Endocrine-Related Cancer ( 2000 ) 7 199 – 226 The EGF-CFC family : novel epidermal growth factor-related proteins in development and cancer

The EGF-CFC gene family encodes a group of structurally related proteins that serve as important competence factors during early embryogenesis in Xenopus, zebrafish, mice and humans. This multigene family consists of Xenopus FRL-1, zebrafish one-eyed-pinhead (oep), mouse cripto (Cr-1) and cryptic, and human cripto (CR-1) and criptin. FRL-1, oep and mouse cripto are essential for the formation of mesoderm and endoderm and for correct establishment of the anterior/ posterior axis. In addition, oep and cryptic are important for the establishment of left-right (L/R) asymmetry. In zebrafish, there is strong genetic evidence that oep functions as an obligatory co-factor for the correct signaling of a transforming growth factor-β (TGFβ)-related gene, nodal, during gastrulation and during L/R asymmetry development. Expression of Cr-1 and cryptic is extinguished in the embryo after day 8 of gestation except for the developing heart where Cr-1 expression is necessary for myocardial development. In the mouse, cryptic is not expressed in adult tissues whereas Cr-1 is expressed at a low level in several different tissues including the mammary gland. In the mammary gland, expression of Cr-1 in the ductal epithelial cells increases during pregnancy and lactation and immunoreactive and biologically active Cr-1 protein can be detected in human milk. Overexpression of Cr-1 in mouse mammary epithelial cells can facilitate their in vitro transformation and in vivo these Cr-1–transduced cells produce ductal hyperplasias in the mammary gland. Recombinant mouse or human cripto can enhance cell motility and branching morphogenesis in mammary epithelial cells and in some human tumor cells. These effects are accompanied by an epithelial-mesenchymal transition which is associated with a decrease in β-catenin function and an increase in vimentin expression. Expression of cripto is increased several-fold in human colon, gastric, pancreatic and lung carcinomas and in a variety of different types of mouse and human breast carcinomas. More importantly, this increase can first be detected in premalignant lesions in some of these tissues. Although a specific receptor for the EGF-CFC proteins has not yet been identified, oep depends upon an activin-type RIIB and RIB receptor system that functions through Smad-2. Mouse and human cripto have been shown to activate a ras/raf/MAP kinase signaling pathway in mammary epithelial cells. Activation of phosphatidylinositol 3-kinase and Akt are also important for the ability of CR-1 to stimulate cell migration and to block lactogenic hormone-induced expression of β-casein and whey acidic protein. In mammary epithelial cells, part of these responses may depend on the ability of CR-1 to transactivate erb B-4 and/or fibroblast growth factor receptor 1 through an src-like tyrosine kinase. Endocrine-Related Cancer (2000) 7 199–226 Online version via http://www.endocrinology.org 1351-0088/00/007–199  2000 Society for Endocrinology Printed in Great Britain Salomon et al.: The EGF-CFC family The EGF-CFC gene family A new family of genes, the EGF-CFC family, that encodes extracellular growth factor-like and/or co-receptor-like proteins has been identified in the human, mouse, frog and zebrafish. Members of this family include human cripto-1 (CR-1) (also known as teratocarcinoma-derived growth factor-1 (TDGF-1)) and criptin (unpublished Human Genome Sciences Inc. patent number S5981215), mouse cripto-1 (Cr-1/tdgf-1) and cryptic,Xenopus FRL-1 and zebrafishone-eyed pinhead(oep) (Table 1) (Ciccodicolaet al. 1989, Donoet al.1993, Kinoshitaet al.1995, Shenet al. 1997, Zhanget al. 1998). CR-1 was serendipitously isolated in a screen for glucose-6–phosphate dehydrogenase as a chimeric cDNA from a human NTERA2/D1 embryonal carcinoma cDNA library (Ciccodicolaet al. 1989). The mouse gene, Cr-1, was subsequently identified and cloned from a mouse embryo cDNA library (Dono et al.1993). The closely related mouse cryptic gene was cloned by differential RT-PCR from a mouse embryoid body-derived mesoderm cDNA library, and criptin from a human pancreatic carcinoma cDNA library (Shenet al. 1997). Similarly, Xenopus laevisFRL-1 was isolated in a functional screening in yeast with the intent of identifying fibroblast growth factor receptor-1 (FGFR-1) activating genes from a Xenopus embryonic mesoderm cDNA library (Kinoshita et al. 1995). More recently, the zebrafish oepgene has been positionally cloned and identified (Zhang et al. 1998). Genomic organization of mouse and human cripto CR-1 maps centromerically to a region on chromosome 3p21.3 that is adjacent to or possibly part of a region which Table 1 EGF-CFC family and their function. EGF-CFC proteins Function Human Cripto-1 (CR-1) Gastrulation/mammary morphogen Cryptin ? Mouse cripto-1 (Cr-1) Primitive streak A/P axial formation and positioning Mesoderm and endoderm formation Cardiomyocyte development cryptic L/R symmetry Cardiac defects/right isomerization Postnatal lethality Xenopus FRL-1 Mesoderm formation Neuroectoderm formation Zebrafish one-eyed pinhead A/P axial formation and positioning (oep) mesoderm and endoderm formation Ventral neuroectoderm L/R symmetry 200 www.endocrinology.org is frequently deleted or exhibits loss of heterozygosity (LOH) in a subpopulation of head and neck, renal, gastric, bladder, breast and lung carcinomas (Dono et al.1991, Sacconet al. 1995, Toddet al. 1996, Sekidoet al. 1998, Cuthbertet al. 1999). Interestingly, theβ-catenin gene (CTNNB1) is located at the chromosomal region 3p21.3–p22 while the activin receptor IIB is located at 3p22 (Kraus et al.1994, Bondestam et al. 1999). These two genes may be involved in intracellular signaling by EGF-CFC proteins (see below). The mouse and human cripto genes consist of six exons and five introns and possess inverted Alu and B1 sequence elements respectively, and AUUU(A)-type Kamen-like sequences in a large 3 ′ untranslated region, suggesting that they encode relatively short-lived mRNA species (Fig. 1) (Donoet al.1991, 1993). The human CR-1 coding sequence containing the six exons is 4.8 kb in length. There is an excellent conservation of the exon-intron structure in the region of exon 4 which contains the EGF-like motif, while exons 1 and 3 of the mouse are 15 and 33 nucleotides shorter respectively than the corresponding human exons (Dono et al. 1991, 1993). The 5 ′ upstream genomic sequences of mouse Cr-1 and human CR-1 between -610 and -1 from the most distal translation start sites are quite dissimilar (Baldassarreet al. 2000). Several TATA and CAAT boxes are present in the mouse Cr-1 promoter region while in the human CR-1 promoter these sequences are missing. These data suggest that regulation of expression of the mouse and human cripto genes may not be entirely similar. Multiple copies of cripto-specific sequences are present in the human and mouse genomes (Dono et al. 1991, Sacconet al. 1995, Liguori et al.1996, 1997, Scognamiglio et al.1999). At least five other human CR-1–related pseudogenes and two mouse Cr-1 pseudogenes have been identified (Scognamiglio et al. 1999). The CR-2, CR-4 and CR-5 genes are truncated at the 5′-end and have accumulated point mutations, deletions and insertions (Scognamiglioet al. 1999). These genes map to chromosomes 2q37, 6p25 and 3q22 respectively while CR-6 maps to 19q13.1. The CR-3 pseudogene which maps to the Xq21–q22 region and the mouse Cr-2 pseudogene are intronless genes that have many characteristics of a retroposon but nevertheless have the potential to code for functional proteins that differ from the proteins encoded by either CR-1 or Cr-1 by only five amino acids. CR-1 and Cr-1 encode major mRNA species of approximately 2.2 kb. In some cases, less abundant transcripts of about 1.7, 3.0, 3.2 and 3.5 kb have been detected in midgestation mouse embryos and in primary and metastatic human colon and hepatic carcinomas, suggesting that theymay arise by the use of different polyadenylation sequences, by alternative splicing or by the use of an alternative initiation site for transcription (Ciccodicolaet al. 1989, Donoet al. 1993, Johnsonet al.1994, Baldassarret al.2000). In this regard, a truncated CR-1 protein of 145 amino acids may be expressed from the 1.7 kb mRNA transcript in metastatic human Endocrine-Related Cancer (2000) 7 199–226 Figure 1 (A) Schematic diagram of the human CR-1 gene and mRNA. Hatched areas represent the six exons, with the red exon encoding EGF-like motif. Numbers represent the number of bases in each region (obtained from Baldassarre et al. 2000). (B) Amino acid sequence alignment of cripto-related proteins, zebrafish oep, mouse cripto, mouse cryptic and Xenopus FRL-1. Green and yellow bars indicate amino acid identity among 4/4 and 3/4 proteins respectively. Blue and red underline represent EGF-like motif and cysteine-rich domains respectively (data obtained from Zhang et al. 1998 and reproduced with permission of the author). (C) Sequence alignment of EGF-like motifs. Green and red shaded areas represent amino acid identity in three or more proteins and in six proteins respectively. Conservation of six cysteines noted in yellow with disulfide bonds between cysteines (Cys) 1 and 3, 2 and 4, and 5 and 6 constitute loops A, B and C respectively. Note absence of residues between Cys 1 and 2 in cripto-related proteins compared with mouse EGF. colorectal carcinomas and in hepatic colon metastases due to the use of a second CUG initiation codon (at leucine 44 in the coding sequence) that eliminates the use of the first two exons and thereby deletes 43 amino acids from the N-terminus of the protein. In the mouse, low levels of Cr-1 mRNA expression can be detected by RNAse protection assays in the adult spleen, heart, lung and in distinct regions of the brain (Dono et al. 1993, Johnsonet al.1994). Adult tissues in the human that express low levels of mRNA transcripts for the 188 amino acid isoform of CR-1 as detected by RT-PCR include