|Home | About | Journals | Submit | Contact Us | Français|
During vesicular transport between the endoplasmic reticulum and the Golgi, members of the TMED/p24 protein family form hetero-oligomeric complexes that facilitate protein cargo recognition as well as vesicle budding. In addition, they regulate each other's level of expression. Despite analyses of TMED/p24 protein distribution in mammalian cells, yeast, and C. elegans, little is known about the role of this family in vertebrate embryogenesis. We report the presence of a single point mutation in Tmed2/p24β1 in a mutant mouse line, 99J, identified in an ENU mutagenesis screen for recessive developmental abnormalities. This mutation does not affect Tmed2/ p24β1 mRNA levels but results in loss of TMED2/p24β1 protein. Prior to death at midgestation, 99J homozygous mutant embryos exhibit developmental delay, abnormal rostral-caudal elongation, randomized heart looping, and absence of the labyrinth layer of the placenta. We find that Tmed2/ p24β1 is normally expressed in tissues showing morphological defects in 99J mutant embryos and that these affected tissues lack the TMED2/p24β1 oligomerization partners, TMED7/p24γ3 and TMED10/p24δ1. Our data reveal a requirement for TMED2/p24β1 protein in the morphogenesis of the mouse embryo and placenta.
Vesicular transport mediates molecular trafficking between the various membrane-bound compartments in a cell and thus underlies all major cellular activities. During vesicular transport members of the Transmembrane emp24 domain (TMED) or p24 protein family of trafficking proteins regulate protein-cargo selection and vesicle budding (Carney and Bowen, 2004; Kaiser, 2000; Strating and Martens, 2009). Transmembrane and secreted proteins synthesized in the endoplasmic reticulum (ER) depend on vesicular transport for proper localization to the Golgi where they are further modified before being transported to their final destinations. Properly folded proteins synthesized in the ER interact with adaptors or chaperones, such as TMED/p24 proteins, which enable them to be packaged into coat protein (COP) II vesicles at ER exit sites (Bremser et al., 1999; Gurkan et al., 2006). After budding off the ER membrane, the vesicles move towards the Golgi via bulk flow or along microtubules (Cai et al., 2007). ER-resident proteins delivered to the Golgi along with cargo proteins are re-packaged into COPI vesicles and returned to the ER. Secretory cargo proteins deposited by COPII vesicles are transported through the Golgi stacks by COPI vesicles or by a process of cisternal maturation to the plasma membrane, endosomal/lysosomal system, or extracellular space.
Members of the TMED/p24 family are structurally related, sharing four distinct functional domains. A short signal sequence targets them to the ER membrane during translation; an N-terminal Golgi dynamics (GOLD) domain carries out cargo recognition; a coiled-coil domain mediates interactions between family members; and a short cytoplasmic tail contains conserved motifs for binding to coat complexes in COPI and COPII vesicles (Anantharaman and Aravind, 2002; Bethune et al., 2006; Blum et al., 1999; Bremser et al., 1999; Carney and Bowen, 2004; Dominguez et al., 1998; Goldberg, 2000; Lavoie et al., 1999; Sohn et al., 1996). Members of the TMED/p24 family fall into four subfamilies based on shared protein identity: α, β, δ and γ. Although these subfamilies are conserved in all animals and fungi, species-specific duplications and/or losses have resulted in varying numbers of genes in each TMED/p24 subfamily (Bouw et al., 2004; Carney and Bowen, 2004; Dominguez et al., 1998; Strating and Martens, 2009; Strating et al., 2009). Ten Tmed/p24 genes are present in mammals: five in the γ subfamily, Tmed1/p24γ1, Tmed3/p24γ4, Tmed5/p24γ2, Tmed6/p24γ5, and Tmed7/p24γ3; three in the α subfamily, Tmed4/p24α3, Tmed9p24α2 and Tmed11/p24α1; one in the δ subfamily Tmed10/p24δ1; and one in the β subfamily, Tmed2/p24β1 (Strating et al., 2009). TMED proteins, which are herein referred to by their assigned MGI nomenclature, are reported to exist as monomers, dimers, oligomers or hetero-oligomers (Barr et al., 2001; Carney and Bowen, 2004; Jenne et al., 2002; Luo et al., 2007; Marzioch et al., 1999). According to other experimental evidence, hetero-oligomers comprising one member of each of the four subfamilies form the functional units required for vesicular transport (Blum et al., 1999; Marzioch et al., 1999)
Genetic and biochemical experiments reveal that interactions between TMED proteins regulate their stability: knockdown or deletion of one TMED protein led to decrease or loss of expression of TMED proteins from different subfamilies (Blum et al., 1999; Carney and Bowen, 2004; Denzel et al., 2000; Fullekrug et al., 1999; Marzioch et al., 1999; Takida et al., 2008; Wen and Greenwald, 1999). A null mutation in Tmed10 resulted in developmental arrest before blastocyst formation and decreased expression of two interacting TMED proteins, TMED9 and TMED3, in livers of heterozygous mice (Denzel et al., 2000). In both yeast and mammalian cell lines, TMED2, the sole member of the β subfamily, is found in a complex containing TMED10 and/or TMED7, as well as TMED9 and it is required for their stability (Barr et al., 2001; Fullekrug et al., 1999; Jenne et al., 2002; Marzioch et al., 1999).
Members of the TMED family localize to membranes of the ER, ERGIC (endoplasmic reticulum-Golgi intermediate compartment) and cis-Golgi as well as to COPI and COPII vesicles. Biochemical and genetic experiments demonstrate that TMED proteins bind to both COPI and COPII proteins and likely function in anterograde and retrograde transport between the ER and the Golgi (Bethune et al., 2006; Bremser et al., 1999; Dominguez et al., 1998; Goldberg, 2000). In yeast, mutations of the Tmed2 homolog, emp24, result in delayed maturation of Gas1p, a GPI-anchored protein, and defective transport of invertase, a soluble secreted protein (Marzioch et al., 1999; Muniz et al., 2000). In mammalian cells, reduction of TMED10 levels by RNAi delayed trafficking of GPI anchored proteins to the plasma membrane (Takida et al., 2008). Thus, TMED proteins appear to be required in both yeast and mammalian cells specifically for movement of GPI anchored proteins to the plasma membrane.
In addition to impairing anterograde protein trafficking, mutations in TMED proteins trigger the ER stress-associated unfolded protein response. Loss of TMED proteins in yeast activates splicing of XBP1 pre-mRNA, leading to synthesis of the transcription factor that regulates the unfolded protein response. Loss of TMED also resulted in increased expression and abnormal secretion of Kar2, the yeast orthologue of Hspa5/Grp78, an ER resident chaperone belonging to the heat shock protein 70 (HSP 70) family (Belden and Barlowe, 2001). Similarly, In mammalian cells knockdown of TMED4 led to upregulation of HSP 70 and to decreased apoptosis (Hwang et al., 2008). In Drosophila, mutation of Logjam, the Tmed3 homolog, caused activation of the NF-κB pathway without the associated splicing of Xbp1 (Boltz and Carney, 2008). Thus, TMED proteins may modulate multiple cellular stress pathways.
In C. elegans, a screen for suppressors of lin-12/Glp alleles encoding Notch receptors with reduced activity, identified mutations in the C.elegans Tmed2 homolog, sel-9 (Wen and Greenwald, 1999). All of the lin-12/Glp alleles suppressed by sel-9 mutations carried missense mutations in their extracellular domain. Whereas in wild type strains altered Notch proteins accumulate within the cell, in the sel-9 mutants altered Notch assembled at the cell membrane where it was functional (Wen and Greenwald, 1999). Trafficking of wild type Notch/Glp did not require sel-9; thus Tmed2 likely functions in quality control, specifically selecting defective Notch/Glp as cargo. Studies in primary rat astrocytes and human endothelial kidney cells found that TMED2 interacts with PAR-2, a G protein-coupled receptor that is activated at the plasma membrane by protease cleavage (Luo et al., 2007). Downstream signaling triggered by PAR-2 activation released TMED2-bound PAR-2, freeing it to traffic from the Golgi to the plasma membrane. In this case, TMED2-cargo selectivity appears to regulate post-Golgi protein trafficking.
Here we report the identification of a null allele of Tmed2 in an N-ethyl N-nitrousurea (ENU) screen for recessive mutations perturbing the morphology of the developing mouse embryo. The mutation was discovered in a mouse line designated 99J and the mutant allele named Tmed299J (herein referred to as 99J). 99J homozygous mutant embryos exhibit developmental delay by E8.5, fail to undergo embryonic turning, and display posterior truncations, abnormal heart looping, and absence of the labyrinth layer of the placenta. The mutant embryos are reabsorbed by mid-gestation. Using positional cloning and sequencing we determined that the Tmed299J allele carries a G:C to T:A transversion in the first exon, generating an alanine to glutamic acid substitution in the TMED2 signal sequence, blocking the production of any TMED2 protein. Failure to complement a gene trap allele of Tmed2 confirmed that the mutation in Tmed299J underlies the phenotypic defects. The levels of TMED7 and TMED10 protein were decreased in 99J heterozygous embryos and absent in the homozygous mutants embryo, revealing that TMED2 regulates the stability of these γ and δ subfamily members. The mid-gestation defects of Tmed299J suggest that the cargos selected and trafficked by TMED2 participate in the morphogenetic processes of early development.
The 99J line was generated by ENU mutagenesis of C57BL/6J mice (Anderson, 2000; Garcia-Garcia and Anderson, 2003; Kasarskis et al., 1998) and backcrossed to C3HeB/FeJ females. The PST809 cell line, with insertion of a gene trap cassette in the 3rd intron of Tmed2, Tmed2GT, was identified on ENSEMBL (http://baygenomics.ucsf.edu/). The insertion was sequenced and verified before injection into C57 blastocysts. The chimeric offspring were bred to C57BL/Crc females and heterozygous F1 (C57/129) mice were bred to 99J heterozygous carriers for complementation analysis. CD1 (Charles Rivers, Canada) females mated to CD1 male were used to generate embryos for wholemount in situ hybridization. All mouse breedings and manipulations were performed in accordance with the Canadian Council on Animal Research. To generate embryos, females were placed with a male overnight and checked for the presence of a vaginal plug in the morning. The day that a plug was detected was considered embryonic day (E) 0.5. Mice and embryos from the 99J ENU line were genotyped by PCR with primers to D5MIT65 and D5MIT213. Mice and embryos from the Pst809 gene-trap line were genotyped with primers from JaxMice that recognize the LacZ gene or with primers which were designed to distinguish the wild type Tmed2 allele from the gene trapped allele. Primers Tmed2In4F and Tmed2In4R amplified the wild type Tmed2 allele, and primers Tmed2In4F and Genetrap1R amplified the genetrap allele.
The mutation was mapped using linkage to flanking simple sequence length polymorphism (SSLP) markers from the MIT database or new markers that we generated (http://mouse.ski.mskcc.org/). Exons, splice site acceptors and splice site donors of 35 genes in the 99J minimal region were sequenced at the McGill Genome Center. Genomic DNA from 99J homozygous mutant embryos (n=6), carrier mice (n=2), and two wild type strains C3H/HeJ (n=1) and C57bL/6J mice (n=1) were analyzed with SeqMan II (DNASTAR).
RNA isolation was performed according to standard Trizol (Invitrogen) protocol. The RNA was treated with DNAse prior to reverse transcription reaction. Reverse transcription was done using the Superscript III RT kit (Invitrogen). The Tmed2F and Tmed2R primers were designed to amplify wild type Tmed2, the Tmed2E3F and GenetrapR primers were used to amplify the genetrap mutation in the PST809 cell line, and two primers were used to amplify Gapdh:
Wholemount in situ hybridization was performed on E6.5, E7.5, E8.5, E9.5, E10.5 and E11.5 embryos and placentas. The yolk sacs were collected for genotyping and the placentas and embryos were fixed in PFA, dehydrated and processed for wholemount in situ hybridization according to standard protocols (Wilkinson et al., 1990). Wholemount embryos were examined on a Leica stereomicroscope MZFL III (Leica, Germany) microscope and photographed using a SPOTcam (RT-Slider, USA) camera after wholemount.
For section in situ hybridization and immunohistochemistry, E8.5, E9.5 and E10.5 deciduas were fixed in PFA, embedded in paraffin, and sectioned at 7 μm. Embryonic tissue was scraped off the slides and used to genotype the embryos. Slides containing 99J homozygous mutant embryos and normal controls were used for in situ hybridization or immunohistochemistry. In situ hybridization was performed as described by Neubuser et al. (1995).
We generated two Tmed2 probes. The exon1 specific probe was made by PCR amplification of exon 1 of the Tmed2 gene from E8.5 decidual cDNA using primers that were designed to be complementary to Ensembl sequence ENSMUSG000000293902 (Forward primer: CGAGGAGTGCTTCTTCGAG; Reverse primer: TCCCGGACTTCCATGTACTC). The full length probe was generated by PCR amplification of exon 1 – 4 of the Tmed2 gene from E9.5 embryonic cDNA using primers (Forward primer: GATGGGCCTCATCTTCGAG and Reverse primer: ACCAAAGGACCACTCTGCTG). PCR amplicons of the correct size were subcloned using the dual promoter TA cloning kit (Invitrogen). The Tmed2-TA clones were sequenced on the ABI Prism 310 Genetic analyzer (Perkin Elmer) before probe generation. To generate digoxigenin labeled probe Tmed2-TA constructs were linearized and transcribed using the T7 promoter or Sp6 promoters to generate antisense and sense probes, respectively. Since the exon1 specific probe reported the same expression pattern as the full length probe, the two probes were used interchangeably. RNA probes for Mash2, Gcm1, Esx1, Pl1 and Tpbp were kind gifts from Dr. J. Cross (U. Calgary).
Immunohistochemistry was performed according to standard protocols (2003). Antibodies against VCAM 1 (Santa Cruz) and α4 integrin (Santa Cruz) were used at 1:100 dilutions for immunohistochemistry on sections of E8.5 normal and 99J homozygous mutant embryos. Experimental slides were imaged on an Axiovert Imager. Z1 (Zeiss, Germany) and pictures were captured on an Axiocam MrC5 (Zeiss, Germany).
Individual E9.5 and E10.5 embryos were dissected out of their extra-embryonic membranes and either lysed immediately or flash frozen at -80°C before western blot analysis. Groups of wild type (2 embryos), 99J heterozygous (2 embryos) or 99J homozygous mutant embryos (4 embryos) were homogenized with a syringe and lysed in buffer containing: 50 mM Tris, 150 mM NaCl, 1% NP-40, 1.5 mM Mgcl2, 10% Glycerol, 1 mM EDTA, 25 mM NaF and 1 mM Na3VO4. After protein quantification, 20 μg of protein was separated on a 10% SDS-PAGE gel and electroblotted onto a PVDF Nitrocellulose membrane (Bio-Rad). The blots were blocked in 5% milk and incubated with primary antibodies recognizing the cytosolic tails of TMED2, TMED7, TMED9, TMED10 (Denzel et al., 2000), HSPA5/GRP78 (Abcam, MA), GRP94 (Abcam, MA) and GAPDH (Abcam, MA) diluted in blocking buffer. The blots were washed and incubated with HRP-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch) secondary antibody. The blots were developed with an ECL plus chemiluminescence kit (Amersham).
To determine if the unfolded protein response pathway is activated in 99J homozygous mutant embryos we designed primers for RT-PCR to amplify the Xbp1 mRNA (Forward primer: GATCCTGACGAGGTTCCAGA and Reverse primer: GGTCCCCACTGACAGAGAAA). The PCR product was digested with Pst1 restriction enzyme which yields two bands (240 bp and 110 bp) in the absence of ER stress and a single 350 bp band in the presence of ER stress. Primary embryonic fibroblast cells (PMEFs) were cultured in the presence or absence of thapsigargin (Sigma), according to standard protocols (Leclerc and Rozen, 2008).
The ENU mouse line 99J was identified in a phenotypic screen for recessive mutations that disrupt embryonic morphology at E9.5 (Garcia-Garcia et al., 2005). 99J homozygous mutant embryos displayed several phenotypic abnormalities at midgestation (Fig. 1) and failed to survive embryogenesis (Table 1). To determine the developmental stage at which 99J homozygous mutant embryos first exhibited morphological defects, litters from 99J heterozygous intercrosses were examined between E7.5 and E12.5. At E7.5, 99J homozygous mutant embryos were indistinguishable from their normal littermates and were present at the predicted Mendelian ratio (Table 1). Starting at E8.5, 99J homozygous embryos appeared developmentally delayed relative to wild type and 99J heterozygous embryos. At this stage, expression of Sonic Hedghehog (Shh) a morphogen that is normally expressed in the node and midline of normal E8.5 embryos was also altered. In fact, Shh was only expressed in the head process of 99J homozygous embryos (n=2; Fig. 1A, B), similar to the characterized expression domain of this gene at E7.5 (Bai et al., 2002). Furthermore, E8.5 mutants also contained fewer somites than normal littermates (0-4 pairs of somites; Table 2). By E9.5, 0-10 somite pairs had formed in 99J homozygous mutants, whereas 20-22 somite pairs had formed in their wild type littermates (Fig. 1C; Table 2). 99J homozygous mutant embryos presented a number of additional abnormalities, including a failure to turn, an open neural tube, a small, irregularly shaped tail-bud and a bulbous allantois (Fig. 1D). The heart in 10 % of E9.5 99J homozygous mutant embryos (n=5/48) lacked detectable beating, indicating that some 99J mutants arrested before E9.5 (Table 1).
99J homozygous mutant embryos exhibited discordant development along the rostral-caudal body axis and abnormal heart formation (Table 1; Fig. 1). The rostral half of an E10.5 99J homozygous mutant embryo resembled that of a normal E9.5 embryo; it contained closed headfolds, otic and optic vesicles with morphological signs of differentiation, and forelimb buds (Fig. 1D, E). In contrast, the caudal end of an E10.5 mutant embryo was reduced in size and consisted primarily of a small, malformed tail bud that was devoid of tissue (Fig. 1E).. In normal E8.5 embryos the linear heart-tube loops in a rightward direction to ensure proper positioning of the future ventricles and atria. Only 23.1% of E10.5 99J homozygous mutant embryos (n = 3/13) showed rightward heart looping, while 53.8% showed irregular looping (n = 7/13; Fig 1G, I) and the heart failed to loop in the remaining 23.1% (n= 3/13; Fig 1H) (Table 1B). Abnormal heart looping is associated with delayed or absent expression of the TGF-β family member, Nodal, at the node (Krebs et al., 2003). However, in situ hybridization revealed Nodal expression in the node of 99J homozygous mutants analyzed between E8.5 and E9.5 (n=4, data not shown). Thus, abnormal heart looping was not due to a failure to initiate Nodal expression at the node. No homozygous mutant offspring were recovered at E11.5 and E12.5 (Table 1).
Embryonic survival after E10.5 requires a functional placenta (Watson and Cross, 2005); therefore we examined the development of extraembryonic regions in 99J homozygous mutant embryos to determine if placental defects contributed to the observed embryonic lethality. Chorioallantoic attachment, the joining of the chorion to the tip of the allantois, marks the morphological appearance of the developing placenta at E8.5 (Cross et al., 2003). Analysis of whole mount embryos at E8.5 revealed that 99J mutants failed to undergo chorioallantoic attachment. However, immunohistochemistry detected normal expression in both mutant and wild type embryos of two proteins known to mediate chorioallantoic attachment: VCAM1, in the allantois and its receptor α4 integrin in the chorion (Kwee et al., 1995; Yang et al., 1995) (data not shown). Thus, the normal distribution of VCAM1 and α4 integrin does not require TMED2; delayed chorioallantoic attachment in the 99J mutant likely reflects an overall developmental delay.
Following spreading of the allantois over the surface of the chorion and subsequent chorioallantoic attachment, branching morphogenesis to form the villi of the labyrinth layer of the placenta is initiated at E9.0 (Fig. 2A, C & E). The labyrinth layer, which mediates nutrient and waste exchange between the maternal and fetal environments, consists of allantois-derived fetal blood vessels, as well as maternal blood sinusoids embedded in a trilaminar trophoblast (Rossant and Cross, 2001; Simmons et al., 2008). Fetal blood vessels can be distinguished from maternal sinusoids on the basis of red blood cell morphology: embryonic red blood cells are nucleated, while maternal red blood cells lack nuclei. Due to developmental delay in the 99J homozygote, we examined mutant embryos at E10.5 to determine whether they had initiated formation of the labyrinth layer. The placenta in normal E10.5 embryos exhibited the expected histological giant cell, spongiotrophoblast, and labyrinth layers (Fig. 2C, E). Although the allantois had begun to spread over the surface of the chorion in 99J homozygous mutant embryos, it showed no evidence of invagination into the chorion to form the labyrinth layer (Fig. 2D, F). Unlike wild type embryos, 99J homozygous mutants contained fetal blood vessels only in the allantois and at the boundary between the allantois and the chorion. The observed absence of the labyrinth layer in the 99J homozygous mutant embryo, and thus of a mechanism for nutrient-waste exchange between the fetal and maternal blood supplies, likely underlies their death at midgestation.
The initiation of branching morphogenesis and the ensuing population of the forming villi with allantois- derived vessels depend on proper expression of the transcription factor Gcm1 in the chorion (Anson-Cartwright et al., 2000; Schreiber et al., 2000). In situ hybridization detected comparable levels of Gcm1 expression in the chorionic plate of normal and 99J homozygous mutant embryos at E8.5 (Fig 3A, B), indicating appropriate differentiation of chorionic ectoderm in 99J. However, only a subset of 99J homozygous mutant embryos between E9.5 and E10.5 expressed Gcm1 (n=2/4). In those embryos, Gcm1 expression was restricted to a small region of the chorionic plate (Fig 3D), while in similarly staged wild type embryos, Gcm1 expression was distributed throughout the chorionic component of the labyrinth layer (Fig. 3C). These data suggest that lack of labyrinth layer formation in the 99J mutant does not result from deficient or inappropriate Gcm1 expression.
Formation of a functional placenta also depends on the differentiation of other trophectoderm-derived cell layers, such as giant cells and the spongiotrophoblast. Histological examination identified giant cells in the placenta of E9.5 mutant embryos (Fig. 2D, F). In situ hybridization to placental lactogen (Pl)1, a marker for trophoblast giant cells (Simmons and Cross, 2005) detected Pl1 expression in a layer of cells immediately below the maternal decidua in both E9.5 normal and 99J homozygous mutant embryos, confirming the presence of giant cells in mutant placentas (Fig 3E, F). Similarly, in situ hybridization to Trophoblast specific protein (Tpbp) alpha, a gene expression marker for spongiotrophoblast cells (Simmons and Cross, 2005), found Tpbpa transcripts in the placenta of both normal and 99J homozygous mutant embryos at E9.5, suggesting that spongiotrophoblast cells differentiate appropriately in the 99J mutant (Fig. 3G, H). However, whereas the Tpbpa positive cells resided immediately above the labyrinth layer in the placenta from normal embryos, they were located immediately adjacent to the apparently undifferentiated chorion in the placenta from 99J homozygous mutant embryos (Fig 3 G-H). These data suggest that the 99J mutant might be specifically impaired in the formation of the labyrinth layer. On the other hand, since distinct differentiated trophoblast cell types arise from precursors present in the E8.5 chorion, prior to elaboration of the labyrinth layer (Simmons et al., 2008), the absence of this layer in the 99J mutant might simply reflect developmental delay.
We initially mapped the 99J mutation to mouse chromosome 5 within a 2 Mb interval flanked by MIT markers D5 MIT65 and D5 MIT139. An evaluation of the 35 genes within this interval revealed that targeted deletion/knockout of four of these genes did not generate embryonic lethality, eliminating them from further consideration as the mutated gene in line 99J (Akhmanova et al., 2005; Hughes et al., 2004; Hyun et al., 2004; Lu et al., 2001; Okada et al., 2002; Reiter and Skarnes, 2006). To further reduce the list of potential candidates, we examined the expression of the remaining 31 genes in the 99J critical interval by RT-PCR. Transcripts from three genes were not detected in either normal or 99J homozygous mutant embryos at E9.5, removing them as viable candidates for the 99J mutation (Table 3). Whole mount in situ hybridization for 24 of the now 28 candidate genes determined that three of them Tmed2, Eif2b1, and D5Bwg0834e- were expressed in both the embryo and placenta at E9.5 (Table 3). Parallel experiments to sequence the coding regions and donor and acceptor splice sites of all 35 genes in the 99J interval identified a single point mutation, a G:C to T:A transversion, in the first exon of Tmed2 in 99J homozygous mutant (n=4) and 99J heterozygous (n=3) embryos (Fig. 4A). The G:C to T:A transversion produced a missense mutation that replaced a highly conserved alanine residue in the signal peptide sequence of TMED2 with glutamic acid (A13E) (Fig. 4A). Mutations in the hydrophobic core of signal sequences are associated with loss or decreased protein levels due to abnormal translocation into the ER and/or protein stability (Karaplis et al., 1995; Kendall et al., 1990; Lanza et al., 2002; Nicchitta et al., 2005; Pidasheva et al., 2005; Symoens et al., 2009; Wiren et al., 1989). Thus, we anticipated that the A13E substitution would result in decreased expression or loss of TMED2 protein. Consistent with this prediction, Tmed2 transcripts were present at relatively normal levels in mutant embryos at E9.5 (Fig. 4B), whereas levels of TMED2 protein were reduced in 99J heterozygotes and undetectable in 99J homozygotes at E9.5-E10.5 (Fig. 4C and data not shown).
To confirm that the Tmed299J allele underlies the recessive midgestation lethality in line 99J, we generated a second mutant allele of Tmed2 for complementation analysis. Using the PST 809 ES cell clone (BayGenomics), we generated a line of mice carrying a gene trap allele of Tmed2, Tmed2GT. The gene trap is inserted into the third intron of Tmed2 (Fig. 5A) where it promotes the synthesis of a fusion protein between the first 71 amino acids of TMED2, including the cargo recognition domain (GOLD), and β-galactosidase. A similarly truncated TMED2 protein has been shown to function as a dominant negative mutant in vitro (Luo et al., 2007). Examination of E10.5 embryos produced from matings between Tmed2GT/+ females and Tmed299J/+ males demonstrated a lack of complementation between the gene trap and ENU alleles of Tmed2. Recovered trans-heterozygous Tmed2GT/99J embryos morphologically resembled Tmed299J/99J mutant embryos: they displayed a reduced size compared to normal littermates, abnormal heart looping and a small, irregularly shaped tail bud (Fig. 5C and data not shown).
To establish the temporal pattern of Tmed2 expression during mouse development, we performed RT-PCR on wild type embryos recovered at different times of gestation between E3.5 and E10.5. Robust and continuous transcription of Tmed2 was observed from implantation at E4.5 through the onset of organogenesis at E10.5 (Fig 6). We investigated the spatial distribution of Tmed2 transcription in E5.5 – E10.5 embryos and placentas by in situ hybridization. Two distinct cDNA probes (see Materials and Methods) revealed specific and identical patterns of Tmed2 expression between E5.5 and E10.5 (Fig. 7 and S1). In E5.5 embryos, we detected uniform Tmed2 expression throughout the embryonic and extraembryonic components of the egg cylinder (Fig. S1A.). In more developmentally advanced E5.5 embryos Tmed2 expression was highest in the extraembryonic regions, including the ectoplacental cone, parietal endoderm, and visceral endoderm (Fig. S1B-D). Nonetheless, Tmed2 expression was still detected in the embryonic ectoderm of all embryos analyzed at this stage (Fig. S1A-D).
The initial ubiquitous expression of Tmed2 became spatially restricted by E6.5. During gastrulation Tmed2 was expressed in a very dynamic pattern. E6.5 embryos expressed Tmed2 at the highest levels in ectoplacental cone and extraembryonic ectoderm and at lower levels in visceral endoderm and epiblast (Fig. 7A and Fig. S1E– H)). In late headfold stage E7.5 embryos Tmed2 was found in ectoplacental cone, amnion, anterior neural folds and underlying head mesoderm as well as in the proximal region of the primitive streak (Fig 7C-D and S1I-L). At E8.5 Tmed2 was broadly expressed throughout the embryo, including a region around the node, the neural folds, and allantois (Fig. 7E). In situ hybridization to sections of E8.5 embryos revealed Tmed2 in lateral plate mesoderm, the somites, the tailbud, the heart and aortic arteries (Fig 7E, S1M-P). Overall Tmed2 was expressed in derivatives of all three germ layers - ectoderm, mesoderm and endoderm - although it was at higher levels in the mesoderm and its derivatives. In the endoderm, Tmed2 expression was detected in the foregut but not in the midgut and hindgut (Fig. S1, M-P). As the embryo completed turning between E8.5-E9.0, the highest levels of Tmed2 were found in the forebrain, otic vesicle, first pharyngeal arch and tail bud (Fig 7E, F). E9.5, embryos expressed Tmed2 in the forebrain, otic and optic vesicles, liver anlage, limb buds, neural tube, and tail bud (Fig 7G). At E10.5 Tmed2 was broadly expressed in most embryonic structures, including the ventricles of the heart (data not shown).
Tmed2 expression was maintained throughout placental development. E6.5 and E7.5 embryos expressed Tmed2 in extraembryonic ectoderm and chorion, respectively (Fig. 7A-D), while E8.5 embryos expressed Tmed2 in the allantois, chorionic plate and giant cells of the now forming placenta (Fig. 8A). At E9.5 the placenta exhibited Tmed2 expression in the labyrinth, spongiotrophoblast and giant cell layers, but not in the attached allantois (Fig. 8B). By E10.5 Tmed2 expression was detected in a subset of giant cells (Fig. 8C), in trophoblast cells surrounding the maternal sinusoids, and in fetal blood vessels in the placenta (Fig. 8D).
Previous studies revealed interdependence among TMED proteins for regulation of their expression levels (Jenne et al., 2002; Luo et al., 2007). Therefore, we performed western blot analysis to assess whether the levels of TMED proteins known to interact with TMED2, TMED7, TMED9 and TMED10 (Jenne et al., 2002; Marzioch et al., 1999), were altered in 99J mutants. We found that the levels of TMED7, TMED9 and TMED10 were reduced in 99J heterozygous embryos at E9.5 and E10.5 (Fig. 9 A-C). While expression levels of TMED9 were further reduced in 99J homozygous mutant embryos at E9.5-E10.5 (Fig. 9A), expression of TMED7 and TMED10 was undetectable in mutant embryos at these same stages. However, mRNA levels for Tmed7, Tmed9, and Tmed10 did not vary between normal and 99J mutant embryos at E9.5-E10.5 (Fig. 5B and data not shown). These data suggest that loss of TMED2 directly affects the stability of TMED7, TMED9 and TMED10 at the protein level.
In yeast, deletion of the Tmed2 and Tmed10 homologs activates the unfolded protein response pathway. The mutant yeast strains upregulate expression of the ER chaperone protein HSPA5/GRP78 and also splicing of Hac-1 RNA, encoding a functional homolog of the transcription factor XBP1 (Belden and Barlowe, 2001; Kuznetsov et al., 1997; Rose et al., 1989; Yoshida et al., 1998; Yoshida et al., 2001). Therefore, we hypothesized that the 99J mutant embryos would display increased secretion of HSPA5/GRP78 and splicing of Xbp1 RNA. However, we failed to detect abnormal activation of the unfolded protein response pathway in 99J mutant embryos. Expression of the ER chaperone proteins HSPA5/GRP78 and HSP90β1/GRP94 did not increase in E10.5 99J mutant embryos (Fig. 10A). In addition, expression and splicing of the Xbp1 transcript appeared unaltered in 99J mutants compared to wild type embryos (Fig 10B). In control assays, efficient splicing of Xbp1 was detected in PMEFs after treatment with the ER stress-inducing agent, thapsigargin (Fig 10B). Thus, our data suggest that the loss of TMED2, TMED7, and TMED10 does not activate the unfolded protein response in mouse embryos.
TMED proteins are highly conserved in eukaryotes and are implicated in vesicular transport. We report that Tmed2 is expressed in embryonic and extraembryonic tissues of the developing mouse embryo and is required for morphogenesis of both the embryo and the placenta. Tmed2 mutant embryos fail to form the labyrinth layer of the placenta and die by midgestation, revealing a previously uncharacterized requirement for vesicular transport in placental development. Minimal information exists on the expression of Tmed genes in mammalian tissues and our study represents the first effort to systematically document expression of a Tmed gene during embryonic development. Of potential significance to TMED function, our study reveals that Tmed2, the single member of the β-subfamily, is expressed tissue-specifically during embryonic development.
It has been proposed that TMED family members form hetero-oligomers, consisting of one member of each subfamily, to facilitate formation of COPI and COPII vesicles (Marzioch et al., 1999). While Tmed2 is expressed at all stages of postimplantation development analyzed in this study, it assumes a tissue-specific expression pattern by E6.5, suggesting a temporally and spatially restricted role in vesicle-trafficking.
Tmed2 mutant embryos exhibit developmental delay and are smaller than their normal littermates. These defects could reflect, in part, abnormal and/or reduced vesicular transport in the yolk sac, a structure that expresses Tmed2 and that mediates nutrient uptake by the early post-implantation embryo (Conway et al., 2003). By E10.5, the nutritional requirements of the embryo can no longer be met exclusively by the yolk sac and the continuation of development requires a functional placenta (Conway et al., 2003; Rossant and Cross, 2001). The labyrinth layer of the placenta, which is composed of fetal blood vessels and maternal blood spaces, now takes over as the primary site for nutrient and waste exchange between the growing fetus and its maternal environment (Watson and Cross, 2005). Mutants that fail to form this labyrinth layer arrest at mid-gestation, likely due to nutritional deficiency (Watson and Cross, 2005). We propose that Tmed2 homozygous mutant embryos die at mid-gestation because of impaired labyrinth layer development.
Chorioallantoic attachment initiates the formation of the labyrinth layer. Downs and Gardner (1995) suggest that chorioallantoic attachment occurs in at least three steps: contact between the allantois and the posterior end of the chorion, translocation of the allantois to the center of the chorion, and invasion of the allantois into the chorion. In Tmed2 mutant embryos, the allantois contacts but does not invade the chorion. Failure of allantois cell invasion into the chorion results in a weak attachment between the two tissues. Moreover, the absence of a labyrinth layer in Tmed2 mutants suggests that allantois cell invasion may function as the trigger for reorganization of chorion cells into the labyrinth layer. Failure of allantois invasion into the chorion of Tmed2 mutant embryos might be a non-specific effect of developmental delay. Alternatively, it may point to the requirement for the delivery of specific cargo proteins to the allantois-chorion interface by TMED2-mediated trafficking.
A large number of studies have shown that the expression of individual TMED proteins is interdependent (Bouw et al., 2004; Denzel et al., 2000; Jenne et al., 2002; Luo et al., 2007; Marzioch et al., 1999; Muniz et al., 2000). Our own studies provide further support for this interdependency; however, the significance and requirement for this remain unclear. We found that loss of TMED2 results in loss of TMED7 and TMED10 expression and a decrease in expression of TMED9. Since mutant embryos lacking TMED10 arrest and die prior to implantation at E4.5 (Denzel et al., 2000), whereas Tmed2 mutants survive up to E10.5, TMED10 expression and function must be independent of stable, zygotic expression of Tmed2 in the pre-implantation mouse embryo. These findings indicate that individual members of the TMED protein family are essential for distinct stage and tissue-specific functions during development.
A large family of molecular motor proteins consisting of cytoplasmic dyneins, axonal dyneins and kinesins are required for transport in the cytoplasm and cilia; reviewed in (Hirokawa and Noda, 2008). In addition, two subunits of the COPII coat complex interact with Dynactin1, the cofactor of the dynein molecular motor (Watson et al., 2005), suggesting that vesicular transport by COPII proteins may require normal expression and function of molecular motor proteins. Since TMED2 also interacts with COPII coat complexes (Bethune et al., 2006; Bremser et al., 1999; Dominguez et al., 1998) transport of TMED2 cargo proteins from the endoplasmic reticulum may be mediated by microtubule based motors. Embryos with mutations in Dynactin1 as well as several molecular motors -mD2lic, Kif3A, and Kif3B - either arrest before E8.0 or exhibit a number of morphological phenotypes, including developmental delay, abnormal heart looping and rostro-caudal truncations, suggesting that these molecules are required for normal development and morphogenesis (Lai et al., 2007; Marszalek et al., 1999; Nonaka et al., 1998; Rana et al., 2004; Takeda et al., 1999). Thus it will be important to explore whether these molecular motors are required for trafficking of TMED2 cargo proteins.
In yeast, deletion of Tmed2 results in increased secretion of the ER stress associated protein HSPA5/GRP78 and upregulation and splicing of the pre-mRNA encoding XBP1, a transcriptional regulator of the ER stress pathway (Belden and Barlowe, 2001). Hence, we expected that 99J homozygous mutant embryos would exhibit increased HSPA5/GRP78 protein levels. However, we did not detect increased levels of HSPA5/GRP78 or HSP90β1/GRP94, indicating that Tmed2 mutations do not contribute to increased ER stress in the developing mouse embryo. Furthermore, we did not find increased expression or splicing of Xbp1 (Yoshida et al., 2001). Our data are consistent with a report by Boltz and Carney (2008) reporting that the loss of TMED proteins in Drosophila does not trigger Xbp1 upregulation and splicing; rather, they observed activation of the NF-κB pathway. Activation of the NF-κB pathway independently of unfolded protein response has been reported in mammalian cells as well (Davies et al., 2009; Pahl et al., 1996); therefore future studies will examine this pathway in Tmed2 homozygous mutant embryos.
TMED2 interacts with COPII vesicles (Bethune et al., 2006; Dominguez et al., 1998; Majoul et al., 2001), which in vitro form a cage around newly synthesized proteins and regulate their anterograde transport from the ER. In addition, two studies, one in C. elegans (Wen and Greenwald, 1999), and one in an astrocyte cell line (Luo et al., 2007), suggest that TMED2 may function at other steps in the secretory pathway, possibly in COPI vesicles within the Golgi compartments.
COPII and COPI vesicles, and by extension members of the TMED protein family, play fundamental roles in the intracellular transport of proteins through the secretory pathway. Moreover, cell-cell interactions, as well as the execution of most cell behaviors, depend on the secretory pathway to deliver intact, functional proteins to their appropriate intracellular and extracellular compartments. Consequently the defects displayed by Tmed2-deficient embryos potentially reflect the defective transport and localization of one or several developmentally critical proteins. Recently an examination of genetic interactions between mutant alleles exhibiting similar phenotypes led to the demonstration that Sec24b, a core component of COPII vesicles, regulates planar cell polarity in the neural tube through selective sorting of Vangl2 (Merte et al.). A similar approach, combined with the generation of a conditional Tmed2 allele, will be applied to identify proteins that specifically rely on TMED2 for vesicular transport during the gastrulation stages of mouse development.
Figure S1. Tmed2 expression during embryonic development. Representative images of wild type embryos after in situ hybridization with antisense (A-P) or sense (E’) riboprobes to Tmed2. (A, B) Saggital sections of E5.5 embryos. (A) Tmed2 is ubiquitously expressed in embryonic (ee) and extraembryonic (eec) ectoderm. (B) In a slightly older E5.5 embryo Tmed2 expression continues to be equally distributed throughout the extraembryonic and embryonic ectoderms but is found at higher levels in visceral (ve) and distal visceral endoderm (dve) as well as in parietal endoderm (pe). (C) A slightly oblique transverse section detects higher levels of Tmed2 in ectoplacental cone (epc) and visceral endoderm compared to epiblast. (D) In a more distal section through the same embryo, expression is high in the embryonic ectoderm. (E, F) Transverse sections of E6.5 embryos show the highest levels of Tmed2 expression in the ectoplacental cone (E, epc) and low levels of Tmed2 throughout the embryonic ectoderm (F). (E’) No signal is detected on sections of the E6.5 embryo in E & F when hybridized to a Tmed2 sense probe. (G, H) sagittal sections of an E6.5 embryo show signal in the posterior region of the embryo and scattered in the embryonic ectoderm (ee) and visceral endoderm (ve). (I –L) Representative images of a sagittal section from an E7.5 embryo showing Tmed2 expression in ectoplacental cone (epc-J), amnion (amn-K, L), anterior mesoderm (m-K), and neuroepithelium (ne-K); lower levels of Tmed2 are detected in posterior mesoderm (m-L) and embryonic ectoderm (ee-L). (M-P) Transverse (M) and sagittal sections (N-P) through E8.5 embryos. (M) Tmed2 is broadly expressed in E8.5 embryos. (N)Tmed2 is expressed in somites (s) and dorsal aorta (da), but not in closed neural tube (nt) and midgut endoderm (en). (O) Tmed2 expression is detected in lateral plate mesoderm (lpm) and not in the endoderm (en). (P) High levels of Tmed2 expression are found in posterior tailbud mesoderm (tb) and not in hindgut (hg). Scale bar in A & B = 20 μm.
We would like to thank Dr. Kathryn Anderson for the 99J ENU line, Dr. John Bergeron and Tommy Nilsson for Tmed antibodies, Dr. Roberta Rivi and Yulan Hu for help with the mice. We thank Dr. Aimee Ryan, Dr. Virginia Papaioannou, Dr. Tomoko Watanabe and Michelle Collins for critical reading of the manuscript. We are grateful to the McGill Genome Center for sequencing the Tmed2 mutation and the Montreal Transgenic Core for generating the Pst809 genetrap allele. LAM was funded by NSERC and CIHR operating grants and FRSQ. LAJM and LL are members of the Research Institute of the McGill University Health Centre, which is supported in part by the FRSQ.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.