Targeted disruption of cubilin reveals essential developmental roles in the structure and function of endoderm and in somite formation

Generation of cubilin^-/-mice

To generate mice with targeted disruption of the cubilin gene (and concomitant knock-in of the EGFP reporter gene) we cloned and fully mapped the exon-intron structure of the 5' portion of the murine cubilin gene. A mouse cubilin gene-targeting vector was designed to create a null mutation through deletion of exons 1–6 (Fig. 1). After electroporation of the construct and G418 positive selection and FIAU negative selection, ten ES clones (out of 132 screened) were identified that had the desired recombination based on Southern analysis (using both upstream and downstream flanking probes) (Fig. 1). Targeted ES clones were injected into C57BL/6J blastocysts and the blastocysts were transferred to foster mothers to obtain chimeric mice. Two male chimeras were obtained from the first targeted ES cell line tested and found to be germ line competent through the generation of heterozygous offspring. As shown in Figure 1C, Southern analysis confirmed that offspring from one of these mice contained the properly targeted cubilin allele.

The cubilin exon 1–6 deletion leads to embryonic lethality

Retrograde genotypic analysis was performed on embryos from heterozygous intercross matings (Table 1). At 7.5 dpc homozygous embryos were obtained at a frequency consistent with a normal Mendelian ratio. From 8.5–10.5 dpc, homozygous embryos were detected, but not at a frequency in accordance with Mendelian expectations (Table 1). After 13.5 dpc, no homozygous embryos were detected. Together, these findings indicate that homozygous embryos die over a range of developmental stages between 7.5 to 13.5 dpc. Consistent with this conclusion was the relatively high number of embryo resorptions observed after 7.5 dpc (Table 1).

To establish that embryos homozygous for the targeted cubilin gene deletion of exons 1–6 were indeed cubilin null, RT-PCR analysis was performed using two primer pairs designed to amplify separate 5' and 3' regions of the cubilin transcript. No cubilin mRNA was detectable by either primer pair using RNA isolated from homozygous 8.5 dpc embryos (Fig. 2). Furthermore, immunohistological analysis confirmed that there was no cubilin detectable in paraffin embedded sections of homozygous 8.5 dpc embryos (data not shown).

Developmental retardation and mesodermal defects of homozygous embryos

Morphological analysis of 8–8.5 dpc embryos revealed that homozygous embryos were considerably smaller than wild-type or heterozygous littermates (Fig. 3). Mutant 8–8.5 dpc embryos were morphologically similar to wild-type 7.5 dpc embryos. However, rather than exhibiting a normal cylindrical shape, mutant 8–8.5 embryos appeared acorn shaped owing to a relatively smaller embryonic component as compared to extraembryonic component.

Histological examination of null 8–8.5 dpc embryos revealed that all mutant embryos underwent gastrulation, but did not form somites (Fig. 3F and 3G). In contrast, 8–8.5 dpc wild-type littermates had formed 3–9 somites. Other mesodermal structures did form in the 8–8.5 dpc mutants, albeit with some variability. For example, an allantois was present in 5 of 7 mutant 8–8.5 dpc embryos examined. Most mutant embryos (5 of 7) also had an amnion and chorion, tissues having mesoderm components. While blood islands formed in the yolk sacs of all 8–8.5 dpc mutants they appeared unusually large, extending into the exocoelomic cavity to a greater extent than wild-type blood islands (Fig. 4). Such blood islands appeared to have a larger than normal number of mesodermal cells. Additionally, blood island hematopoietic development appeared to be retarded as evidenced by the failure of the cells to exhibit the characteristic rounded hematopoietic cell morphology (Fig. 4).

By 9.5 dpc, surviving homozygous embryos had not undergone axial rotation but displayed an overall morphological appearance of wild-type 8–8.5-dpc embryos (Fig. 5). Mutant 9.5 dpc embryos had formed a primitive heart and neural head folds (4 of 4 nulls examined) (Fig. 5) and paired dorsal aortae (data not shown), but lacked somites. Homozygous embryos surviving beyond 9.5 dpc (i.e., 11.5 dpc) were also grossly similar to wild-type 8.0 dpc embryos, but still lacked somites (data not shown). Together these findings indicate that lack of cubilin results in developmental retardation that initiates prior to 8.0 dpc, and a failure to achieve developmental milestones appropriate for a wild-type 8.0 dpc embryo, most notably, failure to form somites. The complete lack of somites at 9.5 dpc suggests that the basis is due to some defect of the paraxial mesoderm rather than merely being a consequence of retarded development.

It is important to note that embryos confirmed by genotyping to be heterozygous were indistinguishable morphologically from wild-type littermates over a range of stages examined extending from 7–11.5 dpc (Fig. 3B). This indicates that the lethality of heterozygous embryos, apparent from genotypic analysis, is occurring beyond 11.5 dpc.

Failure of yolk sac blood vessels to undergo remodeling

To assess the effects on blood vessel formation, embryos from heterozygous matings were immunolabeled with anti-PECAM-1. As shown in Figure 6A, yolk sac blood vessel formation in 9.5 dpc mutants appeared retarded as compared to wild-type littermates (Fig. 6B). Normally by 9.5 dpc the yolk sac vessels of the wild-type embryo have undergone remodeling to form larger diameter vessels (Fig. 6B, inset). By contrast, the pattern of yolk sac vascular development in 9.5 dpc mutants appeared as an interconnected network of small diameter blood vessels (Fig. 6A), a pattern similar to that seen in an 8.5 dpc wild-type embryo yolk sac. In some areas of the extraembryonic yolk sac, PECAM-1 labeling often showed enlarged sinusoidal-like vessels (data not shown), which was consistent with the enlarged blood island-like structures observed in H&E stained sections (Fig. 4).

When the PECAM-1-labeled yolk sac vasculature of 11.5 dpc mutants was compared to that of wild-type embryos it was again evident that the blood vessels of homozygous embryos had not progressed beyond that of 8.5 dpc wild-type embryos. As shown in Figure 6D, yolk sac blood vessels of an 11.5 dpc mutant had not undergone the remodeling that normally occurs after 8.5 dpc.

In addition to the yolk sac vascular remodeling anomalies, the allantoic vasculature of 11.5 dpc mutants was aberrant. Instead of a branched central blood vessel normally present in an 8.5 dpc embryo [17], the allantoic blood vessels of 11.5 dpc mutants were bulbous and lacked branching (Fig. 6D). Since this abnormality was not observed in 9.5 dpc mutants, it can be concluded that dysmorphogenesis of the allantoic vasculature occurs after 9.5 dpc and perhaps contributes to a chorioallantoic placental defect.

Abnormalities of embryonic and visceral endoderm in cubilin mutants

Considering that cubilin is normally present on the apical surface of the columnar VE (from 6.0 dpc to at least 9.5 dpc) and on a population of the squamous cells within the embryonic endoderm (EE) (from 7.3–8.0 dpc) [13], we next evaluated the impact of cubilin deficiency on these embryonic endodermal tissues. Examination of homozygous 8–8.5 dpc embryos revealed a number of endodermal anomalies (Fig. 7). In place of a normal squamous definitive endoderm (Fig. 7C), the mutant embryos had both cuboidal and columnar cells arranged in a stratified cuboidal epithelium (Fig. 7D, arrow) as well as simple columnar epithelium (not shown). Additionally, mutant cells in these epithelia had apical processes that may be either microvilli or cilia. Such luminal surface structures are normally not found on EE cells. Based on the findings, the absence of cubilin inhibits the formation of an epithelium morphologically comparable to that of a normal definitive endoderm at this stage, however, it remains to be determined whether this is a result of a defect in the specification and differentiation of the definitive endoderm.

Examination of the VE of homozygous 8–8.5 dpc embryos revealed it to also be morphologically abnormal. In contrast to the simple columnar VE of normal embryos, the VE of cubilin mutants was a stratified epithelium, comprised of cuboidal and columnar cells. Furthermore, mutant VE cells also lacked the large apical vacuoles that are characteristic of VE cells (Fig. 7E and 7F). Given the morphological similarities between the cells of mutant embryonic endoderm there was not a discernable demarcation between the normally distinct EE and VE epithelia. However, a clear demarcation between the VE and an epithelium that would normally be EE was evident upon immunohistological analysis of GFP reporter expression (from the EGFP cassette inserted into the cubilin gene) and megalin expression in mutant 8.0 dpc embryos (Fig. 8). This indicated that although being morphologically similar, the VE and presumed EE of mutants were distinct at the molecular level. Furthermore, the fact that megalin was appropriately expressed on the apical surfaces of the mutant VE indicated that cubilin is not required to direct cell surface trafficking of megalin.

The visceral endoderm of homozygous embryos does not mediate uptake of maternal-derived HDL

A normal function of VE is to mediate uptake of maternal-derived HDL [8]. We therefore evaluated the capacity of cubilin-deficient VE to mediate uptake of maternal-derived HDL. As shown in Figure 9, the VE of heterozygous 8.0 dpc embryos from DiI-HDL-infused mothers showed strong DiI labeling within VE cells. DiI label in these embryos was present exclusively in VE which is the only yolk sac endodermal component expressing cubilin as evidenced by expression of the cubilin gene reporter, EGFP. By contrast, there was no detectable DiI labeling of the VE of homozygous embryos (Fig. 9D–F). These findings indicate that VE uptake of HDL is a cubilin dependent process.

Targeting vector design and generation of mutant mice

Degenerate deoxyoligonucleotide primers 5'-CTICACCARCCICGIATG-3' and 5'-CCRTTRRATYTCRCAYTC-3' were synthesized based on cDNA sequences conserved in the 5' region of both rat cubilin (GI: 24475743) and human cubilin (GI: 3929528). Preparation of template cDNA from adult mouse kidney total RNA and PCR amplification were done using methods described previously [8]. Cycling parameters for PCR amplification were: 94°C for 5 min and then 30 cycles of 94°C for 1 min, 46°C for 1 min and 72°C for 1.5 min. The expected ~820 bp product was sequenced and found to have 93% identity with a 5' portion of the rat cDNA sequence.

The mouse cubilin cDNA was used as a probe to screen a 129-strain mouse genomic library (Genome Systems, Inc.) and a BAC clone (24267) was isolated. DNA sequencing of ~20 kb of the BAC clone was performed to characterize the intron-exon organization of the 5' portion of the cubilin gene, including the region containing exons 1–10. A search of GenBank showed that the sequence of BAC clone 24267 was contained within the mouse chromosome 2 genomic contig GI:82796355.

The 3' arm of cubilin replacement-type targeting vector was constructed from a 5.4-kb EcoRI fragment containing exons 7 and 8 of the mouse cubilin gene. The 5' arm was constructed from a 4.3 kb ApaI-NaeI fragment containing cubilin exon 1 disrupted by insertion of a enhanced green fluorescent protein (EGFP)-N1 (Clontech, Mountain View, CA), neo^R (loxP floxed) cassette 33 nucleotides 5' to the first ATG (see Additional file 1). A tk negative selection cassette was placed at the 5' end of this construct. The resulting targeting vector was linearized by NotI digestion and electroporated into murine 129/Sv-strain embryonic stem (ES) cells. After electroporation of the construct and G418 positive selection and 1-(2-deoxy-2-fluoro-8-d-arabinofuranosyl)-5-iodouracil (FIAU) negative selection, clones having the desired homologous recombination were identified by Southern analysis (Fig. 1C). One of the targeted ES clones was injected into C57BL/6J blastocysts that were then transferred to foster mothers to obtain chimeric mice. Two male chimeras were obtained that were germ line competent. Southern analysis confirmed that offspring from these mice contained the properly targeted cubilin allele. Mice used in this study were maintained on a mixed 129Sv/C57BL/6 genetic background.

Genotypes of progeny from heterozygote intercrosses were determined by PCR. For embryos between 7.5 and 8.5 days postcoitum (dpc) DNA was isolated from the whole embryo and used in PCR. For embryos between 9.5 and 14.5 dpc, DNA for genotyping was isolated from the yolk sac. Two primer pairs were used for PCR-based genotyping. To detect the wild-type cubilin allele, PCR was performed with tail clip genomic DNA preparations using a cubilin sense strand primer, 5'-GCCAAGTAGACCAGGCTGAC-3' (residues 10422223–10422242 in GI: 82796355, and an antisense strand primer, 5'-GCTTCTGAGCCCAGTGAAAC-3' (residues 10422576–10422595 in GI: 82796355). Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 55°C for 1 min and 72°C for 1 min. The expected amplicon size is 373 bp. To detect the targeted cubilin allele, PCR reactions were performed with an EGFP sense strand primer, 5'-CCTGAAGTTCATCTGCACCA-3' (residues 810–829 in GI:1377911), and EGFP antisense strand primer 5'-TGCTCAGGTAGTGGTTGTCG-3' (residues 1288–1269 in GI:1377911). Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 55°C for 1 min and 72°C for 1 min. The expected amplicon size is 478 bp.

For embryos from intercross heterozygous matings that were paraffin embedded within maternal tissue, the assignment of 'mutant' was based on two criteria: 1) by the absence of immunologically detectable cubilin (anti-cubilin T-16 from Santa Cruz Biotechnology, Santa Cruz, CA) in sections of the paraffin embedded embryos; and 2) the distinct mutant morphological phenotype corresponding to that of embryos confirmed by genotypic analysis to be homozygous nulls.

Immunohistochemistry and histology

Embryos were harvested from pregnant females following timed cubilin+/- intercross matings, and fixed for 40 min in 4% paraformaldehyde/PBS. Hematoxylin and eosin (H&E) staining of paraffin embedded embryo sections (7 μm) was performed using standard techniques. Whole-mount immunohistochemistry for the PECAM1/CD31 was performed as described previously [30]. Antibodies to PECAM (clone Mec-13.3) were purchased from BD PharMingen (San Diego, CA). For GFP and megalin immunohistological staining, sections were stained with rabbit antibodies to GFP purchased from Abcam (Cambridge, MA) or rabbit megalin cytoplasmic domain antibodies described previously [13]. Fluorescently conjugated secondary antibodies were purchased from Jackson ImmunoResearch Labs (West Grove, PA). Stained tissue sections were analyzed using a Leica DMR research-grade microscope equipped with Leica objectives and a SPOT-RT camera (Diagnostic Instruments, Sterling Heights, MI).

RT-PCR analysis

Total RNA was isolated from 8.5 dpc embryos from heterozygous matings using Trizol (Invitrogen). Template cDNA was prepared from 1 mg RNA using Superscript II Reverse Transcriptase (Invitrogen). To assess cubilin transcript levels two sets of PCR reactions were performed using a primer pair targeting the 5' end of the transcript and one targeting the 3' end. The 5' primer pair consisted of the forward primer 5'-ATGATGATGACCTTGGCGAATG-3' (residues 275–296 (exon 2) in GI: 94365995) and the reverse primer 5'-GCAGCCAAAGGGTGTTCCAG-3' (residues 587–606 (exon 6) in GI: 94365995). Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 58°C for 1 min and 72°C for 1 min. The expected amplicon size is 332 bp. The 3' primer pair consisted of the forward primer 5'-TCTCATACACCAACTACCCC-3' (residues 9221–9240 (exon 58) in GI: 94365995) and the reverse primer 5'-AGCAGTCTTGTGAGGGCAGC-3' (residues 10041–10060 (exon 62) in GI: 94365995). Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 58°C for 1 min and 72°C for 1 min. The expected amplicon size is 840 bp. To assess GAPDH transcript levels PCR was performed using a forward primer 5'-CGGTGTGAACGGATTTGGC-3' (residues 70–88 in GI: 47607489) and the reverse primer 5'-GCAGTGATGGCATGGACTGT-3' (residues 581–600 in GI: 47607489). Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 54°C for 1 min and 72°C for 1 min. The expected amplicon size is 531 bp. To assess b-actin transcript levels a forward primer 5'-CGGGACCTGACAGACTACCTC-3' (residues 627–647 in GI: 6671508) and the reverse primer 5'-AACCGCTCGTTGCCAATA-3' (residues 827–844 in GI: 6671508) were used. Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 55°C for 1 min and 72°C for 1 min. The expected amplicon size is 218 bp. To assess EGFP transcript levels a forward primer 5'-ACGTAAACGGCCACAAGTTC-3' (residues 743–762 in GI: 1377911) and the reverse primer 5'-AAGTCGTGCTGCTTCATGTG-3' (residues 910–929 in GI: 1377911) were used. Cycling parameters for PCR amplification were: 98°C for 5 min and then 40 cycles of 98°C for 0.5 min, 58°C for 1 min and 72°C for 1 min. The expected amplicon size is 187 bp.

Analysis of embryonic uptake of maternal derived HDL

Human DiI-labeled HDL (DiI-HDL) was purchased from Biomedical Technologies (Stoughton, MA). DiI-HDL was depleted of apoE-HDL and other heparin-binding particles according to Oram [31], dialyzed against 150 mM NaCl, 50 mM Tris pH 7.4 (TBS) containing 0.3 mM EDTA and filter-sterilized. Lipoprotein concentration was determined by BCA protein assay (Pierce, Rockford, IL). DiI-HDL was diluted into PBS to a final concentration of 0.265 mg protein/ml and 75–125 μl was infused into the saphenous vein of pregnant heterozygous mice at 8.0 dpc according to the procedure of Hem et al. [32]. One hour after infusion, embryos were isolated free of parietal endoderm and analyzed by confocal microscopy.