Developmental genetics of the mouse t-complex.

The proximal third of mouse chromosome 17 is known as the t-complex. The t-haplotype is a variant form of this region containing four tandem inversions compared with the wild-type t-complex, and thus recombination in heterozygotes of the t-haplotype is strongly suppressed along the entire t-complex region. Within this genetically locked t-haplotype, many mutations related to various interesting phenotypes (e.g., taillessness, transmission ratio distortion, recessive lethality) have accumulated, and many mouse geneticists have been attracted to t-haplotype research. Many recessive lethal mutations known as t-complex lethal mutations have been found, and detailed phenotypic analyses have revealed that the functions of t-lethal genes are related to important developmental events. Therefore, identification of the genes responsible for these lethal mutations may contribute to our understanding of the mechanisms of mammalian development. In this review, I introduce the phenotypes of t-lethal mutations and describe recent findings, including our results regarding the molecular identification of a t-lethal gene.


INTRODUCTION
isolated mice with an altered tail morphology in heterozygotes more than 80 years ago. The causative mutation was named Brachyury (T), after the Greek words brakhus oura, meaning "short tail." By outcrossing heterozygous short-tailed T/+ mice with various laboratory and wild stocks carrying a normal tail, Dobrovolskaia-Zavadskaia and Kobozieff (1927) discovered tailless mice. Via intercrosses of the animals, only tailless mice could be obtained (Dobrovolskaia-Zavadskaia and Kobozieff, 1932), and Chesley and Dunn (1936) proved that these tailless mice were maintained in a balanced lethal system, i.e., compound heterozygous for T and another recessive lethal mutation, and thereby designated as T/t (Fig. 1A). The previously named t-allele is now known as the t-haplotype, and is a variant form of the proximal third of mouse chromosome 17, spanning approximately 35 Mb, that includes the major histocompatibility complex (MHC or H-2) (reviewed in Silver, 1985). The t-haplotype contains a characteristic chromosomal structure, and is defined by a series of four tandem inversions (Fig. 1B;Silver, 1993). Because of the presence of inversions, recombination is strongly suppressed between the t-haplotype and wild-type chromosome 17.
The t-haplotype is known as a selfish genetic element because heterozygous male mice preferentially transmit the t chromosome, which is mouse chromosome 17 carrying the t-haplotype, to the next generation (Fig. 1C). In most cases, transmission of the t chromosome occurs at least 90% of the time (often > 99%). This phenomenon is male heterozygote-specific, and is called transmission ratio distortion (TRD) (reviewed in Lyon, 2003). Although the mechanism of TRD has not been completely elucidated, Herrmann and his colleagues identified tcomplex responder and three t-complex distorters, the thaplotype forms of which cooperate to decrease the motility of wild-type sperm and thereby enhance the likelihood of fertilization by t-sperm (Herrmann et al., 1999;Bauer et al., 2005Bauer et al., , 2007Bauer et al., , 2012. During the course of t-haplotype research, hundreds of t-haplotype mouse lines have been isolated from wild mouse populations. Interestingly, most of them carried one lethal mutation (reviewed in Bennett, 1975), and in some cases they had two different lethal mutations that could complement each other in genetic tests (Dunn et al., 1973;Artzt et al., 1979;Klein et al., 1984;Golubić et al., 1987). In this manner, 16 complementing loci have been identified (Klein et al., 1984). Among them, seven complementation groups of t-haplotypes were examined to determine their developmental phenotypes, and were found to exhibit lethality at various stages of development (Table 1) (Bennett, 1975;Guenet et al., 1980). Geneti-Edited by Hiroshi Iwasakii * Corresponding author. E-mail: sugimoto@rtc.riken.jp Fig. 1. Characteristics of the t-haplotype. (A) Compound heterozygotes of T and the t-haplotype (T/t) display a tailless phenotype (arrow), whereas heterozygotes of the t-haplotype have normal tails. Because T/T and t/t are lethal, only tailless T/t mice are born from intercrosses of T/t mice. (B) The chromosomal structure of the t-complex region and the locations of t-lethal genes. The t-haplotype contains four tandem inversions (In(17)1-4) compared with the wild-type t-complex. (C) A schematic diagram of transmission ratio distortion. Male heterozygotes of the t-haplotype preferentially transmit the t-haplotype compared with the wild-type chromosome 17.
cists interested in mammalian development attempted to identify the genes responsible for these lethal mutations. However, it was difficult to identify them by conventional genetic tests because of the presence of recombination suppression, and the low viability and fertility of compound heterozygotes between different complementation groups of t-haplotypes.
Toward the end of 2012, a t-lethal gene, t w5 , was positionally cloned for the first time (Sugimoto et al., 2012). The history of mouse developmental genetics can be viewed as the history of mouse t-haplotype research, and this is a good opportunity to review the developmental genetics of the mouse t-haplotype.

STRUCTURE OF THE t-HAPLOTYPE
In early studies of the t-haplotype, it was determined that recombination within a large region linked to the tailless phenotype was strongly suppressed in heterozygotes of t-haplotypes (Dunn and Caspari, 1945;Dunn and Gluecksohn-Schoenheimer, 1950;Dunn et al., 1962;Lyon and Meredith, 1964). This recombination-suppressed region extended from the vicinity of the centromere to the distal portion of the MHC complex (Fig. 1B). Artzt and co-workers identified a large inversion in the distal half of the t-haplotype (Silver and Artzt, 1981;Artzt et al., 1982;Shin et al., 1983). Using probes for the t-complex region obtained by means of a microdissection cloning technique (Röhme et al., 1984), a second large inversion was found in the proximal region of the t-haplotype, in which T and qk were located (Herrmann et al., 1986). Furthermore, Hammer et al. (1989) discovered two small inversions on the proximal sides of the two large inversions. Eventually, four non-overlapping tandem inversions defined the complete t-haplotype, although its precise boundaries have not yet been identified (Fig. 1B).

DEVELOPMENTAL GENETICS OF MUTATIONS USED AS MARKERS FOR t-HAPLOTYPE RESEARCH
Mutations with visible phenotypes, which occurred in the t-complex region of wild-type chromosome 17, have been used as markers to determine the presence or absence of the t-haplotype. Studies of these mutations have yielded important findings for developmental biology, and I will briefly describe three of them.
The T mutation is the oldest marker for identifying the presence of the t-haplotype because of the taillessness in T/t mice (Fig. 1A). In 1990, Herrmann et al. identified the gene responsible for the T mutation by positional cloning. T is a transcription factor that binds to a specific motif through an N-terminal region known as the T-box (Naiche et al., 2005). By searching for genes carrying the T-box domain, several T-box genes have been identified to date. Studies of T-box genes and their targets revealed that they play a central role in the regulatory networks controlling metazoan embryogenesis (Wardle and Papaioannou, 2008). T is predominantly expressed in the mesoderm cells of gastrulation-stage mouse embryos and is required for mesoderm formation (Wilkinson et al., 1990). The identification of T is undoubtedly one of the most important milestones of developmental biology in both vertebrate and nonvertebrate organisms.
The Fused (Fu) mutation was originally identified as a dominant mutation resulting in tail abnormalities (Reed, 1937), and it has been used as a genetic marker in thaplotype research. In 1997, Zeng et al. identified Axis inhibition protein 1 (Axin1) as the gene responsible for the Fu mutation. Axin1 is now known to have pleiotropic functions in negatively regulating canonical Wnt signaling, in activating TGF-β, SAPK/JNK and p53 signaling, and in promoting Myc degradation (Zhang et al., 1999;Rui et al., 2004Rui et al., , 2007Liu et al., 2006;Arnold et al., 2009;MacDonald et al., 2009). Therefore, Axin1 is one of the key regulators of embryogenesis and organogenesis.
The Quaking (qk) mutation was identified as a recessive mutation producing severe defects in myelination of the central nervous system, which lead to rapid tremors by postnatal day 10 (Sidman et al., 1964). The qk mutation is due to a 1-Mb deletion in the regulatory element of a gene encoding the RNA-binding protein QKI (Ebersole et al., 1996). QKI is believed to regulate pre-mRNA splicing (Wu et al., 2002), mRNA stability (Li et al., 2000;Larocque et al., 2005), nuclear mRNA export (Larocque et al., 2002) and translation repression (Saccomanno et al., 1999). Homozygotes of N-ethyl-N-nitrosourea (ENU)mutagenized or targeted deficient alleles exhibit a deficiency of vasculogenesis in the yolk sac and die at around embryonic day (E)10.5 (Noveroske et al., 2002;Li et al., 2003). Furthermore, a recent report demonstrated that QKI plays an important role in cancer suppression through stabilization of a microRNA (Chen et al., 2012). Thus, QKI appears to have pleiotropic functions in mammalian development.
These mutations are commonly used in t-haplotype research, and many mouse geneticists are interested in their phenotypic features. Thus, it is apparent that thaplotype research accelerated the understanding of the function of these genes.

EFFECTS OF t-LETHAL MUTATIONS ON MOUSE DEVELOPMENT
A schematic diagram of early-to mid-stage embryogenesis is shown in Fig. 2, with the lethal periods of each thaplotype indicated. t-haplotypes have been classified by their complementarity with each other (Bennett, 1975). Previously, phenotypic analyses were conducted in seven complementation groups (Table 1). Homozygotes of these t-haplotypes display interesting phenotypes that appear to be related to various developmental events during mammalian embryogenesis. As in the case of molecular identification of the t w5 lethal mutation (see below), I believe that the genes responsible for these mutations will be identified in the future and provide important clues for understanding mammalian development. Phenotypic information, including explanations for embryonic death, will be helpful for investigating the functions of these genes. Thus, I summarize the developmental phenotypes of t-lethal mutations in chronological order based on the time of lethality (Fig. 2, Table 1). t 12 Complementation group t 12 and t w32 belong to this group. Both t 12 /t 12 and t w32 /t w32 embryos develop to the morula stage but die before implantation (Fig. 2;Smith, 1956;Granholm and Johnson, 1978). The intercellular junctions between cells appeared to be tenuous in homozygous embryos, and blastomeres were rounded, suggesting the absence of compaction (Calarco and Brown, 1968;Hillman et al., 1970). Before any signs of morphological abnormality had arisen, mutant embryos could be identified by electron microscopy as early as the two-cell stage, with accumulated lipid droplets in the nucleus and cytoplasm (Hillman et al., 1970;Hillman and Hillman, 1975). Early cleavage stage development is regulated by maternal factors to some extent, and the major wave of zygotic genome activation occurs around the two-cell stage (Tadros and Lipshitz, 2009). The gene responsible for the lethal mutation of this complementation group is probably expressed as early as the two-cell stage and indispensable for cleavage stage development (Table 1). t 0 Complementation group t 0 is one of the t-haplotypes discovered by Dobrovolskaia-Zavadskaia as tailless mice and was established as a mouse line by Dunn (Dobrovolskaia-Zavadskaia and Kobozieff, 1927;Chesley and Dunn, 1936). Homozygotes of t 0 were implanted in the uterus, but growth retardation was first observed as early as E5.5 ( Fig. 2; Gluecksohn-Schoenheimer, 1940). t 0 /t 0 embryos do not form a proamniotic cavity, and resorption of mutant embryos is initiated at E6.5 (Chesley and Dunn, 1936).
The inner cell masses (ICMs) of homozygotes of this group failed to grow in vitro, and trophectoderm (TE) cells also displayed defects in differentiation; therefore, this lethal mutation appears to be cell-autonomous lethal. The inability to establish embryonic stem cell (ESC) lines from t 0 /t 0 blastocysts is consistent with this expectation (Table 1; Martin et al., 1987). t w73 Complementation group The origin of t w73 is from a wild population in Denmark (Dunn and Bennett, 1971;Dunn et al., 1973). The growth of the ectoplacental cone and the attachment of the TE to the uterus are defective in mutant embryos, suggesting a failure in precise implantation (Bennett, 1975). The visceral endoderm cells were flattened at E6.5 in normal embryos but not in t w73 /t w73 embryos, and there was no evidence of gastrulation. Overall, t w73 /t w73 embryos appear to arrest at around E5.5 ( Fig. 2; Spiegelman et al., 1976). During in vitro culture of blastocysts, there was no difference in the expansion of outgrowths between t w73 /t w73 and control embryos despite the impaired invasiveness of TE cells in mutant embryos (Axelrod, 1985). t wPa-1 Complementation group Three t-haplotypes were isolated from mice captured in southern France (Guenet et al., 1980). One of these lines, t wPa-1 , was confirmed to carry a lethal mutation other than the conventional six tlethal mutations ( Table 1). Homozygotes of t wPa-1 exhibited growth retardation shortly after implantation, at around E5.5 (Fig. 2). Many dead cells were observed in all areas of mutant embryos, but since no further analysis has been conducted the particular tissues affected by this mutation cannot be identified. t w5 Complementation group t w5 is one of the most intensively studied t-haplotypes. Detailed histological analyses demonstrated that embryos homozygous for this group first appeared abnormal at E5.5-E6.5 (Figs. 2 and 3A), when mesoderm formation initiates in normal embryos, and numerous pyknotic cells were observed in the embryonic ectoderm (Bennett and Dunn, 1958;Sugimoto et al., 2003Sugimoto et al., , 2012. Cell division as monitored by a bromodeoxyuridine incorporation assay was markedly decreased in the embryonic ectoderm of t w5 /t w5 embryos compared with that in wild-type embryos, whereas cells in the extraembryonic tissues were actively dividing in both t w5 /t w5 and wild-type embryos, at E6.5 (Sugimoto et al., 2012).
Previously, ESCs were successfully established from t w5 /t w5 blastocysts, and they could differentiate into derivatives of all three primary germ layers by forming teratomas (Magnuson et al., 1982;Martin et al., 1987). Furthermore, the peri-implantation lethality of t w5 /t w5 embryos was rescued when the extraembryonic tissues were supported with normally functioning cells by generating chimeric embryos composed of diploid t w5 /t w5 and tetraploid normal embryos (Sugimoto et al., 2003). These results indicate that t w5 /t w5 embryonic cells can survive in an undifferentiated state and differentiate with the support of wild-type cells; and thus, t w5 lethal mutations are not cell-autonomous lethal.
Although homozygosity for t w5 resulted in severe abnormality in the embryonic ectoderm, it was demonstrated using a horseradish peroxidase incorporation assay that endocytic activity was significantly lower in mutant embryos than in control embryos (Sugimoto et al., 2003). Therefore, the visceral endoderm cells of t w5 /t w5 embryos appear to have impaired functional differentiation, and the defects in the growth and differentiation of the embryonic ectoderm are secondary effects of the dysfunction of the visceral endoderm. Therefore, the gene responsible for the t w5 mutation may regulate the growth and differentiation of the embryonic ectoderm through the interaction between embryonic and extraembryonic (probably visceral endoderm) cells. Molecular identification of the t w5 gene strongly supports this hypothesis (Sugimoto et al., 2012; see below). t 9 Complementation group Primitive streak formation initiated normally in homozygous mutant embryos, but morphological abnormalities, including a low number of mesoderm cells, were clearly detectable as early as E7.0 (Bennett and Dunn, 1960;Moser and Gluecksohn-Waelsch, 1967). Most of the mutant embryos die by E10.5 (Fig. 2), but, interestingly, they have duplicated notochords and neural grooves or neural tubes. By electron microscopy, Spiegelman and Bennett (1974) showed that the mesoderm cells of t 9 /t 9 embryos were abnormally shaped and failed to establish intercellular junctions with other cells, which is one of essential factors for the migration of mesoderm cells during gastrulation (Solnica-Krezel and Sepich, 2012). The failure of cell-cell interactions may result in the unorganized differentiation of the trunk mesoderm, which results in the overgrowth of part of the embryo and neural groove duplication. Martin et al. (1987) successfully established ESC lines from homozygotes of t w18 , another t-haplotype of this group, and they could differentiate into various types of cells, including mesodermal tissues. t w1 Complementation group Homozygotes of this mutation exhibited a variable lethal period ranging from E8.5 to immediately before birth ( Fig. 2; Bennett et al., 1959). At E9.5, pyknosis was occasionally observed in the ventral half of neural tissue. In addition to neural tissue degeneration, abnormalities in chondrification and ossification were observed, particularly in the fetal skull at E13.5 and at later stages, suggesting impaired bone formation. By transplanting ICMs isolated from E4.5 embryos into testes, Axelrod et al. (1981) demonstrated that homozygous-mutant ICMs had the ability to form Fig. 3. Identification of the t w5 -lethal gene. (A) E6.5 embryos of wild-type and t w5 /t w5 . Homozygotes of t w5 exhibit a severe growth defect in the embryonic ectoderm. To define the affected tissues in t w5 /t w5 embryos, the embryonic ectoderm was stained with anti-OCT3/4 antibody (magenta). (B) A map of the sequences used in transgenic rescue experiments. BAC5, the 54-kb PacI-RsrII fragment, and the 20-kb PsiI fragment can rescue the early lethal phenotype of t w5 . (C) Sequence analyses revealed the two-nucleotide insertion in Vps52 of t w5 , which resulted in a frameshift mutation, and the reversion of the frameshift in Vps52 of t w5G . Pa, PacI site; Ps, PsiI sites; Pv, PvuI sites; Rs, RsrII site; Tg, transgene. (Images have been modified from Sugimoto et al., 2012.) teratomas containing differentiated derivatives of all three germ layers, including neuroepithelium. However, none of the mutant teratomas contained either cartilage or bone tissues, whereas heterozygous ICMs could differentiate into these tissues. Therefore, this mutation may affect the differentiation and/or growth of skeletal tissues and also the survival of a certain type of neural cell.

GENETIC MAPPING OF t-LETHAL MUTATIONS
While the t-haplotype has been genetically isolated from the mouse genome because of recombination suppression, it is estimated to have existed in mice for more than one million years (Silver, 1993). Therefore, a number of mutations have accumulated within the thaplotype. Developmental geneticists have long attempted to identify t-lethal genes, and, by considerable efforts, candidate subregions for these t-lethal mutations have been identified (see below) using compound heterozygotes between different complementation groups of t-haplotypes, in which recombination between two different t-haplotypes could occur (Silver and Artzt, 1981), as well as deletion mutations within the t-complex region and/or partial t-haplotypes that arose by rare recombination between the t-haplotype and wild type chromosome 17 (Lyon, 2003). However, it was difficult to finally identify the genes responsible for these lethal mutations because of the low viability and fertility of compound heterozygotes and the insufficiency of appropriate markers among t-haplotypes; thus, more than 80 years were required to successfully identify a t-lethal gene for the first time since the original report of the t-haplotype.
The mutation of the t 9 complementation group was mapped near the MHC complex region, but was proved to be not a single gene mutation but a combination of a deletion and a duplication at the distal end of In(17)4 of the t-haplotype (Búcan et al., 1987). Eventually, Barclay et al. (1996) restricted the critical region for the deletion to 5.4 Mb (Fig. 2, Table 1), which is largely occupied by two gene clusters: the Zinc finger protein (Zfp) gene cluster, and the vomeronasal receptor gene cluster including vomeronasal 1 receptor (V1R) genes, V2R genes, and formyl peptide receptor-like genes (Herrada and Dulac, 1997;Matsunami and Buck, 1997;Ryba and Tirindelli, 1997;Liberles et al., 2009;Rivière et al., 2009). The vomeronasal receptors are believed to primarily detect pheromones; therefore, it is unlikely that the lethality of the t 9 complementation group is the result of the loss of these genes. As previously suggested (Crossley and Little, 1991;Noce et al., 1993;Shannon and Stubbs, 1999), the loss of one or more Zfp gene(s) may result in the embryonic lethality of the t 9 complementation group.
The lethal mutation for the t w12 haplotype, which belongs to the t w1 complementation group, was previously mapped between two markers, tufted (tf) and proviral integration site for Moloney murine leukemia virus kinase 1 (Pim1), in In (17)4 (Ark et al., 1991). Inositol 1,4,5trisphosphate receptor, type3 was identified as the gene responsible for the tf mutation (Ellis et al., 2013), and is located 2.5 Mb distal to Pim1 on the wild-type chromosome 17 (Fig. 2, Table 1). There are 50 genes and a piRNA cluster within this 2.5-Mb region, but it is difficult to suggest candidate genes at present.
The t 12 (t w32 ) mutation has also been mapped near the MHC complex region, and although the candidate region has been restricted to a 3.5-Mb region between D17Mit64 and D17Mit83 (Fig. 2, Table 1), there is currently no more specific information.
Only the t w73 lethal mutation has been mapped to In(17)2 (Babiarz et al., 1982;Artzt, 1984). The t w73 critical region was finally narrowed down to 200 kb (Fig. 2, Table 1; Zwart et al., 2001). Of the eight genes located in this 200-kb region, it was reported that Wilms tumor 1-associating protein (Wtap) is indispensable for early embryogenesis (Fukusumi et al., 2008). Mice deficient for Wtap die by E7.5 because of the impairment of definitive endoderm and mesoderm development. Wtap is ubiquitously expressed in embryos from the two-cell stage to E7.5, but the authors demonstrated that chimeric embryos generated with Wtap-null ESCs and wild-type embryos developed normally beyond the lethal period of Wtap-null embryos even when almost all cells of the chimeric embryos were derived from mutant ESCs. Extraembryonic ectoderm and/or visceral endoderm are probably the tissues in which Wtap is required for early embryogenesis, and Wtap is a plausible candidate for the t w73 mutation.

IDENTIFICATION OF THE GENE RESPONSIBLE FOR THE t w5 RECESSIVE LETHAL MUTATION
Previously, the t w5 lethal mutation was mapped to a 750-kb region, which partially overlaps with the MHC complex on chromosome 17 (Abe et al., 2004). However, it is difficult to further narrow the region using conventional genetic tests because of the paucity of appropriate markers within this 750-kb region, where there are still 42 potential candidate genes. To identify the gene responsible for the t w5 recessive lethal mutation, we performed transgenic rescue experiments with five BAC clones covering the candidate region and restriction fragments of these BACs ( Fig. 3B; Sugimoto et al., 2012). With the first screening, we found that only one BAC clone (BAC5 in Fig. 3B), which contained 16 genes, could rescue the early embryonic lethal phenotype of the t w5 mutation. Finally, we identified a 20-kb genomic fragment as a critical region for the early embryonic lethality of t w5 mutation. This 20-kb fragment covers the whole region of one gene, vacuolar protein sorting 52 (Vps52), and the 5' part of ribosomal protein s18 (Rps18). Com-plementation tests using Vps52-knockout mice demonstrated that Vps52 is the gene responsible for the t w5 lethal mutation: compound heterozygotes of the Vps52knockout allele and the t w5 allele displayed a similar phenotype to t w5 homozygotes at the peri-implantation stage. In exon 2 of Vps52, there is a stretch of nine consecutive guanine nucleotides on the allele of the commonly used laboratory strain, C57BL/6 (B6); however, we found an insertion of two additional guanine nucleotides on the t w5 allele, resulting in a frameshift mutation (Fig.  3C). Importantly, we confirmed that this two-nucleotide insertion in Vps52 does not occur in another t-haplotype, t 12 , indicating that the frameshift mutation is specific to the t w5 allele of Vps52. We previously isolated a mouse line lacking the lethal mutation but carrying all of the features of a complete t-haplotype from a balanced lethal cross of T/t w5 mice, but we could not elucidate the reason why this spontaneous revertant, named t w5G , emerged. Interestingly, sequence analysis demonstrated that the t w5G allele of Vps52 has 12 guanine nucleotides in the mutated site; thus, the shifted frame is restored in the t w5G allele of Vps52 with a one-glycine insertion. To the Fig. 4. Vps52 is required for pluripotent differentiation through cell-cell interactions during early embryogenesis and embryonic stem cell (ESC) differentiation. (A) Epiblast-specific Vps52-knockout embryos can survive beyond the lethal period of the t w5 mutation. (B, C) t w5 /t w5 ESCs can differentiate into visceral endoderm (VE) (B) but cannot grow into highly differentiated cystic embryoid bodies (EBs) (C, upper panel). On the other hand, t w5 /t w5 ESCs transduced with a genomic DNA fragment containing the whole Vps52 gene (t w5 /t w5 ;Vps52) can differentiate into cystic EBs (C, lower panel). (D) Chimeric EB formed after 3 (top) and 7 days (bottom) of differentiation. Magnified images of areas indicated with dashed squares are shown (middle). t w5 /t w5 ESCs transduced with a fluorescent protein (tdTomato)-expressing construct (magenta) and t w5 /t w5 ;Vps52 ESCs (shown in C) were mixed, and EB formation was observed. VE-like cells (black arrowheads) were negative for fluorescence, indicating that they were derived from t w5 /t w5 ;Vps52 ESCs. Inner cells surrounded by the VE-like cells were positive for fluorescence and formed the embryonic ectoderm (EmE)-like epithelial layer (white arrowheads). With the support of Vps52-expressing visceral endoderm cells, t w5 /t w5 ESCs can differentiate into cystic EBs. (E) Model for the function of Vps52 in early mouse embryogenesis. Vps52 is expressed in the VE in peri-implantation embryos, and it indirectly regulates the development of the EmE, probably through intercellular signaling from VE to EmE (orange arrows). Scale bars, 0.5 mm (A) and 0.2 mm (C and D). (Images have been modified from Sugimoto et al., 2012.) best of my knowledge, this is the only report of a reversion of a lethal mutation in laboratory mice.
Vps52 is a gene originally identified from a mutant yeast strain with a defect in protein sorting at the late Golgi compartment, and its product is known to be a component of the Golgi-associated retrograde protein (GARP) complex together with VPS51, VPS53 and VPS54 (Conibear and Stevens, 2000;Conibear et al., 2003). Vps52 is widely conserved from unicellular organisms to mammals, and is believed to have a general role in tethering endosomes to the trans-Golgi network (Bonifacino and Rojas, 2006). On the other hand, Vps52 appears to regulate the growth and/or differentiation of the epiblast (or its derivatives) during mouse development: the development of embryonic tissues was severely affected in t w5 / t w5 embryos ( Fig. 3A; see above), although extraembryonic tissues grew to some extent beyond the lethal period. The facts that t w5 /t w5 ESCs could be established and that tetraploid embryos could rescue the early lethality of t w5 / t w5 embryos indicate that VPS52 plays important roles in extraembryonic tissues rather than in epiblasts (Magnuson et al., 1982;Sugimoto et al., 2003). The results of an epiblast-specific knockout assay for Vps52 also suggested that the expression of Vps52 in epiblasts is dispensable for peri-implantation embryogenesis (Fig. 4A), and the transcription of Vps52 mRNA is actually higher in the visceral endoderm than in epiblasts in peri-implantationstage embryos (Sugimoto et al., 2012). When t w5 /t w5 ESCs were induced to differentiate into embryoid bodies (EBs) in suspension culture, GATA4-positive visceral endoderm cells emerged during the early period of differentiation, as observed in wild-type ESCs (Fig. 4B). However, EBs from t w5 /t w5 ESCs could not differentiate further, and they retained a simple ball-like morphology without a cavity even at the late period. Meanwhile, t w5 / t w5 ESCs carrying the Vps52 expression construct could become highly differentiated cystic EBs (Fig. 4C), similar to wild-type ESCs. Furthermore, t w5 /t w5 EBs could grow into cystic EBs with the support of visceral endoderm cells expressing functional Vps52 (Fig. 4D). These in vitro assays also demonstrated that VPS52 is required for the differentiation of pluripotent cells through cell-cell interactions, probably between the visceral endoderm and the epiblast (Fig. 4E).
The visceral endoderm is known to induce the differentiation of the pluripotent epiblast through intercellular signaling during early embryogenesis (reviewed in Tam and Loebel, 2007). However, the mechanisms by which the signal molecules are delivered from cell to cell, activated in specific cells, and degraded to reduce the signal intensity at the appropriate time are unclear. Recently, it was reported that one such signaling pathway, BMP-Smad, which is known to be required for the initiation of gastrulation, is spatiotemporally restricted during early mouse embryogenesis through the mVam2 (also known as Vps41)-dependent endocytic pathway (Aoyama et al., 2012).
Previously, it was reported that a deficiency of VPS54, another component of the GARP complex, resulted in mid-gestation lethality around E11.5, presumably because of cardiovascular malfunction (Schmitt-John et al., 2005). The phenotypes of Vps52-null embryos are different from those of Vps54 mutants, suggesting that Vps54 is dispensable for peri-implantation embryogenesis; in addition, the function of Vps52 in visceral endoderm is independent of Vps54. To understand the role of the GARP complex as a signal regulator for early embryogenesis, it will be worthwhile to compare the phenotypes associated with deficiencies of the other GARP components, VPS51 and VPS53. The identification of Vps52 as the gene responsible for the t w5 lethal mutation will certainly be a clue for understanding the mechanisms of intercellular signaling during mammalian embryogenesis.

CONCLUDING REMARKS
During the evolution of Mus musculus, many mutations have accumulated within the t-haplotype because they could not be segregated owing to the suppression of recombination. In this paper, I have reviewed the phenotypic features of the lethal mutations contained in each complementation group of t-haplotypes. These mutants display lethality at various periods throughout mouse embryogenesis from the cleavage stage to the perinatal stage, and their phenotypes are related to various developmental events, including compaction of pre-implantation embryos, the initial steps of implantation by trophoblast cells, epiblast differentiation, mesoderm induction, neural development and bone formation. The genes responsible for these lethal mutations must have important functions in mammalian development. Of these t-lethal genes, the t w5 -lethal gene was identified for the first time. The gene responsible, Vps52, had not previously been implicated in embryogenesis, but was demonstrated to play an important role in controlling the differentiation of pluripotent cells through cell-cell interactions, thereby regulating mammalian development. The relationship between the intracellular vesicle transport system and developmental regulation is one of the hot topics of cell biology and developmental biology (Wada, 2013;Wada and Sun-Wada, 2013). In addition to t w5 , the identification of lethal mutations of the other t-haplotypes will provide important information for understanding mammalian development. At present, we can exploit genomic sequencing data, and recent advances in genome engineering technologies have facilitated the easier production of mutant mice (Gaj et al., 2013;Waterston et al., 2002), and the time has therefore come to unravel the mysteries of the t-haplotypes.
During the long history of life science, much important knowledge has been accumulated by research based on mouse genetics and momentous novel findings have been generated at the frontiers of mammalian biology; for example, it has been experimentally demonstrated using inbred mouse strains that the variability of the X chromosome is a key factor for genetic incompatibility in mammals (reviewed in Oka and Shiroishi, 2014). The study of the t-haplotype is one such area of historic research in the field of mouse genetics, and I believe that there are still many buried treasures to be uncovered within the thaplotype.