Murine SEC24D can substitute functionally for SEC24C during embryonic development

The COPII component SEC24 mediates the recruitment of transmembrane cargos or cargo adaptors into newly forming COPII vesicles on the ER membrane. Mammalian genomes encode four Sec24 paralogs (Sec24a-d), with two subfamilies based on sequence homology (SEC24A/B and C/D), though little is known about their comparative functions and cargo-specificities. Complete deficiency for Sec24d results in very early embryonic lethality in mice (before the 8 cell stage), with later embryonic lethality (E7.5) observed in Sec24c null mice. To test the potential overlap in function between SEC24C/D, we employed dual recombinase mediated cassette exchange to generate a Sec24cc-d allele, in which the C-terminal 90% of SEC24C has been replaced by SEC24D coding sequence. In contrast to the embryonic lethality at E7.5 of SEC24C-deficiency, Sec24cc-d/c-d pups survive to term, though dying shortly after birth. Sec24cc-d/c-d pups are smaller in size, but exhibit no other obvious developmental abnormality by pathologic evaluation. These results suggest that tissue-specific and/or stage-specific expression of the Sec24c/d genes rather than differences in cargo export function explain the early embryonic requirements for SEC24C and SEC24D.

. SEC24B-deficient mice exhibit late embryonic lethality at ~ E18.5 due to neural tube closure defects resulting from reduced trafficking of the planar-cell-polarity protein VANGL2 14 . Loss of SEC24C in the mouse results in embryonic lethality at ~ E7.5 12 , with E7.5 Sec24c null embryos exhibiting abnormal gastrulation and thinning of the embryonic ectoderm, suggesting that SEC24C is required in the embryonic ectoderm just prior to gastrulation 12 . In contrast, SEC24D deficiency in mice results in embryonic death at or before the 8-cell stage 15 . Abnormalities in SEC24C function have not been described in humans, whereas bi-allelic point mutations in human SEC24D (the truncating mutation p.Gln205*, p.Ser1015Phe located in a cargo-binding pocket, and p.Gln978Pro located in the gelsolin-like domain) have been reported in patients with a developmental skeletal disorder 16 . The expansion of the number of COPII paralogs over evolutionary time suggests a divergence in cargo recognition function, with the disparate deficient mouse phenotypes resulting from paralog-specific protein interactions with a subset of cargo molecules, or other functions beyond cargo recognition. Consistent with this model, all four mammalian Sec24s are broadly and ubiquitously expressed in all tissues examined 15,17,18 . Nonetheless, subtle differences in the developmental timing and/or tissue-specific patterns of expression as the explanation for the unique phenotypes associated with deficiency for each SEC24 paralog cannot be excluded.
The SEC24C and SEC24D proteins appear to have similar structural domains. To test if the duplicated SEC24C/D genes are fixed in the genome because one of the genes acquired a new function (neofunctionalization) or if the copies of the duplicated gene split the function of the ancestral gene (subfunctionalization) in vivo, we employed dual recombinase mediated cassette exchange (dRMCE) 19 to knock-in the C-terminal 90% of the coding sequence for SEC24D in place of the corresponding SEC24C coding sequence, at the Sec24c locus. Surprisingly, these SEC24D sequences can largely substitute for SEC24C during embryonic development, rescuing the early embryonic lethality previously observed in SEC24C-deficient mice. In contrast, the same SEC24D sequences expressed in the context of the Sec24c gene fail to rescue SEC24D-deficiency. Taken together, these results demonstrate a high degree of functional overlap between the SEC24C/D proteins and suggest that the deficiency phenotypes for each paralog are determined largely by tissue or developmental timing-specific differences in their gene expression programs.

Results
Identification of a dRMCE targeted ES cell clone. Direct microinjection of 102 zygotes generated from a Sec24c +/− X Sec24c +/+ cross with pDIRE and the Sec24c-d replacement construct (Fig. 1) failed to generate any progeny mice with the correctly targeted event; 65 were heterozygous for the Sec24c − allele containing the FRT and loxP sites required for dRMCE 19 and 47 were wild type. While no targeted insertions were observed, 20/102 (20%) pups carried random insertions of the dRMCE replacement vector. A screen of 288 Sec24c GT ES cell clones co-electroporated with pDIRE and the Sec24c-d targeting construct also failed to yield any properly targeted colonies, though 18 random insertions of the Sec24c-d construct (6.25%, Table S1) were observed. However, analysis of a second set of 288 ES cell clones transfected with alternative Cre and FLP expression vectors identified a single clone carrying a potential targeted insertion of the dRMCE replacement construct into the Sec24c locus. A single round of subcloning generated pure clonal ES cell populations carrying the Sec24c c-d allele.
Heterozygosity of the Sec24 c-d allele does not affect embryonic development or gross morphology. One out of the three microinjected ES cell subclones achieved germline transmission, and mice carrying the Sec24c c-d allele were generated. The expected Mendelian ratio of Sec24c +/c-d mice was observed in N2 progeny of backcrosses to C57BL/6J mice (p > 0.38, Table 1A). Sec24c +/c-d mice were indistinguishable from their wild type littermates, exhibiting normal fertility and no gross abnormalities on standard autopsy examination (data not shown). There were also no differences in body weight at 4 and 6 weeks of age (Fig. 2B). While only a Figure 1. Generation of the chimeric Sec24c c-d allele. (A) Schematic representation of dRMCE to generate the Sec24c c-d allele. The replacement vector pUC19-Sec24c-d contains the Sec24c intron 2 splice acceptor (yellow), the Sec24d coding sequence beginning with G 120 (gray), and a stop codon followed by a poly A signal sequence (red). Arrows represent primers used for genotyping, long-range PCR, and RT-PCR (for sequences, see Table S2). (B) The SEC24C-D fusion protein encoded by the dRMCE generated Sec24c c-d allele, which contains the first 57 amino acids of SEC24C followed by the SEC24D sequence corresponding to the remaining ~ 95% of the SEC24C protein.  4). Additionally, the presence of the Sec24d cDNA ("any ins. ") (lanes 5,6), and the loxP and FRT sites (103 bp and 106 bp products in lanes 7,8, respectively) was confirmed. The signal in lane 9 is due to the Sec24c + allele, confirming that ESC clone 12275 is heterozygous for the Sec24c c-d allele. Clone 12275 does not carry any random insertions of pCAGGS-iCre (lane 10) or pCAGGS-Flpo (lane 11). (D) A genotyping PCR assay on mouse genomic DNA from tail clip to distinguish between the wild type and Sec24c c-d allele using primers E, F, and G. (E) Long range PCR confirms correct targeting. Primers GF4 + U were used to amplify the 5′ arm resulting in a 7.6 kb product, and primers F and GR4 were used to amplify the 3′ arm to yield a 5.8 kb product. Primers were located outside the homology arms (GF4 and GR4) and within the Sec24d cDNA (F and U). Neither set of primers yields a band from the Sec24c wt allele.  The Sec24c c-d allele rescues Sec24c −/− mice from embryonic lethality.. Table 1D shows the distribution of offspring from Sec24c +/c-d intercrosses. No Sec24c c-d/c-d mice were observed at weaning (0/163, p < 1.7 × 10 -13 , Table 1D). Genotyping of 31 P0 progeny observed to die shortly after birth, all notably smaller and paler than their surviving littermates ( Fig. 2A), identified 26 as Sec24c c-d/c-d (p < 3.8 × 10 -14 ). The remaining 5 pups were either wildtype (n = 3) or Sec24c +/c-d (n = 3). Taken together with the full set of progeny genotypes at P0 (n = 97), the observed number of Sec24c c-d/c-d offspring is consistent with the expected Mendelian ratios. Sec24c c-d/c-d pups were 20-30% smaller by weight than their littermate controls at P0 (Fig. 2B), were significantly shorter in crown-rump length (Fig. 2C), and often exhibited a hunched appearance involving the shoulder girdle and trunk. Gross autopsy and histologic analyses (performed blinded to genotype) failed to identify any obvious abnormality to account for the neonatal lethality in Sec24c c-d/c-d mice (Fig. 3A). Lungs of both heterozygous and wild type pups exhibited open alveoli lined by flattened alveolar epithelial cells (Fig. 3B), while the alveoli of Sec24c c-d/c-d neonates were open but had thickened walls (5 out of 9) or were uninflated (4 out of 9), and lined by columnar epithelia (Fig. 3B). However, this lung pathology is unlikely to account for the neonatal death, as 7/9 neonates exhibited no breathing movements at delivery. Similarly, Sec24c c-d/c-d embryos did not exhibit a skeletal defect detectable by alcian blue or alizarin red stains (Fig. 3C,D). Furthermore, mice expressing SEC24D from the SEC24C locus demonstrate indistinguishable distribution of the serotonin transporter SERT in the brain by immunohistochemistry compared to wild type littermate controls (Fig. 3E), demonstrating that the SEC24C-D protein is capable of secreting a cargo previously shown to depend on SEC24C (but not on SEC24D) for secretion 20 . At earlier embryonic time points (E12.5 to E17.5), viable Sec24c c-d/c-d embryos were observed at the expected ratios (p > 0.78, Table 1D), and although clearly smaller than their littermate controls, no other significant histopathologic differences were identified at E15.5 or E17.5 upon observation by an independent reviewer blinded to sample genotypes (Fig. 4). Furthermore, the Sec24c c-d/c-d embryos were pink, alive, and moving at E18.5. Thus, expression of SEC24C-D in place of SEC24C is sufficient to support development to term, but not survival past birth.    Table 2 shows the results of intercrosses between Sec24c +/c-d Sec24d +/GT and Sec24d +/GT15 mice. No Sec24c +/c-d Sec24d GT/GT progeny were observed at 2 weeks of age (n = 113, p < 1.43 × 10 -5 , Table 2) or at embryonic time points from the blastocyst stage to birth (n = 94, p < 7.56 × 10 -5 , Table 2), although a single Sec24c +/c-d Sec24d GT/GT embryo was detected at the 8-cell early morula stage. These results are indistinguishable from the pattern previously reported for Sec24d GT/GT mice 15 .
These results indicate that the Sec24c c-d allele fails to complement disruption of Sec24d, a finding that is not surprising given that expression of Sec24d from the Sec24c locus is unlikely to recapitulate the exact expression of Sec24d from its own genomic locus. Sec24c +/c-d Sec24d +/GT mice were viable and healthy and observed in the expected numbers (p > 0.72, Table 1B,C), and like the Sec24c +/c-d mice, there was no significant difference in lifespan between Sec24c +/c-d Sec24d +/GT mice (n = 24) compared to controls. The absence of exon 3 in this transcript (also seen in the Sec24c +/− samples) results in a frame-shift and early termination codon, as previously described 12 .

Discussion
We show that SEC24D coding sequences inserted into the Sec24c locus (Sec24c c-d ) largely rescue the early embryonic lethality observed in Sec24c −/− mice 12 . In contrast, and as expected, Sec24c c-d is unable to substitute for SEC24D in Sec24c +/c-d Sec24d GT/GT mice. Our findings suggest that SEC24D can largely or completely substitute functionally for SEC24C during embryonic development, when expressed from the Sec24c locus. The incomplete rescue of SEC24C deficiency by the substituted SEC24D sequences could be due to imperfect interaction between the residual 57 amino acids of SEC24C retained at the N-terminus with the remaining 992 C-terminal amino acids of SEC24D. Alternatively, the targeting of Sec24d cDNA sequences into the Sec24c locus could have potentially disrupted regulatory sequences important for the control of Sec24c gene expression. We also cannot exclude a "passenger" gene effect 21 of another, incidental mutation in or near the Sec24c c-d locus, since only a single targeted allele was characterized. Finally, it should be noted that there are two alternative splice forms of    26 , N-myc and c-myc 27 , Oxt2 and Oxt1 28 , and members of the Hox gene family, including Hoxa3 and Hoxd3 29 , all of which were carried out by traditional knock-in approaches with cDNA targeting constructs and homologous recombination. The proteins encoded by these genes are involved in key tissue-specific transcriptional and regulatory pathways, where such complementarity at the level of protein function might be expected. Our finding of a similar complementarity between paralogs of a key cytoplasmic structural component present in all eukaryotic cells, has fewer precedents, aside from our recent demonstration of overlapping function for SEC23A and SEC23B 30 . Though we cannot exclude subtle differences in protein function, the remarkable extension of survival from E7.5 to E18.5 and generally normal pattern of embryonic development in Sec24c c-d/c-d mice demonstrate a high degree of functional overlap between SEC24C and SEC24D as well as the critical importance of spatial, temporal, and quantitative gene expression programs in determining the phenotypes of SEC24C and SEC24D deficiency.
Several secretory protein cargos have been shown to exhibit specificity for an individual SEC24 paralog, including the dependence of VANGL2 on SEC24B 14 , SERT, SLC6A14, and autotaxin on SEC24C 20,31,32 , the GABA1 transporter on SEC24D 33 , and others 34 . However, there is also evidence for significant overlap among the cargo repertoires of the mammalian Sec24 paralogs, particularly within the subfamilies 34 . Several cargo exit motifs are recognized by multiple SEC24 paralogs, including the DxE signal on VSV-G, and the IxM motif on syntaxin 5, both of which confer specificity for human SEC24A/B 35 . The human transmembrane protein p24-p23 exhibits a preference for SEC24C or SEC24D and is thought to be a cargo receptor for GPI-anchored CD59, explaining the specificity of the latter for SEC24C/D 36 . Similarly, PCSK9, which is dependent on SEC24A for ER export shows some overlap with SEC24B, both in vivo and in vitro, but none with SEC24C or D 13 .
Taken together with our results, these previous reports suggest significant functional overlap within but not between the SEC24A/B and SEC24C/D subfamilies. This model is also consistent with the ~ 58-60% sequence identity between SEC24A and B and between SEC24C and D, but only ~ 25% between the A/B and C/D subfamilies. This high degree of functional overlap within SEC24 subfamilies also suggests that the deficiency phenotypes observed for loss of function for any of the SEC24 paralogs may be due in large part to subtle differences in their finely tuned expression patterns, despite reports of generally ubiquitous expression for all 4 paralogs 12 . Consistent with these findings, the same Sec24c c-d allele reported here was recently shown to rescue the neuronal phenotype observed in mice with deletion of Sec24c in neuronal progenitor cells 37 .
Humans with compound heterozygous point mutations in SEC24D present with skeletal disorders such as Cole-Carpenter syndrome and severe osteogenesis imperfecta 16 , with the medaka vbi 38 and the zebrafish bulldog 22 mutants exhibiting similar skeletal defects. These results have been interpreted as indicating a specific critical role for SEC24D in the secretion of extracellular matrix proteins 16,22,38 . However, SEC24D-deficient mice exhibit very early embryonic lethality 15 , at a time in development well before the establishment of the skeletal system.
A similar discrepancy in phenotypes between mice and humans has been observed for SEC23B deficiency, which manifests as congenital dyserythropoietic anemia type II in humans [39][40][41] , and perinatal lethality due to profound pancreatic degeneration in mice [42][43][44][45] . In contrast, mice with SEC23A deficiency exhibit a phenotype reminiscent of the human disease (cranio-lenticulo-sutural-dysplasia) resulting from loss of function mutations in SEC23A [46][47][48] . A recent report demonstrated that SEC23A can functionally replace SEC23B when expressed from the endogenous regulatory elements of Sec23b in mice 30 and that the expression of the SEC23 paralogs has shifted during the course of evolution. Consistent with this model, SEC23B is the predominantly expressed paralog in the mouse pancreas with comparable expression of SEC23A/B in mouse bone marrow, while in humans, SEC23B is predominantly expressed in the bone marrow, with comparable expression of SEC23A/B in www.nature.com/scientificreports/ the pancreas. These results likely explain the disparate phenotypes of SEC23A and SEC23B deficiencies within and across species 30 . Our data suggest that evolutionary shifts in the expression programs for the Sec24c and Sec24d genes may explain the disparate phenotypes resulting from SEC24C/D deficiencies between/across vertebrate species, despite considerable overlap at the level of SEC24C/D protein function. Such changes in relative levels of gene expression could result in major differences in dependence on one or the other paralog across tissue types, even among closely related species.

Methods
Cloning of Sec24c-d dRMCE construct pUC19-Sec24c-d. pUC19-Sec24c-d (Fig. 1) was generated by assembling the Sec24c-d cassette (GenBank accession KP896524) which contains a FRT sequence, the endogenous Sec24c intron 2 splice acceptor sequence, a partial Sec24d coding sequence (from G 120 to A 3099 in the cDNA sequence, encoding the SEC24D sequence starting at Val41 and the SV40 polyA sequence present in the Sec24c GT allele 12 . The entire cassette was inserted into pUC19 at the HindIII and EcoRI restriction sites, and the integrity of the sequence was confirmed by DNA sequencing. Plasmid purification and microinjections: pDIRE, the plasmid directing dual expression of both iCre and FLPo 19 was obtained from Addgene (Plasmid 26745). Plasmid pCAGGS-FLPo was prepared by subcloning the FLPo-bovine growth hormone polyadenylation signal sequences from the pFLPo plasmid 49 into a pCAAGS promoter plasmid 50 . Plasmid pCAAGS-iCre was prepared by adding the bovine growth hormone polyadenylation signal to iCre (kind gift of Rolf Sprengel) 51 and subcloning into a pCAGGS promoter plasmid. Plasmids pCAGGS-iCre and pCAGGS-FLPo contain the CAG promoter/ enhancer, which drives recombinase expression of iCre or FLPo in fertilized mouse eggs 52 , and were used as an alternative source of iCre and FLPo for some experiments, as noted. All plasmids, including the Sec24c-d replacement construct described below, were purified using the Machery-Nagel NucleoBond ® Xtra Maxi EF kit, per manufacturer's instructions. All microinjections were carried out at the University of Michigan Transgenic Animal Model Core. Co-injections of pUC19-Sec24c-d with pDIRE were performed on zygotes generated from the in vitro fertilization of C57BL/6J oocytes with sperm from Sec24 +/− male mice 12 . For each microinjection, 5 ng/μl of circular recombinase plasmid mixed with 5 ng/μl of circular donor plasmid was administered 30,53,54 and microinjected zygotes were then transferred to pseudopregnant foster mothers. Tail clips for genomic DNA isolation were obtained from pups at 2 weeks of age. Notably, the FRT and LoxP sites are located in introns 2 and 3 of the Sec24c gt allele, respectively; therefore, we were limited to swapping the C-terminal 90% coding sequence of Sec24c with that of Sec24d.
Genotyping assays. Genotypes of potential transgenic mice and ES cell clones were determined using a series of PCR reactions at the Sec24c locus. All genotyping primers used in this study are listed in Table S2 and those for the Sec24c locus are depicted in Fig. 1, and expected band sizes are given in Table S3. Primers S, T, and U were used to amplify fragments of DNA unique to pUC19-Sec24c-d, which should detect targeted insertion at the Sec24c locus or random insertion elsewhere in the genome. Targeted insertion was detected using primer pairs flanking the FRT 5′ recombination site (primers C + primer D or R) and the loxP 3′ recombination site (primers I or F + primers H or J). Integration of pCAGGS-iCre, pCAGGS-FLPo, or pDIRE was detected with primer sets iCre10F + iCre10R or FLPo8F + FLPo8R. Mice carrying the Sec24d GT allele were genotyped as described previously 15 .

Transient electroporation of ES cells.
ES cell clone EPD0241-2-A11 for Sec24c tm1a(EUCOMM)Wtsi (Sec24c GT allele 12 ) was expanded and co-electroporated with pUC19-Sec24c-d and either pDIRE or pCAGGS-iCre and pCAGGS-FLPo. ES culture conditions for JM8.N4 ES cells were as recommended at http:// www. KOMP. org. Electroporation was carried out as previously described (240 kV, 475 μF) 19 with the exception that pUC19-Sec24c-d lacks a drug selection cassette that can be used to enrich for correct homologous recombination in mouse ES cells. After 1 week, individual ES cell colonies were plated in 96 well plates, 288 from cells transfected with pUC19-Sec24c-d and pDIRE, and 284 from cells transfected with pUC19-Sec24c-d, pCAGGS-iCre, and pCAGGS-FLPo. Cells were expanded and plated in triplicate for frozen stocks, DNA analysis, and G418 screening. To test for G418 sensitivity, cells were grown in selection media containing G418 for 1 week, and then fixed, stained, and evaluated for growth. Genomic DNA was prepared from each ES cell clone as previously described 55 and resuspended in TE.

Subcloning of ES cells.
PCR analysis demonstrated that clone 6-H9 contained a mixed population of cells, some of which were properly targeted with recombinations occurring at the outermost FRT and loxP sites, and others that still contained the parental Sec24c GT allele, consistent with the observed mixed resistance to G418 (Table S1). One round of subcloning produced six different subclones consisting of a pure population of properly targeted ES cells, each with one wild type Sec24c allele and one correctly targeted Sec24c c-d allele (Fig. 1C) and none carrying random insertions of either pCAGGS-iCre or pCAGGS-FLPo. The Sec24c c-d allele is registered with the MGI database as Sec24c tm1Dgi (MGI 5501092), but will be referred to as Sec24c c-d within this text.
Generation of Sec24c +/c-d mice. Three correctly targeted subclones of 6-H9 were used to generate mice carrying the Sec24c c-d allele (Fig. 1D). ES cell clones were cultured as described previously 56 and expanded for microinjection. ES cell mouse chimeras were generated by microinjecting C57BL/6 N ES cells into albino C57BL/6J blastocysts as described 57 and then bred to B6(Cg)-Tyr c-2J /J (JAX stock #000058) to achieve germline transmission. ES-cell-derived F1 black progeny were genotyped using primers G, E, and F (Fig. 1D) www.nature.com/scientificreports/ Sec24c c-d allele was maintained on the C57BL/6J background by continuous backcrosses to C57BL/6J mice. Initial generations were also genotyped to remove any potential iCre and FLPo insertions.
Long-Range PCR. The integrity of the Sec24c locus with the newly inserted SEC24D sequence was confirmed by long-range PCR (Fig. 1E). Genomic DNA from Sec24c +/+ and Sec24c +/c-d mice were used as templates for a long-range PCR spanning the original arms of homology used for construction of the Sec24c GT allele 12 .
Primers used for long range-PCR are depicted in Fig. 1 and listed in Table S2. PCR was carried out using Phusion Hot Start II DNA Polymerase (Thermo Scientific), and products were separated on a 0.8% agarose gel.

RT-PCR.
Total RNA was isolated from a tail clip of Sec24c +/+ , Sec24c +/c-d , and Sec24c c-d/c-d embryos and liver biopsies from Sec24c +/− mice using the RNAeasy kit (Qiagen) per manufacturer's instructions, with the optional DNaseI digest step included. cDNA synthesis and PCR were carried out in one reaction using SuperScript ® III One-Step RT-PCR System with Platinum ® Taq (Invitrogen) following manufacturer's instructions. Primers used for RT-PCR are depicted in Fig. 5 and listed in Table S2.
Timed matings. Timed matings were carried out for intercrosses of Sec24c +/c-d mice. Embryos were harvested at designated time point for genotyping and histological analysis. The embryo age was estimated from the time of coitus and embryonic appearance to within a 1-day range. Since normal numbers of null embryos were observed at each of these embryonic time points, finer resolution of embryonic stage would be unlikely to add significant additional insight. Genotyping was performed on genomic DNA isolated from tail clip from mice > E12.5 or from yolk sacs and embryonic tissue from embryos < E12.5 days of age.
Animal care. All  Histology. Tissues, embryos and pups were fixed in Bouin's solution (Sigma-Aldrich) at room temperature overnight, then transferred to 70% EtOH. Prior to embedding, fixed P0 pups and E17.5-E18.5 embryos were sectioned longitudinally at the midline. Processing, embedding, sectioning and H&E staining were performed at the University of Michigan Microscopy and Image Analysis Laboratory. Immunohistochemistry was performed as previously described 44,45 using SERT antibody (AB9726 Millipore) at a 1:2500 dilution (1 h incubation). Briefly, primary antibodies were applied following antigen retrieval and quenching of endogenous peroxidases. Subsequently, primary antibody was washed and polymer HRP secondary antibody was applied (Biocare, Concord CA). Negative controls were obtained by substitution of the primary antibody with universal negative reagent (Biocare, Concord CA). 3,3-diaminobenzidine was applied to visualize the reactions.
Alcian blue staining. Following deparaffinization and hydration with xylene and graded alcohols, formalin-fixed, paraffin embedded slides were mordanted in 3% Acetic Acid (Rowley Biochemical Inc., E-323-3) for three minutes then stained with Alcian Blue Solution (Rowley Biochemical Inc., E-323-1) for 30 min (pH 2.5). Slides were then washed for 10 min in running tap water, followed by a deionized Water rinse. Slides were counterstained in Nuclear Fast Red Solution (Rowley Biochemical Inc., E-323-2) then dehydrated and cleared through graded alcohols and xylene and coverslipped with Micromount (Leica cat# 3801731, Buffalo Grove, IL) using a Leica CV5030 automatic coverslipper.
Alizarin red staining. Following deparaffinization and hydration with xylene and graded alcohols, formalin-fixed, paraffin embedded sections were stained with 2% Alizarin Red, pH 4.2 (Rowley Biochemical Inc., C-206-1) for 30 s then blotted well prior to quick dehydration in Acetone and Acetone-Xylene (50-50) for 15 s each, then cleared in 3 changes of xylene. Slides were coverslipped as described above.
Embryonic phenotyping. Crown to rump length measurements were obtained using a caliper on fresh specimens or by measuring the length of the embryo in a longitudinal section. Body weights were obtained immediately after birth. To account for normal variations in embryonic length and weight between litters, all measurements were normalized to the average length of Sec24c +/+ and Sec24c +/c-d animals within a given litter (mean of controls = 100%). Values for each individual were then calculated based on that average for controls within the same litter.

Statistical analysis.
To determine if there is a statistical deviation from the expected Mendelian ratios of genotypes from a given cross, the p-value reported is the χ 2 value calculated using the observed ratio of genotypes compared to the expected ratio. All other p-values were calculated using Student's unpaired t-test.