Placental overgrowth and fertility defects in mice with a hypermorphic allele of epidermal growth factor receptor

Abstract

Epidermal growth factor receptor (EGFR) is a member of the ERBB family of receptor tyrosine kinases that has been shown to play an important developmental and physiologic role in many aspects of pregnancy. We have previously shown in mice that Egfr^tm1Mag nullizygous placentas have fewer proliferative trophoblasts than wild-type and exhibit strain-specific defects in the spongiotrophoblast and labyrinth layers. In this study we used mice with the hypermorphic Egfr^Dsk5 allele to study the effects of increased levels of EGFR signaling on placental development. On three genetic backgrounds, heterozygosity for Egfr^Dsk5 resulted in larger placental size with a more prominent spongiotrophoblast layer and increased expression of glycogen cell-specific genes. The C3HeB/FeJ strain showed additional placental enlargement of Egfr^Dsk5 homozygotes with a significant number of homozygous embryos dying prior to 15.5 days post-coitus (dpc). We also observed strain-specific subfertility in Egfr^Dsk5 heterozygous females and pregnancy loss was dependent on maternal factors rather than embryo genotype. Higher levels of phospho-EGFR were detected in the uterus of Egfr^Dsk5 heterozygotes but the structure of Egfr^Dsk5 heterozygous nonpregnant uteri appeared similar to wild-type. Collectively, our results demonstrate that mice with increased levels of EGFR signaling exhibit an extensive level of genetic background-dependent phenotypic variability. In addition, EGFR promotes growth of the placental spongiotrophoblast layer in mice, and EGFR expressed in the uterine stroma may play an underappreciated role in preparation of the uterus for embryo implantation.

Introduction

ERBB family receptor tyrosine kinases are critical mediators of cell signaling in a broad range of developmental and physiologic processes. Epidermal growth factor receptor (EGFR), in particular, plays a role in many aspects of female reproduction and pregnancy. Female mice that are homozygous for a hypomorphic allele of Egfr, Egfr^wa2, exhibit impaired lactation as well as delayed puberty, and three ligands that bind EGFR, Epiregulin (EREG), Amphiregulin (AREG), and Betacellulin (BTC), stimulate in vivo oocyte maturation and cumulus expansion via EGFR activation in ovarian follicles (Fowler et al. 1995; Hsieh et al. 2007; Park et al. 2004; Prevot et al. 2005). In addition, EGFR is expressed in the uterine stroma where it regulates not only uterine development but also embryo implantation (Das et al. 1994; Tong et al. 1996). Uterine grafts derived from Egfr^tm1Mag nullizygous pups develop smaller grafts compared to wild-type and proliferative response to estradiol is diminished in Egfr^tm1Mag nullizygous uterine stroma but not epithelium (Hom et al. 1998). During implantation, expression of EGFR and ligands is observed in the uterus at the site of blastocyst attachment as well as on the surface of the implanting blastocyst (Das et al. 1994; Tong et al. 1996). Although there is redundancy in uterine expression of EGFR ligands, heparin-binding EGF-like growth factor (HBEGF) may play a unique and essential role since Hbegf-null female mice are subfertile due to partial implantation failure (Xie et al. 2007). Finally, EGFR regulates growth and differentiation of the placenta. Egfr^tm1Mag nullizygous embryos exhibit strain-dependent placental defects that range from minor reduction of the spongiotrophoblast layer to severe labyrinth dysmorphogenesis (Sibilia and Wagner 1995; Threadgill et al. 1997). The consequences of increased EGFR signaling in placental development and reproduction can now be determined in mice using a hypermorphic EGFR allele, Egfr^Dsk5 (Dark skin-5 allele).

The Egfr^Dsk5 allele was originally discovered during an N-ethyl-N-nitrosourea (ENU)-mutagenesis screen for visible dominant mutations (Fitch et al. 2003). Egfr^Dsk5 heterozygous and homozygous mice on the C3HeB/FeJ (C3H) background exhibit hyperpigmented footpads, long nails, wavy hair, and a thickened epidermis. The Egfr^Dsk5 mutation was molecularly identified as a Leu863Gln substitution within a region of the kinase domain important for stabilization of the receptor activation loop. When crossed to mice heterozygous for the Egfr^wa2 hypomorphic allele, compound heterozygous mice are wild type in appearance, suggesting that Egfr^Dsk5 is a gain-of-function allele that causes increased levels of EGFR signaling. Livers from Egfr^Dsk5 heterozygous mice have significantly lower levels of total EGFR compared to wild-type livers and Egfr^Dsk5 homozygous livers show a further reduction in total EGFR. However, both Egfr^Dsk5 heterozygous and homozygous livers have a larger proportion of phosphorylated EGFR compared to livers from wild-type mice. These data demonstrate that a negative feedback mechanism limits signaling in Egfr^Dsk5 liver, and possibly other tissues, by downregulating EGFR protein. The hypermorphic Egfr^Dsk5 allele may also be relevant to cancer since the human EGFR L861Q mutation, equivalent to mouse L863Q, was identified in gefitinib-responsive non-small-cell lung tumors (Lynch et al. 2004). When transfected into 32D cells, the L861Q form of EGFR exhibits ligand-independent phosphorylation and escapes ligand-induced receptor downregulation through a mechanism that involves receptor binding of HSP90 (Yang et al. 2006). Although there has been limited physiologic characterization of the Egfr^Dsk5 allele, there have been no reports of lung tumors or increased incidence of any other cancers in Egfr^Dsk5 heterozygous or homozygous mice. This is some-what surprising considering mice homozygous for the hypomorphic Egfr^wa2 allele show a reduction in tumors when crossed to numerous cancer models, including those of the mammary, colon, and skin (Gillgrass et al. 2003; Roberts et al. 2002; Sibilia et al. 2000).

The present study uses Egfr^Dsk5 mice to determine if increased EGFR signaling affects the development of the placenta. We utilized three genetic backgrounds to identify strain-dependent phenotypes related to the Egfr^Dsk5 mutation since genetic background influences phenotypes in mice with null or hypomorphic Egfr alleles. In addition to the original isogenic C3H strain on which Egfr^Dsk5 was generated, we backcrossed the allele to two additional genetic backgrounds, C57BL/6J (B6) and 129S1/SvImJ (129). We report that placental weight is increased in Egfr^Dsk5 heterozygotes and homozygotes on all three backgrounds. The larger placenta does not affect embryonic growth but is accompanied by strain-specific embryonic lethality of some Egfr^Dsk5 homozygotes before 15.5 dpc. In addition, we identified a strain-specific fertility defect in Egfr^Dsk5 heterozygous females and found that older female Egfr^Dsk5 heterozygotes frequently exhibited additional reproductive phenotypes of the uterus and ovary.

Materials and methods

Mice and genetic crosses

The Egfr^Dsk5 allele was generated by random mutagenesis with ENU as previously described and maintained isogenic on the C3H background (Fitch et al. 2003). B6 and 129-Egfr^Dsk5 congenic mice were generated by backcrossing C3H-Egfr^Dsk5 heterozygous stocks to B6 and 129 wild-type strains for ten or more generations. Congenic Egfr^Dsk5 heterozygous mice were then intercrossed to produce litters from each background containing wild-type, Egfr^Dsk5 heterozygous, and Egfr^Dsk5 homozygous congenic embryos and pups. Mice were fed Purina Mills Lab Diet 5058 or 5010 and water ad libitum under specific pathogen-free conditions in an American Association for the Accreditation of Lab Animal Care-approved facility. All experiments were approved by an Institutional Animal Care and Use Committee.

Genotyping

DNA was extracted from adult ear punches or embryo tail biopsies for genotyping by incubating at 95°C in 100 μl of 25 mM NaOH/0.2 mM EDTA for 20 min and then neutralizing with 100 μl of 40 mM Tris-HCl, pH 5.0. For the subsequent genotyping reactions, 1 μl of lysed tissue sample was used per reaction.

Egfr^Dsk5 allele was amplified by PCR with the following primers: DskF, 5′-AGATGGTTCACTCCCTCACG-3′ and DskR, 5′-ATGCTTCCTGATCTACTCCC-3′ (Qiagen, Valencia, CA). PCR conditions were 40 cycles at 94°C for 20 s, 62°C for 20 s, and 72°C for 60 s. PCR products were digested for 3 h at 37°C with AluI and Restriction Enzyme Buffer 2 (NEB) and run on a 3% agarose gel to separate a 220-bp product corresponding to wild-type Egfr and a 150- and 70-bp set of products corresponding to the digested Egfr^Dsk5 allele.

Collection of placenta and uterus samples

Noon on the day that copulation plugs were observed was designated as 0.5 dpc. Pregnant females were euthanized by exposure to a lethal dose of isoflourane and embryos with their corresponding placentas dissected from the uterine horns on the morning of 15.5 dpc or 18.5 dpc into phosphate-buffered saline (PBS). The placenta and extra-embryonic tissues were separated from the embryo by mechanical dissection and a tail biopsy was collected for DNA extraction to determine the genotype of each embryo. Wet weights of embryos and placentas were recorded at the time of dissection. Placentas were preserved in RNAlater (Ambion, Austin, TX) for extraction of RNA or fixed in 10% NBF (neutral-buffered formalin) for histologic analysis. Uteri were collected from nonpregnant virgin mice approximately 3 months of age and fixed in 10% NBF.

Histology

After fixing placentas and uteri in 10% NBF overnight, tissues were washed in PBS, dehydrated in a graded series of ethanols and xylenes, and embedded in paraffin. Seven-micron sections were cut using a Leica RM2165 microtome. Sections were deparaffinized, rehydrated in a graded series of ethanols, and stained with hematoxylin and eosin (H&E) or Periodic acid-Schiff (PAS). Stained sections were dehydrated in a series of ethanols and mounted using permount. Representative histologic images were photographed on a Nikon FXA microscope at a magnification of 1×, 2×, or 10× using a CCD digital camera.

Real-time PCR

Placentas were homogenized in 1.2 ml Trizol using a bead mill (Eppendorf, Westbury, NY) and RNA was isolated according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA). For each sample, 15 μg of RNA was DNAse-treated, followed by a phenol–chloroform extraction. RNA was quantified (NanoDrop Technologies, Wilmington, DE) and 1 μg of each sample was reverse transcribed using the cDNA Archive kit (Applied Biosystems, Foster City, CA). The amount of cDNA corresponding to 20 ng of RNA was used for each 20-μl real-time PCR reaction on an MXP-3000 instrument (Stratagene, La Jolla, CA). Primer and probe sets for Gusb, Eomes, Esrrb, Esx1, Dlx3, Gm52, Tcfeb, Ctsq, Timp2, Glut3, Cx31, and Pdch12 were run according to the manufacturer’s protocol with 2 × Taqman Universal Mastermix (ABI). Probes for 4311, Gcm1, and Pl1 were designed and manufactured in-house (Dr. Kathleen Caron, UNC). Gusb was used as an endogenous control and fold change of each gene of interest was calculated using the ΔΔCt method (Livak and Schmittgen 2001). The average ΔCt of wild-type animals for each strain/allele combination was used as the control value to calculate ΔΔCt values for samples of the same strain and allele. Fold-change values were computed from the ΔΔCt for each sample and converted to a percent increase over the wild-type average fold change for Egfr^Dsk5 heterozygous and homozygous samples.

Western blot

Three-month-old females were treated with 10 μl/g body weight phosphatase inhibitor (5 mM Na₃VO₄, 50 mM H₂O₂) by intraperitoneal injection and sacrificed by CO₂ after 5 min. Whole uteri were collected and snap frozen in liquid nitrogen. Frozen tissue was minced in 5 volumes lysis buffer (10 mM Tris–HCl, pH 7.4, 100 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% NP-40, 10% glycerol, 0.1% SDS, 0.5% sodium deoxycholate, 1 mM PMSF, 10 μg/ml Leupeptin, 10 μg/ml Aprotinin, 1 mM Na₃VO₄, 1 mM NaF) and homogenized in 2-ml tubes for 4 min using a bead mill. Samples were then sonicated for 30 sec and incubated for 1 h on ice. Lysates were cleared by centrifugation for 10 min at 13,000 rpm and protein quantified using the Bradford-based Protein Assay (Bio-Rad, Hercules, CA).

Samples were diluted with 2× sample buffer and boiled for 5 min. Eighteen micrograms of each sample was separated by denaturing 7.5% sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) for 1 h at 200 V and transferred to a PVDF membrane for 1.5 h at 100 V. The membranes were blocked for 1 h in 5% BSA/TBST (10 mM Tris, pH 7.5, 150 mM NaCl, 0.1% Tween 20) for phospho-EGFR detection and 5% milk/TBST for total EGFR and β-actin detection. Primary antibody incubations were overnight at 4°C followed by five TBST washes, and secondary antibody incubations were 1 h at room temperature. The phospho-EGFR antibody (Cell Signaling Technology, Danvers, MA) was diluted 1:1000 in 5% BSA/TBST and the total EGFR antibody (Upstate Biotechnology Inc., Lake Placid, NY) was diluted 1:1000 in 5% milk/TBST. The β-actin antibody (Sigma, St. Louis, MO) was diluted 1:10,000 in 5% milk/TBST. HRP-conjugated secondary antibodies were diluted 1:10,000 in 5% blocking agent/TBST. Following the secondary antibody incubation, blots were washed five times in TBST and protein detected by an enhanced chemiluminescence system (GE Healthcare Life Sciences, Piscataway, NJ).

Statistical analysis

All placenta and embryo weights were analyzed using the Mann-Whitney test. A χ² goodness-of-fit test was performed to determine if the genotype distribution deviated from expected Mendelian ratios. Real-time fold change values were analyzed using Student’s t test.

Results

Egfr^Dsk5 mice have background-dependent pigmentation defects

Strain-dependent phenotypes were evident when the Egfr^Dsk5 allele was backcrossed to 129 and B6 and compared to the original C3H background. Egfr^Dsk5 heterozygotes and homozygotes exhibited slightly wavy coats on all three backgrounds, with the phenotype being most pronounced on the B6 background (Fig. 1a, b). The C3H and 129-Egfr^Dsk5 heterozygotes exhibited pigmented footpads at 3 months of age, a phenotype not manifested by B6-Egfr^Dsk5 heterozygotes (Fig. 1c, d). However, some B6-Egfr^Dsk5 heterozygotes showed very slight footpad pigmentation by 6–8 months of age. We also observed long toenails in the C3H and 129-Egfr^Dsk5 heterozygotes that were frequently dark-colored on the 129 background (Fig. 1e).

Fig. 1.

Strain-specific coat and skin phenotypes observed in Egfr^Dsk5 heterozygotes. a Wild-type and Egfr^Dsk5 heterozygote on B6 background. b Wild-type and Egfr^Dsk5 heterozygote on C3H background. c Footpads from 3-month-old B6 wild-type and Egfr^Dsk5 heterozygote. d Footpads from 3-month-old C3H wild-type and Egfr^Dsk5 heterozygote. Arrowhead indicates excess pigmentation on footpad from Egfr^Dsk5 heterozygote. e Pigmented toenails (arrowheads) on 129-Egfr^Dsk5 heterozygote but not on C3H-Egfr^Dsk5 heterozygote. Pigmentation visible on C3H is from footpad

Egfr^Dsk5 embryos have larger placentas

Since reduced EGFR signaling has a detrimental effect on proper development of the placenta, we investigated whether an increased level of EGFR signaling also affects placental growth. At 15.5 dpc placenta weight was increased 18% in B6-Egfr^Dsk5 heterozygotes (P < 0.001) and homozygotes (P < 0.001) compared to placentas from wild-type littermates (Fig. 2a). On the C3H background, Egfr^Dsk heterozygotes had placenta weights that were increased 17% over wild-type (P < 0.01) and Egfr^Dsk5 homozygotes 55% more than wild-type (P < 0.001; Fig. 2a). The difference in placenta weight between Egfr^Dsk5 heterozygotes and homozygotes on the C3H background was significant; placentas from Egfr^Dsk5 homozygotes weighted 28% more than placentas from Egfr^Dsk5 heterozygotes (P < 0.001). Placenta weight was increased 12% in 129-Egfr^Dsk5 heterozygotes (P < 0.01) and homozygotes (P < 0.01) compared to wild-type (Fig. 2a). None of the strains examined showed differences in fetal weight between the three genotypes, suggesting that increased placental weight did not affect growth of the fetus (Fig. 2b).

Fig. 2.

Weights of placentas and embryos from wild-type, Egfr^Dsk5 heterozygous, and homozygous littermates measured at 15.5 dpc on three genetic backgrounds. a Egfr^Dsk5 heterozygous (n = 37) and homozygous (n = 24) placentas weighed 18% more than wild-type (n = 29) on B6. C3H-Egfr^Dsk5 heterozygous (n = 20) placentas weighed 17% more than wild-type (n = 15) placentas, and homozygous placentas (n = 4) showed a 55% increase in weight compared to wild-type placentas. Egfr^Dsk5 heterozygous (n = 20) and homozygous (n = 15) placentas weighed 12% more than wild-type (n = 9) on 129. b There were no significant differences observed in weight when Egfr^Dsk5 heterozygous and homozygous embryos were compared to wild-type in any of the three strains. ** P < 0.01 compared to wild-type, *** P < 0.001 compared to wild-type unless a different comparison is designated by horizontal bars

Altered expression of trophoblast cell subtype markers in Egfr^Dsk5 placentas

To determine the effect of the Egfr^Dsk5 allele on differentiation of the placental trophoblast, we measured transcript levels for a panel of specific trophoblast cell subtype markers (Dackor et al. submitted). Significant differences were observed in the expression of several genes in Egfr^Dsk5 heterozygous and homozygous placentas when compared to wild-type placentas at 15.5 dpc (Table 1). On the B6 background, labyrinth-expressed genes Gcm1 and Dlx3 were significantly reduced in Egfr^Dsk5 heterozygous (n = 7) and homozygous placentas (n = 4), while 4311, a marker of spongiotrophoblast, was elevated in Egfr^Dsk5 heterozygotes and homozygotes compared to wild-type controls (n = 7). Pdch12, a marker of glycogen cells, and Pl-1, a marker of trophoblast giant cells, were significantly elevated in Egfr^Dsk5 heterozygotes and homozygotes when data from the two genotypes were combined and compared to wild type. In addition, expression of the trophoblast stem cell marker Eomes was significantly elevated in B6-Egfr^Dsk5 homozygotes but not in heterozygotes. There were no significant changes in the expression of Tcfeb, Esx1, Esrrb1, Gm52, Ctsq, Timp2, Glut3, and Cx31.

Table 1.

Percent expression of trophoblast cell subtype markers in Egfr^Dsk5 heterozygous and homozygous placentas compared to wild-type littermates

		C57BL/6J		C3HeB/FeJ		129 Sv
		+/Dsk5 (%)	Dsk5/Dsk5 (%)	+/Dsk5 (%)	Dsk5/Dsk5 (%)	+/Dsk5 (%)	Dsk5/Dsk5 (%)
Gcm1	Lz	58***	73a	97	83	80	90
Dlx3	Lz	77**	79**	74a	65*	89	84
Tcfeb	Lz	97	94	83	78*	93	90
Esx1	TS cells, Lz	94	96	95	98	81	85
Esrrb1	TS cells	109	108	106	136	117a	148*
Eomes	TS cells, Lz	103	118*	86	89	100	109
Gm52	Lz	103	100	79a	73*	85	79
Ctsq	Labyrinth TG	113	122	80	96	94	124
4311	SpT, GT	120a	127**	132	123	137	171
Pdch12	GT	126a	138a	142a	150*	117	144
PL-1	TG	135a	148a	69	87	81	101
Timp2	Decidua	98	116	101	91	91	98
Glut3	Lz	94	95	96	111	89	87
Cx31	GT	90	90	91	94	99	115

^*P < 0.05,

^**P < 0.01,

^***P < 0.001

^aP < 0.05 when +/Dsk5 and Dsk5/Dsk5 data were combined

On the C3H background, expression of Dlx3 was reduced in Egfr^Dsk5 heterozygotes (n = 5) and homozygotes (n = 5) compared to wild type (n = 5). Tcfeb and Gm52, both labyrinth-specific genes, were reduced in Egfr^Dsk5 homozygous placentas, while 4311 expression was increased in both Egfr^Dsk5 heterozygotes and homozygotes compared to wild type but the data did not reach significance. Pdch12 was significantly elevated in Egfr^Dsk5 homozygotes and increased in Egfr^Dsk5 heterozygotes.

The 129 strain also exhibited genotype-associated changes in gene expression. Overall, the labyrinth-specific genes were reduced in Egfr^Dsk5 heterozygotes and homozygotes but the reduction was not significant for any probe alone. Esrrb1, a gene expressed in trophoblast stem cells, was significantly elevated in Egfr^Dsk5 heterozygotes (n = 6) and homozygotes (n = 6) compared to wild type (n = 4, P < 0.05). Similar to B6 and C3H, the expression of 4311 and the expression of Pdch12 were increased in 129-Egfr^Dsk5 heterozygotes and homozygotes but the changes were not significant.

Egfr^Dsk5 placentas have an expanded spongiotrophoblast layer

We examined H&E- and PAS-stained tissue sections to further characterize the overgrowth phenotype observed in Egfr^Dsk5 heterozygous and homozygous placentas. Consistent with the real-time data, we observed an increased layer of spongiotrophoblast in Egfr^Dsk5 heterozygotes and homozygotes compared to wild type (Fig. 3). There were also increased numbers of PAS-positive cells, suggesting that the population of glycogen trophoblasts was larger in Egfr^Dsk5 placentas (Fig. 3g–i). We did not observe any obvious changes in the size or structure of the labyrinth layer for any of the genotypes. The decreased expression of labyrinth markers observed was probably due to a disproportionate increase in the size of the spongiotrophoblast layer.

Fig. 3.

Egfr^Dsk5 heterozygous and homozygous placentas have an expanded spongiotrophoblast layer compared to wild-type. H&E staining of 15.5-dpc placentas from (a) B6 wild-type, (b) Egfr^Dsk5 heterozygote, and (c) Egfr^Dsk5 homozygote. d, e, f Closeup of placentas from a, b, c, respectively. PAS staining of 15.5-dpc placentas from (g) B6 wild-type, (h) Egfr^Dsk5 heterozygote, and (i) Egfr^Dsk5 homozygote. The spongiotrophoblast (sp) compartment is bracketed in each section

Reduced fertility in 129 and C3H-Egfr^Dsk5 heterozygous females

During collection of 15.5-dpc placentas we observed a large number of dead embryos in litters from C3H and 129, but not from B6-Egfr^Dsk5 heterozygous females. In the B6 strain, 93% of embryos were viable, while in C3H and 129, 39 and 21% of embryos were viable, respectively (Fig. 4a). Viable embryos from each strain were genotyped to determine if the genotype distribution deviated from expected Mendelian ratios (Table 2). Neither B6 nor 129 had numbers of viable Egfr^Dsk5 heterozygous and Egfr^Dsk5 homozygous embryos that were different than the expected ratios, although data for 129 approached significance with a higher number of Egfr^Dsk5 homozygotes than expected. For C3H only 11% of viable embryos were Egfr^Dsk5 homozygous, which is significantly different from the expected 25% (P < 0.05). The viability of Egfr^Dsk5 heterozygous embryos was not affected in the C3H cross and the lethality of the Egfr^Dsk5 homozygotes did not fully account for the 50–60% reduction in live embryos observed at 15.5 dpc in C3H.

Fig. 4.

Embryo viability in C3H and 129 litters depends on maternal genotype. a Overall embryo viability (all genotypes) in litters from Egfr^Dsk5 heterozygous intercrosses using three genetic backgrounds. 93% of embryos were viable in litters from B6 intercrosses versus 39% in C3H and 36% in 129 intercrosses. *** P < 0.001 compared to B6. b Embryo viability in C3H crosses between wild-type females and Egfr^Dsk5 heterozygous males was 86 vs. 39% for Egfr^Dsk5 heterozygous intercrosses. Embryo viability in 129 crosses between wild-type females and Egfr^Dsk5 heterozygous males was 88 vs. 36% for Egfr^Dsk5 heterozygous intercrosses and 21% for Egfr^Dsk5 heterozygous females crossed to wild-type males. The number of litters used for each group is indicated above the error bars. *** P < 0.001 compared to wild-type female × Egfr^Dsk5 heterozygous male breeding

Table 2.

Survival of 15.5-dpc embryos from Egfr^Dsk5 intercrosses on three congenic strains

Strain	Average litter size	+/+	+/Dsk5	Dsk5/Dsk5	Total viable	P
C57BL/6J	10.0	31 (30.5%)	44 (44%)	26 (25.5%)	101	0.338
C3HeB/FeJ	6.5	16 (26%)	39 (63%)	7 (11%)	62	0.034
129S1/SvImJ	8.9	10 (20%)	21 (42%)	19 (38%)	50	0.104

Since genotype alone did not explain the lethality of C3H embryos at 15.5 dpc, additional matings were set up for C3H and 129 to investigate the origin of reduced embryonic viability. When wild-type C3H and 129 females were mated to Egfr^Dsk5 heterozygous males from their respective strains, a significant increase in the number of viable embryos was observed compared to results from the Egfr^Dsk5 heterozygous intercrosses (Fig. 4b). For C3H, embryo viability was 86% versus the 39% observed in Egfr^Dsk5 heterozygous intercrosses. For 129, embryo viability was 88% when the female was wild type versus 36% viability in Egfr^Dsk5 heterozygous intercrosses. To confirm that the embryonic lethality was due to maternal factors and/or uterine environment, reciprocal crosses were performed for the 129 strain (Fig. 4b). When Egfr^Dsk5 heterozygous females were mated to wild-type males, embryo viability was 21%, similar to the number observed in Egfr^Dsk5 heterozygous intercrosses (39%).

Levels of phospho-EGFR are higher in Egfr^Dsk5 uteri

Since the fertility defect observed seemed to be dependent on maternal, but not embryonic, genotype, we measured the levels of total and phosphorylated EGFR in uteri from Egfr^Dsk5 heterozygous mice to evaluate whether there is a similar downregulation of EGFR in uterus compared to that previously reported in Egfr^Dsk5 livers. Uteri were collected from phosphatase inhibitor-treated 3-month-old wild type and B6 and C3H-Egfr^Dsk5 heterozygotes and their wild-type littermates. Analysis using Western blots revealed that the levels of total EGFR were similar in samples from wild type and Egfr^Dsk5 heterozygous placentas but phospho-EGFR was significantly higher in the Egfr^Dsk5 heterozygous placental samples (Fig. 5). In addition, phospho- and total EGFR were detected at higher levels in the C3H strain versus B6.

Fig. 5.

Western blot for phospho-EGFR and total EGFR in uteri extracts from B6 and C3H females. Samples are B6 in lanes 1–4 and C3H in lanes 5–8. Even-numbered lanes are wild-type animals and odd-numbered lanes are their Egfr^Dsk5 heterozygous littermates. The top panel shows protein levels of phospho-EGFR, the middle panel shows levels of total EGFR, and the bottom panel shows ACTB as a loading control

Nonpregnant Egfr^Dsk5 and wild-type uteri are similar histologically

Next we examined uteri from virgin random-cycling Egfr^Dsk5 heterozygous and wild-type female littermates at approximately 3 months of age. In all four sets of B6 and C3H littermates examined, Egfr^Dsk5 heterozygous uteri weighed more than wild-type uteri (Fig. 6a). Uteri weights for 129 littermates were not measured. H&E-stained tissue sections revealed no obvious defects in Egfr^Dsk5 heterozygous uterine morphology (Fig. 6c, d). Uteri from B6 and C3H-Egfr^Dsk5 heterozygotes had clearly differentiated luminal and glandular epithelium, stroma, and myometrium that appeared similar to wild-type tissue. In our B6 breeding colony, Egfr^Dsk5 heterozygous female fertility declined at a relatively young age compared to wild-type females (data not shown). We dissected several 5–9-month-old B6-Egfr^Dsk5 heterozygous females and frequently noted the appearance of fluid-filled cysts on one or both ovaries (Fig. 6b). The aged Egfr^Dsk5 heterozygous females exhibited additional sporadic uterine abnormalities not observed in younger animals but these defects have not been fully characterized.

Fig. 6.

Reproductive system phenotypes in Egfr^Dsk5 heterozygous females. a Nonpregnant, random-cycling Egfr^Dsk5 heterozygous females had higher uterus-to-body weight ratios than wild-type females on B6 and C3H backgrounds. b B6-Egfr^Dsk5 heterozygous ovary. Bursal sac is filled with fluid. c Wild-type C3H nonpregnant uterus with myometrium (M), glandular epithelium (GE), stroma (S), and luminal epithelium (LE) labeled (10×). d C3H-Egfr^Dsk5 heterozygous uterus. No obvious histologic differences were observed

Discussion

Similar to mouse models in which EGFR signaling is reduced or abolished, we have shown here that phenotypes resulting from increased EGFR signaling vary by genetic background. Mice that are heterozygous or homozygous for the hypermorphic Egfr^Dsk5 allele display strain-dependent hair, skin, and nail phenotypes. EGFR is known to be involved in progression of several types of cancer, but our initial characterization of B6, 129, and C3H-Egfr^Dsk5 heterozygotes did not reveal an obvious increase in tumor susceptibility in mice younger than 9 months of age. However, we found Egfr^Dsk5 heterozygous and homozygous placentas enlarged and Egfr^Dsk5 heterozygotes subfertile due to several potential strain-specific defects in female reproduction.

EGFR in the placenta

Our examination of placentas from mice with at least one hypermorphic allele of EGFR demonstrates that increased activation of EGFR can result in strain-specific effects on placental growth. In the B6 and 129 strains, heterozygosity and homozygosity for Egfr^Dsk5 resulted in the same increase in placental weight, suggesting that placental growth does not continue beyond a threshold reached with one Egfr^Dsk5 allele. However, it is unknown whether this limitation on growth is a property of the trophoblast population or general negative feedback inhibition of EGFR signaling. There is evidence from studies on Egfr^Dsk5 mouse livers that total EGFR is downregulated, particularly in the homozygote, and a similar mechanism could limit the increase in trophoblasts observed in B6 and 129-Egfr^Dsk5 strains (Fitch et al. 2003). If there is a threshold in growth-promoting effects of Egfr^Dsk5, it is not reached in C3H-Egfr^Dsk5 heterozygotes since placentas from Egfr^Dsk5 homozygous mice are larger than heterozygous placentas. The weight differences were highly significant between Egfr^Dsk5 genotypes, and we observed some embryonic lethality of C3H-Egfr^Dsk5 homozygotes that could be related to the placental overgrowth.

Although placental weights were altered, no significant effects of the Egfr^Dsk5 allele on embryo weights were observed at 15.5 dpc in any of the three strains. Both molecular and histologic analyses showed that the increased placental weights are due to an increase in spongiotrophoblast and glycogen cell populations. Placentas homozygous for the Egfr^tm1Mag-null allele have fewer numbers of proliferating trophoblasts and this phenotype does not correlate with severity of labyrinth defects. Therefore, the increased layer of spongiotrophoblast observed in Egfr^Dsk5 heterozygous and Egfr^Dsk5 homozygous placentas is probably a result of greater trophoblast proliferation and suggests that EGFR plays a major role in promoting cell cycle progression in spongiotrophoblast and glycogen cell precursors. Numerous mouse models have been shown to exhibit a reduction in spongiotrophoblasts, including Mapk14, Erk2, Sos1, Sp1/3, and Akt1, genes known to function downstream of EGFR (Kruger et al. 2007; Mudgett et al. 2000; Qian et al. 2000; Saba-El-Leil et al. 2003; Yang et al. 2003). EGFR may activate spongiotrophoblast-specific factors required for proliferation through one or more of these pathways.

EGFR in the female reproductive tract

Our data indicate that female mice heterozygous for the dominant Egfr^Dsk5 hypermorphic mutation exhibit reduced fertility that may be due to defects in the uterus and/or ovaries. In litters from 129 and C3H-Egfr^Dsk5 pregnant females, embryos of all genotypes showed reduced survival, suggesting that the maternal uterine environment has a detrimental effect on litter viability. The maternal origin of embryo loss was verified when significant embryo loss occurred in crosses between 129-Egfr^Dsk5 heterozygous females and wild-type males but not in crosses between wild-type females and Egfr^Dsk5 heterozygous males. The phenotype was strain-specific; embryo viability was normal in litters from B6-Egfr^Dsk5 heterozygous females.

The precise timing of implantation is mediated by molecular crosstalk between the uterus and blastocyst, and a delay in implantation can result in pregnancy loss, similar to that observed in litters from Egfr^Dsk5 heterozygous females (Wang and Dey 2006). Several mouse models have been described in which blastocyst implantation occurs beyond the normal window of uterine receptivity and characterization of the phenotype has revealed that production of prostaglandins is essential for the process. In mice deficient for prostaglandin endoperoxide synthase 2 (COX2) or cytosolic phospholipase A2 (CPLA2), a provider of arachidonic acid for prostaglandin synthesis, implantation sites are not apparent until day 5.5 and are fewer in number with poor permeability compared to implantation sites in wild-type females (Song et al. 2002; Wang et al. 2004). The implantation delay in these mice has been shown to lead to a later wave of embryonic lethality resulting in smaller litter sizes for both models. Wild-type blastocysts that are transferred into a pseudopregnant uterus on day 5.5 also exhibit this wave of lethality, suggesting that embryonic lethality observed in the COX2- and CPLA2-deficient mice is a result of the delay in implantation rather than a direct consequence of gene deletion in the embryos (Song et al. 2002). Previous studies have indicated that EGFR signaling may be involved in uterine receptivity and implantation timing. Transgenic mice that overexpress either Tgfa or Btc display delayed implantation but it is unclear whether embryonic lethality is associated with the phenotype in these models (Das et al. 1997; Gratao et al. 2008). In addition, it is unknown whether these particular ligands are involved in implantation timing under normal circumstances because expression of Tgfa and Btc transgenes may be controlled in a manner not spatially or temporally similar to expression of the endogenous genes.

Several additional reproductive system anomalies were noted in Egfr^Dsk5 heterozygous females. C3H and B6-Egfr^Dsk5 heterozygous uteri weighed more than uteri from wild-type littermates but differentiation of the tissue appeared similar to that of wild type at the histologic level. At a young age B6-Egfr^Dsk5 heterozygous exhibited normal reproductive capacity; however, we did notice that following their first pregnancy, B6-Egfr^Dsk5 heterozygous female fertility declined. This reduced reproductive capacity could be explained by sporadic uterine abnormalities and fluid-filled cysts on one or both ovaries observed in B6-Egfr^Dsk5 heterozygotes over 5 months of age, similar to the phenotype of Inhibin alpha (Inha) transgenic mice (McMullen et al. 2001). Uterine and/or ovarian defects may render older Egfr^Dsk5 heterozygous females unable to ovulate and/or support implantation of embryos. Several recent studies suggest that increased levels of EGFR activity in female mice may result in uterine hyperplasia. Transgenic mice with mouse mammary tumor virus (MMTV)-regulated overexpression of human EGFR exhibit cystic hyperplasia of uterine glands at 9 months of age (Marozkina et al. 2007). Nine-month-old mice deficient for mitogen-inducible gene-6 (MIG6), an endogenous inhibitor of EGFR signaling, have enlarged uteri with an increased number of uterine glands and hyperplastic glandular and luminal epithelium (Jin et al. 2007). In humans, MIG6 is overexpressed in moderate to severe cases of endometriosis (Burney et al. 2007).

In summary, our study emphasizes extensive strain-dependent phenotypic variation evident in mice with increased levels of EGFR signaling. The same modifiers that influence variability observed in mice with null and hypomorphic alleles of EGFR are probably involved but there may be an additional class of modifiers that control downregulation of activated EGFR. Our data also indicate that EGFR plays a previously underappreciated role in the uterus during pregnancy. The fertility defects we have described in Egfr^Dsk5 heterozygous mice suggest that in humans, an increased level of uterine EGFR signaling may contribute to some cases of infertility.