Somites provide the basic body plan for metameric axial structures in vertebrates, and establish the segmental features through the sequential gene expression in the presomitic mesoderm (PSM). A crucial protein for segment border formation is the bHLH transcription factor Mesp2, the expression of which is restricted to the anterior PSM. A gene candidate that is activated by Mesp2 is Epha4, as its expression pattern resembles Mesp2and is absent in Mesp2-null embryos. We have analyzed the enhancer region of Epha4, which is responsible for its expression in the anterior PSM,and identified an E-box containing region. Subsequent transgenic and transient luciferase analyses successfully determined that the presence of repeated E-box sequences is a minimum essential requirement for the expression in the anterior PSM. We also show that Mesp2 directly binds to the enhancer sequence of Epha4. Furthermore, the forced expression of Mesp2 in somitic cells results in the activation of Epha4 and repression of the caudal gene Uncx4.1, which may trigger the events leading to the formation of abnormal somites and rostralized vertebra. In addition, ectopic Mesp2 expression induces abnormally epithelialized structures, which support to the idea that Mesp2 induces the formation of segmental borders by activating genes that play roles in cellular epithelialization.

INTRODUCTION

Somites are basic structures that underlie the segmental body architecture in vertebrates. The mechanisms involved in the generation of serially segmental units are a fascinating model system that has been used by many developmental biologists to further our understanding of the temporal and spatial control of gene expression. Somite precursors are derived as paraxial mesoderm from the primitive streak or tailbud region and these cells then come under the control of the segmentation clock, in which Notch signal oscillation generates the periodicity (for reviews, see Aulehla and Herrmann, 2004; Bessho and Kageyama, 2003; Maroto and Pourquié,2001; Pourquié,2003; Rida et al.,2004; Saga and Takeda,2001). Notch signaling is suppressed in the anterior PSM by a bHLH protein, Mesp2, and the anterior limits of the Mesp2 expression domain demarcate the next segmental border(Morimoto et al., 2005). Mesp2 is a key transcription factor for both segment border formation and for the generation of rostrocaudal patterning within somites(Saga et al., 1997; Takahashi et al., 2000). The expression of many genes is affected in the Mesp2-null embryo, in which the genes required for rostral property are suppressed but those required for the development of caudal properties are enhanced. Among these genes, only lunatic fringe (Lfng) has so far been shown to be a direct target of Mesp2(Morimoto et al., 2005).

Since the Mesp2 expression domain is very similar to that of Epha4, and this gene is also suppressed in the Mesp2-null embryo(Nomura-Kitabayashi et al.,2002), it was probable that Mesp2 directly activated Epha4 in the rostral compartment of the somites. Furthermore, Epha4 is implicated in segmental border formation in zebrafish(Cooke et al., 2005; Barrios et al., 2003; Durbin et al., 1998), although gene knockout studies indicate that Epha4 is not the sole protein required for segmental border formation in the mouse, as no somitic phenotype has been reported (Dottori et al., 1998; Kullander et al., 2001) (M. Asano, personal communication). The identification of target genes for a transcription factor is necessary to understand fully the genetic networks involved in a particular biological system. However, it is very difficult to achieve these using straightforward methods such as SELEX or immunoprecipitation, particularly in embryonic tissues. As an alternative method, we attempted to first identify the Epha4 enhancer and then test whether Mesp2 directly binds to this region; if it does not bind, we can search the binding protein that might be a direct target of Mesp2. Fortunately, the Epha4 enhancer elements identified showed direct binding to Mesp2, together with E47 (Tcfe2a-Mouse Genome Informatics) in vitro. Moreover, the forced expression of Mesp2 resulted in the reverse phenotype of the Mesp2-null embryo, whereby Epha4 is activated, Uncx4.1 is suppressed and ectopic epithelialization could be observed in the transgenic embryos.

MATERIALS AND METHODS

Construction of lacZ reporter constructs

An 8.8 kb fragment (NheI-XbaI) was isolated from an Epha4-containing bacterial artificial chromosome clone (415B3) and subcloned into the pBluescript vector (Stratagene). A series of deletion constructs were then generated using the appropriate restriction enzymes(Fig. 1A). These genomic fragments were inserted into a lacZ reporter vector, upstream of the hsp promoter (Kothary et al.,1989). E-box deletion and mutant constructs were subsequently generated via PCR using the 630 bp enhancer region (HindIII cut) as a template (Imai et al.,1991).

Formation of E-box multimer constructs

Synthetic oligonucleotides were designed to generate two repeats of 20 bp containing an E-box when annealed (Fig. 2C). These E-box-containing sequences were flanked by BglII and BamHI sites. The complementary oligonucleotides were annealed and phosphorylated with T4 polynucleotide kinase prior to ligation. Ligated DNA were digested with BamHI and BglII,and separated on 2% agarose gels. Multimerized bands were excised and subcloned into the pBluescript vector.

Generation of transgenic mice

All constructs were digested with restriction enzymes to remove vector sequences and then gel purified. Transgenic mice were generated by microinjection of fertilized eggs. Microinjected eggs were transferred into the oviducts of pseudopregnant foster females. The genotypes of the embryos were identified by PCR using DNA prepared from the yolk sac.

Fig. 1.

Identification of a somite-specific Epha4 enhancer.(A) The lacZ transgene constructs used to identify the cis-acting somite enhancer within the Epha4 genomic region. The numbers of transgenic mouse embryos that expressed β-gal in the somites,among the transgene-positive embryos, are indicated on the right. E, EcoRI; H, HindIII; N, NheI; Sa, SacI; Sw, SwaI; P, PmaCI; X, XbaI. (B,C)Lateral view of β-gal activity driven by the 630 bp (HindIII) Epha4 enhancer fragment in a 10.5 dpc embryo. A magnified image in the somitic region of B is shown in C. (D,E) Comparison by in situ hybridization at 10.5 dpc of the transgene expression (lacZ) in a transgenic embryo (D) with endogenous Epha4 expression in the wild-type embryo (E). In situ signals in the anterior PSM are indicated by arrows in the lower panels showing magnified images. (F,G)β-Gal activity driven by the 630 bp enhancer in 10.5 dpc wild-type (F)and Mesp2L/L embryos (G).

Fig. 1.

Identification of a somite-specific Epha4 enhancer.(A) The lacZ transgene constructs used to identify the cis-acting somite enhancer within the Epha4 genomic region. The numbers of transgenic mouse embryos that expressed β-gal in the somites,among the transgene-positive embryos, are indicated on the right. E, EcoRI; H, HindIII; N, NheI; Sa, SacI; Sw, SwaI; P, PmaCI; X, XbaI. (B,C)Lateral view of β-gal activity driven by the 630 bp (HindIII) Epha4 enhancer fragment in a 10.5 dpc embryo. A magnified image in the somitic region of B is shown in C. (D,E) Comparison by in situ hybridization at 10.5 dpc of the transgene expression (lacZ) in a transgenic embryo (D) with endogenous Epha4 expression in the wild-type embryo (E). In situ signals in the anterior PSM are indicated by arrows in the lower panels showing magnified images. (F,G)β-Gal activity driven by the 630 bp enhancer in 10.5 dpc wild-type (F)and Mesp2L/L embryos (G).

Luciferase assay

An Epha4 somite enhancer insert (630 bp HindIII-HindIII fragment) and E-box multimers were cloned into the pGL3-Promoter vector (Promega). NIH3T3 cells were grown at 80%confluency in 24-multiwell plates and transfected with luciferase gene constructs using Lipofectamine Plus (Invitrogen). Cells were harvested 36 hours after transfection and luciferase activities were measured using a Dual Luciferase Assay Kit (Promega). The transfection efficiency was normalized by co-transfection of the Renilla luciferase expression vector pRL-TK (Promega),and the relative luciferase activity was determined as recommended by the manufacturer.

Electrophoretic mobility shift assay (EMSA)

For protein preparation, NIH3T3 cells were grown at 80% confluency in 10 cm dishes and transfected with expression vectors containing either 3×FLAG-tagged Mesp2 or Myc-tagged E47. Nuclear extracts were prepared using Nuclear Extract Kit (Active Motif). The protein concentrations were measured by the Bradford assay (Pierce). EMSA was performed using a DIG gelshift and detection kit (Roche). Binding reactions were carried out by mixing DIG-labeled and unlabeled (for competition experiments) probes with nuclear extracts from NIH3T3 cells. In experiments using antibodies, the nuclear extracts were preincubated with the antibody for 1 hour on ice.

Generation of CAG-CAT-Mesp2 transgenic and Mox1-creknock-in mice

A targeting vector was designed to introduce the Cre gene near to the translational initiation site of the Mox1 gene (see Fig. S1 in the supplementary material) and used to establish the Mox1-cre mouse line, in which Cre recombinase is expressed instead of Mox1 and the gene activity is examined by crossing with a reporter line, R26R(Zambrowicz et al., 1997). To achieve the ectopic expression of Mesp2, a CAG-floxed-CAT-Mesp2transgene was constructed (Yamauchi et al., 1999), in which CAT gene can be excised by Cre recombinase and thus the Mesp2 gene comes under the control of the CAG promoter (see Fig. S1 in the supplementary material). Transgenic mouse lines were established by microinjection of CAG-floxed CAT-Mesp2 DNA as described above.

Analyses of embryos by LacZ staining, in situ hybridization, skeletal and the histological methods

Embryos were fixed and stained in X-gal solution for the detection ofβ-gal activity, as described previously(Saga et al., 1999). For histology analyses, samples stained by X-gal were postfixed with 4%paraformaldehyde, dehydrated in an ethanol series, embedded in paraffin and sectioned at 6 μm. Whole-mount in situ hybridization was performed using InsituPro robot (Intavis). The transcripts were visualized using anti-digoxigenin (DIG) antibodies conjugated to alkaline phosphatase. Color reactions were performed using BM Purple (Roche). Methods employed for section in situ hybridization and for the immunohistological detection of Mesp2 have been previously described (Morimoto et al., 2005). Skeletal preparations by Alcian Blue/Alizarin Red staining have also been described previously(Saga et al., 1997; Takahashi et al., 2000). Probes used for the in situ hybridization detection of Uncx4.1 and Sox9 were kindly provided by Dr Peter Gruss and Dr Veronique Lefebvre, respectively. For the detection of actin filaments, frozen sections were stained with AlexaFluor 488-conjugated phalloidin (Molecular Probes)according to the manufacturer's protocol.

RESULTS

Enhancer analysis of Epha4

Previous enhancer studies have indicated that a somite specific enhancer is not contained within the 7.5 kb region upstream of the Epha4transcriptional start site, in which only rhombomere-specific enhancer activity has been identified (Theil et al., 1998). To identify the somite specific enhancer region of the Epha4 gene, we focused on the more upstream region of the gene. We subsequently found the enhancer within an 8.8 kb fragment, beginning 8 kb upstream of the Epha4 transcriptional start site(Fig. 1A). β-Gal activity was observed in the rostral compartment of the segmented somites(Fig.1B,C) and to confirm that this reflects endogenous Epha4 expression, we compared the expression of lacZ RNA with the endogenous Epha4 transcripts. Among several expression domains of endogenous Epha4, such as limb buds,branchial arches and rhombomeres, an identical expression pattern was revealed in both the anterior PSM and the rostral compartment of the S1 somites(Fig. 1D,E). Further transgenic analyses were conducted using DNA fragments that had been generated by several restriction enzymes. Although we detected one embryo which showed somite-specific expression using Sw-X region, we concentrated our analyses on the P-Sw region, which showed most consistent results. Further deletion identified a HindIII fragment of 630 bp that could sustain endogenous pattern of somite-specific Epha4 expression(Fig. 1A). We established a permanent transgenic line using a lacZ reporter with the 630 bp enhancer. The somite-specific expression was observed during somitogenesis from 8.5 to 11.5 days post-coitum (dpc) as similar to the endogenous one (see Fig. S2 in the supplementary material). However, the transgene expression became weaker in the later stage embryo at 11.5 dpc and the somite-specific expression was not observed with both probes for endogenous Epha4 and lacZ transgene after 13.5 dpc (data not shown). When the expression was examined in the Mesp2-null genetic background(Mesp2L/L)(Takahashi et al., 2000), noβ-gal activity was detected in the Mesp2L/L embryos(Fig. 1F,G), which confirms that the enhancer contains cis elements required for the Epha4activation downstream of the Mesp2.

Fig. 2.

Identification of the functional E-box motifs responsible for the somite-specific activation of Epha4. (A) Sequence of the 630 bp Epha4 somite enhancer (core enhancer) region. The four indicated fragments (fragment 1-4) represent deleted sequences from the core enhancer region. The results of transgenic analyses using these deleted constructs are shown in parentheses. (B) Sequence alignment of somite enhancer regions (fragment 2 and fragment 3) of mouse Epha4 with the corresponding regions of human Epha4. The mutations introduced in each E-box and the results of the subsequent transgenic analyses are shown.(C) Schematic representation of artificially constructed enhancers,containing six E-box motif repeats. (D) The artificial enhancers were cloned upstream of a lacZ reporter vector and the results of the subsequent transgenic analyses (representative images are shown) are indicated in parentheses. For the [E2]6-lacZ artificial enhancer, no somite expression was observed among the 11 transgene-positive embryos. Blue letters indicate putative E-box motifs.

Fig. 2.

Identification of the functional E-box motifs responsible for the somite-specific activation of Epha4. (A) Sequence of the 630 bp Epha4 somite enhancer (core enhancer) region. The four indicated fragments (fragment 1-4) represent deleted sequences from the core enhancer region. The results of transgenic analyses using these deleted constructs are shown in parentheses. (B) Sequence alignment of somite enhancer regions (fragment 2 and fragment 3) of mouse Epha4 with the corresponding regions of human Epha4. The mutations introduced in each E-box and the results of the subsequent transgenic analyses are shown.(C) Schematic representation of artificially constructed enhancers,containing six E-box motif repeats. (D) The artificial enhancers were cloned upstream of a lacZ reporter vector and the results of the subsequent transgenic analyses (representative images are shown) are indicated in parentheses. For the [E2]6-lacZ artificial enhancer, no somite expression was observed among the 11 transgene-positive embryos. Blue letters indicate putative E-box motifs.

Subsequent sequence analysis revealed that this 630 bp region contained eight E-boxes. As Mesp2 belongs to the bHLH family of transcription factors,which are known to bind E-box or N-box motifs, we initially performed deletions of some of the Epha4 E-boxes. We generated four deletion constructs that lack a region (fragments 1-4) of the E-boxes(Fig. 2A) and examined the enhancer activities. Our results clearly showed that the enhancer activity was completely lost when using both fragment 2- and 3-deletion constructs,indicating that both regions are necessary for its expression(Fig. 2A). Both fragments 2 and 3 contain two E-boxes, designated E1-E4(Fig. 2B). The core sequences of E1, E3 and E4 are identical (CAAATG or CATTTG) but only E1 and E3 are conserved in the human EPHA4 gene. To determine whether these E-boxes are crucial, we next introduced mutations into their consensus sequences and transgenic analysis was conducted using the full 630 bp enhancer as a wild-type activity control. The E2 mutation did not affect enhancer activity but a considerable reduction in the activity was observed for the E1 and E3 mutations. In addition, mutation of E4 substantially disrupted enhancer activity, indicating that a mutation in a single E-box abrogates the enhancer activity and that the presence of tandem repeats of the E-box consensus sites is important for the activity. To confirm this, we generated reporter constructs with artificial enhancers, composed of six tandem repeats of the E1, E2, E3 or E4 boxes and flanking sequences(Fig. 2C). Transient transgenic analysis revealed that each of these regions, with the exception of E2, showed weak but specific expression in the somitic region(Fig. 2D). This confirmed that these E-boxes are a minimum requirement for the specificity of Epha4expression in the somitic region and that the consensus sequence is CAAATG.

Fig. 3.

Transactivation of the Epha4 enhancer by the Mesp2/E47 heterodimer.Epha4 somite enhancers (A, 630 bp core sequence; B,six tandem repeats of E-box sequences) were ligated to the pGL3 luciferase vector. Luciferase activity was measured at 36 hours after transfection into NIH3T3 cells. (A) The presence (+) or absence (-) of either Mesp2 or E47 are indicated in each column. (B) Luciferase activity was compared with (black bars) and without (white bars) Mesp2/E47. Mutations in E3(5′-CATTTG-3′), that give rise to E3m (5′-ACGGGT-3′),results in the loss of reporter activity. The results shown are the mean values from three independent experiments. Standard deviations are indicated by error bars.

Fig. 3.

Transactivation of the Epha4 enhancer by the Mesp2/E47 heterodimer.Epha4 somite enhancers (A, 630 bp core sequence; B,six tandem repeats of E-box sequences) were ligated to the pGL3 luciferase vector. Luciferase activity was measured at 36 hours after transfection into NIH3T3 cells. (A) The presence (+) or absence (-) of either Mesp2 or E47 are indicated in each column. (B) Luciferase activity was compared with (black bars) and without (white bars) Mesp2/E47. Mutations in E3(5′-CATTTG-3′), that give rise to E3m (5′-ACGGGT-3′),results in the loss of reporter activity. The results shown are the mean values from three independent experiments. Standard deviations are indicated by error bars.

The Epha4 enhancer is activated by Mesp2 in cultured cells

The identification of E-boxes as a vital component of the Epha4enhancer strongly indicates that this gene is directly regulated by Mesp2, as the bHLH-type transcription factor is known to bind E-boxes. In order to further elucidate whether Mesp2 can activate the Epha4 enhancer, we established a luciferase assay system using NIH3T3 cultured cells. The reporter gene was constructed by ligating the 630 bp Epha4 enhancer region to a luciferase gene. As B-type bHLH transcription factors are known to function as heterodimers with the so called A-type bHLH factors(Ledent and Vervoort, 2001),we analyzed reporter activity with or without E47, which is a typical A-type bHLH factor. As shown in Fig. 3A, Mesp2 and E47 alone exhibited weak transactivating activities when these expression vectors were separately transfected with the reporter construct, whereas strong activity was observed when Mesp2 and E47 were co-transfected. The transactivating activity of Mesp2 alone can be ascribed to its association with endogenous E47. Next, we constructed reporters with six repeats of each of the Epha4 enhancer E-boxes (E1 to E4) used in our transgenic analyses (Fig. 2C). Interestingly, very strong activity was observed when the E3 and E4 constructs were used, and this was only observed upon cotransfection with E47(Fig. 3B). This specificity was confirmed by the lack of activity resulting from a construct with a mutant-type E3 enhancer element. E2 showed no activity, which is consistent with the findings of our transgenic analysis. E1 did not have strong activity,although we obtained positive activity for this E-box via transgenic analysis.

Fig. 4.

The Mesp2/E47 heterodimer binds to the E3 site of the Epha4enhancer. (A) The results of EMSA using nuclear extracts from NIH3T3 cells transfected with FLAG-Mesp2 and/or Myc-E47 and incubated with E3 probe. The quantities of nuclear extracts used (μg) are indicated.(B) A competition assay indicating the specificity of the binding of the Mesp2/E47 (2 μg each) complex to the E3 probe. The addition of 100-fold excess of unlabeled E3 probe, but not the E3m mutant probe, abolished the binding. (C) Evidence for the heterodimer formation of FLAG-Mesp2/Myc-E47. The band containing E3 (arrow) was supershifted by the addition of either anti-FLAG or anti-Myc antibodies. The oligonucleotides used were as follows: E3, 5′-TGGGTCACATTTGTCCAAAA-3′; E3m,5′-TGGGTCAACGGGTTCCAAAA-3′ (E-box is shown in the bold; altered nucleotides in the mutant probe are underlined).

Fig. 4.

The Mesp2/E47 heterodimer binds to the E3 site of the Epha4enhancer. (A) The results of EMSA using nuclear extracts from NIH3T3 cells transfected with FLAG-Mesp2 and/or Myc-E47 and incubated with E3 probe. The quantities of nuclear extracts used (μg) are indicated.(B) A competition assay indicating the specificity of the binding of the Mesp2/E47 (2 μg each) complex to the E3 probe. The addition of 100-fold excess of unlabeled E3 probe, but not the E3m mutant probe, abolished the binding. (C) Evidence for the heterodimer formation of FLAG-Mesp2/Myc-E47. The band containing E3 (arrow) was supershifted by the addition of either anti-FLAG or anti-Myc antibodies. The oligonucleotides used were as follows: E3, 5′-TGGGTCACATTTGTCCAAAA-3′; E3m,5′-TGGGTCAACGGGTTCCAAAA-3′ (E-box is shown in the bold; altered nucleotides in the mutant probe are underlined).

The binding abilities of Mesp2 and E47 to the Epha4 E-boxes were then analyzed using electrophoresis mobility shift assays (EMSA). Nuclear extracts were prepared from NIH3T3 cells, transfected with either FLAG-tagged Mesp2 or Myc-tagged E47, and these were used in the experiments either separately or in combination. The E3 motif was used in the EMSA experiments as it gave the most consistent results in both the transgenic and luciferase assays. As expected, a strong bandshift was observed when both Mesp2 and E47 were combined, although a faint band was detectable when E47 was incubated alone, indicating that it may form a homodimer that can bind the E3 E-box(Fig. 4A). The sequence specificity of the protein-DNA interactions was confirmed by competition assay using both intact and mutated sequences(Fig. 4B). The specificity of the heteroduplex complex was also confirmed by supershift experiments with anti-FLAG and anti-Myc antibodies (Fig. 4C). We also examined the binding specificity by applying other bHLH proteins. A family protein, Mesp1 also showed strong binding, but other bHLH proteins such as paraxis (Tcf15-Mouse Genome Informatics), Myod1 and Twist (Twist1-Mouse Genome Informatics) did not show significant binding to the E3 probe (see Fig. S3 in the supplementary material). These data strongly suggest that Mesp2 forms a heterodimer with E47, binds to the E-boxes within the Epha4 enhancer and then activates Epha4 transcription in the rostral region of somites.

Fig. 5.

Mox1 and Mox1-cre expression and the lineage analysis. Whole-mount in situ hybridization analysis of Mox1expression in 9.5 dpc embryos (A,B) and Mox1-cre in 8.25 (C), 8.5 (D) and 11.5 (E) dpc embryos. (F-I)Whole-mount expression patterns and sagittal sections of β-gal-stained R26R/Mox1-cre double heterozygous embryos at 10.5 dpc. (H,I) The sections were counterstained with Eosin. Arrowheads indicate somite borders forming between S0 and S-1.

Fig. 5.

Mox1 and Mox1-cre expression and the lineage analysis. Whole-mount in situ hybridization analysis of Mox1expression in 9.5 dpc embryos (A,B) and Mox1-cre in 8.25 (C), 8.5 (D) and 11.5 (E) dpc embryos. (F-I)Whole-mount expression patterns and sagittal sections of β-gal-stained R26R/Mox1-cre double heterozygous embryos at 10.5 dpc. (H,I) The sections were counterstained with Eosin. Arrowheads indicate somite borders forming between S0 and S-1.

The overexpression of Mesp2 leads to the activation of Epha4

Epha4 has been implicated in border formation via the repulsive interaction with ephrin molecules, and this occurs during the formation of segmental boundaries in the hindbrain and in the somites(Barrios et al., 2003; Cooke et al., 2005; Durbin et al., 1998). However,loss-of-function experiments have failed to show any functional relevance for Epha4 in the mouse. In order to investigate whether Mesp2 functions as an activator of Epha4, and possibly to reveal the role of Epha4 during somitogenesis, we established a system that achieves the conditional expression of Mesp2 using Cre-loxP. A transgenic mouse line CAG-floxed-CAT-Mesp2 was established in our laboratory, in which the CAT gene is inserted between two loxP sites and can therefore be excised by Cre recombinase. Hence, the Mesp2 gene in this system will come under the control of the CAG promoter after this excision. To activate Mesp2, we generated and then used a Mox1-Cre mouse (see Fig. S1 in the supplementary material). Mox1 expression initiates just prior to segment border formation, in a similar manner to endogenous Mesp2, but its expression persists after segmentation and is relatively higher in the caudal half of the somites(Mankoo et al., 2003; Saga et al., 1997)(Fig. 5A,B). The Creexpression was detected as early as 8.5 dpc and showed the similar pattern to the Mox1 (Fig. 5C-E).

To confirm the presence of Cre recombinase activity, we crossed the Mox1-Cre mouse with the R26R reporter line and examined β-gal activity during the period 8.5-11.5 dpc(Fig. 5F-I; data not shown). The expression of the reporter was found to begin in the paraxial mesoderm and the most prominent levels were restricted to the somitic derivatives, at least up to 11.5 dpc. Some reporter expression in the rostral neural tube and in the intermediate mesoderm was also observed. We detected differences in the initial expression domain between the Mox1(Fig. 5B) or Cretranscripts (Fig. 5E), andβ-gal reporter activity (Fig. 5G), which most likely reflects a time-lag for the activation of the reporter gene following the excision of the CAT gene by Cre recombinase. Histological sections revealed that the reporter activation was initiated in only a few somitic cells just after segmentation, but that theβ-gal expression gradually expanded throughout the entire components of somite derivatives. Hence, this Cre line is a useful system to drive genes in the somitic cell lineage.

Mesp2 activation induces abnormal epithelialization

To activate Mesp2 expression in the somitic lineage, we crossed the CAG-CAT-Mesp2 and Mox1-cre lines. The double heterozygous mice died shortly after birth and their skeletal specimens revealed strong malformations (see below), indicating abnormal somitogenesis. Under a dissection microscope, the morphology of the somites was not found to have been disrupted, which was unexpected from the observations of the skeletal phenotype. Segmental boundaries were observed, although their width was not perfectly equal to the wild type and the surface appeared to be rough. At first, we analyzed Mesp2 expression at 10.5 dpc(Fig. 6A,B). In the wild-type and single heterozygous embryos, Mesp2 is expressed in the anterior PSM as a single band, although the width and the strength of this expression differs from embryo to embryo as shown previously(Fig. 6A)(Takahashi et al., 2000). In double heterozygotes, however, the ectopic expression of Mesp2 could be observed throughout the entire somitic region, in addition to its normal expression pattern in the anterior PSM(Fig. 6B). Moreover, the Mesp2-positive cells often formed clusters and were not localized in specific regions of somites (Fig. 6C,D).

A similar ectopic expression pattern was observed for Epha4,although the levels of ectopic expression were much lower than the endogenous gene expression (Fig. 6E-H). The spotty expression pattern in both Mesp2 and Epha4indicates that the expression is suppressed or the transcripts are destabilized in many cells and only parts of cells maintain the expression. To further investigate the characteristics of the gene expression profiles and morphologies, serial sections were prepared and subjected to staining for Mesp2 protein, Epha4 transcripts and actin filaments(Fig. 6I-N). The segmental borders were found to have generated but fluorescent phalloidin staining revealed cells showing abnormal epithelialized features and broken epithelial sheaths were also evident (Fig. 6N). In the cells nearby, both ectopic Mesp2(Fig. 6K) and Epha4expression (Fig. 6L) could be observed. Although we could not conclude that Mesp2 directly induced Epha4 using the serial sections, these observations indicate that the cells may have acquired repulsive properties that enable them to form abnormal cell borders within somites (Fig. 6O).

Mesp2 activation induces skeletal malformation

The CAG-CAT-Mesp2/Mox1-Cre double heterozygous fetus showed strong skeletal defects, which are restricted in the ribs and vertebra as expected from the somite-specific Mox1 expression(Fig. 7A-H)(Mankoo et al., 2003). The vertebral bodies and the lamina of neural arches were present in these fetuses, although they displayed severe defects in both their morphology and patterning. By contrast, the pedicles of the neural arches were largely lost(Fig. 7C,G). In addition, the proximal region of the ribs did not form properly(Fig. 7D,H). This phenotype contrasts with Mesp2-null embryos and is somewhat similar to Psen1-null mutants (Takahashi et al.,2000), indicating that it is a rostralized phenotype. To gain insight into the morphogenetic failure underlying the skeletal defects observed in the double transgenic mice, cartilage formation was examined by whole-mount staining with Alcian Blue. Strikingly, rib as well as pedicle cartilages were severely affected even in the 13.5 dpc embryo(Fig. 8A,B).

Fig. 6.

Ectopic Mesp2 expression in somites leads to the activation of Epha4. Expression of Mesp2 in 10.5 dpc wild-type(A) and CAG-CAT-Mesp2; Mox1-Cre double heterozygous embryos (B-D). In addition to the normal Mesp2 expression in the anterior PSM, ectopic expression of Mesp2 was observed throughout the entire somitic region (B-D). The expression pattern of Epha4 at 11.5 dpc in wild-type (E,G) and double heterozygous embryos(F,H) is also shown. An expression pattern for Epha4that was similar to Mesp2 was observed in the double heterozygote(F,H). Histological analyses of 10.5 dpc wild-type (M) and double heterozygous (I-L,N) embryos. Ectopic Mesp2 protein (I,K) and Epha4 expression (J,L) were evident in serial sections of double heterozygotes. Magnified images of square parts of I and J are shown in K and L, respectively. Another consecutive section was stained with phalloidin (N)and a similar region of the wild-type embryo is shown in M. In double heterozygotes, abnormal epithelial cells were observed within the somite (N). A paraffin section stained with nuclear Fast Red revealed gaps in epithelialized somites in the double heterozygote at 10.5 dpc (O). Black arrows in I and J indicate the endogenous expression of Mesp2 (I) or Epha4 (J). Red arrows in K,L,N indicate separated cell clusters. White arrow in N indicates an abnormal epithelialized feature. Black arrows in O indicate local gaps. Scale bar: 100 μm.

Fig. 6.

Ectopic Mesp2 expression in somites leads to the activation of Epha4. Expression of Mesp2 in 10.5 dpc wild-type(A) and CAG-CAT-Mesp2; Mox1-Cre double heterozygous embryos (B-D). In addition to the normal Mesp2 expression in the anterior PSM, ectopic expression of Mesp2 was observed throughout the entire somitic region (B-D). The expression pattern of Epha4 at 11.5 dpc in wild-type (E,G) and double heterozygous embryos(F,H) is also shown. An expression pattern for Epha4that was similar to Mesp2 was observed in the double heterozygote(F,H). Histological analyses of 10.5 dpc wild-type (M) and double heterozygous (I-L,N) embryos. Ectopic Mesp2 protein (I,K) and Epha4 expression (J,L) were evident in serial sections of double heterozygotes. Magnified images of square parts of I and J are shown in K and L, respectively. Another consecutive section was stained with phalloidin (N)and a similar region of the wild-type embryo is shown in M. In double heterozygotes, abnormal epithelial cells were observed within the somite (N). A paraffin section stained with nuclear Fast Red revealed gaps in epithelialized somites in the double heterozygote at 10.5 dpc (O). Black arrows in I and J indicate the endogenous expression of Mesp2 (I) or Epha4 (J). Red arrows in K,L,N indicate separated cell clusters. White arrow in N indicates an abnormal epithelialized feature. Black arrows in O indicate local gaps. Scale bar: 100 μm.

Mesp2 is required for the establishment of the rostral properties within somites via the suppression of caudal genes. Therefore, we anticipated that the forced expression of Mesp2 may lead to the suppression of caudal properties, which would be the cause of the skeletal malformation. Uncx4.1 is a molecular marker for caudal somites(Fig. 8C,E,G) and this gene is also known to be required for the pedicle formation of the neural arch(Leitges et al., 2000; Mansouri et al., 2000). In the CAG-CAT-Mesp2/Mox1-Cre double heterozygotes, the caudally restricted expression pattern was not disrupted but the levels of expression were much lower and the stripes were often interrupted(Fig. 8D,F). The histological section revealed the presence of signal-negative regions in the caudal compartments, which was often accompanied by the morphological abnormalities. We noticed local fusion between cells in the caudal compartment and the rostral ones in the posterior somite (Fig. 8H). Such a fusion was never observed in wild-type or single heterozygous embryos (Fig. 8G).

To explore more genes affected in Mesp2-activated embryos and to understand the cause of abnormalities, we examined expressions of several somitic markers at 11.5 dpc. The segmental expression of Pax3 that is the marker of dermomyotome (Fig. 8I)(Denetclaw and Ordahl, 2000)was expanded in the double transgenic embryo(Fig. 8J), which may indicate expansion of the dermomyotomal progenitor. By contrast, Sox9-positive cell lineage appeared to be relatively reduced in the transgenic embryo especially in the thoracic region (Fig. 8K,L), which may account for the underdevelopment of the rib cartilage. The expansion of the rostral compartment of somites was indicated by the expression of Tbx18 (Fig. 8M-P), which is another target candidate of Mesp2 as its expression is lost in the Mesp2-null embryo(Bussen et al., 2004) (Y.T.,unpublished).

Fig. 7.

Ectopic expression of Mesp2 leads to skeletal malformations. Skeletal preparations of 18.5 dpc wild-type (A-D) and CAG-CAT-Mesp2; Mox1-Cre fetuses (E-H) stained with Alcian Blue and Alizarin Red. Lateral (A,E) and dorsal (B,F) views of whole skeletons are shown. Higher magnifications of the lumber region from wild-type(C,D) and double heterozygous fetuses (G,H) are also shown. The lack of pedicles of the neural arches could be observed in the double heterozygotes(G,H). The rib structure was also severely affected in the double heterozygotes (G,H). pd, pedicle; lm, lamina.

Fig. 7.

Ectopic expression of Mesp2 leads to skeletal malformations. Skeletal preparations of 18.5 dpc wild-type (A-D) and CAG-CAT-Mesp2; Mox1-Cre fetuses (E-H) stained with Alcian Blue and Alizarin Red. Lateral (A,E) and dorsal (B,F) views of whole skeletons are shown. Higher magnifications of the lumber region from wild-type(C,D) and double heterozygous fetuses (G,H) are also shown. The lack of pedicles of the neural arches could be observed in the double heterozygotes(G,H). The rib structure was also severely affected in the double heterozygotes (G,H). pd, pedicle; lm, lamina.

These observations are consistent with the idea that ectopic Mesp2 expression is of sufficient strength to activate rostral genes such as Epha4 and Tbx18, and to suppress expression of the caudal gene Uncx4.1.

DISCUSSION

In our current study, we have identified a cluster of E-boxes in the enhancer region of Epha4, incorporating the Mesp2 binding site, in which at least three crucial E-boxes (E1, E3 and E4 in Fig. 2C) are present. The loss of these motifs results in a substantial reduction in gene reporter activity,in both luciferase and transgenic reporter assays, indicating that there is an essential requirement of multiple E-boxes for Epha4 activation by Mesp2. Interestingly, the coexpression of Mesp2 and E47 resulted in higher luciferase activity (tenfold) (Fig. 3B), whereas only weak activity (twofold) was obtained with Mesp2 alone (data not shown). Mesp2 alone could also not bind to E-box containing DNA sequences (Fig. 4A). Therefore, Mesp2 alone or Mesp2 homodimers appear to be inactive on Epha4 somite enhancer.

The core E-box sequence appears to be CAAATG/CATTTG and synthetic enhancers generated by six repeats of the Epha4 enhancer E1, E3 and E4 motifs,and flanking sequences, can recapitulate the segmental expression pattern of this gene in vivo. The differences that we observed in the measured luciferase activities for the multiple E-boxes may reflect the involvement of the sequences flanking the core enhancer region in promoting the binding of bHLH-type transcription factors, which has been observed in other cases(Powell et al., 2004). In addition, other factors may modulate the interactions between Mesp2/E47 and its target sequences. It has been reported that the phosphorylation of E47 is required for the formation of heterodimers with Myod1 and for the subsequent binding to the target sequence (Lluis et al., 2005). The methylation state of target sequences has also been implicated in the binding by another bHLH heterodimer, Max/Myc, in which methylation of the CpG dinucleotide within the E-box has been shown to prevent the access of the bHLH proteins (Perini et al., 2005). Further studies will be required to determine whether such modulations are involved in the binding of the Mesp2/E47 heterodimer to its target sites.

Epha4 is implicated in segmental border formation via its interaction with the Eph ligand ephrin, which is expressed in apposed cells in zebrafish(Barrios et al., 2003; Durbin et al., 1998). However,there is no direct evidence for this in the mouse, as the loss of Epha4 failed to reveal any role for this protein during somitogenesis, which may be due to functional redundancy among the several Eph and ephrin family proteins. In such a situation, a transgenic strategy of forced gene expression is an alternative and effective method. In the current study, we have tried the forced expression of Mesp2 with expectation that Epha4should be induced under the control of Mesp2. The forced expression of Mesp2 not only activates Epha4 expression but also induces the local segregation of somitic cells. Recently, we showed that Mesp2 establishes the segmental boundary by suppressing Notch signaling, which then generates a boundary between the Notch-active and Notch-negative domains(Morimoto et al., 2005). We have also shown that this boundary forms the next somite border. However, the precise molecular mechanisms involved in the generation of these morphological boundaries are not yet fully understood, although Cdc42 and Rac1 are known to play important roles in subsequent epithelial somite formation(Nakaya et al., 2004). Although the direct evidence was not presented, our current data indicate that Mesp2 activates Epha4 in the anteriormost cells in the PSM and that this may activate reverse signaling though ephrin expression in opposing cells and generate a gap during normal somitogenesis. A similar mechanism has previously been proposed for the epithelialization of boundary cells in zebrafish (Barrios et al.,2003; Cooke et al.,2005). Nevertheless, we can not exclude the possibility that pathways other than Epha4 activation by Mesp2 are required for the induction of epithelialization.

Mesp2 is also known as a strong suppressor of the establishment of caudal properties, which is mediated by the suppression of both Notch signaling and Dll1 and Uncx4.1 expression(Nomura-Kitabayashi et al.,2002; Takahashi et al.,2000). We actually did observe suppression of Uncx4.1 in our double heterozygotes, but the segmental pattern of Uncx4.1expression at 10.5 dpc was not found to be severely disrupted. Therefore, our finding of an extremely defective skeletal phenotype in the CAG-CAT-Mesp2/Mox1-Cre mice was somewhat surprising. We postulate that prolonged Mesp2 expression, driven by the CAG promoter, must continuously attenuate Uncx4.1 and the corresponding downstream gene expression in the later stages of development, which would lead to the almost complete suppression of chondrogenesis, as observed in the case of loss of Uncx4.1(Leitges et al., 2000; Mansouri et al., 2000). One of target genes activated by Uncx4.1 and responsible for the phenotype would be Sox9, the product of which is known to be a key regulator of chondrogenesis (Akiyama et al.,2005). However, it remains to be investigated whether this suppression is directly mediated by Mesp2 or by other transcriptional suppressors that are activated by Mesp2.

Fig. 8.

Early defects in chondrogenesis and gene expressions affected in the CAG-CAT-Mesp2;Mox1-Cre double heterozygotes. Skeletal morphology was revealed by Alcian Blue staining in the wild-type (A)and double heterozygous (B) embryos at 13.5 dpc. (C-H) Uncx4.1. expression was reduced in double heterozygotes (D,F) at both 10.5 (C,D) and 11.5 dpc (E,F) compared with wild-type embryos (C,E). The section of 11.5 dpc embryonic tail revealed Uncx4.1-negative cells(arrowheads) in the caudal compartment of somites in the double heterozygote(H). Comparison of expression patterns of Pax3 (I,J), Sox9 (K,L) and Tbx18 (M-P) between wild-type (I,K,M,O) and double heterozygous (J,L,N,P) embryos. Outlines in K and L show Sox9 expression in the rib primordia.

Fig. 8.

Early defects in chondrogenesis and gene expressions affected in the CAG-CAT-Mesp2;Mox1-Cre double heterozygotes. Skeletal morphology was revealed by Alcian Blue staining in the wild-type (A)and double heterozygous (B) embryos at 13.5 dpc. (C-H) Uncx4.1. expression was reduced in double heterozygotes (D,F) at both 10.5 (C,D) and 11.5 dpc (E,F) compared with wild-type embryos (C,E). The section of 11.5 dpc embryonic tail revealed Uncx4.1-negative cells(arrowheads) in the caudal compartment of somites in the double heterozygote(H). Comparison of expression patterns of Pax3 (I,J), Sox9 (K,L) and Tbx18 (M-P) between wild-type (I,K,M,O) and double heterozygous (J,L,N,P) embryos. Outlines in K and L show Sox9 expression in the rib primordia.

We also show in our present study that the Mox1-Cre mouse is a useful tool for inducing either the disruption or activation of genes that are components of the somitic cell lineage. However, there is a delay in gene activation and the activation of Mesp2 was `spotty' and these may be the reason why we did not observe strong segmental defects. Gene reporter activity was also observed in other lineages, including parts of the neural tube and the intermediate mesoderm. Therefore, although a detailed lineage study will be required in future studies, the activity that we observed in our CAG-CAT-Mesp2/Mox1-Cre transgenics proved to be useful for the manipulation of gene expression, at least in the somitic cell lineages. Recently, a similar Cre line (Meox1cre) was reported by another laboratory and the results of their study were consistent with our current data (Jukkola et al.,2005).

Acknowledgements

We are grateful to Dr Baljinder S. Mankoo for generously providing the genomic DNA clones for Mox1 and Drs Alan Rawls, Sachiko Iseki and Atsuko Sehara for cDNA clones encoding paraxis, twist and Myod1,respectively. We also thank Masayuki Oginuma and Dr Kenta Sumiyama for advice on the transgenic mouse analysis. This work was supported by Grants-in-Aid for Science Research on Priority Areas (B), the Organized Research Combination System and National BioResource Project of the Ministry of Education, Culture,Sports, Science and Technology, Japan.

References

Akiyama, H., Kamitani, T., Yang, X., Kandyil, R., Bridgewater,L. C., Fellous, M., Mori-Akiyama, Y. and de Crombrugghe, B.(
2005
). The transcription factor Sox9 is degraded by the ubiquitin-proteasome system and stabilized by a mutation in a ubiquitin-target site.
Matrix Biol.
23
,
499
-505.
Aulehla, A. and Herrmann, B. (
2004
). Segmentation in vertebrates: clock and gradient finally joined.
Genes Dev.
18
,
2060
-2067.
Barrios, A., Poole, R. J., Durbin, L., Brennan, C., Holder, N. and Wilson, S. W. (
2003
). Eph/Ephrin signaling regulates the mesenchymal-to-epithelial transition of the paraxial mesoderm during somite morphogenesis.
Curr. Biol.
13
,
1571
-1582.
Bessho, Y. and Kageyama, R. (
2003
). Oscillations, clocks and segmentation.
Curr. Opin. Genet. Dev.
13
,
379
-384.
Bussen, M., Petry, M., Schuster-Gossler, K., Leitges, M.,Gossler, A. and Kispert, A. (
2004
). The T-box transcription factor Tbx18 maintains the separation of anterior and posterior somite compartments.
Genes Dev.
18
,
1209
-1221.
Cooke, J., Kemp, H. and Moens, C. (
2005
). EphA4 is required for cell adhesion and rhombomere-boundary formation in the zebrafish.
Curr. Biol.
15
,
536
-542.
Denetclaw, W. F. and Ordahl, C. P. (
2000
). The growth of the dermomyotome and formation of early myotome lineages in thoracolumbar somites of chicken embryos.
Development
127
,
893
-905.
Dottori, M., Hartley, L., Galea, M., Paxinos, G., Polizzotto,M., Kilpatrick, T., Bartlett, P. F., Murphy, M., Kontgen, F. and Boyd, A. W. (
1998
). EphA4 (Sek1) receptor tyrosine kinase is required for the development of the corticospinal tract.
Proc. Natl. Acad. Sci. USA
95
,
13248
-13253.
Durbin, L., Brennan, C., Shiomi, K., Cooke, J., Barrios, A.,Shanmugalingam, S., Guthrie, B., Lindberg, R. and Holder, N.(
1998
). Eph signaling is required for segmentation and differentiation of the somites.
Genes Dev.
12
,
3096
-3109.
Imai, Y., Matsushima, Y., Sugimura, T. and Terada, M.(
1991
). A simple and rapid method for generating a deletion by PCR.
Nucleic Acids Res.
19
,
2785
.
Jukkola, T., Trokovic, R., Maj, P., Lamberg, A., Mankoo, B.,Pachnis, V.,Savilahti, H. and Partanen, J. (
2005
). Meox1Cre: a mouse line expressing Cre recombinase in somitic mesoderm.
Genesis
43
,
148
-153.
Kothary, R., Clapoff, S., Darling, S., Perry, M., Moran, L. and Rossant, J. (
1989
). Inducible expression of an hsp68-lacZ hybrid gene in transgenic mice.
Development
105
,
707
-714.
Kullander, K., Mather, N. K., Diella, F., Dottori, M., Boyd, A. W. and Klein, R. (
2001
). Kinase-dependent and kinase-independent functions of EphA4 receptors in major axon tract formation in vivo.
Neuron
29
,
73
-84.
Ledent, V. and Vervoort, M. (
2001
). The basic helix-loop-helix protein family: comparative genomics and phylogenetic analysis.
Genome Res.
11
,
754
-770.
Leitges, M., Neidhardt, L., Haenig, B., Herrmann, B. and Kispert, A. (
2000
). The paired homeobox gene Uncx4.1 specifies pedicles, transverse processes and proximal ribs of the vertebral column.
Development
127
,
2259
-2267.
Lluis, F., Ballestar, E., Suelves, M., Esteller, M. and Munoz-Canoves, P. (
2005
). E47 phosphorylation by p38 MAPK promotes MyoD/E47 association and muscle-specific gene transcription.
EMBO J.
24
,
974
-984.
Mankoo, B., Skuntz, S., Harrigan, I., Grigorieva, E., Candia,A., Wright, C., Arnheiter, H. and Pachnis, V. (
2003
). The concerted action of Meox homeobox genes is required upstream of genetic pathways essential for the formation, patterning and differentiation of somites.
Development
130
,
4655
-4664.
Mansouri, A., Voss, A., Thomas, T., Yokota, Y. and Gruss, P.(
2000
). Uncx4.1 is required for the formation of the pedicles and proximal ribs and acts upstream of Pax9.
Development
127
,
2251
-2258.
Maroto, M. and Pourquié, O. (
2001
). A molecular clock involved in somite segmentation.
Curr. Top. Dev. Biol.
51
,
221
-248.
Morimoto, M., Takahashi, Y., Endo, M. and Saga, Y.(
2005
). The Mesp2 transcription factor establishes segmental borders by suppressing Notch activity.
Nature
435
,
354
-359.
Nakaya, Y., Kuroda, S., Katagiri, Y., Kaibuchi, K. and Takahashi, Y. (
2004
). Mesenchymal-epithelial transition during somitic segmentation is regulated by differential roles of Cdc42 and Rac1.
Dev. Cell
7
,
425
-438.
Nomura-Kitabayashi, A., Takahashi, Y., Kitajima, S., Inoue, T.,Takeda, H. and Saga, Y. (
2002
). Hypomorphic Mesp allele distinguishes establishment of rostrocaudal polarity and segment border formation in somitogenesis.
Development
129
,
2473
-2481.
Perini, G., Diolaiti, D., Porro, A. and Della Valle, G.(
2005
). In vivo transcriptional regulation of N-Myc target genes is controlled by E-box methylation.
Proc. Natl. Acad. Sci. USA
102
,
12117
-12122.
Pourquié, O. (
2003
). The segmentation clock: converting embryonic time into spatial pattern.
Science
301
,
328
-330.
Powell, L. M., Zur Lage, P. I., Prentice, D. R., Senthinathan,B. and Jarman, A. P. (
2004
). The proneural proteins Atonal and Scute regulate neural target genes through different E-box binding sites.
Mol. Cell. Biol.
24
,
9517
-9526.
Rida, P., Le Minh, N. and Jiang, Y. (
2004
). A Notch feeling of somite segmentation and beyond.
Dev. Biol.
265
,
2
-22.
Saga, Y. and Takeda, H. (
2001
). Mesp2: a novel mouse gene expressed in the presegmented mesoderm and essential for segmentation initiation.
Nat. Rev. Genet.
11
,
835
-845.
Saga, Y., Hata, N., Koseki, H. and Taketo, M. M.(
1997
). Mesp2: a novel mouse gene expressed in the presegmented mesoderm and essential for segmentation initiation.
Genes Dev.
11
,
1827
-1839.
Saga, Y., Miyagawa-Tomita, S., Takagi, A., Kitajima, S.,Miyazaki, J. andInoue, T. (
1999
). MesP1 is expressed in the heart precursor cells and required for the formation of a single heart tube.
Development
126
,
3437
-3447.
Sakai, K. and Miyazaki, J. (
1997
). A transgenic mouse line that retains Cre recombinase activity in mature oocytes irrespective of the cre transgene transmission.
Biochem. Biophys. Res. Commun.
237
,
318
-324.
Takahashi, Y., Koizumi, K., Takagi, A., Kitajima, S., Inoue, T.,Koseki, H. and Saga, Y. (
2000
). Mesp2 initiates somite segmentation through the Notch signalling pathway.
Nat. Genet.
25
,
390
-396.
Theil, T., Frain, M., Gilardi-Hebenstreit, P., Flenniken, A.,Charnay, P. and Wilkinson, D. (
1998
). Segmental expression of the EphA4 (Sek-1) receptor tyrosine kinase in the hindbrain is under direct transcriptional control of Krox-20.
Development
125
,
443
-452.
Yamauchi, Y., Abe, K., Mantani, A., Hitoshi, Y., Suzuki, M.,Osuzu, F., Kuratani, S. and Yamamura, K. (
1999
). A novel transgenic technique that allows specific marking of the neural crest cell lineage in mice.
Dev. Biol.
212
,
191
-203.
Zambrowicz, B., Imamoto, A., Fiering, S., Herzenberg, L., Kerr,W. and Soriano, P. (
1997
). Disruption of overlapping transcripts in the ROSA beta geo 26 gene trap strain leads to widespread expression of beta-galactosidase in mouse embryos and hematopoietic cells.
Proc. Natl. Acad. Sci. USA
94
,
3789
-3794.

Supplementary information