2, 3, 4, 5. or 6 days in culture, embryos were processed for nuclear counts by fixation and Hoechst staining, as RESULTS well as for expression of lacZ, dnFGFR, FGFR-4, and FGFR-3 transgenes. Uninjected or lacZ same-stage embryos were used as nega- Effects of expression of the dnFGFR begin at the fifth tive controls. Hoechst-stained embryos were counted in serial Z- axis focal planes at the microscope in blinded experiments. All cell division and are cell autonomous. To assay for effects experiments were done at least twice. of loss of FGF signal input in Xenopus, a dnFGFR has been Immunocytochemistry. To develop fixed pPGKdnFGFR-myc, used previously (Amaya et al., 1991, 1993). This mutant pPGKFGFR-4, pPGKFGF-4, and pPGKFGFR-3-injected embryos, receptor specifically blocks all input of FGF signal when we used the 9E10 clone of mouse anti-human myc hybridoma (Evan expressed to a greater degree than the endogenous receptors et al., 1985) and anti-FGF-4, anti-FGFR-3 (Santa Cruz), and rabbit (Amaya et al., 1991). Due to the dimerization requirement -polyclonal anti-C؅ terminus FGFR-4 antibody as primary antibody, and the ability of all FGF receptors to heterodimerize (John respectively. Rabbits were immunized with the 14 amino acids at son and Williams, 1993), the function of all endogenous the C-terminus of the FGFR-4 and antibodies were purified using FGF receptors is blocked by this dnFGFR. We constructed the oligopeptide as previously described (Partanen et al. 1992). Im- a dnFGFR by completely removing the tyrosine kinase do- munocytochemistry was performed as described (Rappolee et al., main. The dnFGFR, FGFR-4, FGFR-3, and FGF-4 were ex- 1994), except that embryos were fixed for 30 min in fresh 4 % pressed using a strong constitutive promoter (PGK). We paraformaldehyde and permeabilized for 5 min with 0.25% Triton X-100. The anti-human myc supernatant was used neat and anti- added a human myc epitope tag to analyze the expression FGFR-4, anti-FGF-4, and anti-FGFR-3 were used at 1/400. Hoechst of the dnFGFR transgene. (10 mg/ml) was used as a counterstain. LacZ-expressing embryos Circularized plasmids injected at 1000–5000 copies per were developed in a microdrop assay under oil as previously de- embryo achieved transgenesis throughout the period of pre- scribed (Behringer et al., 1993). For standard diascopic and epifluo- implantation development (Fig. 1). Mosaic embryos pro- rescence micrography, a Nikon Microphot FXA microscope with a duced by this procedure are good tools for analyzing the 420-nm barrier filter and Walgreen’s 100 ISO print film were used. timing and location of FGF activity, since possible lethal Objectives were chosen to cover the field for pertinent detail at 4– events in the injected cells are not lethal to the embryo 201. Negatives were scanned into a Power PC computer using (Slack, 1991; Hogan et al., 1994). This approach has been Adobe Photoshop 3.0. used previously to show that the basic helix-loop-helix Assaying cell death in preimplantation embryos. For testing gene, Hxt, transdifferentiates cells into the extraembryonic cell death, pyknotic nuclei were counted after Hoechst staining as previously described (Hogan et al., 1994). For the trypan blue test, ectodermal cell lineage (Cross et al., 1995). embryos at each stage were shifted directly from medium to 0.4% We expressed control lacZ and dnFGFR genes in mosaic trypan blue solution for 10 min and blue cells counted under the embryos. Expression was detected within hours of micro- Zeiss (West Germany) inverted microscope. Same-stage embryos injection at the two-cell stage and continued throughout were fixed and used as positive controls. the culture of preimplantation embryos (Fig. 1). During Analysis of mosaic embryos for position of dnFGFR and lacZ- culture of the mosaic embryo, the total number of cells in expressing cells. Mosaic embryos were made for lacZ or dnFGFR the embryo was not affected during the first 2 days or the as described above. To lower the immunocytochemistry back- first four cell divisions (Table 1). However, during the third ground, lacZ was coinjected with dnFGFR and used to detect the day of embryo culture, or the fifth cell division, a signifi- position of these cells. Five days later embryos were processed for cantly lower cell number was detected in embryos express- lacZ expression, stained with Hoechst, and micrographed. The ing the dnFGFR. We tested whether the effects of the ICMs of the embryos were positioned at 12 o’clock by rotating dnFGFR continued more than 3 days after microinjection. micrographs of Hoechst-stained embryos in the Adobe Photoshop 3 application. To quantify the expression patterns of the transgenes There was a continuing significant deficit in cell number in the extraembryonic ectoderm, each blastocyst was overlaid with in dnFGFR-expressing embryos cultured for an additional a circle divided into eight 45Њ segments so that the top of the blasto- 2 days (Table 1). cysts bisected the line between segments one and eight. We scored We used a full-length FGFR-4 gene to test two parame- segments two to seven as positive if any cell was stained in those ters: whether the effects of overexpression of a nonmito-

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba 108 Chai et al.

dogenous FGFR-4 and patches of FGFR-4 overexpressing cells could be detected by immunocytochemical means (data not shown). Overexpression of FGFR-4 had no effect on the number of cells in mosaic blastocysts at the third day of culture (Table 1). When FGFR-4 was coinjected with the dnFGFR transgene at equimolar concentrations of plasmid, it rescued the mosaic embryo to produce the same number of cells as control embryos. These data suggest that the effects of the dnFGFR are reversible and specific to its activity in blocking endogenous FGF receptors. These data agree with previous mouse data (Wang et al., 1994), but contrast with the mitogenic effects of human FGFR-4 (Vainikka et al., 1994). To assay for the cell autonomous nature of the defect in cell division, embryos were analyzed for the number of cells expressing the lacZ or dnFGFR transgene. The embryos used for assaying total cell number (Table 1) were developed to detect activity of the lacZ transgene and expression of the myc-tagged dnFGFR. We assayed mosaic embryos at 2 and 3 days after microinjection, when the effect on cell division was first seen. During the second day of culture, the lacZ- and dnFGFR-expressing embryos were evenly divided into transgene-expressing cells and the progeny of uninjected cells (Table 2). By the third day, the cells expressing the dnFGFR were about half the number of the progeny of uninjected cells, but statistically not different in number than the previous day. Also, the number of cells in the progeny of uninjected cell was the same in the lacZ- and dnFGFR-expressing embryos on the third day. To assay for the accumulating effects of expression of dnFGFR, we analyzed the number of cells derived from the dnFGFR-injected cell for an additional 2 days. There continued to be no difference in the number of progeny of the uninjected cell in the dnFGFR- and lacZ-injected embryos. This suggests that the FIG. 1. Cells expressing dnFGFR and lacZ are detected through- effects of inhibition of FGF input are cell autonomous. out the period of culture. The expression of lacZ (A, C, E, G) and Endogenous FGF does not prevent cell death. To deter- dnFGFR (B, D, F, H) transgenes starts at the two-cell stage, soon mine if FGF acts to induce cell division, two parameters after microinjection (in one of the two cells) and continues for the were tested. First, it was determined that the number of first 3 days of culture. LacZ was detected by its enzymatic activity cells expressing the dominant negative in mosaic embryos and dnFGFR was using antibody to the myc epitope tag. Embryos did not change significantly after day 2 (Table 2). Second, we are not all to the same scale. I and J are uninjected embryos devel- assayed for cell death in the lacZ- and dnFGFR-expressing oped to detect the expression of lacZ or dnFGFR, respectively. Ar- mosaic embryos. Normalized to the total number of cells, rowheads show expressing patches and arrows show nonexpressing cells. C, days in culture. there was no significant difference in the fraction of pyknotic cells in the ICM after 5 days of culture of lacZ or dnFGFR transgenic embryos (lacZ: 3.7 { 0.5%, n Å 19; dnFGFR: 3.9 { 0.5%, n Å 20; P Å 0.8). Trypan blue was genic FGFR could inhibit cell division and whether the used to detect dead cells on the outside epithelium, the effects of the dnFGFR were reversible. The murine FGFR- extraembryonic ectoderm. At days 2–5 of culture, there 4 is weakly enzymatic, with little or no mitogenic or enzy- were no dead cells in embryos transgenic for dnFGFR (0 matic function in cell lines (Wang et al., 1994; Ornitz et dead cells in 67 embryos) or lacZ (0 dead cells in 71 em- al., 1992). FGFR-4 can phosphorylate itself and some sub- bryos). These data suggest that FGF is not required for the strates such as phospholipase C-g, but not as strongly as prevention of cell death in the preimplantation embryo. FGFR-1 (Wang et al., 1994). FGFR-4 mRNA and protein Division of trophoblast cells requires FGF. Previous are detected throughout the embryo at E3.5 (Rappolee et reports suggested that FGF-4 regulated survival of the ICM al., 1994; Jacobson and Rappolee, unpublished data). Using and the growth of endoderm that arises from it (Feldman et pPGKFGFR-4, we expressed levels of FGFR-4 protein in al., 1995; Rappolee et al., 1994). Our results suggest that mosaic embryos that were greater than the amount of en- FGF signaling controls cell division in the ICM and surpris-

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba FGF as an Essential Regulator of Mouse Embryo Fifth Cell Division 109

TABLE 1 Effects of the Expression of a Mutant and Wild-Type FGF Receptors in the Preimplantation Mouse Embryo

Total cell number/mosaic embryo Days after microinjection

Transgene 0 1 2 3 4 5

LacZ, total cellsa 2 { 0 6.8 { 0.1 19.4 { 0.5 36.1 { 0.9 57.9 { 3.0 78.9 { 4.3 No. embryos 156 135 37 68 26 25 dnFGFR total cells 2 { 0 6.9 { 0.1 18.8 { 0.5 26.9 { 0.9 42.6 { 3.6 58.4 { 4.9 No. embryos 169 129 (P Å 0.7b)41(PÅ0.4b)79(Põ0.001c)25(PÅ0.002c)24(PÅ0.003c) dnFGFR / FGFR-4 — — — 38.9 { 3.3 — — No. embryos 40 (P Å 0.8d) FGFR-4 alone — — — 34.1 { 2.0 — — No. embryos 23 (P Å 0.9e)

a These data were derived from seven sets of microinjections. Total counts are recorded as means { SEM. b At days 1 and 2, the total cell number in lacZ and dnFGFR mosaic embryos was not significantly different. c At days 3, 4, and 5, the total cell number in lacZ and dnFGFR mosaic embryos was significantly different. d,e At day 3, the total cell number in mosaic embryos expressing lacZ, dnFGFR / FGFR-4, and FGFR-4 alone was not significantly different.

ingly in extraembryonic ectoderm. If FGF is required for We made mosaic embryos as described above, oriented cell division in the extraembryonic ectoderm, other effects the ICM at the top of the micrographs (Fig. 2), and then are predicted. Mitotic trophoblasts that produce the extra- scored for the presence of transgenic cells in eight 45Њ seg- embryonic ectoderm are in contact with and adjacent to ments of the embryo. Segments one and eight included the the ICM (Gardner et al., 1973; Cruz and Pedersen, 1985; ICM and segments four and ﬁve were at the opposite abem- reviewed in Pedersen, 1987). Postmitotic progeny of tropho- bryonic pole. The tabulated data (Table 3) show that only blasts are pushed away from a position near the ICM by 1/54 of the dnFGFR-expressing mosaic embryos were posi- population pressure. If expression of the dnFGFR transgene tive in segments one or eight, whereas 46/50 of the lacZ- prevents cell division in these trophoblasts, then dnFGFR expressing embryos were positive in these segments. Extra- expression cannot endure in trophectoderm cells near the embryonic ectoderm cells in segments two to seven contain ICM in mosaic embryos. almost all dnFGFR-expressing cells, whereas segments ex-

TABLE 2 The Cell Autonomous Nature of FGF Signal Input in the Regulation of Cell Division

Cell number/patch in mosaic embryos

Days after microinjection

Transgene 2 3 4 5

LacZ, staineda 9.0 { 0.7 18.3 { 1.1 27.4 { 1.5 36.3 { 1.5 No. embryos 18 27 20 20 LacZ, unstained 8.3 { 0.7 (P Å 0.5b) 17.7 { 1.3 (P Å 0.5b) 29.9 { 1.6 (P Å 0.2b) 41.7 { 2.9 (P Å 0.3b) No. embryos 18 27 20 20 dnFGFR, stained 7.5 { 0.7 8.5 { 0.6 (P Å 0.6c) 9.5 { 0.6 (P Å 0.6c) 9.5 { 0.6 (P Å 0.6c) No. embryos 21 52 23 21 dnFGFR, unstained 9.0 { 0.8 (P Å 0.5d) 16.3 { 1.2 (P õ 0.001d) 33.2 { 1.5 (P õ 0.001d) 45.9 { 2.7 (P õ 0.001d) No. embryos 21 52 23 21

a These data were derived from experiments described in Table 1. After the total cell counts were done, embryos were stained for lacZ gene activity or for the presence of the myc tag on the dnFGFR. The number of cells are recorded as means { SEM. b At days 2, 3, 4, and 5, the number of unstained cells in lacZ and dnFGFR mosaic embryos was not significantly different. c At day 3, 4, and 5, the number of stained cells in dnFGFR mosaic embryos was not significantly different from day 2. d At days 3, 4, and 5, the number of stained cells in dnFGFR or lacZ mosaic embryos was significantly different. At day 2, the number of stained cells in dnFGFR or lacZ mosaic embryos was not significantly different.

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba 110 Chai et al.

FIG. 2. The location of the dnFGFR transgene can be detected in ICM cells and trophectoderm cells, but not in immediate progeny of mitotic trophoblast cells. LacZ (A, B) tends to be expressed (arrowhead) in the ectoderm near the ICM (arrow). dnFGFR (C, D) is expressed in the ectoderm away from the ICM (arrowhead) in mosaic embryos 5 days after microinjection. The dnFGFR is detected in the ICM (arrow). The location of transgene expression was detected by lacZ enzymatic activity.

pressing lacZ are equally distributed with a tendency to entiate. To test whether expression of dnFGFR affects differ- segments one, two, seven, and eight. These results suggest entiation, we microinjected transgenes for dnFGFR, or con- that dnFGFR-expressing extraembryonic ectodermal cells trol neo or lacZ, into the pronuclei of fertilized one-cell become postmitotic and do not remain near the ICM. zygotes to produce expression in every cell in the blastocyst Expression of a dnFGFR does not affect early embryonic (Fig. 3, upper panel). Typically, 90–95% of the microin- differentiation events. The expression of the dnFGFR jected embryos expressed transgene in every cell (data not might cause cells to cease division by causing them to differ- shown). We detected no difference between dnFGFR- or

TABLE 3 The Location of Cells Expressing the dnFGFR in Mosaic Embryos

Number of embryos having at least one cell in a given segment Segment

1 8 Transgene (near ICM) 234567(near ICM) Sample size

LacZ 45 40 28 29 28 37 43 42 50 dnFGFR 1 22 31 37 35 30 21 1 54

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba FGF as an Essential Regulator of Mouse Embryo Fifth Cell Division 111

Hogan et al., 1994). Cavitation occurs at the time that the expression of the dnFGFR results in cessation of cell division. As in the mosaic embryos, no difference in total cell number between lacZ- and dnFGFR-expressing cells was seen 2 days after microinjection (data not shown). However, by the sixth day of this culture period, the embryos transgenic for dnFGFR had about one-third of the number of cells as embryos transgenic for lacZ (Table 4). As in the mosaic embryos, the cessation of cell growth occurred after the fourth cell division. However, the morphology and volume (Fig. 3, lower panel; lacZ: 74.7 { 6.3 arbitrary units3 vs dnFGFR: 70.8 { 5.0 arbitrary units3; P Å 0.6, two experiments, n Å 63) of the embryo at the end of cavitation were the same in dnFGFR- and lacZ-expressing embryos. Em- bryos expressing the dnFGFR in all cells developed ICMs at high frequency (Fig. 3, lower panel), suggesting there was no qualitative requirement of endogenous FGF signaling on allocation to the ICM. These data suggest that early differentiation events oc- curring at the time of the first FGF-mediated cell growth are not affected by the loss of FGF signal input. We also tested for effects of dnFGFR on later differentiation events. The dnFGFR-expressing embryos had reduced hatching from the zona pellucida, did not attach to fibronectin-coated substrates, and had no outgrowth of extraembryonic ectoderm cells (Table 4, day 7). In similar experiments using different media conditions (CZB cf KSOM) and a different strain of mice (CD-1 cf MF-1), no dnFGFR-expressing embryos attached and grew out (0/58), but 35% (22/63) of control embryos transgenic for neo attached and grew out. These data suggest that cavitation is not affected by the expression of the dnFGF receptor. Although cells expressing the dnFGFR stopped dividing in the fifth cell division, morphogenetic events proceeded at a rate that was the same as control embryos that expressed lacZ. Later differentiation events, attachment and outgrowth, were affected by expression of the dnFGFR transgene. FGFR-3 overexpression increases the number of cells in FIG. 3. Preimplantation embryos expressing the dnFGFR in all the blastocyst. FGF-4 protein added outside cultured pre- cells differentiate normally, but produce fewer cells than embryos implantation embryos had no effect on cell number (Rap- expressing lacZ. In the upper panel, a characteristic blastocyst ex- polee et al., 1994), but expression of the dnFGFR suggests presses lacZ (A) in every cell as detected by the enzymatic activity of lacZ. Phase contrast of the same blastocyst (B). In the lower that endogenous FGF signaling is required for cell division panel, Hoechst-stained dnFGFR-expressing blastocysts (C) have at the fifth cell division. To reconcile these finding and to fewer cells than lac Z-expressing embryos (A) (see cell counts in test for the continuing influence of FGF after the fifth cell Table 4), but the morphology of dnFGFR-expressing embryos (D) division, embryos were tested for cell number after FGFR- and lacZ-expressing embryos (B) appears the same. Embryos were 3 or FGF-4 transgenes were overexpressed alone or together injected at 0.5 days after fertilization (E0.5). All embryos were ob- in every cell in the preimplantation mouse embryo. served at 6C. C, days in culture. Note that nuclei in the embryos FGFR-3 mRNA and protein are detected in all cells of the with low cell number (dnFGFR-expressing embryos) appear less E3.5 blastocyst at high levels (Rappolee et al., 1994; Jacob- bright due to spreading of fewer nuclei over a similar volume. In son and Rappolee, unpublished data). It is enzymatically A and C, arrows show ICMs. active and weakly mitogenic and is activated by FGF-4 (Or- nitz et al., 1992, 1996). The overexpression of FGFR-3 in all cells of the early embryo increases the number of cells lacZ-injected transgenic embryos in the initiation, timing, significantly 6 days after microinjection (Table 5), sug- or progression of embryo development of compaction or gesting that enzymatically active FGF receptors can have cavitation (Table 4, days 3–6), two early differentiation significant effects on cell number in the preimplantation events in mouse development (Rappolee and Werb, 1994; mouse embryo. However, overexpression of FGF-4 had no

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba 112 Chai et al.

TABLE 4 Timing and Number of dnFGFR-Expressing Embryos Undergoing Compaction, Cavitation, and Outgrowth

Cumulative number of embryos at given stages of differentiation Days after microinjection

Transgene 3 4 5 6 7

LacZa Compacted 7 38 0 0 Cavitated 0 9 47 47 Total cell number (No. embryos) 72.8 { 6.2 (28) Outgrowing (total No. embryos) 13 (19) dnFGFR Compacted 9 33 0 0 Cavitated 0 10 43 43 Total cell number (No. embryos) 24.1 { 1.3 (29) Outgrowing (total No. embryos) P õ 0.001b 0 (14)

a In two experiments, 94 embryos were microinjected at day 0 and cultured in KSOM for 4 days. The number of embryos that compacted and cavitated was assayed at days 3–6. At day 4, 19 lacZ- and 14 dnFGFR-expressing embryos were removed and cultured in Ham’s F-12 with 10% FCS to obtain outgrowths. At day 6, the embryos remaining in KSOM were ﬁxed and assayed for cell number. Four embryos were lost during handling or degenerated between days 3 and 6 in culture. b The differences in the mean number of cells in blastocysts microinjected with lacZ and dnFGFR were highly signiﬁcant as assayed by Student’s t test.

effect on cell number when injected alone. Also, FGF-4 did the same as FGFR-3. The coinjection of FGFR-3 and FGF- not increase the number of cells when coinjected with 4 resulted in signiﬁcant increases in cell numbers 5 and 6 FGFR-3 compared with embryos injected with FGFR-3 days after injection. alone. In a dose response, the concentration of FGFR-3 in- Taken together, these data suggest that the mouse em- jected was found to be maximal for stimulating cell number bryo continues to be sensitive to FGF after the ﬁfth cell increases (data not shown). The concentration of FGF-4 was division. In mediating cell division, FGF ligands appear to

TABLE 5 The Effects of FGFR-3 Alone and Together with FGF-4 in the Preimplantation Mouse Embryo

Days after microinjection

Transgenea 56

LacZ Total cell number (No. embryos) — 71.4 { 2.2 (23) FGF-4 Total cell number (No. embryos) — 74.6 { 2.2 (29) (P Å 0.306b) LacZ Total cell number (No. embryos) — 72.2 { 2.3 (23) FGFR-3 Total cell number (No. embryos) — 82.4 { 2.7 (26) (P Å 0.007b) LacZ Total cell number (No. embryos) 46.5 { 2.0 (23) 68.0 { 1.7 (26) FGFR-3 / FGF-4 Total cell number (No. embryos) 54.3 { 2.6 (21) (P Å 0.006b) 79.8 { 2.3 (25) (P õ 0.001b)(PÅ0.469c)

a All experiments were done twice and the results shown are cumulative. b The differences in the mean number of cells in blastocysts microinjected with lacZ and FGFR-3 or lacZ and FGFR-3 / FGF-4 were significant as assayed by Student’s t test, but the difference between lacZ- and FGF-4-expressing embryos was not significant. c The difference in the mean number of cells in blastocysts microinjected with FGFR-3 and FGFR / FGF-4 was not significant.

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba FGF as an Essential Regulator of Mouse Embryo Fifth Cell Division 113

In this report, we used a dnFGFR to show that endogenous FGF signaling is required for the cell division of extraembryonic and embryonic cells at the fifth cell division of the mouse embryo. The effect was cell autonomous since the cell division of progeny of uninjected cells in mosaic embryos was unaffected. At the fifth cell division, the two affected mitotic cell populations were responsible for formation of all embryonic and extraembryonic cell lineages. FGF was not required for prevention of cell death in the embryos, but cells expressing dnFGFR went into cell division stasis. The static position in the cell cycle of the dnFGFR-expressing cells is being studied. FGF is not required for essential early differentiation events. FGF was not required for compaction, the first essential differentiation event in the mouse. Growth factors have not been implicated in the mechanism of compaction (Rappolee, 1997). FGF was also not required for initiation or maintenance of cavitation, since cavitation initiates soon after the cells expressing the dnFGFR stopped dividing. Cav- itation can be regulated in vitro by addition of epidermal growth factor (Dardik and Schultz, 1991), but not by addition of FGF-4 (Rappolee et al., 1994; unpublished data). FGF was not required for the initial allocation of ICM FIG. 4. Normal blastocysts grow out more extraembryonic ecto- cells as embryos expressing dnFGFR in all cells developed derm when recombinant FGF-4 is present. (A) In culture with media ICMs. These data suggest that the cessation of cell cycle alone (37.7 { 3.7 outgrowing cells) and with FGF-4 (48.7 { 4.9 does not immediately lead to differentiation of ICM into cells; P Å 0.01). (B) The outgrowths cover a surface area that is endoderm or trophectoderm cells. These studies require proportional to the number of nuclei in the outgrowth. For media testing with markers for each cell lineage and differentia- alone (909 { 37.7 arbitrary units of area) and for FGF-4 (1,604 { tion state to find if FGF mediates quantitative effects. 53.5 arbitrary units of area, P Å 0.01). Other differentiation events that occurred after cavitation appeared to be blocked by blocking FGF signal input. In our studies, blocking FGF input resulted in prevention of attachment and outgrowth of the blastocyst in vitro. How- be saturating and the concentration of endogenous FGF re- ever, the dnFGFR-expressing embryos ceased cell division ceptors is limiting. earlier at the time of cavitation and the effects on later FGF-4 increases outgrowth of extraembryonic ectoderm differentiation events may be indirect. in blastocyst outgrowths. Results presented here suggest The role of FGF in mediating division of extraembryonic that FGF input is required for cell division in extraembry- ectoderm cells. The ICM induces the cell division of adja- onic ectoderm in preimplantation embryos. Can FGF con- cent trophoblasts (Gardner et al., 1973). Transplanted ICMs tinue to influence the growth of extraembryonic ectoderm induce ectopic cell division in extraembryonic ectoderm in after implantation? To test this, FGF-4 protein was added recipient blastocysts. Extraembryonic cells next to the ICM to blastocyst outgrowth culture. FGF-4 increased the num- are mitotic and cells injected with tracers produce postmi- ber of outgrowing cells determined to be trophectoderm totic progeny that are displaced from the pole of the embryo by morphological (had large nuclei) and antigenic (lacked containing the ICM (Copp et al., 1978, 1979; Cruz and Ped- expression of laminin) criteria. The increase in number of ersen, 1985). Our results suggest that FGF is required for cell trophectoderm cells was proportional to the increase in the division in the extraembryonic ectoderm and that increased size of the outgrowth (Fig. 4), suggesting that FGF-4 can act FGF-4 can increase the number of trophectoderm cells. as a growth factor, but does not cause scattering of the ICM-derived FGF-4 may be required for this cell division of trophectoderm cells. These data indicate that FGF-4 contin- polar trophoblast cells. If so, reexamination of the fgf4 null ues to regulate the division of extraembryonic ectoderm mutants may detect a failure of cell division in these cells. cells in cultured embryos after the fifth cell division and Phenotypes of FGF mutants in early mouse development. after implantation. The phenotype of embryos expressing the dnFGFR occurs earlier than the fgf4 null mutant. In the fgf4 null mutant, DISCUSSION the blastocyst appeared normal but degenerated in utero after implantation began. In vitro, the ICM degenerated first FGF mediates cell division by a positive, cell autono- and the endoderm rising from the ICM apparently never mous regulatory signal required at the fifth cell division. formed. Conversely, in the dnFGFR-expressing embryo, the

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba 114 Chai et al. embryo stopped dividing before implantation, although the This implies a role for FGF in embryonic regulation of im- morphology of the embryo appeared normal. It is likely that plantation and the regulation of cell cycle. maternal FGF-4 mRNA transcripts (Rappolee et al., 1994) and protein sustain the fgf4 null embryos and the lethal phenotype occurs when these molecules are depleted. Other ACKNOWLEDGMENTS FGF ligands, transiently expressed in the preimplantation embryo, may prevent preimplantation lethality in the fgf-4 null embryos. In contrast, expression of a dnFGFR inhibits We thank Drs. Daniel Linzer, Elizabeth Puscheck, Eugene Perga- signal transduction by maternal or zygotic FGF ligands and ment, Linda Chen, and Zena Werb for discussion of the manuscript. We thank Claudio Basilico for recombinant FGF-4 and the Develop- receptors. mental Studies Hybridoma Bank for SSEA-1 antibodies. We thank Some differences in the phenotypes in the current report Sheri Dewan, Michelle Kim, Soldrea Roberts, and Anjali Adukia and the fgf4 null study may be due to culture conditions or for excellent technical assistance and Karen Anderson and Crystal different genetic backgrounds. Examples of these differ- Beecher for editorial assistance. This research was supported by ences have been reported. Culture conditions can have dis- grants from NIH (NICHD HD30739-02), the Feinberg Cardiovascu- tinct effects on the phenotype of mouse embryos (Lawitts lar Research Institute at Northwestern University Medical School, and Biggers, 1993; Rappolee and Werb, 1994) and the EGF ACS Illinois Division (95-10), the American Heart Association receptor null is lethal on one genetic background at the (96014080), and the Markey Charitable Trust (D.A.R.). The Devel- peri-implantation stage (Threadgill et al., 1995), but not on opmental Studies Hybridoma Bank is maintained by the Depart- two other backgrounds. ment of Pharmacology and Molecular Sciences, Johns Hopkins Uni- versity, and the Department of Biological Sciences, University of FGF signaling mediates an embryonic checkpoint for im- Iowa, under Contract N01-HD-2-3144. plantation. Intact peri-implantation mouse embryos do not respond to exogenous FGF. Preimplantation mouse embryos cultured from the two-cell stage to the blastocyst stage do not undergo cell division faster or cavitate faster REFERENCES when FGF-4 is added (Rappolee et al., 1994). However, other growth factors such as IGF-II (Rappolee, et al., 1992), CSF- Amaya, E., Musci, T. J., and Kirschner, M. W. (1991). Expression of 1 (Bhatnagar et al., 1995), and TGF-a/EGF (Dardik and a dominant negative mutant of the FGF receptor disrupts meso- Schultz., 1991) cause increases in the rate of cell division derm formation in Xenopus embryos. Cell 66, 257–270. or cavitation over the same period under similar conditions. Amaya, E., Stein, P. A., Musci, T. J., and Kirschner, M. W. (1993). These data suggest that preimplantation embryos are nonre- FGF signalling in the early specification of mesoderm in Xeno- sponsive specifically to the exogenous form of FGF. This pus. Development 118, 477–487. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, may be due to the saturation of receptors by endogenous G. J., Smith, J. A., and Struhl, K. (1993). ‘‘Current Protocols in FGF-4 reported here. Postimplantation embryos also do not Molecular Biology.’’ Wiley, NY. respond to exogenous FGF. Despite the expression of FGF- Basilico, C., and Moscatelli, D. (1992). The FGF family of growth 2 in the uterine epithelium near the site of blastocyst im- factors and oncogenes. Adv. Cancer Res. 59, 115–165. plantation and the expression of FGF in the uterine stroma Behringer, R. R., Crotty, D. A., Tennyson, V. M., Brinster, R. L., and fluids (Wordinger et al., 1992; Carlone and Rider, 1993: Palmiter, R. D., and Wolgemuth, D. J. (1993). Sequences 5؅ of the Brigstock et al., 1990), the fgf4 null embryos degenerate homeobox of the Hox-1.4 gene direct tissue-specific expression quickly after implantation (Feldman et al., 1995). Since of lacZ during mouse development. Development 117, 823–833. FGF-2 and FGF-4 have nearly identical binding specificities Bhatnagar, P., Papaioannou, V. E., and Biggers, J. D. (1995). CSF-1 for the alternate splice forms of all FGF receptors (Ornitz, and mouse preimplantation development in vitro. Development. 121, 1333–1339. et al., 1996), the failure of maternal FGF-2 to rescue the fgf4 Brigstock, D. R., Heap, R. B., Barker, P. J., and Brown, K. D. (1990). null is not due to differential binding to the embryo, but to Purification and characterization of heparin binding factors from exclusion of external FGF. However, when blastocysts or porcine uterus. Biochem. J. 266, 273–282. ICM cells isolated from blastocysts are allowed to adhere Brinster, R. L., Chen, H. Y., Trumbauer, M. E., Yagle, M. K., and to fibronectin and grow out, disrupting the intact embryo, Palmiter, R. D. (1985). Factors affecting the efficiency of intro- FGF-4 has effects upon the embryo (Rappolee et al., 1994; ducing foreign DNA into mice by microinjecting eggs. Proc. Natl. Feldman et al., 1995). Since the synthesis of embryonic Acad. Sci. USA. 82, 4438–4442. FGF-4 is localized in the ICM (Niswander and Martin, 1992; Carlone, D. L., and Rider V. (1993). Embryonic modulation of basic Rappolee et al., 1994), this ligand may only signal cells fibroblast growth factor in the rat uterus. Biol. Reprod. 49, 653– within the embryo. These data suggest that the peri-implan- 665. Copp, A. J. (1978). Interaction between inner cell mass and trophec- tation embryo measures the amount of endogenous FGF in toderm of the mouse blastocyst. I. A study of cellular prolifera- isolation from maternal FGF. tion. J. Embryol. Exp. Morphol. 48, 109–125. The data presented here have a wide range of implica- Copp, A. J. (1979). Interaction between inner cell mass and trophec- tions. The embryo requires positive regulation of cell divi- toderm of the mouse blastocyst. II. The fate of the polar trophec- sion by FGF signaling early in development. FGF is the first toderm. J. Embryol. Exp. Morphol. 51, 109–120. growth factor found that is required before implantation. Cross, J. C., Werb, Z., and Fisher, S. J. (1994). Implantation and the

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba FGF as an Essential Regulator of Mouse Embryo Fifth Cell Division 115

placenta: Key pieces to the development puzzle. Science 266, lian Embryonic Development’’ (J. Rossant and R. A. Pedersen, 1508–1518. Eds.), pp. 3–33. Cambridge Univ. Press, Cambridge. Cross, J. C., Flannery, M. L., Blanar, M. A., Steingrimsson, E., Jen- Rappolee, D. A., Brenner, C. A., Schultz, R., Mark, D., and Werb, kins, N. A., Copeland, N. G., Rutter, W. J., and Werb, Z. (1995). Z. (1988). Developmental expression of PDGF, TGF-alpha, and Hxt encodes a basic helix-loop-helix transcription factor that reg- TGF-beta genes in preimplantation mouse embryos. Science 241, ulates trophoblast cell development. Development 121, 2513– 1823–1825. 2523. Rappolee, D. A., Schultz, G. A., Pedersen, R. A., Sturm, K. A., and Cruz, Y. P., and Pedersen, R. A. (1985). Cell fate in the polar troph- Werb, Z. (1992). Insulin-like growth factor II acts through an ectoderm of mouse blastocysts as studied by microinjection of endogenous growth pathway regulated by imprinting in early cell lineage tracers. Dev. Biol. 112, 73–83. mouse embryos. Genes Dev. 6, 939–952. Dardik, A., and Schultz, R. M. (1991). Blastocoel expansion in the Rappolee, D. A., and Werb, Z. (1994). The role of growth factors in preimplantation mouse embryo: Stimulatory effect of TGF-a and mammalian pregastrulation development. Adv. Dev. Biol. 3, 41– EGF. Development 113, 919–930. 71. Evan, G. I., Lewis, G. K., Ramsay, G., and Bishop, J. M. (1985). Isola- Rappolee, D. A., Basilico, C., Patel, Y., and Werb, Z. (1994). Expres- tion of monoclonal antibodies specific for human c-myc proto- sion and function of FGF-4 in peri-implantation development in mouse embryos. Development 120, 2259–2269. oncogene product. Mol. Cell. Biol. 10, 3610–3616. Rappolee, D. A. (1998). Growth factors and hormones in mamma- Feldman, B., Poueymirou, W., Papaioannou, V. E., DeChiara, T. M., lian development. Growth factors in the mammalian pre- and and Goldfarb, M. (1995). Requirement of FGF-4 for postimplanta- post-implantation embryo. In ‘‘Hormones and Growth Factors tion mouse development. Science 267, 246–249. in Development and Neoplasia (R. Dickson and D. Salomon, Gardner, R. L., Papaioannou, V. E., and Barton, S. C. (1973). Origin Eds.). Wiley, NY of the ectoplacental cone and secondary giant cells in mouse Slack, J. M. W. (1991). ‘‘From Egg to Embryo. Regional Specification blastocysts reconstituted from isolated trophoblast and inner cell in Early Development,’’ pp. 171–195. Cambridge Univ. Press, mass. J. Embryol. Exp. Morphol. 30, 561–572. Cambridge, UK. Glantz, S. A. (1992). ‘‘Primer of Biostatistics: The Program. Version Smallwood, P. M., Munoz-Sanjuan, I., Tong, P., Macke, J. P., 3.0.’’ McGraw-Hill, New York. Hendry, S. H., Gilbert, D. J., Copeland, N. G., Jenkins, N. A., and Hebert, J. M., Basilico, C., Goldfarb, M., Haub, O., and Martin, Nathans, J. (1996). Fibroblast growth factor (FGF) homologous G. (1990). Isolation of cDNAs encoding four mouse FGF family factors: New members of the FGF family implicated in nervous members and characterization of their expression patterns during system development. Proc. Natl. Acad. Sci. USA 93, 9850–9857. embryogenesis. Dev. Biol. 138, 454–463. Stark, K. L., McMahon, J. A., and McMahon, A. P. (1991). FGFR-4 Hogan, B., Beddington, R., Constantini, F., and Lacy, B. (1994). ‘‘Ma- and a new member of the FGF receptor family, expressed in the nipulating the Mouse Embryo: A Laboratory Manual,’’ 2nd ed. definitive endoderm and skeletal muscle lineages of the mouse. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Development 113, 641–651. Johnson, D. E., and Williams, L. T. (1993). Structural and functional Stewart, C. L., Kaspar, P., Brunet, L. J., Bhatt, H., Gadi, I., Kontgen, diversity in the FGF receptor multigene family. Adv. Cancer Res. F., and Abbondanzo, S. J. (1992). Blastocyst implantation depends 60, 1–46. on maternal expression of leukaemia inhibitory factor. Nature Lawitts, J. A., and Biggers, J. D. (1993). In ‘‘Guide to Techniques in 359, 76–79. Mouse Development’’: (P. M. Wassarman and M. L. De Pam- Threadgill, D. W., Dlugosz, J., Hansen, L. A., Tennenbaum, T., Lichti, U., LaMantia, C., Mourton, T., Herrup, K., Harris, R. C., philis, Eds.). Academic Press, San Diego. Barnard, J. A., Yuspa, S. H., Coffey, R. J., and Magnuson, T. (1995). McBurney, M. W., Sutherland, L. C., Adra, C. N., Leclair, B., Rud- Targeted disruption of mouse EGF receptor: Effect of genetic nicki, M. A., and Jardine, K. (1991). The mouse Pgk-1 gene pro- background on mutant phenotype. Science 269, 230–234. moter contains an upstream activator sequence. Nucleic Acids Vainikka, S., Joukov, V., Wennstrom, S., Bergman, M., Pelicci, Res. 19, 5755–5761. P. G., and Alitalo, K. (1994). Signal transduction by fibroblast Niswander, L., and Martin, G. R. (1992). Fgf-4 expression during growth factor receptor-4 (FGFR-4). J. Biol. Chem. 269, 18320– gastrulation, myogenesis, and limb and tooth development in the 18326. mouse. Development 114, 755–768. Wang, J-K., Gao, G., and Goldfarb, M. (1994). Fibroblast growth Ornitz, D. M., and Leder, P. (1992). Ligand specificity and heparin factor receptors have different signaling and mitogenic poten- dependence of fibroblast growth factor receptors 1 and 3. J. Biol. tials. Mol. Cell. Biol. 14, 181–184. Chem. 267, 16305–16311. Wordinger, R. J., Moss, A. E., Gray, D., Chang, I. F., and Jackson, Ornitz, D. M., Xu, J., Colvin, J. S., McEwen, D. G., MacArthur, T. L. (1992). Immunohistochemical localization of basic fibro- C. A., Coulier, F., Gao, G., and Goldarb, M. (1996). Receptor spec- blast growth factor within the mouse uterus. J. Reprod. Fertil. ificity of the fibroblast growth factor family. J. Biol. Chem. 271, 96, 141–152 15292–15297. Yee, S. P., and Rigby, P. W. (1993). The regulation of myogenin Partanen, J., Vainikka, S., Korhonen, J., Armstrong, E., and Alitalo, expression during the embryonic development of the mouse. K. (1992). Diverse receptors for FGFs. Prog. Growth Factor Res. Genes Dev. 7, 1277–1289. 4, 69–83. Received for publication December 11, 1997 Pedersen, R. A. (1987). In ‘‘Experimental Approaches to Mamma- Accepted January 21, 1998

AID DB 8858 / 6x3a$$$101 05-18-98 21:56:03 dba