ABSTRACT
The murine cripto gene encodes a 171-aminoacid epidermal growth factor-related protein, with 93% similarity to its human counterpart in the ‘EGF-like’ domain. The murine cripto mRNA contains two B1 repeats in its 3 non-coding region and a 163-nucleotide homology to the human mRNA.
The mouse cripto gene is expressed at low level in specific organs of the adult animal such as spleen, heart, lung and brain. In situ hybridization analysis during murine embryogenesis (day 6.2 to day 10.5) reveals a very restricted expression pattern. cripto transcripts are first detected in a few epiblastic cells at day 6.5. During gastrulation, the transcripts are expressed in the forming mesoderm and later during development cripto gene expression is restricted to the truncus arteriosus of the developing heart. This expression pattern suggests a role for cripto gene in the determination of the epiblastic cells that subsequently give rise to the mesoderm.
INTRODUCTION
The increasing number of growth factors that have been isolated are grouped into families according to sequence or structural similarities. The ‘EGF family’ includes Epidermal Growth Factor (EGF) (Carpenter and Cohen, 1990), Transforming Growth Factor-α (TGF-α) (Derynck et al., 1984; Lee et al., 1985), Vaccine and Shope Virus Growth Factors (VVGF and SVGF, respectively) (Brown et al., 1985; Chang et al., 1987). Recently, several additional ‘EGF-like’ growth factors with structural similarities have been identified: the human amphiregulin (AR) (Shoyab et al., 1989), the human CRIPTO protein (Ciccodicola et al., 1989), the rat Schwannoma-derived Growth Factor (SDGF) (Kimura et al., 1990) and the Drosophila gene spitz (spi) (Rutledge et al., 1992).
The ‘6 cystein-motif’, which is characteristic of all the members of the ‘EGF family’, is also found in a large number of the proteins (including cell surface and extracellular proteins) that are involved in cell-adhesion during development. Among these are the products of several Drosophila genes: Notch (Kidd et al., 1986), crumbs (Knust et al., 1987), slit (Rothberg et al., 1988), Delta (Vässin et al., 1987) and tolloid (Lepage et al., 1992).
Polypeptide growth factors (PGFs) are currently the subject of great interest mainly because they are expressed in cells during specific developmental processes. During organogenesis and postnatal development, the major role of all these gene products appears to be regulation of cell growth and differentiation by specific protein-protein interactions. However, their expression pattern during early vertebrate development suggests that they might act as signaling molecules in such processes as gastrulation and pattern formation (for reviews see Smith, 1989; Melton, 1991; Stern, 1992). Functional evidence comes from experimental studies in Xenopus laevis where PGFs have been implicated as endogenous mesoderm-inducing factors. In particular, fibroblast growth factor FGF-type PGFs induce mostly posterior/ventral mesoderm (for reviews see Smith, 1989; Melton, 1991), whereas transforming growth factor TGF-β-type (for reviews see Smith, 1989; Melton, 1991) as well as wnt-type PGFs induce dorsal/anterior structures (Sokol et al., 1991). Similar PGFs seem to play a role in mesoderm induction in other vertebrate as well. In mouse embryos, mRNA encoding Fgf-5 (Hébert et al., 1991), Fgf-4 (Niswander and Martin, 1992) and int-2 (Wilkinson et al., 1988) have been found sequentially expressed during gastrulation, suggesting that these PGFs may be involved in mesoderm formation.
We have previously characterized a human cDNA encoding a 188-aminoacid protein named CRIPTO (Ciccodicola et al., 1989; Dono et al., 1991). The central portion of the CRIPTO protein is structurally similar to the members of the ‘EGF family’ of growth factors. Overexpression of CRIPTO leads to transformation of the mouse mammary epithelial NOG-8 cell line (Ciardiello et al., 1991a). Here, we report the isolation of a mouse homolog encoding a 171-aminoacid protein, which is 93% homologous to the human protein in its ‘EGF-like’ region.
The human CRIPTO mRNA contains an Alu repeat and a non-canonical polyadenylation site (AGTAAA) in its long 3′ untranslated region. Interestingly, the murine cripto mRNA contains, in the same region, two B1 repeats (homologous to the human Alu family). Furthermore, the last 163 nucleotides of the human and murine mRNAs show 69.3% similarity, suggesting an evolutionary conserved role for these non-coding sequences.
Previously we have shown that the CRIPTO gene is expressed in human and murine teratocarcinoma cells but not in retinoic acid (RA)-treated cells (Ciccodicola et al., 1989). We describe here the expression pattern of the murine cripto gene in the adult mouse and during embryogenesis. Our data provide the first molecular evidence that the cripto gene is expressed during early vertebrate development. The cripto mRNA is present just prior to the onset of gastrulation. During gastrulation its expression seems restricted to the cells of the forming mesoderm. This intriguing expression pattern suggests important roles during vertebrate embryogenesis and possibly during mesoderm induction.
MATERIALS AND METHODS
Isolation of the murine cripto transcripts
A cDNA library (Robertson et al., 1986; kindly provided by Dr Françoise Poirier) was screened using the coding region of the human CRIPTO gene (Ciccodicola et al., 1989) as probe. Phage DNA was analyzed by restriction mapping using standard procedures (Maniatis et al., 1990). The two longest DNA inserts were subcloned in the pGEM-1 vector (Promega Corp.) and sequenced by the dideoxy method, (Hattori and Sakaki, 1986). The sequencing was performed in both directions using various synthetic primers (Applied Biosystems).
For analysis of the deduced open reading frame, various sequence analysis programs were used (Kyte and Doolittle, 1982; von Heijne, 1986).
RNA analysis
Total RNA was extracted from cells and tissues by the guanidinium isothiocyanate method (Chirgwin et al., 1979) and poly(A)+ RNA was purified by oligo(dT)-cellulose chromatography (Aviv and Leder, 1972). RNAse protection experiments were performed as previously described (Rebagliati et al., 1985). Single-stranded antisense RNA probes were synthesized using a cripto cDNA fragment (from nucleotide 357 to 662) cloned in pGEM4Z (Promega). PCR amplification (Saiki et al., 1988) was performed using oligo(dT)-primed cDNAs prepared from 5 μg of poly(A)+ RNA. One fifth of the cDNAs were amplified using two oligos: one corresponding to nucleotides 1149 to 1169 and the other complementary to nucleotides 1482 to 1462.
Mice embryos
C57/B16 mice were mated between 11 pm and 7 am. Day 0.5 d.p.c. was assumed as the middle of the day of vaginal plugging. Pregnant females were killed by cervical dislocation and embryos were collected in ice-cold PBS under a dissection microscope (Zeiss SV11) and fixed in 4% paraformaldehyde/PBS1X overnight. Embryos between gestational days 8.5 and 10.5 were staged according to the somite numbers. Embryos between day 6 and 8 were subsequently staged more carefully on the basis of the size and morphology.
In situ hybridization
RNA probes were synthetically produced as described above. For in situ hybridization, transcription reactions with T7 or SP6 polymerases (Riboprobe kit, Promega Biotec) were carried out in presence of [35S]CTP (Amersham).
The template was removed with RNAse-free DNAse and the labeled RNA was purified through a Sephadex G-50 column. The transcripts were hydrolyzed to an average length of 150 nucleotides by random alkaline hydrolysis. The probes were used at a working concentration of 1×105 cts/minute/μl in hybridization mix (Wilkinson and Green, 1990).
In situ hybridization was carried out as described by Wilkinson and Green (1990) with minor modifications. 30 μl of the appropriate probe in hybridization mix was added to each slide. Hybridization was carried out overnight at 55 °C. Slides were subsequently washed under stringent conditions (65 °C, 2×SSC, 50% formamide) and treated with RNAse to remove unhybridized and non-specifically bound probe. Autoradiography was performed with Kodak NTB2 emulsion. Exposure times were between 10 and 15 days. After developing, sections were stained in 0.02% toluidine blue and mounted in DPX. Sections were examined and photographed using a Zeiss SV11 microscope with both dark-field and brightfield illumination.
RESULTS
Characterization of the murine cripto mRNA
We used a cDNA fragment containing the complete coding portion of the human CRIPTO cDNA to screen a cDNA library prepared from stem cells of 3-day-old mouse embryos (Robertson et al., 1986; kindly provided by Dr Françoise Poirier). Ten clones were isolated and characterized. The two cDNAs containing the longest inserts were completely sequenced. Fig. 1 shows the nucleotide sequence and the deduced open reading frame encoding the 171-aminoacid cripto protein. The 3′ untranslated sequence (1.1 kb) contains a tandem of two B1 repeats (positions +880 to +1149 ; underlined in Fig. 1).
Comparison of the human and murine CRIPTO genes
An overall comparison of the murine and human cDNA reveals a similar overall structure (Fig. 2A) consisting of a relatively short coding sequence (564 and 513 nucleotides in human and in mouse, respectively) and a long 3′ untranslated region. The 3′ untranslated regions of both species contain a short repetitive sequence (Alu repeat in the human and B1 repeat in the mouse). Alignment of the nucleotide and aminoacid sequences reveals significant homologies in two regions. The coding sequences (80% overall homology) are very highly conserved in the region encoding the ‘EGF-like’ segment (34/36 matching aminoacids; Fig. 2B). The second region of homology is located at the 3′ end of the non-coding sequences (69.3% homology; Fig. 2C) encompassing an 18-nucleotide identity region around the unusual polyadenylation site (AGTAAA).
A hydrophobicity plot (Kyte and Doolittle, 1982) reveals that the 171-aminoacid murine open reading frame (ORF), unlike its human counterpart, contains a hydrophobic domain in its N-terminal region. This domain is indicative of a signal peptide (Fig. 1B). This result is further confirmed using the algorithm of von Heijne (1986), which predicts a cleavage site between the putative signal peptide and the mature protein. One of the four predicted cleavage sites occurs between aminoacids 17 and 18 of the murine cripto ORF (Fig. 1C).
Temporal and spatial expression pattern of the murine cripto gene
To gain insight into the possible functions of the evolutionary conserved cripto gene, we analyzed a variety of adult tissues and different embryonic stages by both RNase protection and Polymerase Chain Reaction (PCR) analyses.
As shown in Fig. 3A, cripto mRNA is expressed at low levels in spleen, heart, lung and brain (except the cortex) of the adult animal (12- to 50-fold less than in F9 cells; Fig. 3A). No expression is seen in testis, ovary, intestine, liver, stomach, muscle, kidney or seminal vescicles. During embryogenesis, low levels of cripto transcripts are detected in all stages analysed (Fig. 3B).
To study the temporal and spatial expression pattern of the cripto gene, in situ hybridization experiments were carried out. cripto transcripts were first detected during gastrulation. Mouse gastrulation starts at about 6.5 days of development with the formation of the primitive streak at the posterior side of the egg cylinder (Snell and Stevens, 1966). Prior to this, the embryo consists of the primitive ectoderm (or epiblast) and the primary embryonic endoderm. In the prestreak embryo (day 6-6.2), the cripto gene is expressed in a few cells in the region where embryonic ectoderm cells will invaginate to form the primitive streak (Fig. 4A). After onset of primitive streak formation, the cripto gene is expressed in the most posterior part of the embryo where cells are committed to form mesoderm (Figs 4B, 5A). As gastrulation proceeds, the primitive streak extends approximately halfway to the distal tip of the embryo and mesoderm cells are present in the proximal portion of the egg cylinder (Fig. 4h; Dush and Martin, 1992). Later, the mesoderm layer extends around approximately two thirds the circumference of the egg cylinder and the primitive streak has extended to a point that is near the distal tip of the embryo (Fig.4I; Dush and Martin, 1992). cripto mRNA is detected in both the ectoderm and cells of the growing mesoderm during these stages (Figs 4C,D and Fig. 5B,C). In late-streak embryos (Figs 4I, 5G, 6F), the streak extends to the distal tip of the egg cylinder and the head process. The head processes is formed by a group of cells that extends anterior to the primitive streak along the anterior-posterior axis of the embryo and is partially exposed to the yolk sac cavity through a hiatus in the endoderm cell layer (Poelmann, 1981). The pattern of expression of cripto transcripts in late-streak embryos is similar to that seen at the midstreak stage (Figs 4E,F, 5D, 6A). In addition, hybridization signals are also detected in the head process region (Figs 4F, 6A).
During later embryonic development, expression of the cripto gene becomes restricted to the heart whereas all the other mesodermal derivatives are negative (Fig. 6B and data not shown). The cripto gene expression is now very localized in the cells of the developing truncus arteriosus (Fig. 6C). The hybridization signal declines during subsequent developmental stages (Fig. 6D,E). After gestational day 10.5 no cripto transcripts are detected by in situ hybridization. PCR analysis of later stages (Fig. 3B) reveals continued expression of the cripto gene, but so far the expressing cells have not been identified.
DISCUSSION
We have previously reported the isolation of the human EGF-related growth factor gene called CRIPTO (Ciccodicola et al., 1989). It was subsequently demonstrated that multiple copies of this gene are present in both the human and murine genome (Dono et al., 1991). The corresponding murine cripto cDNA was cloned to start studying the expression pattern and possible functions during embryonic and postnatal development.
The murine cripto gene encodes a putative secreted growth factor
The murine cripto cDNA encodes a 171-aminoacid protein containing a domain resembling EGF/TGFα. In contrast to the human protein (Ciccodicola et al., 1989), the murine cripto protein contains a putative signal peptide that is probably cleaved to produce the mature secreted 25×103Mr protein. The central portion of the murine and human proteins are highly homologous, whereas their N-termini are divergent. The regions encoding the EGF-like domains are completely identical with exception of a conservative aminoacid substitution (isoleucine to methionine; Fig. 2B). The most 3′ regions of the human and mouse cripto mRNAs are highly conserved (69.3% homology; Fig. 2B). Usually, the 3′ UTR of eukaryotic mRNAs are not evolutionarily conserved. However, in several cases, all or a portion of the 3′ sequences are conserved, implying a functional role in post-transcriptional regulation. Most eukaryotic mRNAs have a poly(A) tail added at their 3′ ends following termination of transcription. Two cis-acting elements are known to define the site of polyadenylation: a AAUAAA sequence (positioned 20 –30 nucleotides prior to the 3′ end of the mRNA) is followed by a variable GU-rich motif. These sequences represent a minimum consensus required for poly(A) addition. Post-transcriptional gene regulation is controlled by elements upstream to the poly(A) addition site and other elements in trans, which regulate mRNA stability and translation (for recent reviews see Proudfoot, 1991; Wickens, 1990). The conserved parts of the 3′ UTR of the human and murine CRIPTO transcripts contain the poly(A) signals and an 18-nucleotide identity region. Preliminary results suggest an important role for this region in post-transcriptional regulation of CRIPTO mRNA expression or stability.
Is the cripto gene product involved in induction during mesoderm formation ?
In mammals, the mesoderm forms by migration and delamination of the ectoderm cells through a linear structure called primitive streak (Slack, 1991). Recently, various evidences indicate that soluble peptide growth factors function as signaling molecules during mesoderm induction in Xenopus (e.g. noggin, FGFs, activins and Wnts; for reviews see Smith, 1989; Melton, 1991; Smith et al., 1993) and mouse (e.g. TGF-β2 and FGF5; Slager et al., 1991; Hébert et al., 1991). TGF-β2 is expressed in the embryonic endoderm at the pre- and early streak stages (Slager et al., 1991). FGF5 is transcribed in all epiblastic cells (Hébert et al., 1991). Recently, it has been shown that the homeobox gene goosecoid, a marker for the region of Spemann’s organizer, is specifically expressed at the most anterior part of the primitive streak during the early gastrula stage (Blum et al., 1992). This region could, therefore, correspond to the murine homolog of Spemann’s organizer. Candidate molecules involved in regulation of the goosecoid gene could be TGF-β2 and FGF5 which are expressed at the time of goosecoid induction in the gastrulating mouse embryos (Slager et al., 1991; Hébert et al., 1991). But to date it is not clear which one, if any, could provide the endogenous signal.
The data reported here show that the cripto gene, which encodes a putative secreted growth factor, is expressed in mesodermal cells during the early stages of their morphogenetic movements. cripto mRNAs are already present in pre-streak embryos in a patch of cells of the epiblast (Fig. 4A). Subsequently cripto mRNAs are found in the ectoderm at the site of primitive streak formation (early streak stage; Fig. 4B). Cell labeling experiments showed (Lawson et al., 1991) that embryonic mesoderm is derived from all areas of the epiblast with the exception of its distal tip and the region just anterior to it. The expression of the cripto gene during early-streak is confined to the region where early primitive streak cells appear, suggesting that it could be an early marker of mesoderm induction. During later stages, cripto mRNAs are present in all mesodermal cells and in the ectoderm region proximal to the primitive streak (Fig. 4C-F). Interestingly, cripto transcripts are strongly expressed in the cells migrating through the primitive streak (Fig. 5B-D). It is tempting to speculate that cripto proteins are required for the initial formation of the mesodermal germ layer. They could function as molecules directly involved in specifying epiblastic cell fate or in rendering epiblastic cells competent to respond to inductive or positional signals. Therefore, it will be interesting to study whether cripto proteins can affect the expression of genes active during mesoderm formation such as Brachyury (Hermann et al., 1990) and goosecoid (Blum et al., 1992). In view of its known mitogenic activity (Ciccodicola et al., 1989; Ciardiello et al., 1991a), the cripto proteins could alternatively be involved in stimulating cell proliferation during gastrulation. In addition, recently has been demonstrated that the retinoic acid is produced in the Hensen’s node and in the neighbouring primitive streak cells (Hogan et al., 1992). In contrast, cripto is down regulated in human teratocarcinoma cells treated for 24 hours with retinoic acid (Ciccodicola et al., 1989). Recent data in human teratocarcinoma cells show that cripto repression starts at 16 hours after retinoic acid addition (data not shown) and, in vivo, cripto expression disappears about 36 hours after the start of gastrulation. These data could suggest a possible action of the morphogen retinoic acid on cripto repression. However, more information is needed to understand the possible role(s) of the cripto gene during gastrulation.
Possible role of the cripto gene in heart morphogenesis
Analysis of 8.5- to 10.5-day-old mouse embryos revealed surprisingly restricted expression of cripto transcripts in the truncus arteriosus of the developing heart (Fig. 6C-E). Expression in the truncus arteriosus ceases before terminal differentiation occurs (data not shown). The heart differentiates and becomes functional before the other organs. The four chambers of the developing heart form as a result of extensive differential growth and rotation of the immature cardiac tube (and not by active folding; Snell and Stevens, 1966). Therefore, the temporary and highly restricted expression pattern of the cripto gene may be important for appropriate regulation of growth and differentiation of the developing cardiac rudiment.
Interestingly, only the truncus arteriosus, a structure connecting the developing ventricle and the ventral aorta, expresses the cripto gene (Fig. 6B-E). It is possible that the cripto gene products are secreted and therefore the secretion into the circulatory system could distribute the protein(s) to its target site(s) elsewhere in the developing embryos.
This study establishes that the cripto gene, like EGF and TGF-α (Twardzik, 1985; Liscia et al., 1990), is expressed differentially in both the adult mouse and the developing embryo. It has been previously demonstrated that the human CRIPTO mRNA is preferentially expressed in human colorectal carcinoma and not in normal colonic mucosa (Ciardiello et al., 1991b), in human gastric carcinomas (Kuniyasu et al., 1991) but not in normal mucosa. These data suggest that this putative growth factor-like protein may function in an autocrine fashion to regulate tumor cell growth. In this regard, it is worth noting that the human CRIPTO gene can transform NIH-3T3 fibroblast cells and NOG-8 mouse mammary epithelial cells in vitro (Ciardiello et al., 1991a). Its expression pattern during embryonic development further supports the hypothesis that the cripto gene product could be involved in regulation of the two interdependent processes of proliferation and differentiation.
ACKNOWLEDGEMENTS
We are grateful to Dr Françoise Poirier for the generous gift of the cDNA library. We thank Dr Edoardo Boncinelli, David Salomon and Dr Rolf Zeller for their critical reading of the manuscript, Dr Domenico Maglione for implementing the algorithm of von Heijne and Mrs Maria Terracciano for her technical assistance. This work was supported by grants from the ‘Associazione Italiana per la Ricerca sul Cancro’ (AIRC), to A. S. and M. G. P., the ‘Progetto Finalizzato Ingegneria Genetica’ CNR to M. G. P., the ‘Progetto Finalizzato Biotecnologie e Biostrumentazione’ CNR, the ‘Fifth AIDS Project’ of the Ministero della Sanità to A. S and M. G. P. R. D. was supported by an AIRC fellowship.