Cell transformation by the middle T-antigen of polyoma virus - Nature

ã

Oncogene (2001) 20, 7908 ± 7916 2001 Nature Publishing Group All rights reserved 0950 ± 9232/01 $15.00 www.nature.com/onc

Cell transformation by the middle T-antigen of polyoma virus Natalia Ichaso1 and Stephen M Dilworth*,1 1

Department of Metabolic Medicine, Imperial College School of Medicine, Hammersmith Hospital, Du Cane Road, London W12 0NN, UK

The polyoma virus region expressed early in the lytic cycle encodes three proteins, or T-antigens, that together cause the infected cell to enter the cell cycle and so provide a suitable cellular environment for replication of the viral genome. Under some circumstances infection does not kill the cell, but the T-antigens are still produced, resulting in the cell becoming transformed and tumorigenic. Most of this transforming action is exerted by the middle T-antigen, which has the ability to convert established cell lines to an oncogenic state. Middle T is a membrane bound polypeptide that interacts with a number of the proteins used by tyrosine kinase associated receptors to stimulate mitogenesis, so MT can be considered as a permanently active analogue of a receptor. Through a de®ned series of interactions, MT assembles a large multi-protein complex at the cell membrane, consisting of MT, the core dimer of protein phosphatase 2A, an src-family tyrosine kinase, and via phosphotyrosines, ShcA, phosphatidylinositol (3') kinase, and phospholipase Cg-1. Tyrosine phosphorylation stimulates PI3K and PLCg-1 enzymatic activity, and on ShcA creates binding sites for Grb2 with its associated Sos1 and Gab1. This activates p21ras, and hence, the MAP kinase cascade. Consequently, MT can be used as a model for studying cell transformation and growth factor receptor signalling pathways. Oncogene (2001) 20, 7908 ± 7916. Keywords: middle T-antigen; cell transformation; signalling pathways; tyrosine kinases; PP2A Introduction In 1953, Gross reported that a cell free ®ltrate from spontaneously formed mouse leukaemia cells contained an infectious agent that could cause parotid gland tumours when inoculated back into mice (Gross, 1953). Procedures were soon established to grow this agent in culture, and it was found that mice injected with large amounts developed a variety of tumours, leading to it being named polyoma virus (PyV) (Eddy et al., 1958). This was one of the ®rst demonstrations that a DNA virus could cause tumours. Since then, PyV has become a popular model for studying viral replication and cell transformation. Despite being classi®ed as a tumour virus, PyV exists as a lytic virus in most wild mouse *Correspondence: SM Dilworth; E-mail: [email protected]

populations, and causes few harmful eects. Like most viruses, PyV encodes a potent activity that stimulates an infected quiescent cell to enter the cell cycle, which provides a suitable environment for replication of the viral genome and production of viral progeny. However, when expression of these mitogenic proteins is uncoupled from the lytic cycle, their activity is potentially tumorigenic. It is probably the ease with which these proteins can be expressed outside a lytic cycle that separates the tumour viruses from the nontumour inducing viral types. For PyV, expression of the T-antigens without cell lysis occurs under a number of circumstances; for example, during infection of rat cells, when viral DNA replication fails as a consequence of the inability of PyV large T-antigen (LT) to bind rat DNA polymerase a. In a small number of these abortively infected cells the viral DNA becomes integrated into the host genome, and stable expression of the T-antigens occurs. PyV is a small, double stranded, closed circular DNA virus that is closely related to similar viruses isolated from hamsters, rhesus monkeys, African green monkeys, bovines, budgerigars, and humans. All have a similar structure (Figure 1) with the approximately 5 kb genome divided into two regions that are transcribed either early or late in the lytic cycle. The late expressed area produces the viral capsid proteins, whereas the early region encodes the polypeptides that are required for viral replication, and are responsible for the transforming properties of each virus, the Tantigens. The early region encodes at least two proteins, a large and small T-antigen (LT and ST) created by alternate splicing patterns. What sets the hamster and murine polyoma viruses apart from the others is the synthesis of a third protein, the middle Tantigen (MT), and it is this polypeptide that is responsible for the enhanced transforming properties of these two viruses. Molecular techniques now allow each T-antigen to be expressed independently, and this has shown that the murine and hamster T-antigens have separate, complementary transforming activities. LT cannot transform cells in culture, but has the ability to `immortalize' primary cells (Rassoulzadegan et al., 1983). MT, however, is able to transform established cells to a fully transformed, tumorigenic phenotype (Treisman et al., 1981), but cannot overcome the senescent properties of primary cells. The combination of these two activities (plus some enhancing contributions from ST) is responsible for

Polyoma virus middle T-antigen N Ichaso and SM Dilworth

Figure 1 A schematic map of the polyoma virus genome. The closed circular DNA of polyoma virus is shown, and the regions transcribed into mRNA indicated around the outside. The coding regions are represented by boxes, with the name of the protein produced marked within, the mRNA non-coding regions are designated by solid lines, and the sequences removed by splicing in broken lines. The early and late regions are marked on the inside of the circle, together with the number of kilobases of DNA. Also marked are the positions of the regions required for enhancer activity of early gene transcription, and the origin of DNA replication

the eective tumorigenic properties of each virus (Rassoulzadegan et al., 1982). Here we will examine just the transforming actions of MT. There have been a number of recent reviews that document fully how the role of MT has been uncovered (Dilworth, 1995; Dunant and Ballmer Hofer, 1997; Kiefer et al., 1994; Messerschmitt et al., 1997; Nicholson and Dilworth, 2001), so we will limit ourselves here to presenting only the current position regarding MT function, and then concentrate on some of the areas where uncertainty still exists. We apologize to the many authors whose excellent early work has not been referenced because of space constraints. MT Properties MT is a 421 amino acid protein that is a highly eective oncogene that transforms established cells in culture, and can cause a variety of tumours in animals, even without expression of LT or ST (reviewed in Kiefer et al., 1994). The majority of these in vivo tumours are endothelial in origin, but targeted expression can be used to form tumours in a variety of cell types, particularly breast tissue (Dankort and Muller, 2000). Most DNA tumour viral oncogenes do not have any intrinsic enzymatic activity of their own. Instead they bind to cellular proteins and alter their regulation. Most of these interactions mimic the mechanisms used by the cell

to control these proteins, but in a fashion that cannot be restricted by other cellular factors. Consequently, the targets of DNA viral oncogenes are usually the key proteins that control cell cycle progression, and much can be learnt from examining how viruses interact with, and alter the activities of, these proteins. In the case of MT, the proteins targeted (Figure 2) are those involved in the signalling pathways used to promote cell growth (or dierentiation induction, etc) by activated tyrosine kinase associated receptors (TKRs). Consequently, in many respects, MT can be considered as a permanently active analogue of a growth factor receptor. Hamster PyV MT (HamMT) has dierent oncogenic properties in vivo, inducing a range of lymphoid tumours and skin epitheliomas (through infection with the intact virus, or transgenic expression), although it can still transform ®broblasts in culture (Scherneck and Feunteun, 1990). However, the two proteins seem to work in a similar fashion, so murine MT will be described here, and the dierences with HamMT noted at appropriate places.

7909

Interaction between MT and cellular components MT associates with cellular components in a de®ned series of binding reactions (summarized in Figures 2 and 3) that are responsible for the transforming eect, so each interaction will be considered in the order that they probably occur in the cell. Membrane binding MT is associated with most cellular membranes (Dilworth et al., 1986), though localization to a precise perinuclear membrane site may be an initial step (Zhu et al., 1998). MT is probably synthesized on free polysomes, and then attaches to membranes via a 22 residue stretch of hydrophobic amino acids found near its C-terminal end. Flanking this region are groups of basic residues that also in¯uence membrane location (Dahl et al., 1992; Elliott et al., 1998). Without these hydrophobic amino acids, MT no longer binds to membranes, fails to associate with most of the cellular proteins that it normally binds, and does not transform. However, this hydrophobic region does more than simply mediate membrane interaction, as replacement with the membrane targeting domains of vesicular stomatitis virus glycoprotein G (Templeton et al., 1984), cytochrome b5 (Zhu et al., 1998), or the myristylation signal of pp60c-src (Elliott et al., 1998) all locate MT to a membrane site, but fail to restore transformation. In addition, mutations within the hydrophobic domain can also abolish transforming activity without disrupting membrane binding (reviewed in Markland and Smith, 1987), and a signi®cant homology exists between the regions in murine and hamster MT. Although one could Oncogene


7910

Figure 2 The proteins found associated with MT, and the sequences recognized. MT is shown schematically, with the region common with ST shown shaded, and the sequences unique to MT in solid. The proteins bound to MT are marked above and below MT spanning the amino acid sequences required for each interaction. The CXCXXC motifs (C3) required for PP2A binding are also marked. The position of the hydrophobic amino acids required for membrane association are shown on the right

PP2A association

Figure 3 The sequence of binding reactions between MT and cellular proteins. The series of interactions between MT and host proteins are shown in square boxes, and the results of activating each signalling pathway in oval shapes. To the side of each arrow are the events required to promote each event

imagine reasons why a speci®c sequence is required, such as targeting precise cell membrane sites, or the need to interact with proteins within the membrane environment, so far little information about the role of this region has been found. Interestingly, the CAAX acylation motif from H-ras has been found to functionally replace the hydrophobic region (Elliott et al., 1998), so suggesting other targeting sequences can perform the same function. Oncogene

All of the ST species encoded by polyoma virus family members interact with the core dimer of protein phosphatase 2A (PP2A), which consists of the 35 kDa catalytic (C) polypeptide and the regulatory 65 kDa (A) component. As the N-terminus of MT is essentially ST, not surprisingly, MT also binds PP2A. AC (Pallas et al., 1990; Walter et al., 1990). Mutations of a number of residues in MT that are highly conserved within the viral ST family, notably a CXCXXC motif found twice in the centre of the region, eliminate binding to PP2A. MT interacts with the HEAT repeats 2 to 8 of PP2A.A (though the presence of the catalytic subunit of PP2A can further improve binding) (Ruediger et al., 1992). It is still not clear what eect MT has on the phosphatase activity of PP2A, with some studies suggesting it is inhibited, and others stimulated. It is unlikely that an overall inhibition of phosphatase activity within the cell occurs, as MT would have to bind nearly all of the PP2A available to have a signi®cant eect, and this is clearly not the case. Inhibition of a speci®c (membrane bound?) class of PP2A is possible, but there appears to be no change in membrane bound phosphatase activity in MT expressing cells (Ulug et al., 1992), so again this is improbable. PP2A normally consists of a trimer, an AC core dimer plus a variable B subunit that is thought to in¯uence substrate speci®city and subcellular distribution of the phosphatase (Mumby, 1995; Janssens and Goris, 2001). If MT is acting as a virally encoded B subunit, it is feasible that substrate speci®city is altered as a consequence of binding to MT. However, there is no evidence yet that the phosphatase activity of PP2A associated with MT acts on any protein outside, or even inside, the MT complex. Some studies even suggest that the phosphatase activity is not required to form a MT complex (Ogris et al., 1999). What is clear is that all MT mutations that aect PP2A association also prevent interaction with most of the other proteins that bind wild type MT, with the


exception of Hsc70, and possibly 14-3-3 (see later). These MT mutants also show a complete loss of transforming activity. Recently, our group has found that PP2A binding also in¯uences the membrane location of the complex (Brewster et al., 1997). MT is usually found in all internal membranes, with a small amount reaching the peripheral cell membrane. Disruption of PP2A binding alone prevents MT reaching the cell periphery. Exactly how this occurs, and whether it is important for transformation, remains to be shown. MT mutants that just bind PP2A (Brewster et al., 1997) should enable the role of this interaction to be examined further. Hsc70 binding Mutants in the N-terminal half of MT that fail to bind PP2A associate instead with a prominent 70 kDa polypeptide. This protein has been identi®ed as Hsc70 (Pallas et al., 1989; Walter et al., 1987), and a region of MT with homology to the DNA J protein of E. coli, including a highly conserved HPDKGG sequence, is responsible for this association. Hsc70 can act as a molecular chaperone (reviewed in Helmbrecht et al., 2000), so may be involved in assembling the MT containing, multimeric complex. If this is the case, Hsc70 interaction with MT mutants that fail to bind PP2A may represent a trapped intermediate of the assembly process. However, as this region is shared with both LT and ST, it is also possible that the J domain is required for the function of the other Tantigens, and has no role in MT. Deletion of the HPDKGG motif in MT has a small (our unpublished observations), or no (Campbell et al., 1995; Glenn and Eckhart, 1995), eect on MT induced foci formation, so interaction with Hsc70 is not critical for transformation of cultured cells. src-Family binding PyV MT is associated with three src-family members in ®broblasts, pp60c-src (Courtneidge and Smith, 1983), pp62c-yes (Kornbluth et al., 1987), and pp59c-fyn (Cheng et al., 1988a; Horak et al., 1989; Kypta et al., 1988), though only one polypeptide is bound to each MT molecule. All of the associations seem to occur in the same manner, so pp60c-src will be used generically. Not all of the MT polypeptides within the cell interact with pp60c-src; the amount present in a complex varying between 5 and 50% depending on the cell line used. Whether there is a dierence between the MT molecules that bind pp60c-src and those that don't is not clear, though some reports suggest that the amount of complex formation is promoted by PKC activation, so phosphorylation may be involved in establishing the association (Matthews and Benjamin, 1986). It seems likely that prior association with PP2A is required for MT to bind pp60c-src, as the triple complex contains equimolar amounts of MT, PP2A and src-family polypeptides (Glover et al., 1999), and all MT mutants that disrupt PP2A binding also fail to interact with

pp60c-src. A region of MT (amino acids 185 to 210) containing the N-terminal part of the MT unique sequences, is also required for pp60c-src binding, but has no eect on the interaction with PP2A (Brewster et al., 1997). Point mutagenesis has shown that two motifs in this area are involved in pp60c-src association, each containing a basic patch followed by either a serine or threonine residue (Glover et al., 1999). To interact with pp60c-src, the serine and threonine probably have to be unphosphorylated, so it is tempting to speculate that the PP2A bound to MT is required to keep these residues in this state. However, there is no evidence so far that this region ever becomes phosphorylated in cells. The development of a system that recreates the MT-PP2A-pp60c-src complex in vitro has established that the ®rst 220 amino acids of MT alone can interact with pp60c-src (our own unpublished data). In cells, such an MT fragment does not bind pp60c-src, probably because it fails to associate with membranes, and as a consequence, does not locate to a site where pp60c-src is present. Such limitations do not exist in vitro. Most interactions with pp60c-src occur through its SH2 and SH3 domains (Erpel and Courtneidge, 1995). However, association with MT is unusual in that it requires just the kinase domain plus a small part of the tail region of pp60c-src (Dunant et al., 1996, and our own unpublished observations), though src-kinase activity is not required. Association between MT and just the kinase plus tail fragment occurs more eciently than with the whole pp60c-src molecule, suggesting the SH2 and SH3 domains can hinder binding. Within the tail region, the regulatory tyrosine phosphorylation site, Y527, is not required (Cheng et al., 1988b), but the region linking the kinase domain to this residue is essential. This area is highly conserved within all members of the src-family, so it is feasible that this sequence may also interact with a cellular protein that regulates pp60c-src activity. Recently, the active forms of the G proteins Gas and Gai have been found to bind to the kinase domain of pp60c-src, and to stimulate its kinase activity (Ma et al., 2000). This raises the exciting prospect that MT may have utilized this host regulatory mechanism to stimulate src-family tyrosine kinase activity. Although the mechanisms through which pp60c-src binds to MT have still not been de®ned, the eects of the association on kinase activity are clear. Inside normal cells, pp60c-src is largely in an inactive conformation, Y527 is phosphorylated, and an internal association between P-Y527 and the SH2 domain inhibits kinase activity (Xu et al., 1997). Once bound to MT, Y527 is dephosphorylated, Y416 is phosphorylated (probably autocatalytically) and kinase activity is increased approximately 20-fold. It has not yet been determined whether Y527 is dephosphorylated by the PP2A bound to MT, or whether it is exposed to a cellular tyrosine phosphatase as a consequence of association with MT. It is also not known whether MT binds to the active form of pp60c-src and prevents it from being inactivated, or if it can activate a preexisting inactive pp60c-src molecule. However, we have

7911

Oncogene


7912

shown in vitro that addition of excess csk (the kinase that phosphorylates Y527) makes little dierence to the formation of a MT-pp60c-src complex (unpublished observations). It seems feasible, then, that MT can directly activate repressed pp60c-src. Once activated, the associated pp60c-src phosphorylates MT on a number of tyrosine residues, all of which are found in the C-terminal unique region between residue 220 and the C-terminus (unpublished observations). Three major sites have been identi®ed, Y250, Y315 and Y322. Pp60c-src also phosphorylates most of the other polypeptides that associate with these sites (see below). It has not been established whether cellular proteins outside the complex are also critical substrates of the MT bound pp60c-src activity, though it is clear that a number of cellular proteins do become tyrosine phosphorylated in cells expressing MT (Yonemoto et al., 1987). However, this may not be as a result of direct phosphorylation by MT associated pp60c-src. HamMT is dierent to PyV MT in its src-family member binding pro®le, associating mainly with pp59c-fyn (Courtneidge et al., 1991). There is no homology between the region of PyV MT required to bind pp60c-src and any area of HamMT, suggesting a dierent binding strategy is used. It is also intriguing to note that whereas binding to murine MT activates pp60c-src and pp62c-yes, it does not change pp59c-fyn activity. It is possible, then, that this re¯ects a dierence in the normal regulation of pp59c-fyn that HamMT has overcome by employing an alternate strategy. The SH2 domain of pp59c-fyn is required to associate with HamMT, and mutation of HamMT Y324 reduces binding to a low level (Dunant et al., 1997). This suggests that at least in part the binding and activation of pp59c-fyn by HamMT occurs via an SH2 domainphosphotyrosine interaction. However, there is additional evidence that the N-terminal part of HamMT is also involved (Goutebroze et al., 1997). ShcA binding The ShcA family of proteins bind to sequences surrounding Y250 of MT (Campbell et al., 1994; Dilworth et al., 1994). Although both the PTB and SH2 domains of ShcA are able to interact with this region of MT in vitro, it is the PTB domain that appears to mediate interaction with this NPTY motif in vivo (Blaikie et al., 1997). Both the p66 and p52 forms of ShcA bind to MT, but the p42 species interacts poorly (Dilworth et al., 1994). Similar to the position with ShcA bound to TKRs, MT associated ShcA is phosphorylated (presumably by the MT bound pp60c-src) on at least three tyrosines, the YY sequence at position 239 ± 240 (Gotoh et al., 1996; van der Geer et al., 1996), and a single Y at position 317 (Salcini et al., 1994). Both of these sites are within the collagen homology 1 (CH1) region of ShcA, and can then interact with the SH2 domain of the adapter molecule Grb2. Through its SH3 domains, Grb2 is also known to associate with a number of other proteins, including

Oncogene

the p21ras guanine nucleotide exchange factor, Sos1, and another docking molecule, Gab1. Relocation of Sos1 to a membrane site is sucient to activate p21ras and the MAP kinase (MAPK) pathway (Aronheim et al., 1994). Grb2, Sos1, and Gab1 all become tyrosine phosphorylated as a consequence of binding MT (Nicholson et al., manuscript submitted), but the end results of this are unknown. Removal of the ShcA binding site decreases transformation induction by MT dramatically, suggesting MAP kinase activation is a signi®cant factor in MT tumorigenesis (Druker et al., 1992; Markland et al., 1986). In addition, both rac1 and CDC42 are known to be required for MT transformation (Chen et al., 1999; Urich et al., 1997), and it is likely that these are also downstream of p21ras. Recently, we have been able to show that for transformation, insertion into MT of just the 239/240 and 317 phosphorylation sites of ShcA can functionally replace the ShcA binding region (Nicholson et al., manuscript submitted). This demonstrates that these sites are responsible for all of the signal output generated by the ShcA associated with MT. Interestingly, both YY239 and Y317 appear to bind Grb2 to similar levels, but the YY239 region is much more ecient as an ShcA replacement. The reason for the increased eectiveness of this area may lie in an ability to promote binding of proteins to the associated Grb2 SH3 domains, or may indicate that an additional polypeptide interacts with this motif. As these two sites functionally replace ShcA binding, the sole role of Y250 in MT is to interact with ShcA, and not some additional protein. HamMT has no sequence that is homologous to the Y250 region of MT, so probably does not bind to ShcA. Whether this in¯uences the type of tumours that are induced by HamMT has not yet been established. 14-3-3 binding The MT complex contains some members of the 14-3-3 family of proteins (Pallas et al., 1994). They bind to phosphorylated S257 in MT, though it is not clear yet which kinase is responsible for phosphorylating this residue (Cullere et al., 1998). The 14-3-3 family are dimeric proteins that can link other polypeptides together, and there is some evidence that they may cause multimerization of the MT complex (Senften et al., 1997). However, removal of the S257 site does not aect transformation of cells in culture, so this may not be signi®cant. A dramatic reduction in the number of salivary gland tumours induced in mice by a viral mutant with an S257 defect in MT has been observed (Cullere et al., 1998), suggesting 14-3-3 binding may play a role in only some cell types. PI3K association Immunoprecipitates formed with anti-MT antibodies contain an activity that phosphorylates phosphatidylinositol on the 3' OH group. This correlates with the binding, and tyrosine phosphorylation, of an 85 kDa


polypeptide to phosphorylated tyrosine 315 in MT (Courtneidge and Heber, 1987; Kaplan et al., 1987; Talmage et al., 1989). This 85 kDa species contains two SH2 domains, and is associated with an 110 kDa polypeptide that expresses the PI3K catalytic activity. Both SH2 domains can bind phosphorylated Y315 in MT, and possibly also phosphorylated Y322. There is some evidence that a bipartite interaction occurs between the two SH2 domains of PI3K and MT Y315 and Y322 (Yoakim et al., 1992), and this may explain why a number of studies have indicated that mutation of MT Y315 to phenylalanine does not always abolish binding of PI3K (Cohen et al., 1990a; Dilworth et al., 1994). This may also account for the variation observed in the transforming eciency of such mutants (Markland and Smith, 1987). However, deletion of the whole region completely removes PI3K binding activity, and severely aects transformation induction. It is possible that additional activities have been aected by these large deletions, though. As a consequence of binding to MT, PI3K is translocated from an essentially cytoplasmic location to a membrane site in close proximity with its substrates (Cohen et al., 1990b), which may be sucient to increase the amount of phosphorylated lipids in the cell. In addition, there is evidence that interaction between the p85 SH2 domains and phosphotyrosine sequences is able to stimulate PI3K activity of the 110 kDa subunit (Carpenter et al., 1993), so binding to MT probably increases PI3K action by two routes. However, no role for tyrosine phosphorylation of p85 has so far been found. A number of other forms of PI3K have been isolated, but there is no evidence yet whether these types have a role in MT transformation (Vanhaesebroeck et al., 1997). MT transformation causes an increase in the amount of PI(3,4)P2 and PI(3,4,5)P3 (the products and eectors of PI3K activity) in cells, and this rise is abolished by removal of the PI3K binding site (Gorga et al., 1990; Serunian et al., 1990; Ulug et al., 1990). However, other non-transforming mutations in MT, notably removal of the ShcA binding site, also inhibit the MT mediated increase in the level of PI3K products in cells, despite MT precipitates still containing normal levels of p85 and PI3K enzymatic activity (Ling et al., 1992). This may indicate that a large degree of cross talk and co-operation exists between the ShcA-MAPK and PI3K stimulated pathways, possibly through the interaction of PI3K and activated p21ras (Rodriguez Viciana et al., 1994), or through the binding of PI3K to Gab1 via Grb2 and ShcA. Once produced, the products of PI3K have been implicated in activating some PKC family members, and there is evidence that MT transformation does activate PKC (Marcellus et al., 1991). In addition, PI3K products stimulate the kinases PDK1 and PKB. PKB then phosphorylates a number of proteins, including transcription factors, such as E2F and the Forkhead family, and so in¯uences transcription pro®les. PKB also phosphorylates the regulator of apoptosis, Bad (reviewed in Alessi and Cohen,

1998; Downward, 1998; Hemmings, 1997; Meier and Hemmings, 1999). Again, MT is known to increase PKB activity (Meili et al., 1998; Summers et al., 1998), but the role this has in MT induced transformation is not yet clear. Finally, PI3K activity is also linked to actin ®lament reorganization and membrane ruing through the action of rho and rac. MT transformation depends on racl and CDC42 activity (Chen et al., 1999; Urich et al., 1997), but it is not yet clear why and how this dependence is mediated.

7913

PLCg-1 binding Phosphorylation of Y322 in MT creates a binding site for the SH2 domain of phospholipase Cg-1 (Su et al., 1995). As a consequence of association with MT, PLCg-1 is tyrosine phosphorylated and this probably stimulates its enzymatic activity. However, removal of Y322 from MT has only a small in¯uence on its transforming activities in high serum concentrations, though its eect is more noticeable in low serum conditions. Due to the possibility that Y322 may also be involved in binding and activating PI3K, however, it is not yet clear exactly what role PLCg-1 binding plays in transformation by MT.

Remaining questions Although we know a great deal about the signi®cant components of MT induced transformation, it is also clear that there are still many questions outstanding. Some of the major ones include: Do we know all of the proteins that bind MT? Most of the major species found bound to MT in immunoprecipitates separated on 1D or 2D gels have been identi®ed (see Figure 4). However, there are other species present at lower levels, particularly in in vitro kinase labelling reactions, which have yet to be characterized. Whether these have any in¯uence on MT transformation is unclear. Most of the polypeptides that are known to associate with MT have been de®ned by mutations that disrupt binding. However, there is still one MT mutation that has no identi®ed binding defect, dl1015 (Magnusson et al., 1981). Although there are reports that dl1015 MT may bind less pp60c-src, PI3K, or ShcA, when MT expression levels are carefully matched we have been unable to observe any dierence in protein binding capacity (unpublished observations). Despite this, dl1015 MT transforms poorly and cannot induce a sustained increase in PI3K products (Ling et al., 1992). Whether this indicates that a protein interacts with this region that we have yet to identify, or signi®es that the interplay between the proteins that are known to be bound is in some way defective, remains to be determined. Oncogene


7914

stream. However, when a MT mutant that fails to bind ShcA is transfected together with another plasmid containing a mutation that disrupts association between MT and PI3K, transformation induction is poor (although some reports have obtained up to 20% of wild type levels) (Chen et al., 1999; Urich et al., 1997). This suggests both sites have to be on the same molecule to function. It is not immediately clear why this should be so, though some requirement for organizing a local concentration of signalling components at the membrane seems a reasonable possibility. Do PP2A and pp60c-src act on proteins outside the MT complex?

Figure 4 Proteins bound to MT. An autoradiograph of an SDS ± PAGE separation of immunoprecipitates formed with a series of anti-MT monoclonal antibodies is shown. Wild type MT expressing, and control, cells were labelled with S35-methionine, then lysed and precipitated with a series of anti-MT monoclonal antibodies that recognize dierent epitopes on the MT molecule (Dilworth and Grin, 1982; Dilworth and Horner, 1993). The polypeptides were then separated by SDS containing polyacrylamide gel electrophoresis, and the proteins detected by autoradiography. The migration position of MT is shown to the right of the gel, the major proteins associated with MT to the left. Polypeptides of unknown identity are indicated with a question mark. The PAb number of each monoclonal antibody used is indicated above each lane.The use of a number of antibodies to detect MT associated proteins ensures the reactions are with MT, and provides an initial characterization of the site recognized by each species. Proteins immunoprecipitated from control lysates as well as MT expressing cells, such as the polypeptide migrating slightly slower than PI3K 85 kDa band in lanes 3 and 8, probably re¯ect cross reaction between the antibody and cellular proteins, so are not signi®cant

How do the multiple signalling pathways stimulated by MT interact? There is ample evidence from MT itself, as well as from receptor signalling studies, that there is a large degree of cross talk between the ShcA-MAPK, and the PI3K pathways. This may occur at the level of PI3K interacting with p21ras, or extra binding of PI3K to the Gab1 that is associated with Grb2. However, despite being indicative, none of these suggestions really explains why two actions are required to increase the level of PI3K products in cells when a PI3K activity is clearly associated with, and stimulated by, phosphorylated Y315 and Y322 of MT. In addition, both pathways appear to be initiated as separate entities on the information we have available so far, but interact further downOncogene

Although we know that PP2A activity is probably altered by interaction with MT, and pp60c-src is stimulated and phosphorylates MT itself, it is still not clear whether this is the sole role of these proteins in MT induced transformation. For example, a number of proteins are known to become tyrosine phosphorylated during MT transformation (Yonemoto et al., 1987), but it is not apparent whether this occurs as a result of direct action by pp60c-src or if it is merely a consequence of the transformation process itself. Conclusions The middle T-antigen of polyoma virus can in many respects be considered as a permanently active analogue of a growth factor receptor. MT interacts with, and activates, src-family tyrosine kinases, that then phosphorylate MT to provide binding sites for a series of SH2/PTB domain containing proteins that stimulate signalling pathways, exactly like an activated receptor. MT probably does not bind some receptor interacting proteins, such as GAP, but these tend to be those involved in inactivating signalling, which the virus does not require. The similarities between MT and TKRs can be used not only to analyse how transformation occurs, but also how signalling pathways function in inducing cell changes, including growth and dierentiation. In this respect, it is interesting to note that MT can replace receptor activity in a number of model systems, including dierentiation induction (Kennedy et al., 1998; Metcalf et al., 1987; Muser et al., 1989; Platko et al., 1998). The analogy between MT and tyrosine kinase associated receptor activation is not complete, which is probably an indication that dierent receptors stimulate a de®ned range of pathways to achieve their eect, whereas MT has primarily developed to promote mitogenesis. However, it is often possible to learn as much from systems that are similar, but not identical, as it is from exact analogues. So, the virus that is rapidly approaching the 50th anniversary of its discovery has still an important role to play in helping us to uncover how cell cycle control is mediated.


Acknowledgments We are extremely grateful to Dr Nick Dibb for his help in writing this review. Work in the authors' laboratory is

funded by the Cancer Research Campaign (UK), the BBSRC, and the AICR.

7915

References Alessi DR and Cohen P. (1998). Curr. Opin. Genet. Dev., 8, 55 ± 62. Aronheim A, Engelberg D, Li N, Al-Alawi N, Schlessinger J and Karin M. (1994). Cell, 78, 949 ± 961. Blaikie PA, Fournier E, Dilworth SM, Birnbaum D, Borg JP and Margolis B. (1997). J. Biol. Chem., 272, 20671 ± 20677. Brewster CE, Glover HR and Dilworth SM. (1997). J. Virol., 71, 5512 ± 5520. Campbell KS, Auger KR, Hemmings BA, Roberts TM and Pallas DC. (1995). J. Virol., 69, 3721 ± 3728. Campbell KS, Ogris E, Burke B, Su W, Auger KR, Druker BJ, Schahausen BS, Roberts TM and Pallas DC. (1994). Proc. Natl. Acad. Sci. USA., 91, 6344 ± 6348. Carpenter CL, Auger KR, Chanudhuri M, Yoakim M, Schahausen B, Shoelson S and Cantley LC. (1993). J. Biol. Chem., 268, 9478 ± 9483. Chen Y, Freund R, Listerud M, Wang Z and Talmage DA. (1999). Oncogene, 18, 139 ± 148. Cheng SH, Harvey R, Espino PC, Semba K, Yamamoto T, Toyoshima K and Smith AE. (1988a). EMBO J., 7, 3845 ± 3855. Cheng SH, Piwnica Worms H, Harvey RW, Roberts TM and Smith AE. (1988b). Mol. Cell. Biol., 8, 1736 ± 1747. Cohen B, Liu YX, Druker B, Roberts TM and Schahausen BS. (1990a). Mol. Cell. Biol., 10, 2909 ± 2915. Cohen B, Yoakim M, Piwnica Worms H, Roberts TM and Schahausen BS. (1990b). Proc. Natl. Acad. Sci. USA., 87, 4458 ± 4462. Courtneidge SA, Goutebroze L, Cartwright A, Heber A, Scherneck S and Feunteun J. (1991). J. Virol., 65, 3301 ± 3308. Courtneidge SA and Heber A. (1987). Cell, 50, 1031 ± 1037. Courtneidge SA and Smith AE. (1983). Nature, 303, 435 ± 439. Cullere X, Rose P, Thathamangalam U, Chatterjee A, Mullane KP, Pallas DC, Benjamin TL, Roberts TM and Schahausen BS. (1998). J. Virol., 72, 558 ± 563. Dahl J, Thathamangalam U, Freund R and Benjamin TL. (1992). Mol. Cell. Biol., 12, 5050 ± 5058. Dankort DL and Muller WJ. (2000). Oncogene, 19, 1038 ± 1044. Dilworth SM. (1995). Trends Microbiol., 3, 31 ± 35. Dilworth SM, Brewster CE, Jones MD, Lanfrancone L, Pelicci G and Pelicci PG. (1994). Nature, 367, 87 ± 90. Dilworth SM and Grin BE. (1982). Proc. Natl. Acad. Sci. USA., 79, 1059 ± 1063. Dilworth SM, Hansson HA, Darnfors C, Bjursell G, Streuli CH and Grin BE. (1986). EMBO J., 5, 491 ± 499. Dilworth SM and Horner VP. (1993). J. Virol., 67, 2235 ± 2244. Downward J. (1998). Curr. Opin. Genet. Dev., 8, 49 ± 54. Druker BJ, Sibert L and Roberts TM. (1992). J. Virol., 66, 5770 ± 5776. Dunant N and Ballmer Hofer K. (1997). Cell Signal, 9, 385 ± 393. Dunant NM, Messerschmitt AS and Ballmer Hofer K. (1997). J. Virol., 71, 199 ± 206. Dunant NM, Senften M and Ballmer Hofer K. (1996). J. Virol., 70, 1323 ± 1330.

Eddy BE, Stewart SE and Berkeley W. (1958). Proc. Soc. Exp. Biol. and Med., 98, 848 ± 851. Elliott J, Jones MD, Grin BE and Krauzewicz N. (1998). Oncogene, 17, 1797 ± 1806. Erpel T and Courtneidge SA. (1995). Curr. Opin. Cell. Biol., 7, 176 ± 182. Glenn GM and Eckhart W. (1995). J. Virol., 69, 3729 ± 3736. Glover HR, Brewster CE and Dilworth SM. (1999). Oncogene, 18, 4364 ± 4370. Gorga FR, Riney CE and Benjamin TL. (1990). J. Virol., 64, 105 ± 112. Gotoh N, Tojo A and Shibuya M. (1996). EMBO J., 15, 6197 ± 6204. Goutebroze L, Dunant NM, Ballmer Hofer K and Feunteun J. (1997). J. Virol., 71, 1436 ± 1442. Gross L. (1953). Proc. Soc. Exp. Biol. Med., 83, 414 ± 421. Helmbrecht K, Zeise E and Rensing L. (2000). Cell Prolif., 33, 341 ± 365. Hemmings BA. (1997). Science, 275, 628 ± 630. Horak ID, Kawakami T, Gregory F, Robbins KC and Bolen JB. (1989). J. Virol., 63, 2343 ± 2347. Janssens V and Goris, J.(2001). Biochem. J., 353, 417 ± 439. Kaplan DR, Whitman M, Schahausen B, Pallas DC, White M, Cantley L and Roberts TM. (1987). Cell, 50, 1021 ± 1029. Kennedy AP, Sekulic A, Irvin BJ, Nilson AE, Dilworth SM and Abraham RT. (1998). J. Biol. Chem., 273, 11505 ± 11513. Kiefer F, Courtneidge SA and Wagner EF. (1994). Adv. Cancer Res., 64, 125 ± 157. Kornbluth S, Sudol M and Hanafusa H. (1987). Nature, 325, 171 ± 173. Kypta RM, Hemming A and Courtneidge SA. (1988). EMBO J., 7, 3837 ± 3844. Ling LE, Druker BJ, Cantley LC and Roberts TM. (1992). J. Virol., 66, 1702 ± 1708. Ma YC, Huang J, Ali S, Lowry W and Huang XY. (2000). Cell, 102, 635 ± 646. Magnusson G, Nilsson MG, Dilworth SM and Smolar N. (1981). J. Virol., 39, 673 ± 683. Marcellus R, Whit®eld JF and Raptis L. (1991). Oncogene, 6, 1037 ± 1040. Markland W, Oostra BA, Harvey R, Markham AF, Colledge WH and Smith AE. (1986). J. Virol., 59, 384 ± 391. Markland W and Smith AE. (1987). Biochim. Biophys. Acta., 907, 299 ± 321. Matthews JT and Benjamin TL. (1986). J. Virol, 58, 239 ± 246. Meier R and Hemmings BA. (1999). J Recept Signal Transduct Res, 19, 121 ± 128. Meili R, Cron P, Hemmings BA and Ballmer Hofer K. (1998). Oncogene, 16, 903 ± 907. Messerschmitt AS, Dunant N and Ballmer Hofer K. (1997). Virology, 227, 271 ± 280. Metcalf D, Roberts TM, Cherington V and Dunn AR. (1987). EMBO J., 6, 3703 ± 3709. Mumby M. (1995). Seminars in Cancer Biology, 6, 229 ± 237. Muser J, Kaech S, Moroni C and Ballmer Hofer K. (1989). Oncogene, 4, 1433 ± 1439. Oncogene


7916

Oncogene

Nicholson PR and Dilworth SM. (2001). Viruses, Cell Transformation and Cancer: Perspectives in Medical Virology. Grand, RJA (ed.). Elsevier Science B. V.: Amsterdam, pp 85 ± 128. Ogris E, Mudrak I, Mak E, Gibson D and Pallas DC. (1999). J. Virol., 73, 7390 ± 7398. Pallas DC, Fu H, Haehnel LC, Weller W, Collier RJ and Roberts TM. (1994). Science, 265, 535 ± 537. Pallas DC, Morgan W and Roberts TM. (1989). J. Virol., 63, 4533 ± 4539. Pallas DC, Shahrik LK, Martin BL, Jaspers S, Miller TB, Brautigan DL and Roberts TM. (1990). Cell, 60, 167 ± 176. Platko JD, Forbes ME, Varvayanis S, Williams MN, Brooks III SC, Cherington V and Yen A. (1998). Exp. Cell. Res., 238, 42 ± 50. Rassoulzadegan M, Cowie A, Carr A, Glaichenhaus N, Kamen R and Cuzin F. (1982). Nature, 300, 713 ± 718. Rassoulzadegan M, Naghashfar Z, Cowie A, Carr A, Grisoni M, Kamen R and Cuzin F. (1983). Proc. Natl. Acad. Sci. USA, 80, 4354 ± 4358. Rodriguez Viciana P, Warne PH, Dhand R, Vanhaesebroeck B, Gout I, Fry MJ, Water®eld MD and Downward J. (1994). Nature, 370, 527 ± 532. Ruediger R, Roeckel D, Fait J, Bergqvist A, Magnusson G and Walter G. (1992). Mol. Cell. Biol., 12, 4872 ± 4882. Salcini AE, McGlade J, Pelicci G, Nicoletti I, Pawson T and Pelicci PG. (1994). Oncogene, 9, 2827 ± 2836. Scherneck S and Feunteun J. (1990). Arch Geschwulstforsch, 60, 271 ± 278. Senften M, Dilworth S and Ballmer Hofer K. (1997). J. Virol., 71, 6990 ± 6995. Serunian LA, Auger KR, Roberts TM and Cantley LC. (1990). J. Virol., 64, 4718 ± 4725.

Su W, Liu W, Schahausen BS and Roberts TM. (1995). J. Biol. Chem., 270, 12331 ± 12334. Summers SA, Lipfert L and Birnbaum MJ. (1998). Biochem. Biophys. Res. Comm., 246, 76 ± 81. Talmage DA, Freund R, Young AT, Dahl J, Dawe CJ and Benjamin TL. (1989). Cell, 59, 55 ± 65. Templeton D, Voronova A and Eckhart W. (1984). Mol. Cell. Biol., 4, 282 ± 289. Treisman R, Novak U, Favaloro J and Kamen R. (1981). Nature, 292, 595 ± 600. Ulug ET, Cartwright AJ and Courtneidge SA. (1992). J. Virol., 66, 1458 ± 1467. Ulug ET, Hawkins PT, Hanley MR and Courtneidge SA. (1990). J. Virol., 64, 3895 ± 3904. Urich M, Senften M, Shaw PE and Ballmer Hofer K. (1997). Oncogene, 14, 1235 ± 1241. van der Geer P, Wiley S, Gish GD and Pawson T. (1996). Curr. Biol., 6, 1435 ± 1444. Vanhaesebroeck B, Leevers SJ, Panayotou G and Water®eld MD. (1997). Trends Biochem. Sci., 22, 267 ± 272. Walter G, Carbone A and Welch WJ. (1987). J. Virol., 61, 405 ± 410. Walter G, Ruediger R, Slaughter C and Mumby M. (1990). Proc. Natl. Acad. Sci. USA., 87, 2521 ± 2525. Xu W, Harrison SC and Eck MJ. (1997). Nature, 385, 595 ± 602. Yoakim M, Hou W, Liu Y, Carpenter CL, Kapeller R and Schahausen BS. (1992). J. Virol., 66, 5485 ± 5491. Yonemoto W, Filson AJ, Queral Lustig AE, Wang JY and Brugge JS. (1987). Mol. Cell. Biol., 7, 905 ± 913. Zhu W, Eicher A, Leber B and Andrews DW. (1998). Oncogene, 17, 565 ± 576.