Catenin Degradation and Selection of APC Mutations in Colorectal Tumours from FAP Patients

Home , Oncogene

Oncogene (2010) 29, 1663–1671 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Contribution of the 15 amino acid repeats of truncated APC to b-catenin degradation and selection of APC mutations in colorectal tumours from FAP patients

EM Kohler1, K Brauburger, J Behrens and J Schneikert

Nikolaus-Fiebiger-Center for Molecular Medicine, University of Erlangen-Nu¨rnberg, Erlangen, Bayern, Germany

The adenomatous polyposis coli (APC) protein is a leads to the development of polyps in the gastrointest- negative regulator of the mitogenic transcription factor inal tract that ultimately evolve as aggressive and fatal b-catenin by stimulating its proteasomal degradation. tumours in affected people. Besides gastric, duodenal This involves several APC domains, including the binding and colorectal cancer, familial adenomatous polyposis sites for axin/conductin, the recently described b-Catenin coli (FAP) patients may also develop desmoid tumours Inhibitory Domain (CID) and the third 20 amino acid (Kotiligam et al., 2008). repeat (20R3) that is a b-catenin-binding site. The four 15 Many functions have been attributed to adenomatous amino acid repeats (15R) and the 20R1 are also b-catenin- polyposis coli (APC) (Nathke, 2006; Akiyama and binding sites, but their role in b-catenin degradation Kawasaki 2006; Aoki and Taketo 2007; Schneikert has remained unclear. We show here that binding of and Behrens, 2007a; McCartney and Nathke, 2008), but b-catenin to the 15R of APC is necessary and sufficient to it is best understood as a component of the canonical target b-catenin for degradation whereas binding to the wnt signalling pathway, controlling the fate of the 20R1 is neither necessary nor sufficient. The first 15R transcription factor b-catenin (Clarke, 2006; Clevers, displays the highest affinity for b-catenin in the 15R-20R1 2006; Taketo, 2006; Polakis, 2007; Fodde and Brabletz module. Biallelic mutations of the APC gene lead to 2007). In the presence of a proliferative signal provided colon cancer in familial adenomatous polyposis coli by an extracellular ligand from the wnt family, cytosolic (FAP) and result in the synthesis of truncated products b-catenin accumulates, translocates into the nucleus and lacking domains involved in b-catenin degradation but still stimulates the transcription of target genes (Behrens having a minimal length. The analysis of the distribution et al., 1996). Thereby, b-catenin translates the wnt signal of truncating mutations along the APC sequence in colo- into changes in the transcriptional state that ultimately rectal tumours from FAP patients revealed that the first manifest by cell proliferation. In the absence of a wnt 15R is one target of the positive selection of mutations signal, b-catenin is destroyed in the proteasome where that lead to tumour development. it is conveyed only after a critical phosphorylation Oncogene (2010) 29, 1663–1671; doi:10.1038/onc.2009.447; step catalysed in the so-called degradation complex published online 7 December 2009 (Kimelman and Xu, 2006). In this complex, axin/ conductin recruits the kinases glycogen synthase kinase Keywords: adenomatous polyposis coli; APC; b-catenin; 3b and casein kinase 1 that sequentially phosphorylate colon cancer; mutation b-catenin (Liu et al., 2002). This process is drastically enhanced by the presence of APC (Munemitsu et al., 1995) that displays several domains involved in b- catenin degradation (Figure 1). The SAMP repeats Introduction constitute binding sites for axin/conductin (Behrens et al., 1998). The seven 20 amino acid repeats (20R) are Familial adenomatous polyposis coli is an autosomal b-catenin-binding sites (Rubinfeld et al., 1993). Among dominant genetic disorder characterized by the germline the latter, a role in b-catenin degradation has been transmission of a mutation in the adenomatous poly- demonstrated in detail for the third 20R (20R3) only posis coli gene (Groden et al., 1991, 1993; Olschwang (Rubinfeld et al., 1997; Kohler et al., 2008). In contrast, et al., 1993). Somatic inactivation of the remaining allele the 20R2 cannot bind to b-catenin (Liu et al., 2006; Kohler et al., 2008). Phosphorylation of the 20R1 and the 20R3 increases their affinity for b-catenin (Tick- Correspondence: Dr J Schneikert, Nikolaus-Fiebiger-Center for Molecular Medicine, University Erlangen-Nu¨rnberg, Glu¨ckstrasse 6, enbrock et al., 2003; Ha et al., 2004; Xing et al., 2004; Erlangen, Bayern, 91054, Germany. Liu et al., 2006; Ferrarese et al., 2007). The b-Catenin E-mail: [email protected] Inhibitory Domain (CID) domain is located between the 1Current address: Department of Psychiatry and Psychotherapy, 20R2 and the 20R3 and is essential in truncated APC to University of Erlangen-Nu¨rnberg, Schwabachanlage, Erlangen, Germany. target b-catenin for degradation (Kohler et al., 2009). Received 15 June 2009; revised 23 October 2009; accepted 3 November The four 15 amino acid repeats (15R-A to D) are also b- 2009; published online 7 December 2009 catenin-binding sites (Su et al., 1993; Eklof Spink et al., Functional analysis of adenomatous polyposis coli EM Kohler et al 1664

Figure 1 Degradation of b-catenin can be independent of its interaction with APC. SW480 cells were transiently transfected with the indicated truncated APC expression constructs, encoding products fused to YFP at their N terminus. The CID domain, the 15 amino acid-, 20 amino acid- and SAMP repeats are shown as boxes. A black, M-labelled box indicates a mutated repeat that cannot bind to b-catenin, except the 20R3 for which the affinity is only decreased, although drastically. Definitive inactivation of the 20R3 requires additional mutations, which is indicated by MM (Kohler et al., 2008). Flag, empty control vector. To analyse the transcriptional activity of b-catenin, cells were co-transfected with either a b-catenin-dependent luciferase reporter gene (TOPglow) or a control reporter gene lacking functional b-catenin-binding sites (FOPglow), together with a b-galactosidase expression vector as an internal control to correct for variations in the transfection efficiency. Cells were harvested 24–36 h post transfection and cell extracts were submitted either to western blotting (WB) using either an anti-GFP or an anti-b-actin antibody, or to a luciferase assay (TOPglow/ FOPglow). Data are represented as the ratio of TOPglow to FOPglow values corrected by the b-galactosidase values and are the mean ( þ /À s.d.) of three independent experiments. In parallel, transfected cells were fixed and stained with an anti-b-catenin antibody. The column on the right (IF) indicates for each YFP–APC construct the percentage of transfected cells negative for b-catenin. At least hundred cells were scored for each construct. The intracellular localization of the YFP–APC constructs is also shown.

2001; Liu et al., 2006), but it is not known whether they 2009). These animals do not develop tumours in the have a role in b-catenin degradation. gastrointestinal tract. In some cases however, the CID APC mutations as found in the germline of FAP domain that allows truncated APC to target b-catenin patients lead to premature termination of translation for degradation even in the absence of the 20R3 and the and therefore to the synthesis of truncated products SAMP repeats in the appropriate cell line, can be (Miyoshi et al., 1992; Powell et al., 1992; Miyaki et al., retained in a colorectal tumour if additional molecular 1994, 2008; Segditsas and Tomlinson, 2006). Mutations events lead to its inactivation (Kohler et al., 2009). Thus, of both APC alleles in FAP tumours are interdependent, APC truncated shortly after the CID can be used that is, the site of the germline mutation determines the experimentally to target efficiently b-catenin for degra- nature and the location of the somatic mutation dation if it is expressed in a permissive cell line. (Lamlum et al., 1999; Rowan et al., 2000; Albuquerque Despite the negative selection, colon cancer cells et al., 2002; Crabtree et al., 2003). This results in a always retain a truncated APC of a minimal length. In complex distribution of APC mutations along the 95% of the cases the fragment extends up to at least the sequence, which is partially explained by the necessity end of the 20R1 (Kohler et al., 2008). At the molecular of removing or inactivating important domains involved level, the reasons of this positive selection are not in the degradation of b-catenin (that is, the SAMP known, but its strength indicates that the retained frag- repeats, the 20R3 and the CID domain). As a conse- ment probably fulfils an essential function. Actually, quence, b-catenin is constitutively stabilized. The SAMP RNA interference experiments have shown that APC repeats and the 20R3 are almost systematically excluded truncated shortly after the 20R1 and lacking the CID from truncated APC (Kohler et al., 2008). The necessity domain, is required for optimal cell proliferation of removing both domains is illustrated by the avail- (Schneikert and Behrens, 2006) and can still exert a ability of mice heterozygous for APC isoforms truncated control on the transcriptional activity of b-catenin either shortly after the first SAMP repeat (APC þ /1638T (Schneikert et al., 2007b). This is in line with the recent mice) (Smits et al., 1999) or between the 20R3 and analysis of APC þ /1322T mice (Pollard et al., 2009) the first SAMP repeat (APC þ /1572T mice) (Gaspar et al., carrying an APC truncation falling shortly after the

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1665 20R1. These mice display a more severe polyposis but for degradation (Kohler et al., 2009). All transfected polyps with lower amounts of nuclear b-catenin than in constructs were expressed at a similar level and mice, in which truncated APC does not contain any displayed an identical diffuse and cytoplasmic intracel- b-catenin-binding site. The positive selection rule is not lular localization (Figure 1). YAPC1641 efficiently absolute and tolerates 5% of exceptions in which APC is downregulated the transcriptional activity of b-catenin truncated before the 20R1. In these cases, it is not and none of the transfected cells stained positive for known which APC domain is the target of the positive b-catenin. Thus, yAPC1641 targets b-catenin efficiently selection. for degradation. Mutating any of the b-catenin-binding In this communication, we provide evidence that the sites, either alone or in a combined manner led only to a 15R are important b-catenin-binding sites involved in modest reduction of the activity of yAPC1641. Even a APC-mediated degradation of b-catenin and the first construct in which all repeats were mutated to abolish 15R is one parameter of the positive selection of APC- any possibility of b-catenin binding (as in yAPC1641- truncating mutations in tumours from FAP patients. 15Rm-20R1m-20R3 mm) (Kohler et al., 2008) was still displaying a considerable b-catenin downregulating activity (Figure 1). In contrast, this activity was lost Results when the SAMP repeat was deleted from the latter construct (as in yAPC1519-15Rm-20R1m-20R3 mm). We The contribution of the b-catenin-binding sites of APC concluded that the SAMP repeat, which provides an to b-catenin degradation was first investigated in an interaction surface for axin/conductin confers a high b- APC construct tagged at its N terminus with YFP and catenin-degrading efficiency to APC even in the absence extending up to amino acid 1641, shortly after the first of any functional b-catenin-binding site. axin/conductin-binding site (Figure 1). Such a construct In colon cancer, the SAMP repeats and the 20R3 are was previously shown to inhibit b-catenin-dependent lost. However, our previous results have shown that transcription and target b-catenin for degradation truncated APC lacking these domains but retaining the (Kohler et al., 2009). Mutations that either inactivate CID can still efficiently target b-catenin for degradation b-catenin binding to the 20R1 and the four 15R, or upon expression in a permissive cell line (Kohler et al., drastically reduce the association of b-catenin to the 2009). Therefore, to reveal a possible involvement of the 20R3 as described previously (Kohler et al., 2008) were 15R in b-catenin degradation, we used a construct introduced in the construct, either individually or in all lacking both SAMP repeat and 20R3 but extending possible combinations. The resulting plasmids were up to amino acid position 1466, that is, after the transiently transfected in SW480 colon cancer cells that CID (Figure 2). When transiently expressed in SW480 contain relatively high levels of cytosolic b-catenin, and cells, yAPC1466 targeted b-catenin for degradation and expression was monitored by western blotting using inhibited its transcriptional activity, as previously an anti-GFP antibody. The fluorescence of YFP described (Kohler et al., 2009). Introducing mutations allowed also revealing the intracellular localization of that abolished b-catenin binding to the four 15R (as in the YFP–APC fusions. The consequences on b-catenin- yAPC1466–15Rm) led to a very strong reduction of mediated transcription were measured by co-transfect- the b-catenin downregulating activity of yAPC1466 ing a b-catenin-dependent luciferase reporter construct (Figure 2). We confirmed that the 15R are important (see Materials and methods section). In parallel, the determinants of b-catenin degradation by mutating fraction of transfected cells negative for b-catenin was them in additional constructs extending up to amino determined by immunofluorescence staining using an acid positions 1558, 1519 and 1502 (supplementary anti-b-catenin antibody. The latter two assays reflect Figure 1). Thus, the 15R of truncated APC are necessary the ability of an APC construct to target b-catenin for targeting b-catenin to degradation. Mutation of the

Figure 2 The 15R are necessary and sufficient for b-catenin degradation mediated by truncated APC in SW480 cells, whereas the 20R1 is neither necessary nor sufficient. Transcriptional and immunofluorescence assays using the indicated constructs. Experimental procedures are as described in the Figure 1 legend.

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1666 20R1 in yAPC1466 (as in yAPC1466–20R1m) had only a yAPC1466–15RAmCmDm–20R1m were both reducing marginal effect, if any (Figure 2). Thus, we concluded the level of b-catenin and inhibiting b-catenin-mediated that the 15R are sufficient to target b-catenin for transcription, although they could not reach the activity degradation, whereas the 20R1 is not necessary to of yAPC1466–20R1m (Figure 3a). Thus, the 15R-A and perform the function. YAPC1466–15Rm containing only 15R-B are involved in b-catenin degradation and the 20R1 as a functional binding site did not affect can work independently of the other repeats. Combining the transcriptional activity of b-catenin or its level both 15R-A and 15R-B in the same construct (Figure 2). High-affinity b-catenin binding to the 20R (yAPC1466–15RCmDm-20R1m) resulted in a better effi- has been shown previously to require their phosphor- ciency. There was some quantitative discrepancy be- ylation (Tickenbrock et al., 2003; Ha et al., 2004; Xing tween the reporter assay and the b-catenin degradation et al., 2004; Liu et al., 2006; Ferrarese et al., 2007). We assay, but the data matched in a qualitative way. When verified that this phosphorylation event indeed occurs in lower amounts of expression vectors were transfected, SW480 cells (supplementary Figure 2) and concluded the b-catenin downregulating activity of the construct that the 20R1 is not sufficient for targeting b-catenin for containing the 15R-B decreased but not that of the degradation. Finally, the simultaneous mutations of construct containing the 15R-A, indicating different both 15R and 20R1 (as in yAPC1466–15Rm-20R1m) activities of the repeats (Figure 3b). Under these fully inactivated the molecule (Figure 2), indicating conditions, the construct combining both 15R-A and that b-catenin binding to yAPC1466 is necessary to 15R-B was more efficient than the construct containing target b-catenin for degradation. the 15R-A alone, suggesting a cooperation between the In the subsequent experiments, we investigated the repeats (Figure 3b). individual contributions of each 15R to the b-catenin To understand the difference of b-catenin down- downregulating activity of yAPC1466 (Figure 3). In a regulating activity associated with the 15R-A and 15R-B first step, we designed the yAPC1466–15RBmCmDm- of APC, we performed immunoprecipitation experi- 20R1m and yAPC1466–15RAmCmDm-20R1 m constructs ments using cell extracts from transiently transfected in which all b-catenin-binding sites were mutated except HEK293T cells, that we mixed with extracts from SW480 the 15R-A or the 15R-B, respectively (Figure 3a). In cells to provide a source of b-catenin (Figure 3c). We addition, we also created the yAPC1466–15RCmDm– found that the constructs displaying a functional 15R-A 20R1m construct, in which both 15R-A and 15R-B were binding efficiently to b-catenin, similarly to were preserved. YAPC1466–15RBmCmDm–20R1m and yAPC1466–20R1m. In contrast, the construct retaining

Figure 3 The 15R-A is more potent than the 15R-B to bind b-catenin, target it for degradation and inhibit its transcriptional activity. (a), Transcriptional assay using 1 mg of the indicated constructs and immunoﬂuorescence assay conducted as described in the Figure 1 legend. (b), Transcriptional assay using 10 ng of the indicated constructs. (c), Immunoprecipitation experiment using extracts from HEK293T cells transiently transfected with the constructs indicated in (b). Cell extracts were prepared 40 h post-transfection and mixed with an extract provided by SW480 cells (input). Immunoprecipitation was performed using an anti-GFP antibody (IP-GFP) followed by western blottings using either an anti-GFP or an anti-b-catenin antibody, as indicated. *Po0.02, using the Student’s t-test.

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1667 the 15R-B only (yAPC1466–15RAmCmDm-20R1m) failed assays where they may have reached values near the to bind b-catenin. Thus, the 15R-A has a higher affinity Kd of the 15R-B (Liu et al., 2006). In contrast, a for b-catenin than the 15R-B. construct containing both mutated 15R-A and 15R-B To compare the binding affinity of the 15R-A with was strongly affected, confirming the involvement of the that of the other repeats, we repeated the immunopre- 15R-B in b-catenin downregulation. In addition, mutat- cipitation experiment using constructs in which the 15R ing the 15R-C virtually abolished the inhibitory activity were sequentially mutated (Figure 4a). Mutation of of APC toward b-catenin and therefore revealed the 15R-A alone (as in yAPC1466–15RAm) led to a the contribution of the 15R-C. As already shown above pronounced reduction of the b-catenin-binding activity. (Figure 2), yAPC1466 lacking all functional b-catenin- Further mutating the remaining 15R (as in yAPC1466– binding sites was fully inactive, whereas mutation of 15RAmBm, yAPC1466–15RAmBmCm and yAPC1466– the 20R1 had only a marginal effect (Figure 4b). All 15Rm) did not exacerbate substantially the consequence constructs were expressed at a similar level and of mutating the 15R-A alone. We found that mutating displayed indistinguishable intracellular localizations the 20R1 led to a relatively modest reduction of the (Figure 4b). We concluded that the 15R-B and the b-catenin-binding activity (compare yAPC1466 and 15R-C confer a b-catenin downregulating activity to yAPC1466–20R1m in Figure 4a), although we could yAPC1466. confirm phosphorylation of the 20R1 in HEK293T cells In tumours from FAP patients, most truncated APC too (data not shown). Finally, the introduction of molecules retain the 20R1 (Figure 5). Out of 227 inactivating mutations in all b-catenin-binding sites tumours in which APC does not display the CID abolished effectively the interaction of truncated APC domain, 211 contain at least one APC mutation with b-catenin (as in yAPC1466–15Rm-20R1m). The occurring after position 1283, that is, after the 20R1 strong decrease in b-catenin binding achieved upon (Lamlum et al., 1999; Albuquerque et al., 2002; Crabtree mutation of the 15R-A alone indicated that this repeat et al., 2003; Miyaki et al., 2008). When a germline displays the highest affinity for b-catenin among all mutation occurs before the 20R1 (before position 1271), b-catenin-binding sites in yAPC1466. This conclusion most second hits are located after it, indicating a was confirmed in additional immunoprecipitation selective pressure towards its retention. Inversely, when experiments using constructs in which the 20R1 and the the truncated APC product resulting from the germline 15R were sequentially deleted (supplementary Figure 3). mutation already contains the 20R1, the second allele When the constructs in which the 15R were sequen- is lost in most cases. LOH proceeds essentially by tially mutated were transiently transfected in SW480 homologous recombination (Gupta et al., 1997; Haigis cells (Figure 4b), we observed that mutation of the 15R- and Dove 2003) resulting in the duplication of the allele A alone affected only slightly the b-catenin down- containing the truncating mutation. Importantly, as regulating activity of yAPC1466 in both transcriptional homologous recombination is spontaneously much and immunofluorescence assays. This indicates a com- more frequent than any other mutational event (Hong pensatory effect from the remaining 15R despite the et al., 2007), LOH indicates that there is no other strong reduction of b-catenin binding. The discrepancy selective pressure beside the loss of the second allele and between the immunoprecipitation (Figure 4a) and therefore the biochemical features of truncated APC the transcriptional and immunofluorescence assays resulting from the germline mutation are sufficient to (Figure 4b) likely comes from the fact that both APC allow tumorigenesis. However, 16 exceptions do not and b-catenin concentrations were higher in the latter respect the 20R1 rule, because in these cases all

Figure 4 (a) The 15R-A provides the highest affinity b-catenin-binding site in truncated APC. Immunoprecipitation experiments using the indicated constructs and performed as described in the legend of Figure 3c. (b), The 15R-C is involved in b-catenin degradation, whereas the 15R-D is not sufficient. Transcriptional and immunofluorescence assays using the constructs indicated in a, and performed as described in the Figure 1 legend. *Po0.02, using the Student’s t-test.

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1668 20R3 and the SAMP repeats but containing the CID domain. A detailed analysis of the individual participa- tions of the 15R to APC-mediated degradation of b-catenin revealed that they can work independently of each other but their relative contributions to the overall activity of the 15R region are not equivalent. The 15R-A turned out to display the highest affinity toward b-catenin among all binding sites of the 15R-20R1 module (Figure 4 and supplementary Figure 3). It was also sufficient to provide a relatively good b-catenin downregulating activity to truncated APC (Figure 3). The importance of the 15R-A is illustrated by the recent discovery of a rare substitution at amino acid position 1026 in the 15R-A, that has been shown to alter b-catenin binding to the repeat and to co-segregate systematically with attenuated FAP (Menendez et al., 2008). The 15R-B could also mediate b-catenin degra- Figure 5 The 15R-A is a parameter of the selection of APC- dation on its own (Figure 3a) but to a lesser extent than truncating mutations in colorectal tumours from FAP patients. The diagram shows the distribution of the positions of somatic the 15R-A, likely due to a lower affinity for b-catenin mutations in the APC protein sequence as a function of the (Figure 3b). Finally, we could also reveal a contribution position of the germline mutation. This analysis includes only of the 15R-C, but not of the 15R-D (Figure 4). It is still tumours in which an APC-truncating mutation always occurs possible that the 15R-D plays a role in b-catenin before the CID domain (before position 1417) and reported in degradation and that a low affinity for b-catenin may (Lamlum et al., 1999; Rowan et al., 2000; Albuquerque et al., 2002; Crabtree et al., 2003) (227 tumours out of 328). This was made for prevent detecting binding and downregulating activities. the sake of simplicity, because the CID is also a parameter of the Alternatively, these functions of the 15R-D may depend interdependent bi-allelic selection of APC mutations. Truncations on the other 15R. occurring between residues 1270–1284 (two tumours) were Together, our results are compatible with two models excluded, as the consequence on b-catenin binding to the 20R1 was not clear. Among all tumours shown here, there was no of b-catenin degradation mediated by the 15R of APC. mutation falling between positions 1020 and 1053. The dark bars The high affinity of the 15R-A for b-catenin may allow indicate tumours with truncating mutations always occurring APC to process low concentrations of b-catenin whereas before the 20R1. LOH, loss of heterozygocity. Po10À8 using the the other 15R with a lower affinity to engage simulta- Fischer’s exact test for the comparison of tumour distribution in neously several b-catenin molecules when its concentra- every germline category. tion increases. Alternatively, b-catenin binding to the 15R-A first may stimulate cooperative binding to the other remaining 15R, as suggested by the evidence of truncating mutations fall before the 20R1 (before a cooperative activity between the 15R-A and 15R-B position 1271) (Figure 5). When we analysed the (Figure 3b). distribution of the somatic APC mutations in these tumours as a function of the position of the germline hit The 20R1 is neither necessary nor sufficient for (Figure 5), we found that 14 out of 16 tumours express b-catenin degradation at least one truncated APC molecule containing at The lack of effect observed in the immunofluorescence least the 15R-A. In addition, when the germline and transcriptional assays after mutating the 20R1 mutation falls after position 1053 and therefore allows (Figure 2) may have resulted from the conditions of the retention of the 15R-A, we observed that the second ectopic expression where the 20R1 would be dispensa- allele is lost in the majority of cases (8 cases out of 11 ble. However, truncated APC containing mutant 15R tumours), indicating that the presence of the 15R-A is but wild-type 20R1 could not target b-catenin for sufficient for tumorigenesis. We concluded that trun- degradation (Figure 2), although the 20R1 could be cated APC are selected for the presence of the 15R-A phosphorylated and thus could bind relatively efficiently within the group of exceptional tumours. to b-catenin (Figure 4). In comparison, mutation of the 15R-B clearly revealed its involvement in b-catenin downregulation, although we could not detect any b-catenin binding to it (Figure 4). Thus, the 20R1 is either Discussion not involved in b-catenin degradation, or its function depends on b-catenin binding to the 15R region first. Individual contributions of the 15R to b-catenin degradation By combining site-directed mutagenesis and deletion The contributions of the 20R1-3 to b-catenin analysis of APC with transient transfection assays, we degradation are not equivalent have demonstrated that b-catenin binding to the 15R is The comparison of the individual contribution of the necessary and sufficient to efficiently target b-catenin for 20R1 to b-catenin degradation with that of the 20R2 degradation in a truncated APC molecule lacking the and the 20R3 reveals that the repeats are not equivalent.

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1669 In contrast to the 20R1, the 20R2 cannot bind to expression of axin2/conductin alone leads already to b-catenin (Liu et al., 2006; Kohler et al., 2008) whereas efficient b-catenin degradation in SW480 cells expressing the 20R3 is sufficient for b-catenin binding (Kohler an inactive endogenous truncated APC lacking all et al., 2008) and degradation (supplementary Figure 1). SAMP repeats and the CID domain (Behrens et al., The APC sequence contains four supplementary 20Rs 1998), we think that both upregulation of endogenous that still await a functional analysis, as it is not possible axin2/conductin and ectopic expression of a truncated to predict their involvement in b-catenin degradation in APC construct containing a SAMP repeat, but lacking view of their sequence homology and even their affinity functional b-catenin sites is sufficient for the b-catenin for b-catenin, by analogy with the 20R1 that displays a degradation machinery to bypass b-catenin binding to relatively high affinity for b-catenin, though is not APC, which is thought nevertheless to occur under sufficient to target it for degradation. physiological conditions. Actually, the multiplicity of One model of APC-mediated degradation of SAMP repeats and b-catenin-binding sites in APC and b-catenin relies on the mutually exclusive abilities of the fact that either domain is not strictly necessary to axin and the 20R3 to bind to b-catenin and to compete achieve a high b-catenin degrading activity confirm for it (Xing et al., 2003; Ha et al., 2004). According to the recently described robustness of the degradation this model, axin-bound b-catenin is phosphorylated by complex (Peterson-Nedry et al., 2008). CK1 and GSK3b. Subsequent phosphorylation of the 20R3 increases its affinity for b-catenin, allowing the The 15R-A is a parameter of the selection of truncated displacement of phosphorylated b-catenin from axin APC in colorectal cancer and its transfer onto APC. Our results indicate that the In colorectal tumours, truncating mutations affecting phosphorylated 20R1 does not mediate b-catenin APC occur after the 20R1 in the vast majority of cases degradation and others have shown that it displays a (Lamlum et al., 1999; Albuquerque et al., 2002; Crabtree lower affinity for b-catenin than the phosphorylated et al., 2003; Miyaki et al., 2008). In tumours that do not 20R3 (Liu et al., 2006) that can perform the function. respect this rule, the 15R-A seems to be the target of the This suggests that the phosphorylated 20R1 cannot selection of APC-truncating mutations (Figure 5). The target b-catenin for degradation because it may not be selective pressure on APC mutations is different in able to compete with axin for b-catenin binding. colorectal, duodenal, desmoid and breast tumours. However, our results reveal also that the 15R-B can Colorectal APC mutations have the tendency to fall promote b-catenin degradation although it displays a before the CID domain (Kohler et al., 2009), whereas lower affinity for b-catenin than the 20R1. Therefore, truncated APCs from duodenal (Groves et al., 2002) and the differences of affinities of the 20R1 and the 20R3 for desmoid (Latchford et al., 2007) tumours rather keep b-catenin do not explain their functional differences in the CID domain while simultaneously avoiding the the context of the model. Alternatively, the 15Rs and the presence of the 20R3 (Kohler et al., 2009). In contrast, 20Rs may have different roles in b-catenin degradation. the presence of the 20R3 is clearly selected in truncated APC from breast tumours (Kuraguchi et al., 2009). Although there is increasing evidence that colorectal The SAMP repeat compensates for the lack of tumours derive from crypt stem cells (Barker et al., b-catenin binding 2009), we suggest that the 16 tumours in which the When a SAMP repeat is present in truncated APC, selection has been acting on the 15R-A may have arose b-catenin binding to APC is not necessary anymore to from a specific cell type. target b-catenin for degradation (Figure 1). This result Altogether, this work has revealed the involvement contradicts apparently the conclusions of other authors of the 15R in b-catenin degradation mediated by (Kawahara et al., 2000) who showed that an internal truncated APC. The 15R-A provides the highest affinity APC fragment spanning amino acids 1342–1887, and b-catenin-binding site in the 15R-20R1 module and is therefore retaining the SAMP repeats, requires the 20Rs one target of the selection of APC-truncating mutations to target b-catenin for degradation in SW480 cells. The in colorectal cancer. It would be interesting now to presence of the N terminus of APC in our constructs, determine the consequences of mutating the 15Rs which seems to contribute to b-catenin downregulation and/or the 20Rs on tumour location and develop- (Kohler et al., 2009), may provide a compensation to the ment in truncated APC from the currently available lack of b-catenin binding. mouse models, namely the APC1638T (Smits et al., 1999), Our observation that the presence of a SAMP repeat APC þ /1572T (Gaspar et al., 2009) and APC þ /1322T (Pollard probably alleviates the requirement of b-catenin-binding et al., 2009) mice. As the affinity of the 15R-A for stems from the combination of two facts. First, although b-catenin is higher than that of the 20R1, it is puzzling we also performed some of our transcriptional assays that most colorectal tumours are selected for the under limiting conditions (data not shown), the con- presence of the 20R1 in truncated APC. centration of exogenous APC construct was probably higher than the level reached by endogenous full length APC in natural conditions. Second, our constructs were Materials and methods transfected in the SW480 colorectal cancer cell line known to express relatively high levels of axin2/ Cells conductin. As it has also been established that ectopic Cells are described in Supplementary information.

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1670 Antibodies 1). The internal control pUHD16.1 encoding the b-galactosi- The rabbit antibody H102 against b-catenin was purchased dase was transiently transfected together with either FOPglow from Santa Cruz (Heidelberg, Germany). Secondary antibo- or TOPglow plasmids at an equimolar ratio (500 ng each). dies coupled to either horseradish peroxidase or Cy3 were The total amount of other co-transfected constructs was from Dianova (Hamburg, Germany), the anti-b-actin from 150 ng unless otherwise specified. The transcriptional activity Sigma (Taufkirchen, Germany) and the anti-GFP from Roche measured 24 h post transfection is defined as the ratio (Mannheim, Germany). of TOPglow and FOPglow luciferase values normalized to the b-galactosidase values. FOPglow values were not signifi- APC Plasmids cantly affected by the expression of any YFP–APC construct. The plasmids expressing the YFP–APC fusion proteins were constructed by standard molecular biology methods, using Immunofluorescence staining, immunoprecipitation pCMV–APC and YFP–APC (Krieghoff et al., 2006) as a and western blotting template for PCRs and recipient vector, respectively. The Immunofluorescence staining, immunoprecipitation and specific mutations introduced in the 15R and the 20R are western blotting were performed according to (Behrens et al., described in detail in (Kohler et al., 2008). The sequence of any 1998). Data acquisition and processing, as well as experimental of these plasmids is available upon request. pcDNAflag procedures for supplementary Figure 2, are described in (Behrens et al., 1998) was used as a control vector. Supplementary information.

Transfection Plasmids were transfected into cells overnight using 5 ml Conflict of interest polyethylenimine (1 mg/ml) per mg DNA. For transient transfection of plasmids, 2 mg total DNA/200 000 cells/35 mm dish were used. Expression was allowed to proceed for 36 h for The authors declare no conflict of interest. subsequent western blotting, or 24 h for immunofluorescence.

TOP/FOP reporter assays Acknowledgements (van de Wetering et al., 1997). The TOPglow reporter consists of a tandem repeat of four TCF/LEF1-binding sites inserted in We thank G Daum for technical, and A Doebler for secretarial front of a TATA box (Staal et al., 2002), driving the expression assistances. We thank also I Wacker for critical reading of of luciferase in a b-catenin-dependent manner. In the FOPglow the paper. This work was supported by grants from the reporter, the four binding sites are mutated to abolish the Interdisziplina¨res Zentrum fu¨r Klinische Forschung (IZKF) binding of TCF/LEF1 (T-cell factor/lymphoid enhancer factor and the Sander Stiftung to JB.

References

Akiyama T, Kawasaki Y. (2006). Wnt signalling and the actin Ferrarese A, Marin O, Bustos VH, Venerando A, Antonelli M, cytoskeleton. Oncogene 25: 7538–7544. Allende JE et al. (2007). Chemical dissection of the APC Repeat 3 Albuquerque C, Breukel C, van der Luijt R, Fidalgo P, Lage P, Slors multistep phosphorylation by the concerted action of protein FJ et al. (2002). The ‘just-right’ signaling model: APC somatic kinases CK1 and GSK3. Biochemistry 46: 11902–11910. mutations are selected based on a specific level of activation of the Fodde R, Brabletz T. (2007). Wnt/beta-catenin signaling in beta-catenin signaling cascade. Hum Mol Genet 11: 1549–1560. cancer stemness and malignant behavior. Curr Opin Cell Biol 19: Aoki K, Taketo MM. (2007). Adenomatous polyposis coli (APC): a 150–158. multi-functional tumor suppressor gene. J Cell Sci 120: 3327–3335. Gaspar C, Franken P, Molenaar L, Breukel C, van der Valk M, Smits Barker N, Ridgway RA, van Es JH, van de Wetering M, Begthel H, R et al. (2009). A targeted constitutive mutation in the APC tumor van den Born M et al. (2009). Crypt stem cells as the cells-of-origin suppressor gene underlies mammary but not intestinal tumori- of intestinal cancer. Nature 457: 608–611. genesis. PLoS Genet 5: e1000547. Behrens J, Jerchow BA, Wurtele M, Grimm J, Asbrand C, Wirtz R Groden J, Gelbert L, Thliveris A, Nelson L, Robertson M, Joslyn G et al. (1998). Functional interaction of an axin homolog, conductin, et al. (1993). Mutational analysis of patients with adenomatous with beta-catenin, APC, and GSK3beta. Science 280: 596–599. polyposis: identical inactivating mutations in unrelated individuals. Behrens J, von Kries JP, Kuhl M, Bruhn L, Wedlich D, Grosschedl R Am J Hum Genet 52: 263–272. et al. (1996). Functional interaction of beta-catenin with the Groden J, Thliveris A, Samowitz W, Carlson M, Gelbert L, Albertsen transcription factor LEF-1. Nature 382: 638–642. H et al. (1991). Identification and characterization of the familial Clarke AR. (2006). Wnt signalling in the mouse intestine. Oncogene 25: adenomatous polyposis coli gene. Cell 66: 589–600. 7512–7521. Groves C, Lamlum H, Crabtree M, Williamson J, Taylor C, Bass S Clevers H.. (2006). Wnt/beta-catenin signaling in development and et al. (2002). Mutation cluster region, association between germline disease. Cell 127: 469–480. and somatic mutations and genotype-phenotype correlation in Crabtree M, Sieber OM, Lipton L, Hodgson SV, Lamlum H, Thomas upper gastrointestinal familial adenomatous polyposis. Am J Pathol HJ et al. (2003). Refining the relation between ‘first hits’ and ‘second 160: 2055–2061. hits’ at the APC locus: the ‘loose fit’ model and evidence for Gupta PK, Sahota A, Boyadjiev SA, Bye S, Shao C, O’Neill JP et al. differences in somatic mutation spectra among patients. Oncogene (1997). High frequency in vivo loss of heterozygosity is primarily a 22: 4257–4265. consequence of mitotic recombination. Cancer Res 57: 1188–1193. Eklof Spink K, Fridman SG, Weis WI. (2001). Molecular mech- Ha NC, Tonozuka T, Stamos JL, Choi HJ, Weis WI. (2004). anisms of beta-catenin recognition by adenomatous polyposis coli Mechanism of phosphorylation-dependent binding of APC to revealed by the structure of an APC-beta-catenin complex. Embo J beta-catenin and its role in beta-catenin degradation. Mol Cell 15: 20: 6203–6212. 511–521.

Oncogene Functional analysis of adenomatous polyposis coli EM Kohler et al 1671 Haigis KM, Dove WF. (2003). A Robertsonian translocation Nathke I.. (2006). Cytoskeleton out of the cupboard: colon cancer and suppresses a somatic recombination pathway to loss of hetero- cytoskeletal changes induced by loss of APC. Nat Rev Cancer 6: zygosity. Nat Genet 33: 33–39. 967–974. Hong Y, Cervantes RB, Tichy E, Tischfield JA, Stambrook PJ. (2007). Olschwang S, Laurent-Puig P, Groden J, White R, Thomas G. (1993). Protecting genomic integrity in somatic cells and embryonic stem Germ-line mutations in the first 14 exons of the adenomatous cells. Mutat Res 614: 48–55. polyposis coli (APC) gene. Am J Hum Genet 52: 273–279. Kawahara K, Morishita T, Nakamura T, Hamada F, Toyoshima K, Peterson-Nedry W, Erdeniz N, Kremer S, Yu J, Baig-Lewis S, Wehrli Akiyama T. (2000). Down-regulation of beta-catenin by the M. (2008). Unexpectedly robust assembly of the Axin destruction colorectal tumor suppressor APC requires association with Axin complex regulates Wnt/Wg signaling in Drosophila as revealed by and beta-catenin. J Biol Chem 275: 8369–8374. analysis in vivo. Dev Biol 320: 226–241. Kimelman D, Xu W. (2006). beta-catenin destruction complex: Polakis P. (2007). The many ways of Wnt in cancer. Curr Opin Genet insights and questions from a structural perspective. Oncogene 25: Dev 17: 45–51. 7482–7491. Pollard P, Deheragoda M, Segditsas S, Lewis A, Rowan A, Howarth Kohler EM, Chandra SH, Behrens J, Schneikert J. (2009). Beta- K et al. (2009). The APC1322T mouse develops severe polyposis catenin degradation mediated by the CID domain of APC provides associated with submaximal nuclear beta-catenin expression. a model for the selection of APC mutations in colorectal, desmoid Gastroenterology 136: 2204–2213. and duodenal tumours. Hum Mol Genet 18: 213–226. Powell SM, Zilz N, Beazer-Barclay Y, Bryan TM, Hamilton SR, Kohler EM, Derungs A, Daum G, Behrens J, Schneikert J. (2008). Thibodeau SN et al. (1992). APC mutations occur early during Functional definition of the mutation cluster region of adenomatous colorectal tumorigenesis. Nature 359: 235–237. polyposis coli in colorectal tumours. Hum Mol Genet 17: Rowan AJ, Lamlum H, Ilyas M, Wheeler J, Straub J, Papadopoulou A 1978–1987. et al. (2000). APC mutations in sporadic colorectal tumors: A Kotiligam D, Lazar AJ, Pollock RE, Lev D. (2008). Desmoid tumor: a mutational ‘hotspot’ and interdependence of the ‘two hits’. Proc disease opportune for molecular insights. Histol Histopathol 23: Natl Acad Sci USA 97: 3352–3357. 117–126. Rubinfeld B, Albert I, Porfiri E, Munemitsu S, Polakis P. (1997). Loss Krieghoff E, Behrens J, Mayr B. (2006). Nucleo-cytoplasmic distribu- of beta-catenin regulation by the APC tumor suppressor protein tion of \{beta\}-catenin is re. J Cell Sci 119: 1453–1463. correlates with loss of structure due to common somatic mutations Kuraguchi M, Ohene-Baah NY, Sonkin D, Bronson RT, Kucherlapati of the gene. Cancer Res 57: 4624–4630. R. (2009). Genetic mechanisms in Apc-mediated mammary Rubinfeld B, Souza B, Albert I, Muller O, Chamberlain SH, Masiarz tumorigenesis. PLoS Genet 5: e1000367. FR et al. (1993). Association of the APC gene product with beta- Lamlum H, Ilyas M, Rowan A, Clark S, Johnson V, Bell J et al. catenin. Science 262: 1731–1734. (1999). The type of somatic mutation at APC in familial Schneikert J, Behrens J. (2006). Truncated APC is required for cell adenomatous polyposis is determined by the site of the germline proliferation and DNA replication. Int J Cancer 119: 74–79. mutation: a new facet to Knudson’s ‘two-hit’ hypothesis. Nat Med Schneikert J, Behrens J. (2007a). The canonical Wnt signalling 5: 1071–1075. pathway and its APC partner in colon cancer development. Gut Latchford A, Volikos E, Johnson V, Rogers P, Suraweera N, 56: 417–425. Tomlinson I et al. (2007). APC mutations in FAP-associated Schneikert J, Grohmann A, Behrens J. (2007b). Truncated APC desmoid tumours are non-random but not ‘just right’. Hum Mol regulates the transcriptional activity of beta-catenin in a cell cycle Genet 16: 78–82. dependent manner. Hum Mol Genet 16: 199–209. Liu C, Li Y, Semenov M, Han C, Baeg GH, Tan Y et al. (2002). Segditsas S, Tomlinson I. (2006). Colorectal cancer and genetic Control of beta-catenin phosphorylation/degradation by a dual- alterations in the Wnt pathway. Oncogene 25: 7531–7537. kinase mechanism. Cell 108: 837–847. Smits R, Kielman MF, Breukel C, Zurcher C, Neufeld K, Jagmohan- Liu J, Xing Y, Hinds TR, Zheng J, Xu W. (2006). The third 20 amino Changur S et al. (1999). APC1638T: a mouse model delineating acid repeat is the tightest binding site of APC for beta-catenin. critical domains of the adenomatous polyposis coli protein involved J Mol Biol 360: 133–144. in tumorigenesis and development. Genes Dev 13: 1309–1321. McCartney BM, Nathke IS. (2008). Cell regulation by the Apc protein Staal FJ, Noort Mv M, Strous GJ, Clevers HC. (2002). Wnt signals are Apc as master regulator of epithelia. Curr Opin Cell Biol 20: 186–193. transmitted through N-terminally dephosphorylated beta-catenin. Menendez M, Gonzalez S, Obrador-Hevia A, Dominguez A, Pujol EMBO Rep 3: 63–68. MJ, Valls J et al. (2008). Functional characterization of the novel Su LK, Vogelstein B, Kinzler KW. (1993). Association of the APC APC N1026S variant associated with attenuated familial adenoma- tumor suppressor protein with catenins. Science 262: 1734–1737. tous polyposis. Gastroenterology 134: 56–64. Taketo MM. (2006). Wnt signaling and gastrointestinal tumorigenesis Miyaki M, Konishi M, Kikuchi-Yanoshita R, Enomoto M, Igari T, in mouse models. Oncogene 25: 7522–7530. Tanaka K et al. (1994). Characteristics of somatic mutation of the Tickenbrock L, Kossmeier K, Rehmann H, Herrmann C, Muller O. adenomatous polyposis coli gene in colorectal tumors. Cancer Res (2003). Differences between the interaction of beta-catenin with 54: 3011–3020. non-phosphorylated and single-mimicked phosphorylated 20-amino Miyaki M, Yamaguchi T, Iijima T, Takahashi K, Matsumoto H, acid residue repeats of the APC protein. J Mol Biol 327: 359–367. Yasutome M et al. (2008). Difference in characteristics of APC van de Wetering M, Cavallo R, Dooijes D, van Beest M, van Es J, mutations between colonic and extracolonic tumors of FAP Loureiro J et al. (1997). Armadillo coactivates transcription driven patients: variations with phenotype. Int J Cancer 122: 2491–2497. by the product of the Drosophila segment polarity gene dTCF. Cell Miyoshi Y, Nagase H, Ando H, Horii A, Ichii S, Nakatsuru S et al. 88: 789–799. (1992). Somatic mutations of the APC gene in colorectal tumors: Xing Y, Clemens WK, Kimelman D, Xu W. (2003). Crystal structure mutation cluster region in the APC gene. Hum Mol Genet 1: of a beta-catenin/axin complex suggests a mechanism for the beta- 229–233. catenin destruction complex. Genes Dev 17: 2753–2764. Munemitsu S, Albert I, Souza B, Rubinfeld B, Polakis P. (1995). Xing Y, Clements WK, Le Trong I, Hinds TR, Stenkamp R, Regulation of intracellular beta-catenin levels by the adenomatous Kimelman D et al. (2004). Crystal structure of a beta-catenin/ polyposis coli (APC) tumor-suppressor protein. Proc Natl Acad Sci APC complex reveals a critical role for APC phosphorylation in USA 92: 3046–3050. APC function. Mol Cell 15: 523–533.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene