int-1 - a proto-oncogene involved in cell signalling

In vertebrates, successful development of the primary germ layers and the various organs to which they give rise is critically dependent on cell signalling (see papers in this volume). However, until quite recently, the molecular nature of signalling processes has been a matter for conjecture. In view of the sophisticated molecular and genetic description of Drosophila, it is tempting to draw parallels between Drosophila and vertebrate development. In this paper, we review the evidence that implicates the int-1 gene in cell signalling in both Drosophila and vertebrates.

The int-1 gene was first identified following an analysis of mouse mammary tumour virus (MMTV) induced tumours, which occur quite frequently in certain strains of mice (for review, see Nusse, 1986, 1988). Reasoning that some of these tumours may result from proviral activation of cellular proto-oncognes, Nusse and Varmus (1982) analyzed sites of MMTV integration. The first locus isolated using this approach was termed int-1 (int=integration). Since their initial report, several unrelated genes with similar modes of activation have been described (int-2, Peters et al. 1983; int-3 and int-4, Nusse, 1988). From analysis of many independent integrations, a clear picture of the role of MMTV has emerged. Proviral integration sites are clustered 5’ and 3’ of the int-1 gene with viral transcription in the opposite orientation to that of the int-1 gene (Nusse et al. 1984). Thus, MMTV activation of int-1 (and other genes of this class) presumably results from juxtaposition of MMTV enhancer sequences in close proximity to the int-1 locus. This arrangement leads to expression of int-1 in the mammary gland where, in normal circumstances, int-1 is apparently not expressed (Nusse and Varmus, 1982; Nusse et al. 1984). Importantly, MMTV integration always leaves the normal int-1 coding sequences intact. Therefore, inappropriate expression of the normal int-1 protein is implicated in cellular transformation (van Ooyen and Nusse, 1984). An in vivo role for MMTV directed int-1 expression in mammary tumorigenesis has been directly demonstrated in transgenic mice (Tsukamoto et al. 1988). However, it is clear that int-1 will not transform a broad range of cell types in culture, but only certain mammary epithelial cell lines (Brown et al. 1986; Rijsewijk et al. 1987a). Thus, abnormal expression of int-1 can have profound consequences for the phenotype of a cell, but the outcome depends on the cellular context in which int-1 is expressed.

Analysis of genomic (van Ooyen and Nusse, 1984) and cDNA (Fung et al. 1985) sequences have established that mouse int-1 encodes a 370 amino acid, 41K polypeptide. There are several interesting features of this sequence. The amino terminal 48 amino acids are extremely hydrophobic and probably encode a signal peptide sequence which is cleaved at a potential signal peptidase cleavage site after amino acid 27 (Brown et al. 1987, Fig. 1). The remainder of the protein is highly cysteine rich (6%), with over 50% of these residues in the carboxyl-terminal 20% of the protein (Fig. 1). There are four potential N-linked glycosylation sites, all of which are probably used (Fig. 1, Brown et al. 1987; Papkoff et al. 1987).

Papkoff et al. (1987) have used antisera against int-1 protein to demonstrate by crude cellular fractionation that int-1 protein is sequestered in membrane vesicles, but not detectably secreted. Using another approach, we have examined expression of int-1 in cos cells. We inserted a small oligonucleotide encoding a ten amino acid residue of human c-myc in frame in the int-1 coding sequence (Fig. 2). The int-1 myc protein is specifically recognized by a monoclonal antibody 9E10 (Evan et al. 1985; Munro and Pelham, 1987). Immunohistochemical localization detects int-1 myc protein throughout the endoplasmic reticulum (McMahon, unpublished observations, Fig. 3). A control construct in which chick lysozyme is tagged with myc and the recently identified endoplasmic reticulum retention signals shows a similar localization (Munro and Pelham, 1987, Fig. 3). Similar results are obtained with untagged int-1 protein using anti int-1 antibodies (McMahon, unpublished observations). However, as in the studies of Papkoff et al. (1987), we failed to detect int-1 secreted into the medium (Fig. 4), even under conditions in which significant amounts of the lysozyme-myc protein are inappropriately secreted (Fig. 4). Indirect evidence suggests that this is because int-1 or int-1 myc protein, when expressed at high levels, forms in insoluble precipitates in the endoplasmic reticulum (McMahon, unpublished observation). Taken together, these results suggest that int-1 enters the secretory pathway, is glycosylated, and may, under normal circumstances, be secreted.

The human (van Ooyen et al. 1985), frog (Noorder-meer et al. 1989) and Drosophila (Baker, 1987; Rijse-wijk et al. 1987b; Cabrera et al. 1987) int-1 genes have been cloned and sequenced. Analysis of the predicted protein sequence indicates that even in distantly related species, int-1 is highly conserved. Mouse and human int-1 have 99% amino acid similarity, mouse and frog 68%, and mouse and Drosophila 54%. While human, frog and mouse genes encode similar-sized proteins, Drosophila int-1 contains an additional 85 amino acids (Rijsewijk et al. 1987a). All proteins have a hydrophobic leader sequence that conforms to a consensus signal peptide sequence, and all 23 cysteine residues in the mouse sequence are conserved in human, frog and Drosophila int-1. Thus it is tempting to speculate that the high degree of sequence homology may reflect functional conservation in the normal role of int-1 protein in invertebrate and vertebrate species.

int-1 in Drosophila is encoded by the gene wingless (Rijsewijk et al. 1987b). Viable flies carrying wg mutations were originally identified (Sharma, 1973) by a recessive mutation that causes a homeotic transformation of wing structures into a mirror-image duplication of the notum, another thoracic structure (Sharma and Chopra, 1976; Morata and Lawrence, 1977). Analysis of mosaic clones of wg mutant cells in a wild-type background indicates that wg is non-cell autonomous (Morata and Lawrence, 1977; Wieschaus and Riggle-man, 1987), since small mutant clones appear normal. This suggests that a small patch of wg cells in a wild-type background is rescued by the normal product produced in neighboring cells. With the identification of additional lethal alleles of wg (Babu, 1977; Baker, 1987), it became apparent that wg was involved in the establishment of segmentation early in development. Specifically, wg is a member of the segment-polarity class of genes which are required for normal pattern in each segment (Nüsslein-Volhard and Wieschaus, 1980). Loss of wg and other segment-polarity genes results in deletion of a part of the normal pattern, and its replacement by a mirror-image duplication of anterior denticle bands (Nüsslein-Volhard and Wieschaus, 1980; Wieschaus and Riggleman, 1987). Thus, naked cuticle is replaced by additional denticles.

The cloning of wg (Baker, 1987; Rijsewijk et al. 1987b) has provided clues as to how the gene is regulated and to its role in segmentation. By the early gastrula stages, wg RNA expression is detected in a series of approximately 14 stripes in the embryonic epidermis (Fig. 5, Baker, 1987; Rijsewijk et al. 1987b). With the establishment of parasegmental grooves at the extended germ band stage, it is apparent that stripes of wg expression are located in the posterior one quarter of each parasegment, in what will become the posterior region of the compartment which consists of each anterior half-segment (Fig. 5, Baker, 1987). Thus, although wg RNA is located in only a small region of the presumptive segmental unit, loss of wg expression affects large areas of the segment. Taken together with the non-cell autonomy of wg clones in mosaics, these data strongly suggest that the wg protein product is involved in signalling between cells that establishes normal segmental pattern. This conjecture is supported by the analysis of the role of wg in regulating another segmentation gene, engrailed (en).

en is required for specification of posterior developmental compartments (Kornberg et al. 1985). Examination of the expression of wg and en indicates that they have similar but non-overlapping expression in each parasegment, wg is located immediately anterior to the parasegmental groove at extended germ band stages, whereas en is expressed in the adjacent cells that lie posterior to the groove (Baker, 1987). In wg null mutants, normal expression of en is established. However, in the absence of wg, en expression decays prematurely shortly after germ band elongation (Fig. 5, DiNardo et al. 1988), indicating that wg is required for the maintenance or the reinitiation of en expression in adjacent cells (DiNardo et al. 1988). Recently, secretion of int-1 protein and its uptake by en expressing cells has been directly observed in Drosophila embryos (Nusse, personal communication). Thus, the evidence suggests that wg is a secreted protein able to influence the establishment of segmental pattern, at least in part by regulation of en expression.

Normal int-1 expression has been best characterized in the mouse but, to date, observations have been confined to RNA expression. Int-1 is not expressed in detectable amounts in any adult tissue examined except the testis (Jakobovits et al. 1986; Shackleford and Varmus, 1987). Here expression is localized in postmeiotic, round spermatids (Shackleford and Varmus, 1987). During development, int-1 expression is restricted to the developing neural plate and neural tube (Wilkinson et al. 1987; Shackleford and Varmus, 1987). An extensive series of in situ hybridization experiments has provided a detailed picture of int-1 expression during mouse development (Wilkinson et al. 1987). At midgas-trulation stages, prior to neural plate formation, int-1 RNA is not detected. One day later the fetus (∼4 –5 somites) has elevating head folds and a neural plate extending caudally. In the presumptive brain regions, int-1 expression is widespread over the anterior neural folds, but becomes localized to the tips of the neural folds more posteriorly. More caudal regions of the neural plate do not express int-1 at this time. Following neural tube closure, int-1 is localized at the dorsal midline of the neural tube, in presumptive mid-brain and along the developing spinal cord (Fig. 6). In the hind brain, int-1 is expressed at the edge of the rhombic lips at the junction of the pseudostratified neuroepithelium and the thin roof of ependymal cells that cover the hind brain. Thus, the expression of int-1 in the hind brain may represent the equivalent cells that express int-1 in other regions that are not covered by this ependymal layer. Several small patches of int-1 RNA are detected in more rostral regions, but the predominant site of expression is at the dorsal midline. Expression is still detected at the dorsal midline of the developing spinal cord at 16 · 5 days (Wilkinson & McMahon, unpublished observation), and may be expressed beyond this time. However, the cells expressing int-1 are a non-mitotic population and, as the neural tube expands during development, elongation and flattening of these cells make it difficult to detect low levels of RNA. Essentially, the pattern of int-1 expression established by 10 · 5 days is maintained for several more days of fetal development.

Expression of int-1 during frog development has been examined by Northern blot analysis (Nordermeer et al. 1989). Expression was first detected at the neurula stage, consistent with a role for int-1 in frog neural tube development. However, in situ hybridization will be required to provide a more detailed analysis of RNA localization.

Expression of int-1 in the mouse at the dorsal midline correlates with several aspects of neural tube development. The dorsal midline is generated by the fusion of the neural folds, and it is from this region that the neural crest is derived. However, it seems unlikely that int-1 is involved in either of these events. Neural tube closure occurs at caudal positions in the 9 ·5 day embryo, preceding the rostral caudal extension of int-1 expression into this region (Wilkinson and McMahon, unpublished observation). Further, neural crest cells arise from the dorsal midline between 8-5 and 10-5 days of development (Tan and Morriss-Kay, 1985), whereas int-1 expression continues in this region several days after their migration, int-1 is not delectably expressed in migrating neural crest cells (Wilkinson et al. 1987; Wilkinson and McMahon, unpublished observation).

A role for int-1 in neuronal migration seems plausible as the evidence presented so far indicates that int-1 is probably secreted and may alter gene expression in responsive cells. In this respect, it is noteworthy that substrate pathways for neuronal migration have been proposed to guide axonal migration in the dorsal columns that flank the areas of int-1 expression (Katz et al. 1980; Katz and Lasek, 1981; Willis and Cogges-hall, 1978).

An alternative view of the role of int-1 in vertebrates, analogous to its role in Drosophila, is to suggest that int- 1 is involved in specifying position about the midline. As illustrated in Fig. 6, int-1 expression occurs symmetrically around the dorsal-ventral axis of symmetry. Thus, cells left or right of the axis of symmetry would see equivalent levels of secreted int-1 protein and presumably respond in the same fashion, in an attempt to address the role of int-1 in vertebrate development, we have recently begun to manipulate expression in vivo. These initial experiments clearly indicate that inappropriate expression of int-1 has a dramatic and rather unexpected effect on the normal axial specification of Xenopus embryos (McMahon and Moon, in preparation).

We have used the approach described by Harvey and Melton (1988) to express mouse int-1 in Xenopus embryos ectopically (McMahon and Moon, 1989). Fertilized eggs were injected with int-1 RNA and their subsequent development was monitored. Almost all embryos injected with int-1 RNA, but none injected with a variety of control RNAs, developed with a specific phenotype. At early neurula stages, embryos show a bifurcation of their anterior neural tube as a result of the duplication of the embryonic axis (Fig. 7). Therefore, ectopic expression of int-1 seems to interfere with the machinery for axial specification, which normally produces only one embryonic axis. This result is surprising, as it implies that int-1 is influencing processes operating during early gastrulation, which precedes normal int-1 expression in both the mouse and frog (Wilkinson et al. 1987; Noordermeer et al. 1989). However, it strongly suggests that inappropriate expression of int-1 can interfere with positional signalling in vertebrate development and suggests that int-1 normally operates in some aspect of this.

If int-1 is secreted and acts to regulate gene expression in responsive cells, what is the pathway of int-1 action? By analogy with growth factors, int-1 might be expected to operate through a receptor-mediated pathway. At present, there is no conclusive evidence for an int-1 receptor but, from the action of wg in Drosophila, a clear prediction can be made. Null mutants in an int-1 receptor should have a similar phenotype to segmentpolarity mutants. However, unlike wg mutants which are non-cell autonomous, receptor mutants should exhibit cell autonomy. Several mutations have been described that resemble wg (Wieschaus and Riggleman, 1987; Perrimon and Mahowald, 1987), some of which are cell autonomous (Wieschaus and Riggleman, 1987). Clearly, this aspect of the action of int-1 will be under intense scrutiny in the future.

A major goal for developmental biologists is to elucidate those mechanisms that are shared by organisms separated by millions of years of evolution. We have attempted to draw parallels between the role of int-1 in positional signalling in Drosophila and its role in vertebrate development. Although at present we have little more than a rudimentary understanding of int-1 function in vertebrates, the future looks promising. The ability to manipulate int-1 expression in the frog provides a rapid and simple means by which ideas regarding the function and properties of int-1 protein may be tested. Similar experiments are also possible in transgenic mice and some are under way. However, perhaps the single most exciting possibility is the generation of homozygous mouse embryos which lack int-1 function (Thomas and Cappechi, 1987; Mansour et al. 1988). These experimental approaches together should provide conclusive evidence for or against a conserved role for int-1 in positional signalling.

We would like to thank Janet Champion for the int-1 myc expression studies, Roel Nusse and Olivier Destrée for communication of data prior to publication, Hugh Pelham for the gift of the lys-myc construct, Gerald Evan for the antibody 9E10, Jeff Mann for critical reading of the manuscript, and Sharon Perry for the preparation of this manuscript.