A mouse homolog of the Drosophila pair-rule gene Odd Oz (Odz4) maps to the critical region of the l7Rn3 locus on mouse chromosome 7. Here we show that Odz4 is an excellent candidate for this allelic series because (1) it spans the entire critical region, (2) the phenotypes correlate with embryonic expression, (3) the complex genetic inheritance of the alleles is consistent with complex transcriptional regulation, and (4) one allele has a mutation in a conserved amino acid. Odz4 uses five alternate promoters that encode both secreted and membrane-bound proteins. Intragenic complementation of the l7Rn3 alleles is consistent with these multiple-protein isoforms. Further, the allelic series shows that Odz4 is required to establish the anterior-posterior axis of the gastrulating mouse embryo and is necessary later for mesoderm-derived tissues such as somites, heart, and skeleton. Sequencing of RT-PCR products from five of the six alleles reveals a nonconservative amino acid change in the l7Rn3^m4 allele. This amino acid is important evolutionarily, as it is conserved to Drosophila. Together, our data indicate that Odz4 is mutated in the l7Rn3 allele series and performs roles in the mouse brain, heart, and embryonic patterning similar to those of its Drosophila counterpart.

---

THE TENEURIN/ODD OZ (TEN/ODZ) protein family contains many members with orthologs identified in mammals, vertebrates, insects, and nematodes (MINET and CHIQUET-EHRISMANN 2000). The first two members of this gene family (Ten^a and Ten^m) were identified in Drosophila during a search for homologs of Tenascin, a component of the extracellular matrix (BAUMGARTNER and CHIQUET-EHRISMANN 1993; BAUMGARTNER et al. 1994). This family of transmembrane protein receptors shares a common N terminus, called the TENEURIN intracellular domain, as well as 5–8 TENASCIN-type EGF-like repeats and >20 tyrosine/aspartic acid (YD) repeats, which are related to the rearrangement hotspot (RHS) domains found in bacteria. Orthologs have been identified in seven different species (MINET and CHIQUET-EHRISMANN 2000). While only one Ten/Odz gene, R13F6.4, was identified in Caenorhabditis elegans, two homologs have been reported in Drosophila (Ten^a and Ten^m) and zebrafish (Ten^m3 and Ten^m4), three have been located in chicken (Ten1, Ten2, and Ten4), and four genes have been cloned from both mouse (Odz1-4) and human (TEN1-4). Although only one gene has been identified so far in the rat (Neurestin ), it is likely that additional homologs will be annotated in the near future.

The full-length TEN/ODZ proteins range in size from 2515 to 2825 amino acids (MINET and CHIQUET-EHRISMANN 2000). There is 37–41% identity between the vertebrate orthologs and the Drosophila Ten^m or Ten^a proteins, indicating that (1) the Drosophila and vertebrate gene duplications occurred after the species diverged and (2) either of the Drosophila homologs could be orthologous to one of the vertebrate Ten/Odz genes. Ten^m/Odz is expressed in seven stripes in the blastoderm stage of early embryonic development, which is consistent with the expression of the fly pair-rule genes. During later development, expression is dynamic, but most prominent in the central nervous system (CNS) and heart (BAUMGARTNER et al. 1994; LEVINE et al. 1994). Ten^a also demonstrates a dynamic pattern of developmental expression and is prominent in the CNS, hindgut, and brain (FASCETTI and BAUMGARTNER 2002).

One of the mammalian homologs, Odz4 (formerly Doc4), was induced following endoplasmic reticulum stress response in mouse fibroblasts (WANG et al. 1998). In an effort to identify the function of Odz in mammals, OOHASHI et al. (1999) cloned the four mouse homologs, Odz1–4. These studies demonstrated that all four genes were highly expressed in brain, suggested that the mammalian genes functioned as type II transmembrane proteins, and indicated that each gene contained at least three alternatively spliced transcripts. During brain development, these four genes exhibit distinct, but overlapping, expression profiles (ZHOU et al. 2003).

Ethylnitrosourea (ENU) mutagenesis screens in the mouse identify novel mutants that have defects in early embryonic development (NOLAN et al. 1997; HARDISTY et al. 1999; RINCHIK and CARPENTER 1999; HRABE de ANGELIS and FLASWINKEL 2000; ISAACS et al. 2000; NOLAN et al. 2000; SOEWARTO et al. 2000; ALESSANDRINI et al. 2001; GRAW et al. 2002; VIVIAN et al. 2002; KILE et al. 2003). A series of alleles at l7Rn3 was isolated following ENU mutagenesis of a large deletion complex encompassing the Tyrosinase (Tyr) locus on mouse chromosome 7 (RINCHIK and CARPENTER 1999). The l7Rn3 complementation group is composed of six lethal mutations (m1–m6) that demonstrate interesting genetic behaviors, including intragenic complementation and a maternal effect (RINCHIK and CARPENTER 1999).

Odz4 fulfills all of the candidate gene criteria for l7Rn3. Odz4 spans the entire 700-kb critical region for l7Rn3 and encodes an 12-kb full-length transcript that produces different alternatively spliced isoforms. Here, we reveal that Odz4 has five putative promoters and 44 exons, which include 9 additional exons not previously annotated. Many of these new exons are contained in tissue-specific or developmentally restricted transcripts. We also demonstrate that the l7Rn3 mutants display a wide range of phenotypes, each of which is indicative of disruption in mesoderm differentiation. Finally, mutation analysis indicates that the m4 allele contains a nonconservative missense mutation in an amino acid that is conserved through Drosophila. Taken as a whole, our data indicate that Odz4 plays a major role in mouse gastrulation.

Mouse strains and embryo analysis:

BALB/cRl-males were treated with ENU to generate the l7Rn3 complementation group, which is tightly linked with the Tyrosinase locus. The recovery, breeding scheme, and initial identification of this mutant allelic series was previously described (RINCHIK and CARPENTER 1993, 1999). These alleles have been designated m1 (l7Rn3^1777SB), m2 (l7Rn3^677SB), m3 (l7Rn3^2292SB), m4 (l7Rn3^4324SB), m5 (l7Rn3^6105SB), and m6 (l7Rn3^2521SB). We bred the mutations onto the FRCH stock, which is maintained at Baylor College of Medicine. The FRCH animals are dark chinchilla and have curly fur and whiskers. This strain, Tyr^c-ch fr/Tyr^c-ch fr, carries the chinchilla allele of the Tyrosinase locus and the spontaneous mutant, frizzy, which flank the l7Rn3 mutations. Heterozygotes, which are light chinchilla and have normal fur, will have a Tyr^c m +/Tyr^c-ch + fr genotype. Homozygous mutants are obtained from intercrossing heterozygotes and have a Tyr^c m +/Tyr^c m + genotype. Two deletion stocks, Tyr^c-ch/Tyr^c-26DVT and Tyr^c-ch/Tyr^c-1FDFoHrc, can be used to generate hemizygous mutants.

Embryos were obtained from timed matings. The day of the vaginal plug was designated 0.5. Embryos were visually observed and photographed under a dissecting microscope for stage classification.

Genotyping:

Embryos and adult tissues were genotyped with D7Mit352 and/or D7Al7. DNA was isolated from adult tissues using proteinase K digestion and phenol/chloroform extraction according to well-established protocols. DNA was obtained from embryos following a modified proteinase K (Life Technologies, Gaithersburg, MD) digestion. Briefly, embryos were incubated in 1x PCR buffer (Life Technologies) supplemented with 0.08 mg/ml proteinase K for 2–3 hr at 55° and then heated to 95° for 10 min to deactivate the proteinase K. Embryos dissected at or before E9.5 were incubated in 25 µl of 1x PCR buffer, while embryos older than E9.5 were digested in a final volume of 100 µl. Genotyping was performed on 10–200 ng of genomic DNA. After an initial denaturing step at 95° for 5 min, D7Mit352 was amplified with the following cycling parameters: 30 cycles of 94° for 30 sec, 58° for 30 sec followed by 72° for 30 sec, with a final 5-min incubation at 72°. Products were size fractionated on 5% metaphor, 1x TBE or 6% native polyacrylamide, 1x TBE gels. Acrylamide gels were electrophoresed for 18 hr at 35 V. D7Al7 was amplified according to the following cycling conditions: 95° for 5 min, followed by 35 cycles of 94° for 30 sec, 62° for 30 sec, and 72° for 30 sec. After a final extension at 72° for 5 min, products were size fractionated on 3% agarose, 0.5% TBE gels. Primer sequences are listed in Table 1.

View this table: In this window In a new window   TABLE 1 Primers used in this study

 

Deletion mapping:

Over 50 nested deletions span the albino deletion complex (RUSSELL and RUSSELL 1959). Complementation testing between these deletions and the mutants was used to determine which deletions included l7Rn3. A deletion was determined to overlap the mutation if 0 albino pups were born in 30 total progeny from a cross between heterozygous mutant and hemizygous deletion animals (RINCHIK et al. 1990). Once the deletion interval was defined, genomic DNA and cDNA probes from the region were used to delineate the extent of the deletion interval. To accomplish this, Mus musculus females hemizygous for the chinchilla allele of the Tyrosinase locus (Tyr^c-ch/Tyr^c-deletion) were crossed to M. spretus males (+^SPT/+^SPT), where Tyr^c-deletion designates a deletion. The deletions arose on the C3Hf/Rl and 101/Rl backgrounds, but were bred to the chinchilla stock. Most progeny will be either Tyr^c-ch/+^SPT or Tyr^c-deletion/+^SPT. However, it was possible that the originating C3Hf/Rl or 101/Rl genome was carried across the deletion. Therefore, polymorphisms were found for the M. musculus (c^ch/c^ch, C3Hf/Rl and 101/Rl) and M. spretus (+^SPT/+^SPT) alleles by Southern analysis according to well-established techniques (LOSSIE et al. 1993). The absence of the M. musculus-specific fragment in DNA from Tyr^c-deletion/+^SPT F₁ animals indicates that a probe is deleted and therefore localized within the l7Rn3 critical region.

Deletion interval gene annotation:

We examined the annotated sequence to identify genes that were localized to the critical interval using the NCBI, University of California at Santa Cruz, and Ensembl databases (WATERSTON et al. 2002). These data were generated in part through the use of the Celera Discovery System and Celera's associated databases (KERLAVAGE et al. 2002). ESTs and mRNAs that had short stretches of homology to the l7Rn3 domain and multiple hits across the genome were not included.

Northern blot analysis:

Northern blots of polyadenylated [poly(A)⁺] RNA from adult mouse tissues (CLONTECH, Palo Alto, CA) and whole embryos (CLONTECH) were hybridized with Odz4 cDNAs according to standard protocols (CHURCH and GILBERT 1984). Odz4 was visualized using probes that span different regions of the gene (see Figure 4). Blots were stripped and reprobed with Gapdh as a loading and mRNA integrity control.

View larger version (56K): In this window In a new window Download PPT slide   FIGURE 4.— Northern analysis of Odz4. (A) Representation of the Odz4 cDNA. (A) The two cDNA probes used for Northern analysis (3.11 and 6.2.5). The 3.11 probe consists mostly of 5' UTR with some protein coding sequence, while the 6.2.5 probe is derived entirely from the coding region. (B) Developmental and adult expression of 5' Odz4. During embryogenesis, a 4.5-kb Odz4 transcript is the first to appear, which is detected by this probe throughout embryonic development and into the adult stage. The large 12-kb mRNA is first detected at E 11.5. Although this message is detected throughout embryogenesis, it is silenced in the majority of adult tissues, with the exception of the brain. The 2.0-kb transcript is detected during midgestation and persists in most tissues through adulthood. However, it is the predominant transcript detected in the liver. In addition, adult kidney and heart show additional alternatively spliced products. (C) The Odz4 coding region shows different expression patterns. Using a probe that spans the transmembrane domain and will hybridize to both intracellular and extracellular regions of Odz4, we demonstrate that the protein-coding region has significantly different expression patterns. A 12-kb mRNA is detected first, with a 10-kb transcript appearing by E11.5. By late gestation, we also detect a 6.5-kb message. There is no evidence that this probe detects the 4.5- and 2.0-kb transcripts. There is also a very different expression profile in the adult. In the brain, a large band corresponds to the 10- and 12-kb mRNAs. In addition, the 4.5-kb transcript is also detected. At this exposure (24 hr), no transcripts are detected in other tissues. A longer exposure indicates that the 10- and 12-kb transcripts are detected in testes, kidney, and heart (data not shown). (D) Gapdh was used as a loading and mRNA integrity control. The following tissues and embryonic timepoints were used for analysis: embryonic day 7.5 (E7.5), E11.5, E15.5, E17.5, testes (T), kidney (K), skeletal muscle (M), liver (Li), lung (Lu), spleen (S), brain (B), and heart (H).

 

RT-PCR:

Odz4 RT-PCR and sequence analyses were performed on total RNA and poly(A)⁺ RNA isolated from E7.5 to E13.5 embryos using RNAStat-60 (total RNA; TEL-TEST, Friendswood, TX) or Poly(A) Pure (mRNA; Ambion, Austin, TX). RNA was reverse transcribed with either the Superscript II or III preamplification system (Life Technologies). The resulting PCR products were electrophoresed on agarose gels, purified by QIAquick extraction (QIAGEN, Valencia, CA) or glasswool, and sequenced using an ABI377 sequencer and BigDye v3.0 chemistry (Perkin-Elmer, Foster City, CA).

In situ hybridization:

Embryos were recovered as described (NAGY et al. 2003). Embryos were fixed in 4% paraformaldehyde in PBS for 15 min to 4 hr (depending on the size of the embryo) at 4° and dehydrated through a methanol series prior to processing by previously published methods (ECHELARD et al. 1993). Digoxygenin-labeled antisense probe was generated according to the manufacturer's instructions (Boehringer Mannheim, Indianapolis).

Deletion mapping and candidate gene identification:

Over 50 nested deletions span the 24-Mb Tyrosinase deletion complex, which extends from the Feline sarcoma oncogene (Fes) to the Hemoglobin beta chain 1 (Hbb-b1) locus on mouse chromosome 7 (RUSSELL 1989). The l7Rn3 complementation group lies 9.7 Mb distal to Tyrosinase within this complex, and three of the deletions are depicted here (Figure 1A). Six phenotypically distinct ENU-induced mutations define the l7Rn3 locus (RINCHIK and CARPENTER 1999). Five of these mutant lines (m1, m3, m4, m5, and m6) were propagated in our laboratory for further analysis. The m2 mutants have a phenotype that is similar to the m1 mutants and were not included in these studies. Two molecular markers in this region, D7Mit352 and D7Al7, were used for genotyping. Previous studies showed that all of the l7Rn3 mutant lines failed to complement the Tyr^c-26DVT and Tyr^c-1FDFoHrc deletions (RINCHIK and CARPENTER 1999). Further complementation studies in our lab between the four lethal mutant lines (m1, m3, m4, and m5) and the nested deletions demonstrate that the Tyr^c-39SAS deletion complements our mutants (data not shown). Therefore, the l7Rn3 allele series is located between the breakpoints of the Tyr^c-39SAS and the Tyr^c-1FDFoHrc deletion regions (Figure 1A).

View larger version (18K): In this window In a new window Download PPT slide   FIGURE 1.— Physical map of l7Rn3. (A) Physical map of l7Rn3 in relation to the Tyrosinase locus. Genes and molecular markers are indicated along the 9.7-Mb region. Three nested deletions (Tyrc-26DVT, Tyrc-1FDFoHrc, and Tyrc-39SAS) that span this region are depicted with the deleted intervals shaded. The endpoints of Tyrc-39SAS and Tyrc-1FDFoHrc define the critical region for l7Rn3. (B) Transcript map of the 0.8-Mb l7Rn3 region. The l7Rn3 critical region contains five expressed sequences (genes A–E). Genes A and C have an open reading frame. Gene A is also known as Odz4, Odd oz homolog 4. The longest ORF of Odz4 predicts a 2825-amino-acid protein, while the coding region for gene C is only 133 amino acids long. Several alternatively spliced mRNAs originate from Odz4. These range in size from 1.5 to 12 kb. Genes B, D, and E are expressed in many different tissues and at different timepoints during development, but do not contain significant ORFs. (C) The Odz4 gene. Black vertical lines represent previously identified Odz4 exons, with the original nomenclature indicated at the bottom. The new exon numbers are located on top of the horizontal bar depicting the Odz4 genomic locus. Blue vertical lines represent newly identified exons that do not change the protein coding sequence, while red vertical lines are new exons that are predicted to change the protein structure. Arrows indicate transcriptional start sites, and ATG sites are indicated. The map locations of Fes, Tyr, Frizzled 4 (Fzd4), Discs large homolog 2 (Dlgh2), and Hbb-b1 are shown.

 
One method utilized at the University of California at Santa Cruz to integrate the transcriptome with the computer-based exon prediction programs is SGP, which compares mRNA sequences and spliced ESTs to identify the most likely gene boundaries in a given region (KAROLCHIK et al. 2003). This algorithm requires at least 97.5% identity for inclusion, eliminates small blocks of sequence, aligns the transcripts according to predicted splice junctions, and merges overlapping blocks into clusters. SGP analysis predicts that five genes are located within the 700-kb domain encompassing l7Rn3 (Figure 1B; Table 2, SGP column). The largest of these, Odz4, spans the entire critical region (Figure 1, A and B). The other "genes" were predicted on the basis of evidence from a single mouse EST clone (Table 2; Figure 1B, genes B–E). We excluded ESTs and mRNAs that had incomplete stretches of homology and multiple hits across the genome, as this "hybridization" pattern is indicative of a repetitive element or short protein domain.

View this table: In this window In a new window   TABLE 2 Summary of the transcripts identified in the 12-Mb domain

 

Annotation of the Odz4 genomic locus:

The Odz4 locus is very complex. Many mRNAs, ESTs, and predicted transcripts lie within this 735-kb domain. To determine which of these encode unique genes, we compiled RT-PCR data from each of the annotated exons to determine its expression profile in early postimplantation embryos as well as in adult tissues.

Eight mouse and four human mRNAs delineate the Odz4 gene (Table 2). These include three RIKEN cDNAs (AK031375, AK053790, and AK039472) that partially overlap Odz4, a RIKEN clone that denotes an alternative 3' UTR for Odz4 (exon 20; AK078539), two mouse (D87034 and AK122490) and three human ELM2 (Expressed in low-metastatic cells; AB037723, AL080120, and AK023935) sequences, a human ODZ4 transcript (AK056531), as well as two known Odz4 cDNAs (AF059485 and AB025413; WANG et al. 1998; OOHASHI et al. 1999). The positions of these exons were determined by BLAST alignment with the mouse genome.

The full-length Odz4 gene is predicted from two overlapping mouse cDNAs (AF059485 and AB025413; WANG et al. 1998; OOHASHI et al. 1999). Although these transcripts have slightly different cDNAs, they predict the same basic protein structure (Figure 2A). Since no two cDNAs are identical, we predicted that different tissues and developmental stages would have unique Odz4 messages. We also postulated that there could be different protein isoforms that correlated with these temporal and spatial transcripts.

View larger version (12K): In this window In a new window Download PPT slide   FIGURE 2.— Alternative splicing predicts five different ODZ4 protein isoforms. (A) Overview of the full-length protein predicted from cDNA AF059485. The black bar represents the entire protein, with domains indicated by different colored regions. Protein prediction models indicate that the Teneurin domain (green oval) is intracellular, while the EGF-like (epidermial growth factor-like) repeats (fuscia ovals), NHL (NCL1, HT2A, and LIN41 consensus protein domain; blue rectangles), and RHS repeats (yellow ovals) are extracellular. (B) The locations of the exons encoding specific protein domains are indicated by vertical black bars, and numbered when transitions are present. (C) Protein predicted from an RT-PCR product amplified from exons 2 to 20 (AK078539). (D) Protein predicted from an amplicon between exons 1 and 12. (E) Protein predicted from two eye cDNAs (AK053970 and AK039472). This protein would encode an entirely new protein with one EGF-like domain and a carboxypeptidase D domain. (F) Protein predicted from two 3' cDNAs (AB037723 and AL080120), which can be produced from an internal promoter.

 
We identified five new alternatively spliced exons that are located upstream of the start codon (Figure 1C, blue exons 2, 3, 4, 5, and 7A). These exons undergo extensive alternative splicing and use three different promoters. It is likely that the different promoters confer tissue-specific expression patterns, as transcripts originating from exon 1 are detected in multiple tissues and stages of development, while mRNAs starting from exon 2 display three distinct expression patterns. Amplification products between exons 2 and 6, 7, or 15 are found only in adult brain, while RT-PCR between exons 2 and 13 indicates that this amplicon is expressed only in adult ovary (Figure 3B) and E11.5 embryos. However, RT-PCR products that contain exon 2–6, 22, 24, 30, 31, 32, 33, and 34 are observed in most tissues. No amplification products are obtained between exon 1 and exon 2, suggesting that these exons are alternative transcriptional start sites for Odz4.

View larger version (66K): In this window In a new window Download PPT slide   FIGURE 3.— RT-PCR demonstrates tissue-specific expression of Odz4. (A) Lanes 2–7. This RT-PCR product between exons 6 and 13 is detected only in adult brain. (B) Lanes 8–11. In the adult, a transcript between exons 2 and 13 is expressed only in the ovary. (C) Lanes 2–17. An amplicon between exons 25 and 29 is expressed in every tissue examined and contains two testis-specific transcripts. Even lanes, (+) RT; odd lanes, (–) RT. Ut, uterus.

 
Two additional upstream exons are encoded by the 2.5-kb AK031375 transcript (Figure 1C, exons 3 and 7A). This full-length message was isolated from E13.0 testis, includes exons 3, 6, and 7A, and introduces an intragenic poly(A) signal. A large CpG island overlaps exon 3, which indicates that it could be used as another transcription start site. In addition, RT-PCR amplification between exons 1 and 7 indicates that there are three major amplicons in most tissues (Figure 3C). The smaller fragments harbored different combinations of exons 1–6, although none of these alternatively spliced products contained exon 2.

We identified five alternatively spliced exons (12, 20, 24, 25, and 29 shown in red) that could generate multiple protein isoforms (Figure 1C; InterProScan; ZDOBNOV and APWEILER 2001). Exon 12 (Figure 1C, Table 2, chr7_1650.1) is detected only in the adult brain (Figure 3A) and is predicted to result in a truncated protein that contains a partial Teneurin N intracellular domain (Figure 2D).

The 1.7-kb cDNA encoding exon 20 (Figure 1C; AK078539) was isolated from tissue surrounding the Mullerian duct of an E12.0 female embryo. RT-PCR indicates that this exon is expressed at low levels in adult brain and in the embryo (data not shown). Exon 20 contains many translation stop codons and does not appear to harbor any known protein domains. In silico examination indicates that an alternative poly(A) site is located within the first 450 bp of this exon. Protein prediction software indicates that the short protein isoform will include the Teneurin N intracellular domain and the transmembrane domain (Figure 2C; ZDOBNOV and APWEILER 2001).

Exons 24 and 25A were identified from two eye cDNAs (AK053790 and AK039472). These transcripts overlap to produce two additional noncoding Odz4 exons that are alternatively spliced in testis (Figure 3C). AK039472 is predicted to produce a novel protein that would be a short, secreted peptide with three EGF repeats and a carboxypeptidase D domain (Figure 2E). The other eye cDNA does not predict any significant protein isoforms.

Two human ELM2 cDNAs (AB037723 and AL080120) have 90% identity to the murine Odz4 gene. These transcripts overlap the last eight exons of Odz4 to create an additional 4 kb of 3' UTR. Protein prediction programs predict that these transcripts will produce a secreted form of ODZ4 that contains only an RHS domain (Figure 2F). RHS domains were first identified in bacteria, and members of the TEN/ODZ protein family are the only vertebrate proteins known to have this type of protein domain.

A crossover eliminates other cDNAs and ESTs as candidates:

Eight small mRNAs and ESTs that do not encode alternatively spliced products of Odz4 are located within the 735-kb genomic locus (Table 2). Some of these are transcribed in the same orientation as Odz4, while others are predicted to be expressed from the opposite strand. Four expressed sequences were predicted by SGP analyses to encode genes within the Odz4 locus (Figure 1B, genes B–E; Table 2). Only one of these sequences, gene C, contains a significant ORF. Gene C (SGP Chr7.1262; AK077305; BB617637; BB020862) is 1.9 kb in length and spans four exons. It was isolated from an adult pituitary gland, and the predicted protein is 133 amino acids in length.

While examining hemizygous animals for phenotype characterization, we encountered two animals with a BALB/cRl genotype (i.e., m1/deletion) at D7Mit352, which is located 1 Mb upstream of the Odz4 locus. Since these animals were viable, they must have contained a recombination event between D7Mit352 and the m1 mutation. Further analysis at several polymorphic markers located across the Odz4 gene indicated that the recombination occurred proximal to Odz4 in one of the progeny, but distal to Odz4 exon 9 in the other animal (data not shown). These data demonstrate that the m1 mutation occurs distal to exon 9 and eliminates all of the upstream genes as candidates for the l7Rn3 allelic series.

Northern analysis of Odz4:

Northern analyses using two probes from different regions of the gene detect multiple patterns of Odz4 expression that vary depending on tissue type and stage of development. Embryonic expression using a probe from the 5' end of Odz4 demonstrates that a 4.5-kb message is detected by E7.5 (Figure 4A, probe 3.11). As embryogenesis progresses, a 12.0-kb transcript is seen by E11.5, and all of the three major Odz4 isoforms are active between E15.5 and E17.5. However, in most adult tissues (testes, liver, lung, and heart) this probe detects only the 4.5- and 2.0-kb transcripts, indicating that this 12-kb message is primarily embryonic. The liver and brain have unique expression profiles in the adult animal. A 2.0-kb transcript is exceptionally abundant in the liver, while all three major transcripts are readily detected in the brain. In addition, alternative splicing is apparent in these tissues, as the kidney and skeletal muscle show unique Odz4 transcripts. This probe contains exons 6–11 and includes 5' UTR, as well as part of the Teneurin N intracellular domain.

A probe from the middle portion of the gene (probe 6.2.5) shows a much different pattern of expression. This probe contains exons 13–21 and sequence for part of the Teneurin N intracellular domain and the tenascin-like EGF-like repeats. Although a 12.0-kb transcript is detected with this probe in the same developmental timescale as the 3.11 probe, the 4.5-kb message is not observed (Figure 4B). However, the 6.2.5 probe also hybridizes to a 7.0-kb transcript (E17.5) and a 10.0-kb Odz4 mRNA later in development (E15.5 and E17.5). In the adult, this probe primarily detects brain-specific transcripts of 12.0, 10.0, and 4.5 kb. Longer exposures indicate weak expression of the 12.0- and 10.0-kb messages in testes, kidney, and heart (data not shown). Northern data using a probe from the 3' end of Odz4 show a similar expression pattern to probe 6.2.5 (data not shown). The differences in expression patterns between the 5' UTR/amino terminal and coding region probes indicate that Odz4 is subject to complex transcriptional regulation, generating multiple mRNA products that are active during specific embryonic and adult time points.

Dynamic expression of Odz4 during embryogenesis:

Since Northern analysis shows that Odz4 could have tissue-restricted patterns of expression, we examined its expression early in development using whole-mount in situ hybridization of normal, as well as m1 mutant, embryos. We found that Odz4 is ubiquitously expressed in the epiblast and extraembryonic regions as early as E6.5 (Figure 5A). However, this expression is weak and was observed only after an overnight incubation. By E7.5, Odz4 is highly expressed in the mesoderm of the developing embryo and extraembryonic tissues (Figure 5, B and C). At E8.5, Odz4 is expressed mainly in the neural ectoderm (Figure 5E), while expression in older embryos (E 11.5) is confined to the tail bud and limbs (Figure 5F). Analysis of m1 mutant embryos at E8.5 (Figure 5D) demonstrates that expression of Odz4 is markedly reduced compared to age-matched (Figure 5B) or E6.5 (Figure 5A) embryos.

View larger version (109K): In this window In a new window Download PPT slide   FIGURE 5.— Whole-mount in situ hybridization of Odz4. We used a probe corresponding to nucleotides 148–946 of Odz4 for these experiments (OOHASHI et al. 1999). This region overlaps the Teneurin intracellular protein domains. These experiments were performed in wild-type embryos (A, B, C, E, and F) and m1 mutant (D) embryos. (A) Gastrulation stage. Odz4 is expressed in a gradient in the epiblast of E6.5 embryos. (B) By E7.5, expression is confined to the mesodermal layer of the embryonic and extraembryonic tissues. (C) Magnification of the transition between the embryonic and extraembryonic regions. Odz4 expression is restricted to the mesoderm in embryonic and extraembryonic lineages. (D) Expression of Odz4 in m1 mutants. At E8.5, Odz4 demonstrates weak expression in the epiblast of m1 mutants. When compared to E6.5 embryos (A), it is clear that the m1 mutants have markedly reduced expression of Odz4. (E) Odz4 expression at E8.5. Odz4 expression is abundantly expressed in the developing head folds. In addition, there is weak expression in the posterior somites and presomites. (F) Expression of Odz4 in midgestation. By E11.5, Odz4 is expressed in rapidly dividing tissues, such as the tail bud and limb. The embryonic (Em), extraembronic (Ex), mesoderm (M), allantois (Al), chorion (Ch), head folds (Hf), somites (So), tail-bud (Tb), and limb-bud (Lb) regions are indicated.

 

Phenotypes of the l7Rn3 mutant alleles:

We established the time of death and the appearance of developmental landmarks that are associated with each of the l7Rn3 mutants to determine if the phenotypes correlated with the Odz4 expression patterns. The most severe allele, m1, arrests at E6.5 with a phenotype reminiscent of a pregastrulation stage embryo (Figure 6, A and B; Table 2). Mutants are resorbed by E9.5, do not produce embryonic or extraembryonic mesoderm, arrest before primitive streak formation, and lack a morphological A-P axis (H. NAKAMURA and M. J. JUSTICE, unpublished results). Animals homozygous for the m2 allele have a phenotype that is similar to the homozygous m1 mutants (data not shown). Phenotypic and mutational analysis of the m2 allele was not pursued.

View larger version (46K): In this window In a new window Download PPT slide   FIGURE 6.— Phenotypes of the l7Rn3 mutants. (A) Wild-type animal at E6.5. (B) m1/m1 mutants are delayed compared to normal littermates and have not started to undergo gastrulation by E6.5. (C) Normal E9.5 embryo. (D) m3/m3 embryo at E9.5. Compared to normal control, the m3 homozygotes demonstrate developmental delay with incomplete fusion of the neural tubes. The neural tubes are also misformed. (E) Compared to control embryos, m4/m4 homozygous mutants are severely delayed at E9.5. They also show incomplete somitogenesis and are not able to undergo cardiogenesis. (F) m5/m5 mutants are also delayed at E9.5 compared to control animals. They also have defects in somite formation and heart development that are less severe than those in the m4/m4 animals. (G) Control animal at E13.5. (H) m6 lethal animals (l7Rn3m6/Tyrc-26DVT) die during midgestation with a variety of phenotypes, including defects in turning, neural tube abnormalities, vascularization problems, and limb-bud malformations. (J) The reciprocal cross (Tyrc-26DVT/l7Rn3m6) produces viable animals that have subtle skeletal abnormalities. (K) Skeletons from control animals. (L) The T6 and T7 vertebrae are fused in Tyrc-26DVT/l7Rn3m6 animals. A summary of the phenotypes depicted is shown (bottom).

 
The m3/m3 homozygotes, which demonstrate developmental delay by E7.5, undergo gastrulation, generating mesoderm in embryonic and extraembryonic regions to establish an A-P axis (Figure 6, C and D). These mutants die later than the other alleles (between E9.5 and E11.5) with open head folds, kinky open neural tubes, and incomplete somitogenesis (Table 3). The m4/m4 mutants demonstrate a gradual delay in embryogenesis that is first evident at E5.0. By E8.5, m4/m4 homozygotes lag behind by 1 full day compared to normal littermates. Although m4/m4 homozygotes initiate head-fold elevation, the embryos fail to elongate and appear to have general mesodermal defects that cause uncoordinated somite segregation (Figure 6, C and E; Table 3). Although a rudimentary allantois forms in the m4 mutants, it makes only a partial fusion with the chorionic plate, which most likely leads to embryonic failure between E9.5 and E11.5.

View this table: In this window In a new window   TABLE 3 Time of death of the Odz4 mutants

 
Animals homozygous for the m5 allele show a delay in embryonic development that is similar to the m4 mutants. However, they progress slightly further in development than the m4 allele and are able to initiate heart formation. Like the m4 mutants, the allantois in the m5/m5 homozygotes fails to grow and does not fuse with the chorion, resulting in death between E9.5 and E11.5 (Table 3; Figure 6, C and F).

The m6 allele demonstrates two distinct phenotypes, which range from embryonic lethal to fully viable, depending upon which parent transmits the mutated chromosome (Figure 6, G and H). Two conditions must be met for the severe phenotype to be observed. First, the m6 mutation must be transmitted from the mother. Second, the paternally inherited allele must be deleted for the Odz4 critical region (i.e., these animals must have a m6/Tyr^c-26DVT genotype). These embryos die between E8.5 and E17.5 and have a broad range of phenotypes, which include developmental delay and kinky neural tubes (Figure 6, G and H; Table 3). In addition, some show clear abnormalities in vascularization and limb bud elongation. When the m6 mutation is transmitted from the father, a much different phenotype is observed. These animals survive to adulthood and many appear normal, although 30% are runted and scruffy at weaning. In addition, these mutants have skeletal abnormalities, which are discussed below.

Together, these data would suggest that m1 represents a loss-of-function allele and that the m3, m4, m5, and m6 alleles are hypomorphic mutations. To test this possibility, we examined the embryonic phenotype of compound heterozygotes (Table 4). Previous data showed that each of the m2–m6 alleles fail to complement m1 (RINCHIK and CARPENTER 1999). With the exception of the m1/m3 animals, all heterozygotes compound with m1 demonstrate the phenotype of the less severe allele, suggesting that these alleles are hypomorphic. When compound heterozygotes are formed by mating either m1/+ or m4/+ animals with m3/+ heterozygotes, the double mutants show a phenotype that is reminiscent of the more severe m1 or m4 phenotype. However, in matings with m5/+ heterozygotes, the m3/m5 compound heterozygotes are viable.

View this table: In this window In a new window   TABLE 4 Complementation analysis of allelic embryonic phenotypes

 
We examined the viable mutants (m3/m5, Tyr^c-26DVT/m6, and m6/m6) and found that subtle skeletal abnormalities occur in both the m3/m5 compound heterozygotes and the m6/m6 or deletion/m6 viable animals (Figure 6, J and K). An aberrant spinous process of the C2 vertebrae occurs in 60% of the compound m3/m5 heterozygotes, while 40% of m6/m6 homozygous and 18% of Tyr^c-26DVT/m6 hemizygous mutants display this phenotype. Twenty percent of the m6/m6 homozygous mutants and 27% of the Tyr^c-26DVT/m6 hemizygous mutants also display fused T6/T7 thoracic or L3/L4 lumbar vertebrae (Figure 6). These fusions are not detected in the m3/m5 compound heterozygotes.

Mutation analysis:

To identify the ENU-induced lesions, we sequenced RT-PCR products from homozygous mutants. Overlapping amplicons from the Odz4 coding region were sequenced directly. We identified a G-to-C transversion that causes an Ala 2642 to Pro missense mutation in the m4 allele (Figure 7). This mutation occurs in the globular region of the C terminus, which is predicted to be involved in ligand binding (OOHASHI et al. 1999). This amino acid appears to be important evolutionarily, as it is conserved among all four members of the mouse ODZ proteins, as well as in the human ODZ1, Drosophila TENA, rat NEURESTIN, chicken TENEURIN 1, and zebrafish TENM4 proteins. We sequenced genomic DNA from heterozygous animals in an attempt to identify the remaining mutations. However, we did not find mutations in the coding sequence of m1, m3, m5, or m6. It is possible that these mutations lie in unidentified exons or regulatory elements or that the sequencing of heterozygous DNA did not reveal these lesions.

View larger version (35K): In this window In a new window Download PPT slide   FIGURE 7.— Mutation analysis identified a mutation in the m4 allele. Sequence analysis of RT-PCR products identified an Ala 2642 Pro mutation in the m4 allele. This mutation resides in the globular domain in the C terminus in an amino acid that is conserved among all mouse homologs and in Drosophila Tena, but not Tenm.

 

We report here that mutations in Odz4 are responsible for the l7Rn3 allele series. This conclusion is based on the following criteria: (1) Odz4 maps to the l7Rn3 critical region; (2) the large size and complex transcriptional regulation of Odz4 explains the phenotypic diversity observed within the l7Rn3 allele series; (3) Odz4 is expressed before or during the developmental time points at which the phenotypic anomalies are manifested; and (4) we find mutations within and/or defects in expression of Odz4 in two of the l7Rn3 alleles.

While generating the l7Rn3 mutants, Rinchik found 31 alleles that formed 10 complementation groups along the Tyrosinase deletion complex (RINCHIK and CARPENTER 1999). Seven mutants clustered within the 7.1-kb Myosin heavy chain VIIa (Myo7a) gene (GIBSON et al. 1995). In contrast, only three alleles were isolated from the 2.0-kb Embryonic ectoderm development (Eed) gene (RINCHIK and CARPENTER 1999). Since the l7Rn3 complementation group contained six mutations, we proposed that the causative gene would produce a relatively large mRNA. After narrowing the critical region for l7Rn3 to an 700-kb domain, it was evident that that Odz4 was the best candidate gene. The 9.7-kb Odz4 transcript spanned the entire domain and was the only large protein-coding gene found in the l7Rn3 critical region.

Our current studies demonstrate that the previously identified Odz4 exons (WANG et al. 1998; OOHASHI et al. 1999) have a broad expression pattern and are detected in most adult and embryonic tissues. However, some of the newly discovered Odz4 exons are expressed in complicated developmental and tissue-specific patterns. The most striking example is exon 2, which is part of at least three distinct Odz4 transcripts. Initial evidence revealed that exon 2 was not coamplified with exon 1, indicating that these two 5' UTR exons are alternative transcriptional start sites for Odz4. Further investigation demonstrated that exon 2 is ubiquitously expressed in most transcripts. However, in two alternatively spliced messages exon 2 is detected only in adult brain or in ovary and E11.5 embryos. Similar expression profiles are seen with the other newly identified Odz4 exons. Taken as a whole, these data support our hypothesis that Odz4 uses multiple enhancer elements and alternative start sites to direct its complex tissue-specific and developmental expression profiles.

What is the purpose of all of these different transcripts? One possibility is that they increase the protein diversity of ODZ4. Alternative splicing, along with the use of multiple polyadenylation sites, is one method that mammalian genes can use to maximize the protein function and tissue distribution (BOUE et al. 2003). In addition, different tissue types could easily express specific transcripts by using alternative splicing of various cassettes of exons that would be best suited to that tissue (LEE and IRIZARRY 2003). This is an intriguing possibility, as exons that encode the tenascin-like EGF-like repeats have conserved intron/exon structure among several of the different Odz orthologs (MINET and CHIQUET-EHRISMANN 2000). Furthermore, some of the Odz4 mRNAs lack the characteristic EGF-like repeats, indicating that they can be spliced in cassette form.

Hundreds of pockets of high homology between mouse and human lie along the entire Odz4 genomic locus (KAROLCHIK et al. 2003). These regions lie in discrete portions of sequence and most are located within introns. It is probable that these short domains of high human/mouse homology harbor regulatory elements that are crucial for establishing the correct temporal-spatial expression pattern of the many Odz4 transcripts (HARDISON 2003). ODZ4 also has several RHS/YD repeats located in the C-terminal region of the protein. Although these elements are prevalent in prokaryotes, in eukaryotes, they are found only in ODZ family members (MINET and CHIQUET-EHRISMANN 2000). In prokaryotes, there are three RHS subfamilies, and new members are generated by intermolecular recombination between different family members (ZHAO and HILL 1995). It is possible that ODZ protein diversity is increased by RNA-RNA trans-splicing events among the four ODZ homologs. Trans-splicing, which occurs when two different mRNAs recombine to generate a chimeric cDNA, has been recently detected in mammals (reviewed in FEDOROVA and FEDOROV 2003). This is an interesting idea, given that two chimeric cDNAs that include Odz4 have been identified in mouse and human (LIU et al. 1999; ADELAIDE et al. 2000). It is clear from our expression studies that many different ODZ4 protein isoforms can be generated from alternative splicing of Odz4, and it is likely that this is a major mechanism for maximizing the diversity of partners with which ODZ4 can interact during development.

Developmental expression of Odz4 occurs in a spatio-temporal pattern that mimics the defects seen in the l7Rn3 complementation group. Odz4 is ubiquitously expressed at E6.5, but is restricted to the mesoderm by E7.5. At E6.5, the m1 mutants are smaller than their wild-type littermates and have not started to gastrulate. By E7.5, the embryos are markedly delayed, have not begun to gastrulate, and are unable to develop beyond this point. Each of the other mutant alleles also demonstrates developmental delay, indicating that gastrulation may be impaired in all members of the Odz4 allelic series. One reason for their developmental delay could be that they have mutations that affect cell proliferation and/or cell death, which could alter gastrulation. In this scenario the null mutants (i.e., m1) would be unable to progress through gastrulation because they completely lack the ODZ4 isoform(s) that is crucial for mesoderm induction. We predict that the hypomorphic alleles would have partially functional ODZ4 protein isoforms or low levels of ODZ4 protein that could overcome this block at gastrulation. However, these hypomorphic mutants would subsequently fail when other underlying biological processes required different ODZ4 protein products. Since ODZ4 forms homo- and heterodimers with other family members (FENG et al. 2002), it is possible that some mutations are in protein domains that are necessary for dimerization and/or ligand binding. Alternatively, the developmental delay could perturb crucial timed events that may eventually lead to embryonic death.

Each of the Odz4 alleles fails to complement the m1 mutation, as well as the Tyr^1FDFoHrc deletion (RINCHIK and CARPENTER 1999). Crosswise pair matings of the m1, m4, and m5 alleles result in the compound heterozygotes having the phenotype of the less severe allele, as expected for hypomorphic mutations. However, mutant phenotypes from complementation analyses using the m3 allele indicate that this allele behaves differently. In matings with m1 and m4, the compound heterozygous mutants have the phenotype of the more severe (m1 or m4), not the less severe, m3 allele, indicating that this mutation is not a protein hypomorph. Notably, m3 homozygotes often survive to E11.5, showing neural ectoderm abnormalities that are distinct from the mesodermal anomalies of the other alleles. Importantly, m3 hemizygotes (m3/deletion) have a phenotype that is similar to the m3 homozygotes (data not shown). However, when the m3 heterozygotes are mated with m5 heterozygotes, the resulting compound heterozygous progeny are viable with a phenotype that is similar to the paternal m6 defects.

Although intragenic complementation is rarely reported in mammals, the observation of a phenotype, albeit less severe, is expected. It is possible that the m3 lesion affects a tissue-specific transcript that is also disrupted in the m1 and m4 alleles, but not in the m5 allele. In this scenario, one would assume that the transcriptional levels of Odz4 are crucial for normal development. Alternatively, since ODZ4 is predicted to produce secreted peptides, and it is probable that the protein is cleaved into multiple protein fragments (DGANY and WIDES 2002). It is interesting to speculate that the individual mutations may affect more than one protein product. This may account for some or all of the phenotypic variation that we see among the different mutants and could explain why we see intragenic complementation between the m3 and m5 alleles. Another explanation could stem from the necessity of ODZ4 to dimerize with itself and other family members (FENG et al. 2002). Although none of the alleles has an obvious heterozygous phenotype, it is possible that the lesion in the m3 mutation could have deleterious effects in combination with the m1 and m4 alleles. The behavior of the m3 allele indicates that protein interactions are the most likely cause of the interesting genetic interactions between m3 and the other alleles.

The non-Mendelian pattern of inheritance of the m6 allele is unique. The lethal phenotype of the maternally inherited m6 defect initially manifests as developmental delay, but then proceeds to specific abnormalities in older embryos, including kinky neural tubes, incomplete turning, limb abnormalities, and vasculature defects. Since these phenotypes are observed anytime from E8.5 onward, the expression of Odz4 in the mesoderm (E7.5), neural ectoderm (E8.5), tail bud (E11.5), and limbs (E11.5) is very consistent with these abnormalities. The viable phenotype (paternal m6 defect) is also consistent with defects in Odz4. These animals are often runted at birth and weaning and can have skeletal abnormalities. Since the skeleton is derived from the somites, a subtle defect in somite differentiation could easily give rise to vertebrate fusions and abnormalities.

What could cause this differing pattern of inheritance? The first possibility is that genomic imprinting could account for the differential phenotype. This is an attractive hypothesis, as only maternal inheritance of the m6 mutation results in a lethal phenotype. However, the genetics involved in the m6 allele are more complicated, as this phenotype manifests only in a hemizygous form (i.e., m6/deletion); the m6/m6 homozygous animals are viable and have the same phenotype as the paternally inherited m6 hemizygotes. Another strong possibility is that Odz4 is a mammalian maternal effect gene. Although common in Drosophila, there are few examples of maternally expressed oocyte-specific genes in mammals (TONG et al. 2004). If the m6 allele lies in a transcript that is laid down in the oocyte during oogenesis, it is possible that the presence of a nonfunctional maternal m6 allele coupled with one copy of the mutated zygotic allele could explain the phenotypic differences observed in the m6 mutants.

The finding of a nonconservative amino acid substitution in the m4 allele of l7Rn3 indicates that Odz4 is a very strong candidate for the allele series. We predict that this Ala 2642 Pro substitution, which stems from a G-to-C transversion in the globular domain, would disrupt important protein-protein interactions. We hypothesize that disruption of cellular contacts would cause a delay in mesoderm induction as well as failure of the allantois to migrate and fuse with the chorion. Both of these phenotypes are seen in the Odz4 mutants. However, we have been unable to find another mutation within the coding region in the other mutants by sequencing RT-PCR products or heterozygous genomic DNA.

To find a mutation by sequencing RT-PCR products, it is crucial to analyze the correct tissue, and this method often fails to identify all mutations (WALKOWICZ et al. 1999). In lethal alleles, we can sequence only template made from RNA derived from whole embryos. If the lesion lies in a rare message, the other normal transcripts could easily mask the mutation. Although not well publicized, it has been shown that detection of a lesion in heterozygous DNA is not always possible (PAPATHANASIOU et al. 2003). In this case, a dominant mutation was detected by sequencing homozygous RT-PCR products, but could not be identified by sequencing heterozygous DNA. In heterozygous DNA, it is possible that the wild-type allele amplifies more efficiently, that the incorporation of the BigDye-labeled nucleotide is biased, or that a heterozygous lesion is missed in reading sequence traces. A final possibility is that the mutations in the other alleles are located in unidentified tissue-specific exons or within the noncoding regions that are crucial for directing tissue-specific or temporal expression of Odz4. These mutations could reside outside of the critical region within other, as-yet-unidentified Odz4 transcripts. Since our crossover data have eliminated genes that lie proximal to the translation start site, the possibility that these mutations would be in other genes is unlikely. Although many lesions for ENU-induced allelic series remain elusive, none have been reported in defined regulatory elements to date (GUENET 2004).

Here, we have shown that Odz4 is involved in determining the anterior-posterior axis in mice. The Drosophila Odz (Ten^m) ortholog displays a pair-rule phenotype, with expression occurring in alternating parasegments. TENM is a transmembrane domain protein that is cleaved into multiple membrane-bound and soluble protein isoforms (DGANY and WIDES 2002), each of which may play an important role during development. The other fly ortholog, Ten^a, is expressed in the CNS, brain, and hindgut during development (FASCETTI and BAUMGARTNER 2002). Phylogenetic analysis indicates that either of these two genes could be the ortholog of Odz4 (MINET and CHIQUET-EHRISMANN 2000). In the mouse, Odz4 has three additional homologs, which are all members of the TENEURIN protein family. TENEURINS, cell-surface signaling molecules that are highly expressed in the developing CNS, may play a role in limb development, somite formation, and the patterning of neural connections (RUBIN et al. 1999; TUCKER et al. 2001). Although no proteins outside of ODZ family members share significant homology distal to the transmembrane region, the EGF repeats and transmembrane domain located within the N terminus share significant homology with several members of the NOTCH pathway, including NOTCH, SERATE, CRUMBS, and DELTA (BEN-ZUR et al. 2000; DGANY and WIDES 2002). NOTCH signaling is necessary for development of the pancreas (an endoderm-derived organ) and neuronal lineages (ectoderm-derived tissues), as well as hematopoietic cells, the skeleton, and vascula-ture (mesoderm-derived tissues; reviewed in HARPER et al. 2003). Further, NOTCH signaling plays a major role in somite segmentation (reviewed in RIDA et al. 2004), a process that is clearly disrupted in our mutants. Therefore, it is tempting to speculate that we have discov-ered a new member of the extended NOTCH signaling cascade.

The authors thank Temenit Asgedom for excellent technical support and Alec Wang for kindly supplying the 3.2.5 and 6.11 Odz4 cDNAs. We also thank Sally Camper, Richard Behringer, Tom Meehan, and Kathy Hentges for critically evaluating this manuscript. This work was supported by National Institutes of Health (NIH) fellowship 1F32HD44407 to A.C.L., a Kleberg Foundation grant to M.J.J., and an NIH grant U01HD39372 to M.J.J.

¹ These authors contributed equally to this work.

² Present address: Department of Biochemistry, Cellular and Molecular Biology, University of Tennessee, Knoxville, TN 37996.

ADELAIDE, J., M. CHAFFANET, M. J. MOZZICONACCI, C. POPOVICI, N. CONTE et al., 2000 Translocation and coamplification of loci from chromosome arms 8p and 11q in the MDA-MB-175 mammary carcinoma cell line. Int. J. Oncol. 16: 683–688.[Medline]

ALESSANDRINI, F., T. JAKOB, A. WOLF, E. WOLF, R. BALLING et al., 2001 Enu mouse mutagenesis: generation of mouse mutants with aberrant plasma IgE levels. Int. Arch. Allergy Immunol. 124: 25–28.[CrossRef][Medline]

BAUMGARTNER, S., and R. CHIQUET-EHRISMANN, 1993 Tena, a Drosophila gene related to tenascin, shows selective transcript localization. Mech. Dev. 40: 165–176.[CrossRef][Medline]

BAUMGARTNER, S., D. MARTIN, C. HAGIOS and R. CHIQUET-EHRISMANN, 1994 Tenm, a Drosophila gene related to tenascin, is a new pair-rule gene. EMBO J. 13: 3728–3740.[Medline]

BEN-ZUR, T., E. FEIGE, B. MOTRO and R. WIDES, 2000 The mammalian Odz gene family: homologs of a Drosophila pair-rule gene with expression implying distinct yet overlapping developmental roles. Dev. Biol. 217: 107–120.[CrossRef][Medline]

BOUE, S., I. LETUNIC and P. BORK, 2003 Alternative splicing and evolution. BioEssays 25: 1031–1034.[CrossRef][Medline]

CHURCH, G., and W. GILBERT, 1984 Genomic sequencing. Proc. Natl. Acad. Sci. USA 81: 1991–1995.[Abstract/Free Full Text]

DGANY, O., and R. WIDES, 2002 The Drosophila odz/ten-m gene encodes a type I, multiply cleaved heterodimeric transmembrane protein. Biochem. J. 363: 633–643.[CrossRef][Medline]

ECHELARD, Y., D. J. EPSTEIN, B. ST-JACQUES, L. SHEN, J. MOHLER et al., 1993 Sonic hedgehog, a member of a family of putative signaling molecules, is implicated in the regulation of CNS polarity. Cell 75: 1417–1430.[CrossRef][Medline]

FASCETTI, N., and S. BAUMGARTNER, 2002 Expression of Drosophila Ten-a, a dimeric receptor during embryonic development. Mech. Dev. 114: 197–200.[CrossRef][Medline]

FEDOROVA, L., and A. FEDOROV, 2003 Introns in gene evolution. Genetica 118: 123–131.[CrossRef][Medline]

FENG, K., X. H. ZHOU, T. OOHASHI, M. MORGELIN, A. LUSTIG et al., 2002 All four members of the Ten-m/Odz family of transmembrane proteins form dimers. J. Biol. Chem. 277: 26128–26135.[Abstract/Free Full Text]

GIBSON, F., J. WALSH, P. MBURU, A. VARELA, K. A. BROWN et al., 1995 A type VII myosin encoded by the mouse deafness gene shaker-1. Nature 374: 62–64.[CrossRef][Medline]

GRAW, J., J. LOSTER, D. SOEWARTO, H. FUCHS, A. REIS et al., 2002 V76D mutation in a conserved gD-crystallin region leads to dominant cataracts in mice. Mamm. Genome 13: 452–455.[CrossRef][Medline]

GUENET, J.-L., 2004 Chemical mutagenesis of the mouse genome: an overview, pp. 9–24 in Genetica: Mutagenesis of the Mouse Genome, edited by M. J. JUSTICE and M. A. BEDELL. Kluwer Academic, Dordrecht, The Netherlands.

HARDISON, R. C., 2003 Comparative genomics. PLoS Biol. 1: E58.[Medline]

HARDISTY, R. E., P. MBURU and S. D. BROWN, 1999 ENU mutagenesis and the search for deafness genes. Br. J. Audiol. 33: 279–283.[Medline]

HARPER, J. A., J. S. YUAN, J. B. TAN, I. VISAN and C. J. GUIDOS, 2003 Notch signaling in development and disease. Clin. Genet. 64: 461–472.[CrossRef][Medline]

HRABE DE ANGELIS, M., and H. FLASWINKEL, 2000 Genome wide large scale production of mutant mice by ENU mutagenesis. Nat. Genet. 25: 444–447.[CrossRef][Medline]

ISAACS, A. M., K. E. DAVIES, A. J. HUNTER, P. M. NOLAN, L. VIZOR et al., 2000 Identification of two new Pmp22 mouse mutants using large-scale mutagenesis and a novel rapid mapping strategy. Hum. Mol. Genet. 9: 1865–1871.[Abstract/Free Full Text]

KAROLCHIK, D., R. BAERTSCH, M. DIEKHANS, T. S. FUREY, A. HINRICHS et al., 2003 The UCSC Genome Browser Database. Nucleic Acids Res. 31: 51–54.[Abstract/Free Full Text]

KERLAVAGE, A., V. BONAZZI, M. DI TOMMASO, C. LAWRENCE, P. LI et al., 2002 The Celera Discovery System. Nucleic Acids Res. 30: 129–136.[Abstract/Free Full Text]

KILE, B. T., K. E. HENTGES, A. T. CLARK, H. NAKAMURA, A. P. SALINGER et al., 2003 Functional genetic analysis of mouse chromosome 11. Nature 425: 81–86.[CrossRef][Medline]

LEE, C. J., and K. IRIZARRY, 2003 Alternative splicing in the nervous system: an emerging source of diversity and regulation. Biol. Psychiatry 54: 771–776.[CrossRef][Medline]

LEVINE, A., A. BASHAN-AHREND, O. BUDAI-HADRIAN, D. GARTENBERG, S. MENASHEROW et al., 1994 Odd Oz: a novel Drosophila pair rule gene. Cell 77: 587–598.[CrossRef][Medline]

LIU, X., E. BAKER, H. J. EYRE, G. R. SUTHERLAND and M. ZHOU, 1999 Gamma-heregulin: a fusion gene of DOC-4 and neuregulin-1 derived from a chromosome translocation. Oncogene 18: 7110–7114.[CrossRef][Medline]

LOSSIE, A. C., M. S. BUCKWALTER and S. A. CAMPER, 1993 Lysyl oxidase (Lox) maps between Grl-1 and Adrb-2 on mouse chromosome 18. Mamm. Genome 4: 177–178.[CrossRef][Medline]

MINET, A. D., and R. CHIQUET-EHRISMANN, 2000 Phylogenetic analysis of teneurin genes and comparison to the rearrangement hot spot elements of E. coli. Gene 257: 87–97.[CrossRef][Medline]

NAGY, A., M. GERTENSTEIN, K. VINTERSTEN and R. R. BEHRINGER, 2003 Manipulating the Mouse Embryo: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

NOLAN, P. M., D. KAPFHAMER and M. BUCAN, 1997 Random mutagenesis screen for dominant behavioral mutations in mice. Methods 13: 379–395.[CrossRef][Medline]

NOLAN, P. M., J. PETERS, M. STRIVENS, D. ROGERS, J. HAGAN et al., 2000 A systematic, genome-wide, phenotype-driven mutagenesis programme for gene function studies in the mouse. Nat. Genet. 25: 440–443.[CrossRef][Medline]

OOHASHI, T., X. H. ZHOU, K. FENG, B. RICHTER, M. MORGELIN et al., 1999 Mouse ten-m/Odz is a new family of dimeric type II transmembrane proteins expressed in many tissues. J. Cell Biol. 145: 563–577.[Abstract/Free Full Text]

PAPATHANASIOU, P., A. C. PERKINS, B. S. COBB, R. FERRINI, R. SRIDHARAN et al., 2003 Widespread failure of hematolymphoid differentiation caused by a recessive niche-filling allele of the Ikaros transcription factor. Immunity 19: 131–144.[CrossRef][Medline]

RIDA, P. C., N. LE MINH and Y. J. JIANG, 2004 A Notch feeling of somite segmentation and beyond. Dev. Biol. 265: 2–22.[CrossRef][Medline]

RINCHIK, E. M., and D. A. CARPENTER, 1993 N-ethyl-N-nitrosourea-induced prenatally lethal mutations define at least two complementation groups within the embryonic ectoderm development (eed) locus in mouse chromosome 7. Mamm. Genome 4: 349–353.[CrossRef][Medline]

RINCHIK, E. M., and D. A. CARPENTER, 1999 N-ethyl-N-nitrosourea mutagenesis of a 6- to 11-cM subregion of the Fah-Hbb interval of mouse chromosome 7: completed testing of 4557 gametes and deletion mapping and complementation analysis of 31 mutations. Genetics 152: 373–383.[Abstract/Free Full Text]

RINCHIK, E. M., D. A. CARPENTER and P. B. SELBY, 1990 A strategy for fine-structure functional analysis of a 6- to 11-centimorgan region of mouse chromosome 7 by high-efficiency mutagenesis. Proc. Natl. Acad. Sci. USA 87: 896–900.[Abstract/Free Full Text]

RUBIN, B. P., R. P. TUCKER, D. MARTIN and R. CHIQUET-EHRISMANN, 1999 Teneurins: a novel family of neuronal cell surface proteins in vertebrates, homologous to the Drosophila pair-rule gene product Ten-m. Dev. Biol. 216: 195–209.[CrossRef][Medline]

RUSSELL, L. B., 1989 Functional and structural analyses of mouse genomic regions screened by the morphological specific-locus test. Mutat. Res. 212: 23–32.[CrossRef][Medline]

RUSSELL, W. L., and L. B. RUSSELL, 1959 The genetic and phenotypic characteristics of radiation-induced mutations in mice. Radiat. Res. 1(Suppl.): 296–305.[CrossRef]

SOEWARTO, D., C. FELLA, A. TEUBNER, B. RATHKOLB, W. PARGENT et al., 2000 The large-scale Munich ENU-mouse-mutagenesis screen. Mamm. Genome 11: 507–510.[CrossRef][Medline]

TONG, Z. B., L. GOLD, A. DE POL, K. VANEVSKI, H. DORWARD et al., 2004 Developmental expression and subcellular localization of mouse MATER, an oocyte-specific protein essential for early development. Endocrinology 145: 1427–1434.[Abstract/Free Full Text]

TUCKER, R. P., R. CHIQUET-EHRISMANN, M. P. CHEVRON, D. MARTIN, R. J. HALL et al., 2001 Teneurin-2 is expressed in tissues that regulate limb and somite pattern formation and is induced in vitro and in situ by FGF8. Dev. Dyn. 220: 27–39.[CrossRef][Medline]

VIVIAN, J. L., Y. CHEN, D. YEE, E. SCHNEIDER and T. MAGNUSON, 2002 An allelic series of mutations in Smad2 and Smad4 identified in a genotype-based screen of N-ethyl-N- nitrosourea-mutagenized mouse embryonic stem cells. Proc. Natl. Acad. Sci. USA 99: 15542–15547.[Abstract/Free Full Text]

WALKOWICZ, M., Y. JI, X. REN, B. HORSTHEMKE, L. B. RUSSELL et al., 1999 Molecular characterization of radiation- and chemically induced mutations associated with neuromuscular tremors, runting, juvenile lethality, and sperm defects in jdf2 mice. Mamm. Genome 10: 870–878.[CrossRef][Medline]

WANG, X. Z., M. KURODA, J. SOK, N. BATCHVAROVA, R. KIMMEL et al., 1998 Identification of novel stress-induced genes downstream of chop. EMBO J. 17: 3619–3630.[CrossRef][Medline]

WATERSTON, R. H., K. LINDBLAD-TOH, E. BIRNEY, J. ROGERS, J. F. ABRIL et al., 2002 Initial sequencing and comparative analysis of the mouse genome. Nature 420: 520–562.[CrossRef][Medline]

ZDOBNOV, E. M., and R. APWEILER, 2001 InterProScan—an integration platform for the signature-recognition methods in InterPro. Bioinformatics 17: 847–848.[Abstract/Free Full Text]

ZHAO, S., and C. W. HILL, 1995 Reshuffling of Rhs components to create a new element. J. Bacteriol. 177: 1393–1398.[Abstract/Free Full Text]

ZHOU, X. H., O. BRANDAU, K. FENG, T. OOHASHI, Y. NINOMIYA et al., 2003 The murine Ten-m/Odz genes show distinct but overlapping expression patterns during development and in adult brain. Gene Expr. Patterns 3: 397–405.[CrossRef][Medline]