Embryos lacking an epidermal growth factor receptor (EGFR) exhibit strain-specific defects in placental development that can result in mid-gestational embryonic lethality. To determine the level of EGFR signaling required for normal placental development, we characterized congenic strains homozygous for the hypomorphic Egfrwa2 allele or heterozygous for the antimorphic EgfrWa5 allele. Egfrwa2 homozygous embryos and placentas exhibit strain-dependent growth restriction at 15.5 days post-coitus while EgfrWa5 heterozygous placentas are only slightly reduced in size with no effect on embryonic growth. Egfrwa2 homozygous placentas have a reduced spongiotrophoblast layer in some strains, while spongiotrophoblasts and glycogen cells are almost completely absent in others. Our results demonstrate that more EGFR signaling occurs in EgfrWa5 heterozygotes than in Egfrwa2 homozygotes and suggest that Egfrwa2 homozygous embryos model EGFR-mediated intrauterine growth restriction in humans. We also consistently observed differences between strains in wild-type placenta and embryo size as well as in the cellular composition and expression of trophoblast cell subtype markers and propose that differential expression in the placenta of Glut3, a glucose transporter essential for normal embryonic growth, may contribute to strain-dependent differences in intrauterine growth restriction caused by reduced EGFR activity.
EPIDERMAL growth factor receptor (EGFR) is the prototypical member of the ERBB family of receptor tyrosine kinases and is known to regulate many aspects of cellular biology including cell proliferation, survival, differentiation, and migration (reviewed in Yarden and Sliwkowski 2001). Eleven known ligands bind the extracellular region of ERBB-family receptors, and activation of the tyrosine kinase domain occurs following receptor homo- or heterodimerization. The resulting biological responses are dependent upon specific signaling cascades initiated by ERBBs and can be influenced by the particular ligand–ERBB combination (Yarden and Sliwkowski 2001). Studies using cultured cells have underscored the importance of EGFR in modulating various cellular processes, while animal models have been able to demonstrate that EGFR is required for numerous developmental and physiological processes (Casalini et al. 2004). In vivo studies have shown that EGFR is particularly important for normal placental development in mice; placentas from Egfr nullizygous (Egfrtm1Mag/tm1Mag) embryos exhibit strain-specific defects that result in differential embryonic lethality (Sibilia and Wagner 1995; Threadgill et al. 1995). Two additional Egfr alleles result in reduced EGFR signaling in mice: the recessive hypomorphic Egfrwa2 and dominant antimorphic EgfrWa5 alleles (Luetteke et al. 1994; Fowler et al. 1995; Du et al. 2004; Lee et al. 2004). These alleles can provide insight into the level of EGFR signaling required for normal placental development.
Egfrwa2 is a classical spontaneous mutation that arose in 1935 that causes a distinct wavy coat phenotype in the homozygote (Figure 1; Keeler 1935). This recessive mutation was subsequently found to be a single nucleotide transversion resulting in a valine → glycine substitution in the highly conserved kinase domain of EGFR (Luetteke et al. 1994; Fowler et al. 1995). Since mice homozygous for the Egfrtm1Mag null allele die before or shortly after birth depending on genetic background, the hypomorphic Egfrwa2 allele has been the primary model used to study the effect of attenuated EGFR signaling in a variety of adult physiological and disease states. In addition to eye and hair phenotypes, the adult Egfrwa2 homozygous mouse exhibits delayed onset of puberty, abnormal ovulation, enlarged aortic valves and cardiac hypertrophy, decreased body size, defects in mammary gland development and lactation, increased susceptibility to colitis, and impaired intestinal adaptation following small bowel resection (Fowler et al. 1995; Helmrath et al. 1997; Chen et al. 2000; Egger et al. 2000; O'Brien et al. 2002; Prevot et al. 2005; Hsieh et al. 2007). Despite the widespread use of the Egfrwa2 allele, there are limitations in using Egfrwa2 homozygous mice to clearly define the physiological roles of EGFR. Egfrwa2 has traditionally been maintained in cis, tightly linked with a hypomorphic Wnt3a allele, Wnt3avt (vestigal tail), making phenotypic analysis of reduced EGFR signaling by itself difficult. Furthermore, Egfrwa2 has also typically been maintained on a mixed genetic background and since the Egfr nullizygous phenotype is similarly influenced by genetic modifiers, a mixed background could mask phenotypes that become evident when Egfrwa2 mice are inbred.
The EgfrWa5 allele arose in a large, genomewide N-ethyl-N-nitrosourea mutagenesis screen for dominant visible mutations in the mouse. EgfrWa5 heterozygous mice were first identified by their open eyelids at birth and by development of a wavy coat, similar to the phenotype of Egfrwa2 homozygous mice (Figure 1). EgfrWa5 failed to complement the Egfrtm1Mag null allele and was shown to function as an antimorph since EgfrWa5, but not Egfrtm1Mag, heterozygotes exhibit eyelid and coat phenotypes (Lee et al. 2004). A single nucleotide missense mutation was found in the EgfrWa5 allele that results in an Asp → Gly substitution in the highly conserved DFG domain of the EGFR kinase catalytic loop (Du et al. 2004; Lee et al. 2004). Although EgfrWa5 heterozygotes are viable, EgfrWa5 homozygotes die prenatally and exhibit placental defects identical to those from Egfrtm1Mag homozygous null embryos. Placentas from EgfrWa5 heterozygotes on a mixed background show variable reduction in the spongiotrophoblast layer and minor abnormalities in the labyrinth region, but no effects on embryo survival have been reported.
In vitro studies with EgfrWa5 suggest that it encodes a kinase-dead EGFR since no phosphorylation of EGFRWa5 is detected following stimulation with ligands. In agreement with the genetic data showing that EgfrWa5 is an antimorph, in vitro studies have demonstrated that the EGFRWa5 receptor can inhibit phosphorylation of EGFR and MAPK in a dose-dependent manner (Lee et al. 2004). In Chinese hamster ovary cells expressing an equimolar ratio of EGFR and EGFRWa5 receptors, <10% of wild-type phosphorylation levels were observed by Western blot analysis.
The Egfr allelic series available in the mouse has high utility for studying gene function since EGFR is involved in a multitude of developmental processes and human diseases. Although both Egfrwa2 and EgfrWa5 alleles result in reduced EGFR signaling, the activity and phenotypic consequences of Egfrwa2 homozygosity has not been compared to that of EgfrWa5 heterozygosity when both are on the same genetic backgrounds. Adult EgfrWa5 heterozygous mice appear highly similar to Egfrwa2 homozygotes, but crosses with the ApcMin intestinal tumor model have shown that a more substantial reduction in tumor number occurs when the ApcMin mutation is bred onto the Egfrwa2 homozygous vs. EgfrWa5 heterozygous background (Roberts et al. 2002; Lee et al. 2004). These results suggest that EgfrWa5 heterozygous mice retain higher levels of EGFR activity than Egfrwa2 homozygous mice; however, the data are confounded by the fact that the crosses were performed using different mixed genetic backgrounds.
This study reports a comprehensive genetic analysis of reduced EGFR signaling in Egfrwa2 homozygotes and EgfrWa5 heterozygotes in placental development and embryonic growth for three congenic backgrounds, C57BL/6J (B6), 129S1/SvImJ (129), and BTBR/J-T+, tf/tf (BTBR). Wild-type placenta weight, embryo weight, and mRNA levels of genes selected for their trophoblast-specific expression were found to be highly strain dependent. Egfrwa2 homozygous placentas are reduced in size in all three strains, and a proportion of 129-Egfrwa2 homozygotes die before 15.5 days post-coitus (dpc). Egfrwa2 homozygous embryos also display background-dependent intrauterine growth restriction (IUGR) in late gestation, which is most severe on 129 and BTBR backgrounds and models EGFR-associated IUGR in humans. EgfrWa5 heterozygous placentas exhibit a minor reduction in size on all three backgrounds with no impact on embryonic growth. These results suggest that reduced levels of EGFR signaling can interfere with normal placental development and that embryo development is affected only after placental size is sufficiently reduced. In addition, our data show that the level of EGFR signaling in EgfrWa5 heterozygous mice is higher than in Egfrwa2 homozygotes and suggests that different Egfr allele combinations can be generated to “genetically titer” total EGFR activity in vivo.
MATERIALS AND METHODS
Mice and genetic crosses:
Congenic Egfrwa2 lines were generated by backcrossing mixed C57BL/6JEiC3H-a/A-Egfrwa2/wa2 Wnt3avt/vt mice obtained from The Jackson Laboratory to B6, 129, and BTBR wild-type inbred strains for ≥10 generations. Removal of the linked Wnt3avt allele, 20 cM distal to Egfr on chromosome 11, was verified by PCR-based genotyping. Congenic Egfrwa2 heterozygous mice were then intercrossed to produce litters from each background containing wild-type and Egfrwa2 heterozygous and homozygous congenic embryos and pups.
Congenic EgfrWa5 mice were generated by backcrossing heterozygous EgfrWa5 mice from a mixed genetic background to inbred B6, 129, and BTBR strains for ≥10 generations. Congenic heterozygous mice were then crossed to male or female wild-type animals of the same strain to produce litters containing wild-type and Egfrwa5 heterozygous congenic embryos and pups.
Mice were fed Purina Mills Lab Diet 5058 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 the University of North Carolina Institutional Animal Care and Use Committee.
DNA was extracted from adult ear punches or embryo tail biopsies for genotyping by incubating at 95° in 100 μl of 25 mm NaOH/0.2 mm EDTA for 20 min before neutralizing with 100 μl 40 mm Tris–HCl, pH 5.0. For the subsequent genotyping reactions, 1 μl of lysed tissue sample was used per reaction.
The Egfrwa2 allele was amplified by PCR with the primers Wa2F (5′-TACCCAGAAAGGGATATGCG-3′) and Wa2R (5′-GGAGCCAATGTTGTCCTTGT-3′) (Qiagen). PCR conditions were 30 cycles at 94° for 30 sec, 60° for 60 sec, and 72° for 60 sec. PCR products were digested for 3 hr at 37° with Fok I and restriction enzyme buffer 2 (NEB) and run on a 3% agarose gel to separate a 230-bp product corresponding to wild-type Egfr and a 130- and 100-bp set of products corresponding to the digested Egfrwa2 allele.
The EgfrWa5 allele was detected by real-time PCR with the primers WA5F (5′-GTGAAGACACCACAGCATGTC-3′) and WA5R (5′-CTCTTCAGCACCAAGCAGTTTG-3′) along with the 5′ VIC-labeled probe WA5V1 (5′-AAGATCACAGATTTTGG-3′) to detect wild-type Egfr and the 5′ FAM-labeled probe WA5M1 (5′-AGATCACAGGTTTTGG-3′) to detect EgfrWa5 (ABI). Genotyping was performed on an MXP-3000 real-time PCR instrument (Stratagene) with 2X Taqman Universal PCR Master Mix (Applied Biosystems). PCR conditions were 95° for 10 min followed by 40 cycles of 92° for 15 sec and 60° for 1 min. Amplification of the wild-type allele was detected by comparative quantification of VIC-labeled PCR products, and amplification of the EgfrWa5 allele was detected by comparative quantification of FAM-labeled PCR products with positive and negative EgfrWa5 adult tissue used as a reference sample.
Collection of placenta 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 were dissected from the uterine horns on the morning of 15.5 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) for extraction of RNA or fixed in 10% neutral buffered formalin (NBF) for histological analysis.
After fixing placentas in 10% NBF overnight, tissues were washed in PBS, dehydrated in ethanols and xylenes, and embedded in paraffin. Seven-micrometer 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 histological images were photographed on a Nikon FXA microscope at a magnification of ×1.25, ×10, or ×12 using a CCD digital camera.
Placentas were homogenized in 1.2 ml Trizol using a bead mill (Eppendorf), and RNA was isolated according to the manufacturer's protocol (Invitrogen). For each sample, 15 μg of RNA was DNAse treated, followed by a phenol–chloroform extraction. RNA was quantified (Nanodrop), and 1 μg of each sample was reverse transcribed using the cDNA archive kit (Applied Biosystems). 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). Assays-on-demand 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 2X Taqman Universal Mastermix (ABI). Probes for 4311, Gcm1, and Pl1 were designed and synthesized in-house. Gusb was used as an endogenous control, and fold change of each gene 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 percentage increase over the wild-type average fold change for EgfrWa5 heterozygous and Egfrwa2 heterozygous and homozygous samples.
All placenta and embryo weights were analyzed using the Mann–Whitney test. A genotype by strain interaction test was performed using a standard least-squares regression analysis in JMP (SAS) for placenta and embryo weights. A χ2 goodness-of-fit test was performed to determine if genotype distributions deviated from expected Mendelian ratios. Real-time PCR fold change values were analyzed using the Student's t-test.
Egfrwa2 homozygous placentas are reduced in size on all genetic backgrounds:
To determine the effect of the Egfrwa2 allele on placental development, placentas were collected at 15.5 and 18.5 dpc from each of the three congenic strains by intercrossing respective Egfrwa2 heterozygous mice. At 15.5 dpc, the placental weight of Egfrwa2 homozygotes was reduced 24% in B6 (P < 0.001), 19% in 129 (P < 0.001), and 39% in BTBR (P < 0.01) vs. wild-type and heterozygous littermates (Figure 2A). At 18.5 dpc, the placental weight of Egfrwa2 homozygotes was reduced 24% in B6 (P < 0.05), 37% in 129 (P < 0.01), and 28% in BTBR (P < 0.01) vs. wild-type and heterozygous littermates (Figure 2B). For all strains and time points examined, wild-type and Egfrwa2 heterozygous placental weights did not differ. The extent of placental growth restriction in Egfrwa2 homozygotes was significantly different between strains at 15.5 dpc (P < 0.001) but not at 18.5 dpc.
Egfrwa2 homozygous embryos display strain-dependent intrauterine growth restriction:
To assess the effect of changes in placental size on embryonic growth, wild-type and Egfrwa2 heterozygous and homozygous embryos were collected at 15.5 and 18.5 dpc for each strain. At 15.5 and 18.5 dpc, there were no significant differences in B6 embryo weight between the genotypes (Figure 2, C and D). At 18.5 dpc, there was a small, nonstatistically significant, Egfrwa2 dose-dependent effect on embryo weight for the B6 background (Figure 2D), indicating that the 24% reduction in placental weight in B6-Egfrwa2 homozygotes had little to no effect on embryonic growth.
At 15.5 dpc, 129-Egfrwa2 homozygous embryos did not weigh significantly differently from wild-type embryos but heterozygous embryos weighed 12% more than wild-type and Egfrwa2 homozygotes (P < 0.01 and P < 0.05, respectively; Figure 2C). In contrast, at 18.5 dpc, severe growth restriction was observed in 129-Egfrwa2 homozygotes with Egfrwa2 homozygous embryos weighing 34% less than wild-type and heterozygous littermates (P < 0.01 and P < 0.001, respectively; Figure 2, D and F), similar to the 37% reduction in placental weight. At 18.5 dpc, there were no differences in embryo weight between 129 wild-type and Egfrwa2 heterozygous embryos (Figure 2D).
At 15.5 dpc, BTBR was the only strain with significant embryonic growth restriction mirroring its more severe placental phenotype. Egfrwa2 homozygous embryos weighed 18% less than wild-type and heterozygous littermates (P < 0.05 and P < 0.01, respectively; Figure 2C). By 18.5 dpc, an even more severe embryonic growth restriction was observed in BTBR, with Egfrwa2 homozygous embryos weighing 32% less than wild-type and heterozygous littermates (P < 0.01 and P < 0.001, respectively; Figure 2D). There were no differences in embryo weight between BTBR wild-type and Egfrwa2 heterozygous embryos at either developmental stage. We found that embryo and placenta weights were highly correlated in 18.5-dpc BTBR-Egfrwa2 homozygous embryos (R2 = 0.78) and to some extent in 129-Egfrwa2 homozygotes (R2 = 0.45), suggesting that fetal growth restriction was caused by the placental phenotype (Figure 2E). Although embryonic growth restriction in Egfrwa2 homozygotes was not different among B6, 129, and BTBR strains at 15.5 dpc, the interaction between background and genotype approached significance at 18.5 dpc (P = 0.0545), consistent with an interaction between background and genotype relating to placental weight at 15.5 dpc and resulting in altered embryo weight later in gestation.
129-Egfrwa2 homozygous embryo survival is reduced at 15.5 dpc:
To determine the effect of Egfrwa2 on embryo survival, viable 15.5-dpc embryos were genotyped for each strain and evaluated for deviation from expected Mendelian ratios (Table 1). For B6, 32% of 75 viable embryos were Egfrwa2 homozygous, which was not significantly different than the expected 25%. However, a significant deviation from Mendelian ratios was observed on the 129 background as only 14% of 86 viable embryos were Egfrwa2 homozygous at 15.5 dpc (P < 0.01); a similar percentage of homozygotes was also observed at weaning (Table 1). This result suggests that a significant number of Egfrwa2 homozygous embryos die prior to 15.5 dpc. Although survival of BTBR embryos was similar to B6, three BTBR embryos were found dead at 15.5 dpc and all three were Egfrwa2 homozygous, suggesting that there may be some loss of Egfrwa2 homozygotes prior to 15.5 dpc in the BTBR background as well. There were also fewer than expected numbers of BTBR-Egfrwa2 homozygous weanlings observed in the breeding colony (data not shown).
EgfrWa5 heterozygous embryos have a small reduction in placental size but no change in embryo weight:
To measure the effect of the EgfrWa5 allele on growth of the placenta and embryo, litters were collected from crosses between EgfrWa5 heterozygous and wild-type mice for the same three strains. Placenta weight of EgfrWa5 heterozygotes at 15.5 dpc was reduced by 9% in B6 (P < 0.001) and 129 (P < 0.001) vs. wild-type littermates (Figure 3A), but embryo weight was not affected (Figure 3B). Unlike placenta from BTBR-Egfrwa2 homozygotes, EgfrWa5 heterozygous placentas were more modestly affected at 15.5 dpc, showing only a 5% reduction in placenta weight with no difference in embryo weight compared to wild-type littermates (P < 0.05; Figure 3, A and B). Placenta and embryo weights were also measured at 18.5 dpc in the 129 strain but no significant differences were observed between 129-EgfrWa5 and wild-type littermates at this later developmental stage (Figure 3C).
Viable 15.5-dpc embryos were genotyped for each strain to determine if the genotype distributions deviated from expected Mendelian ratios (Table 2). For B6 and 129, 53% and 51% of viable embryos, respectively, were EgfrWa5 heterozygotes. Although the BTBR strain exhibited the smallest change in placental weight, only 40% of viable embryos were EgfrWa5 heterozygotes (P < 0.05), suggesting that, although the reduction in placental size was more modest than in Egfrwa2 homozygotes, there was still an effect on placental function.
129 and BTBR Egfrwa2 homozygous placentas have few spongiotrophoblasts:
Placentas from 18.5-dpc embryos were stained with H&E for general morphological characterization and with PAS to identify glycogen-containing cells of the spongiotrophoblast layer. Wild-type B6 placentas had a very thick layer of spongiotrophoblast with numerous protrusions into the labyrinth region (Figure 4A). The B6-Egfrwa2 homozygous placentas exhibited a reduction in spongiotrophoblasts compared to wild type (Figure 4B), but there were many glycogen-positive cells present (Figure 4C). Overall the B6 strain showed very intense PAS staining of the spongiotrophoblast, indicating an abundance of glycogen-storing cells in this layer. BTBR and 129 wild-type placentas exhibited a smaller layer of spongiotrophoblast compared to wild-type B6 (Figure 4, D and G), but the layer was well developed and stained strongly for PAS in all wild-type placentas examined. In contrast, there were only a few small clusters of spongiotrophoblasts in the BTBR and 129-Egfrwa2 homozygous placentas (Figure 4, E and H). Closer examination of these clusters revealed some PAS staining (Figure 4, F and I, arrowheads).
There were no detectable differences in the structure of the labyrinth region between 129 wild-type and Egfrwa2 homozygous 18.5-dpc placentas (Figure 4, J and K). Also, no obvious reduction was observed in the spongiotrophoblast layer of 129-EgfrWa5 heterozygous placentas (Figure 4L) when compared to wild type (Figure 4G).
Expression of markers for specific trophoblast cell subtypes differed in Egfrwa2 homozygous vs.
Egfrwa2 homozygous and EgfrWa5 heterozygous placentas were molecularly characterized using a real-time PCR screen. The relative expression of trophoblast cell subtype markers Gcm1, Dlx3, Tcfeb, Esx1, Esrrb, Eomes, Gm52, Ctsq, 4311, Pdch12, Pl1, Timp2, Glut3, and Cx31 were measured by quantitative PCR (Table 3) and compared to an endogenous control, Gusb, in 15.5-dpc placentas. Five to 10 placentas were analyzed for each genotype and strain. Significant differences between Egfrwa2 homozygous and wild-type placentas were found in the expression of several placental genes (Table 4). In B6 and 129, several labyrinth-expressed genes, Gcm1, Dlx3, and Tcfeb, were expressed 30–40% higher in Egfrwa2 homozygous placentas compared to control littermates (P < 0.001–0.05). In 129 Egfrwa2 homozygous placentas, Gm52 was 164% of wild-type levels (P < 0.01). Pdch12, a marker of glycogen cells, and 4311, a marker of spongiotrophoblast, were significantly reduced in both B6 and 129 Egfrwa2 homozygotes with 4311 expressed at particularly low levels in the 129 homozygotes (17% of wild type, P < 0.0001). In B6 homozygotes, a gap junction protein expressed in the glycogen trophoblast Cx31 was 64% of wild-type levels (P < 0.001), and a decidua marker, Timp2, was expressed at 74% of wild-type levels (P < 0.01). In 129 homozygotes, expression of a marker of sinusoidal labyrinth giant cells, Ctsq, was 78% of wild-type levels (P < 0.01). The only significant change in expression between Egfrwa2 and wild-type placentas was for the trophoblast giant cell marker Pl1, which was expressed at 175% of wild-type levels in 129-Egfrwa2 heterozygous placentas. This difference may be related to the higher weight of the 129-Egfrwa2 heterozygous placentas compared to placentas from wild-type littermates.
Changes in CT (ΔCT) values that were significantly different between Egfrwa2 homozygous and wild-type placentas were analyzed with Cluster and visualized with TreeView. Samples from the Egfrwa2 crosses clustered strongly by strain for all probes analyzed. Consistent with analysis of differences in fold change in gene expression, ΔCT values also clustered by genotype for the 129 background, with the Egfrwa2 homozygous samples showing high expression of labyrinth-specific genes and low expression of spongiotrophoblast-specific genes (Figure 5A). The B6 samples showed some genotype-specific clustering but not as strongly as 129 samples did.
Compared to placentas from Egfrwa2 homozygotes, there were fewer differences observed in expression between EgfrWa5 heterozygous and wild-type placentas (Table 3). For the B6 background, EgfrWa5 heterozygous expression of Gcm1, Dlx3, and Esx1 was ∼20% higher than in wild type (P < 0.01–0.05). BTBR-EgfrWa5 heterozygotes showed higher expression of Gm52 compared to wild type (133%, P < 0.05), and there were no significant expression differences between EgfrWa5 heterozygous and wild-type placentas on the 129 background. The ΔCT values for genes from EgfrWa5 samples clustered strongly by strain but not by genotype (Figure 5B).
Wild-type placenta weights, embryo weights, and expression of trophoblast markers are strain dependent:
Wild-type placenta and embryo weights were compared at 15.5 dpc for the three strains (Figure 6). B6 placentas and embryos were the largest of the three strains at 15.5 dpc with an average weight of 98.3 and 385 mg for the placenta and embryo, respectively. Consistent with reduced survival of Egfr mutant embryos on 129 and BTBR backgrounds, wild-type 129 placentas and embryos were the smallest with an average of 73.9 mg for the placenta and 318.6 mg for the embryo, while BTBR placentas had an average weight of 82.3 mg and the embryos an average weight of 332.5 mg. Placenta and embryo weights were significantly different in all strain comparisons (P < 0.001).
Clustering the ΔCT values of the EgfrWa5 data set by sample revealed interesting strain-specific differences in wild-type placenta gene expression. BTBR samples were not included in the cluster analysis because the endogenous control, Gusb, was expressed at a significantly higher level in BTBR placentas compared to 129 and B6 on the basis of total RNA levels (P < 0.0005). Placentas from the 129 background showed high expression of the labyrinth-specific genes Tcfeb, Dlx3, Gm52, Gcm1, Esx1, and Ctsq compared to B6 (Figure 5B), and B6 placentas showed higher expression of Eomes, 4311, Glut3, Pl1, Esrrb, and Pdch12 than 129.
EGFR and intrauterine growth restriction:
Numerous studies have provided evidence that EGFR and its ligands are important for normal growth of the placenta and embryo. Overexpression of the EGFR ligand, EGF, has been found to reduce fetal growth in both humans and mice. In humans, a polymorphism in the 5′ untranslated region of EGF that results in increased EGF expression has been associated with lower birth weight and fetal growth restriction in pregnant women from Western Europe (Dissanayake et al. 2007). In addition, transgenic mice that overexpress EGF are born at half the weight of their littermates and have lower levels of serum IGFBP3 (Chan and Wong 2000). Interestingly, reduced EGF and EGFR phosphorylation have also been associated with low birth weight. Several groups have found associations between IUGR and diminished placental EGFR expression and/or activation in human pregnancies (Fujita et al. 1991; Fondacci et al. 1994; Gabriel et al. 1994; Faxen et al. 1998; Calvo et al. 2004). In pregnant mice, reduction of maternal EGF by sialoadenectomy results in growth restriction of embryos (Kamei et al. 1999). Also, EGFR-deficient mouse embryos exhibit placental defects that are dependent on strain and result in embryonic growth restriction and lethality (Sibilia and Wagner 1995; Threadgill et al. 1995). The effects of genetically reduced, but not abolished, EGFR signaling on placental development and embryo growth has not been previously reported. In this study, we examined the strain-specific effects of two reduced-function alleles of Egfr on placental and embryonic growth and the expression of trophoblast cell subtype markers in the placenta.
Egfrwa2 homozygous placentas were significantly smaller than wild type on all three genetic backgrounds examined, but the growth of Egfrwa2 homozygous placentas and embryos during late gestation differed by strain (Figure 7). Compared to B6 and BTBR, growth of the 129-Egfrwa2 homozygous placenta and embryo slowed the most during this time period and 129-Egfrwa2 homozygous embryos showed severe growth restriction at 18.5 dpc, a phenotype not observed at 15.5 dpc. The BTBR-Egfrwa2 homozygous placenta grew relatively more than Egfrwa2 homozygotes on other backgrounds between 15.5 and 18.5 dpc; however, BTBR-Egfrwa2 homozygous embryos were more growth restricted at 18.5 dpc compared to 15.5 dpc. In contrast, no significant changes were observed in the growth rate of B6-Egfrwa2 homozygous placentas and embryos across late gestation.
Trophoblasts in the placental labyrinth facilitate maternal-fetal exchange of nutrients required for normal embryonic growth and development. The Egfrwa2 homozygous labyrinth appeared well differentiated at the histological level, suggesting that defects in this placental layer do not significantly contribute to the growth restriction observed in Egfrwa2 embryos. Trophoblasts in the junctional zone, spongiotrophoblast, glycogen cells, and trophoblast giant cells have been shown to synthesize various hormones that regulate embryonic growth directly and/or indirectly through modulation of the maternal physiological response to pregnancy. Pronounced growth restriction was observed in 129 and BTBR-Egfrwa2 homozygous embryos, the two strains that also exhibit severely reduced spongiotrophoblast and glycogen cells. Additionally, placental size was correlated with embryonic growth in these strains, further suggesting that the placental growth restriction causes the embryonic growth phenotype. In contrast, B6-Egfrwa2 homozygous embryos were not growth restricted, and their placentas contained a more robust layer of spongiotrophoblast and glycogen cells.
Origins of genetic background-dependent placental phenotypes:
Strain-dependent placental phenotypes have been previously reported in Egfrtm1Mag nullizygous embryos, but the specific role of modifying genes remains unknown. In this study, we demonstrated that strain-dependent differences also exist in wild-type placentas and embryos. Weights of placentas and embryos are significantly different among B6, 129, and BTBR strains with B6 exhibiting the largest and 129 the smallest placentas and embryos. Histological comparison of wild-type placentas from the three strains showed that the numbers of spongiotrophoblasts and the intensity of PAS staining also varied by strain.
Real-time PCR data comparing the expression of trophoblast cell subtype-specific genes in 129 and B6 suggest that, in addition to a difference in size, placentas from the strains may consist of different proportions of trophoblast layers, and/or the level of gene expression may vary. Clustering of ΔCT values revealed that even with Egfr alleles that affect placental composition, the data still clustered most strongly by strain rather than by genotype. Placentas from 129 embryos showed a relatively high expression of a set of labyrinth-specific genes while B6 exhibited the highest expression of a separate set of genes that included Eomes, 4311, Pl1, and Glut3. The relatively high expression of Glut3 in B6 is interesting, considering the role of GLUT3 in embryonic growth. Embryos heterozygous for a null allele of Glut3 display late gestational IUGR, and placental Glut3 expression is reduced in growth-restricted embryos from EGF-deficient sialoadenectomized dams (Kamei et al. 1999; Ganguly et al. 2007). Elevated expression of Glut3 in B6 placentas may allow Egfrwa2 homozygous embryos to escape the severe growth restriction observed in the 129 and BTBR backgrounds.
These strain-specific differences in wild-type embryos are not surprising, given the fact that the placenta is an organ affected strongly by natural selection (Coan et al. 2005; Angiolini et al. 2006). Many imprinted genes play a role in growth and development of the placenta, and during the derivation and maintenance of distinct mouse strains, polymorphic genes that influence placental growth may be fixed in different combinations. The unique placental composition and/or expression of genes known to play important roles in the trophoblast differentiation observed in standard wild-type laboratory mouse strains is interesting considering the large number of transgenic and mutant models with reported placental defects leading to embryonic lethality (Rossant and Cross 2001; Watson and Cross 2005). For some of these models, such as the Egfrtm1Mag nullizygous mouse, the embryonic lethal phenotype is dependent on genetic background, suggesting that the causative placental defects probably vary by strain (Strunk et al. 2004). The inherent strain-specific differences that we have observed in wild-type placenta indicate that the response of the placenta to genetic changes may be determined, in part, by strain-specific trophoblast characteristics.
EgfrWa5 heterozygotes retain more EGFR signaling than Egfrwa2 homozygotes:
In contrast to Egfrwa2 homozygotes, we observed only small decreases (5–10%) in placental weight of EgfrWa5 heterozygotes for the three strains with no significant effect on embryonic weight. Histological changes were not obvious in EgfrWa5 heterozygous placenta and, molecularly, EgfrWa5 heterozygous placentas were more similar to wild type than were Egfrwa2 homozygotes. Also unlike Egfrwa2 homozygotes, there was no embryonic lethality of 129-EgfrWa5 heterozygotes prior to 15.5 dpc. However, we did observe significantly fewer EgfrWa5 heterozygotes than expected for the BTBR background, but the reason remains to be determined. Together, our placenta and embryo weight measurements, histology, and gene expression data show that the EgfrWa5 heterozygous phenotype is less severe than the Egfrwa2 homozygous phenotype.
Recent reports provide evidence for an asymmetric dimer model of EGFR activation (Zhang et al. 2006). Studies have shown that in an ERBB dimer one of the receptors, the activator, acts to hold the other, the activated receptor, in a conformation that promotes its activation and subsequent auto-phosphorylation. The N-lobe of the activated receptor makes critical contacts with the C-lobe of the activator, and mutations that disrupt this interaction generally result in reduced or abolished phosphorylation. A kinase-dead EGFR, such as from the EgfrWa5 allele, is capable of acting as the activator but not the activated receptor. According to this model, EGFR signaling in the EgfrWa5 heterozygote would occur normally through the wild-type EGFR dimer, to some extent through the EGFRWa5/EGFR dimer, and not at all through the EGFRWa5/EGFRWa5 dimer. However, in vitro experiments have shown that EGFRWa5 acts as a dominant negative and has a more severe effect on wild-type EGFR phosphorylation than does a kinase-dead receptor. EGFR phosphorylation is reduced by ∼90% when cells express equal amounts of EGFR and EGFRWa5 (Lee et al. 2004). Thus, the EgfrWa5 mutation not only renders the receptor kinase-dead but also affects receptor activation through an additional mechanism, perhaps by modifying conformation of the receptor that it encodes and/or disrupting assembly of higher-order receptor oligomers.
Estimates of EGFRwa2 receptor-signaling capabilities have varied from 10% to almost wild-type levels of activity, depending on the cell type analyzed and the experimental approach (Luetteke et al. 1994). The Egfrwa2 mutation lies upstream of EgfrWa5 in an α-helix portion of the receptor N-lobe (Luetteke et al. 1994; Fowler et al. 1995). The effect of the Egfrwa2 mutation on EGFR phosphorylation is not well understood, but it is possible that the mutation compromises contact with the C-lobe portion of the activator directly or indirectly by altering conformation of the activated receptor. Du et al. (2004) proposed that Egfrwa2 homozygotes and EgfrWa5 heterozygotes have approximately the same reduction in EGFR signaling. On the basis of the more severe phenotype observed in Egfrwa2 homozygous placentas, we propose that the following levels of EGFR signaling occur in the Egfr allelic series. The levels are arranged in order from wild-type levels of EGFR activity to complete absence of EGFR activity (“≈” indicates allele combinations predicted to produce similar levels of EGFR activity):
Our data also demonstrate that tissue-specific requirements for EGFR signaling can be determined using the allelic series. We have shown that normal development of the placenta requires less EGFR activity than morphogenesis of hair follicles since the Egfrwa2 and EgfrWa5 mouse share the same wavy coat phenotype but not the same degree of placental defects.
In conclusion, our study highlights strain-dependent variation in placental development as well as the effect of diminished EGFR signaling on placental and embryonic growth. IUGR is a common condition with profound consequences for the fetus, including elevated risk for perinatal mortality and increased incidence of reduced cognitive function, diabetes, and heart disease later in life (Barker et al. 2002). It is known that a large number of IUGR cases are caused by placental defects, but the precise developmental mechanisms are not well understood. Egfrwa2 homozygous embryos may serve as a model for investigating growth restriction arising from placental dysfunction. In addition, EgfrWa5 heterozygotes can be used to study levels of EGFR signaling intermediate between the wild-type and the Egfrwa2 homozygote.
This work was supported by National Institutes of Health grant HD39896 to D.W.T. and HD046970 to K.M.C.
Communicating editor: L. Harshman
- Received April 25, 2009.
- Accepted June 10, 2009.
- Copyright © 2009 by the Genetics Society of America