Members of the Myc-Max-Mad network are essential regulators of cellular growth, proliferation, and differentiation (Lüscher, 2001; Zhou and Hurlin, 2001). Proteins of this family are characterized by a conserved basic helix-loop-helix leucine zipper (bHLH-Zip) domain that facilitates DNA binding and protein–protein interactions (Lüscher and Larsson, 1999). Myc protein activity correlates with growth and proliferation, and requires the association with Max to bind to E-box DNA regulatory sequences (CACGTG). The Mad proteins (Mad1, Mxi1, Mad3, and Mad4), and the distantly related and larger Mnt (also termed Rox), can antagonize the transforming activity of Myc by sequestering Max, as well as through the binding and repression of a subset of Myc-target genes (Baudino and Cleveland, 2001; O'Hagan et al., 2000; Hurlin et al., 1997; Meroni et al., 1997). Active repression by the Mads is achieved via recruitment of a repressor complex containing HDAC activity to the N-terminal mSin3 interaction domain (SID) (Ayer et al., 1996; Sommer et al., 1997; Laherty et al., 1997).
Myc gene activity is primarily found in proliferating cells and down-regulated during differentiation (Ayer and Eisenman, 1993; Hurlin et al., 1995). In contrast, the Mads are sequentially induced during the transition from proliferation to differentiation, suggesting that the Mads initiate or enhance this process (Ayer and Eisenman 1993; Hurlin et al., 1994, 1995). Correspondingly, Myc-Max complexes bound to the promoters of hTERT and cyclin D2 in proliferating cells are exchanged for Mad1-Max complexes during differentiation (Xu et al., 2001; Bouchard et al., 2001). However, the Mads may have functional roles in vivo other than to antagonize Myc activity. The Mad knockout mice (Mad1, Mxi1, and Mad3), with the exception of Mnt, exhibited only subtle effects on growth and differentiation and did not give rise to tumors as would be predicted if the Mads were to function as Myc antagonists in vivo (Grandori et al., 2000; Hurlin et al., 2003). In addition, the Mad and Myc proteins are most likely to have both shared and unique target genes (O'Hagan et al., 2000).
In Xenopus, several members of the Myc-Max-Mad network have been characterized to date, including c-Myc, L-Myc, N-Myc, Max, and Mad4 (Schreiber-Agus et al., 1993; King et al., 1993, 1996; Newman and Krieg, 1999). Max is expressed at constant levels throughout embryogenesis. In contrast, the Myc and Mad genes show a more restricted expression. All Myc transcripts are maternally active, but are quickly degraded. While transcripts of L-Myc are not detected in later stages of embryogenesis, those of c-Myc are found at late gastrula stages and early neurula stages in the transversal and lateral neural folds that will later give rise to the neural crest (Bellmeyer et al., 2003; Newman and Krieg, 1999). Consistent with its unique expression pattern, c-Myc plays an essential role in the specification of neural crest (Bellmeyer et al., 2003). Zygotic N-Myc transcripts are first detected in the anterior neural plate and in the migrating neural crest at early neurula stages (Bellmeyer et al., 2003; Vize et al., 1990). Xmad4, the only Xenopus Mad gene isolated to date, is transiently expressed in the cement and hatching glands and later in the developing liver and pronepheros (Newman and Krieg, 1999). In situ hybridization for the murine Mads has also been described, but a detailed analysis in embryonic stages prior to organogenesis was hindered by the low expression levels (Queva et al., 1998).
To gain further insight into the function of this important family of genes in early development, we have isolated and characterized Xenopus homologues of Mad1, Mad3, and Mnt. These novel Xenopus genes are expressed in different and restricted patterns suggesting distinct functions for these proteins in embryogenesis.
RESULTS AND DISCUSSION
Screening of Xenopus cDNA phage libraries led to the isolation of the entire coding regions of Xmad1, Xmad3, and Xmnt (GenBank accession numbers: AY964104, AY964105, and AY964106). The derived amino acid sequences exhibited between 56–72% overall amino acid identity with the corresponding mouse sequences (Fig. 1). Higher homology was observed within the SID repressor and bHLH domains (72–100%).
The temporal expression patterns of the Xenopus Mads during early embryonic development were investigated by RT-PCR analysis with RNA isolated from various stages of Xenopus development. As shown in Figure 2, maternal transcripts are present for all Mad members analyzed. Transcripts of Xmad1 and Xmad3 are slightly decreased after MBT but remain relatively constant throughout development. In contrast, only weak levels of zygotic transcripts are detected for XMnt during early phases of embryogenesis.
The developmental expression patterns of the Xenopus Mads were determined by whole-mount in situ analysis. Early transcripts of Xmad1 are localized throughout the animal hemisphere of early cleavage stage embryos and become localized above the dorsal blastopore lip during early gastrulation (Fig. 3A and B). As gastrulation proceeds, the expression of Xmad1 extends along the dorsal midline (Fig. 3C–E). A transverse section of a stage-17 embryo reveals Xmad1 transcripts are present in the notochord and in the floor plate (Fig. 3F). At this stage, transcripts are also detected in the cement gland, where they are maintained in later developmental stages (Fig. 3G and H). Additional expression is also detected in the pineal gland, as well as the trigeminal and geniculate ganglia (Fig. 3H). As shown by the section through the trunk region of a stage-27 embryo, transcripts are maintained in the floor plate and, in addition, present in the hypochord and the marginal zone of the ventral neural tube where postmitotic neurons are located (Fig. 3I).
In Xenopus embryos, cell proliferation is absent during gastrulation in the involuting dorsal axial mesoderm and in the presumptive notochord, as well as in the cement gland during neurula stages (Saka and Smith, 2001). Thus, consistent with cell culture experiments where ectopic expression of Mad1 results in the inhibition of proliferation in a variety of cells including fibroblasts, adipocytes, and glioblastomas, Xmad1 expression during early development correlates with cells that have undergone terminal differentiation (Gehring et al., 2000; Bejarano et al., 2000; Roussel et al., 1996; Roy and Reisman, 1995; Pulverer et al., 2000). The expression of Xmad1 in the Spemann organizer during gastrulation and later in the notochord and in the floor plate of the neural tube parallels that of the TGF-β superfamily ligand, Xnr4 (Joseph and Melton, 1997). This observation, together with the finding that Mad1, but not other members of the Mad family, are directly induced in keratinocytes by the TGF-β ligands Activin A and TGF-β suggests that Xmad1 may be a direct target of Xnr4 during embryogenesis (Werner et al., 2001).
Similar to Xmad1, expression of Xmad3 is localized throughout the animal hemisphere prior to MBT (data not shown). Strong and specific staining is not observed until stage 20 in the eye vesicle (Fig. 3J and K) and later in the neural tube, olfactory placode, midbrain, and hindbrain (Fig. 3L and M). In addition to expression in the CNS at stage 29, expression is also visible in the pronephros, otic placodes, and tailtip (Fig. 3N). A transversal section reveals that Xmad3 expression in the eye is localized to the retina and transcripts are broadly present throughout the midbrain and hindbrain (Fig. 3O and P). In murine cells undergoing differentiation in vitro and in vivo, Mad3 is primarily coexpressed with Myc in proliferating cells (Hurlin et al., 1995; Quéva et al., 1998). However, in the context of neurogenesis, we observe that the expression of Xmad3 was not limited to the proliferating cells of the ventricular layer, but detected throughout all layers of the hindbrain (Fig. 3P).
XMnt transcripts are first detected by whole mount in situ hybridization at neurula stages, and are localized anteriorly in the neural plate, the neural crest, and weakly in the spinal cord (Fig. 3Q). As development proceeds, staining becomes stronger and localized throughout the CNS, eye vesicle, and the streams of migrating branchial and hyoid neural crest (Fig. 3R–T). The expression of Xmnt in the migrating cranial neural crest cells correlates with the cleft palates and craniofacial deformities observed in Mnt-deficient mice (Sadaghiani and Thiebaud, 1987; Toyo-oka et al., 2004). In addition, transcripts are also present weakly in the cement gland. A sagittal section of a stage-27 embryo shows strong expression in the retina and spinal chord, as well as weak expression in the forebrain and midbrain (Fig. 3V). As shown in the transversal section, transcripts are predominately located in the outermost marginal layer of the ventral hindbrain, where terminally differentiated neurons are located. (Fig. 3W).
The expression pattern analysis conducted in this study and that of the previously described Xmad4 demonstrate that members of the Mad family of transcriptional repressors have partially overlapping expression domains in some tissues, but predominately exhibit distinct domains of expression during Xenopus embryogenesis. This lends itself to the notion that the Mads function not as redundant group of genes but have distinct functions, some of which may be independent from their ability to antagonize Myc activity. The identification of Mad homologs in Xenopus will aid in the elucidation of the function of the Mads and their relationship with Myc and cellular growth and differentiation.
Isolation of Xenopus Mad1, Mad3, and Mnt
A 708-bp sequence was PCR amplified from cDNA using primers derived from a Xenopus EST sequence (BG038912). A Xenopus tadpole head ZAPλ phage cDNA library was screened with the PCR fragment using the ECL Direct Nucleic Acid Labeling and Detection System (Amersham Biosciences). The isolated Xmad1 pBKCMV clone contained a 1,685-bp insert encoding for a protein of 221 aa, 188 bp of 5′-UTR, and 831 bp of 3′-UTR.
A Xenopus oocyte cDNA ZAP phage library was screened with the coding region of the mouse Mad3. The isolated Xmad3 pBKCMV clone contained 523 bp of XMad3 cDNA, but was missing approximately 51 bp of the 5′-coding region. To obtain the additional sequence information, a vector binding primer (5′-CGCGCCTGCAGGTCGACACTA-3′) and a XMad3 sequence specific primer (5′-GGCATGAGCTGGAGTAGAGTA-3′) was used to amplify the Xmad3 sequence from a Xenopus tadpole head library and the resulting PCR product subcloned into the pGemT vector (Promega, Madison, WI). The 833-bp sequence contained the entire open reading frame of Xmad3.
A 3′ primer derived from the EST Sequence (BJ071360) and a vector binding primer were used to amplify a 1.5-kb fragment from a Xenopus tadpole head cDNA phage library and subcloned in pGemT (Promega). The amplified fragment was used to screen the same library by filter hybridization (see above). The isolated XMnt pBKCMV clone contained 238 bp of 5′-UTR, an open reading frame of 1,722 bp and 733 bp of 3′-UTR.
Xenopus Embryo Collection and WMISH
Xenopus laevis embryos were obtained by HCG-induced egg-laying, dejellied in 2% cysteine pH 8.0, washed and cultured in 0.1X MBSH. Embryos were fixed in MEMFA at the desired stage according to Nieuwkoop and Faber (1967). The spatial expression patterns were determined by whole mount in situ hybridization (Harland, 1991) using a DIG labelled antisense probe. Embryos were embedded in gelatine and 30-μm sections were prepared using a vibratome.
Total RNA was extracted from the various embryonic stages (Qiagen RNeasy Kit, Chatsworth, CA) and cDNA prepared using random hexamer primers and MuLV reverse transcriptase (Perkin-Elmer, Norwalk, CT). PCR was performed with Taq polymerase using the following gene specific oligonucleotide primer pairs: Histone H4 (26 cycles), forward: 5′-CGGGATAACATTCAGGGTATCACT-3′ and reverse: 5′-ATCCATGGCGGTAACTGTCTTCCT-3′; Xmad1 (28 cycles), forward: 5′-GAGATAAGAGAGGGGATAAAGG-3′and reverse: 5′-CCTCTCCTTGCTATTGTAAGGC-3′; Xmad3 (28 cycles) forward 5′-ACGCTCAGTCTACTACACAGG-3′ and reverse 5′-GGCATGAGCTGGAGTAGAGTA-3′; Xmad4 (28 cycles) forward 5′-GCAACCTCCATATCTAGG-3′and reverse 5′-ATATAATGGAGCCCATGCTGTC-3′.
We thank K. Ditter for excellent technical assistance and Andreas Nolte for DNA sequencing. This work was supported by a grant from the DFG (CMPB) to T.P. and from the University of Goettingen to K.H.