The tumor suppressor protein APC (Adenomatous Polyposis Coli) is localized in the cytosol and in the nucleus. In this study, we demonstrate that the nuclear APC protein level is high in cells in the basal crypt region of the normal colorectal epithelium. Strikingly, the APC protein staining resembles the staining pattern of a nuclear proliferation marker. As a first step towards a possible role of the nuclear APC protein, we provide data showing the direct interaction of the nuclear APC protein with DNA. A nuclear APC isoform precipitates with matrix-immobilized DNA. Vice versa, the immunoprecipitation of APC from nuclear lysates results in co-precipitation of genomic DNA. Using recombinant APC fragments we mapped three DNA binding domains: one within the β-catenin binding and regulatory domain, and two in the carboxyterminal third of the APC protein. All these three domains contain clusters of repetitive S(T)PXX sequence motifs that were described to mediate the DNA interaction of many other DNA binding proteins. In analogy to S(T)PXX proteins, the APC protein binds preferentially to A/T rich DNA sequences rather than to a single DNA sequence motif.
The tumor suppressor gene APC is inactivated in the majority of human colorectal cancers (for review see Kinzler and Vogelstein, 1996). Insights into the function of the APC gene product came from studies that identified APC interacting proteins and from immunocytochemical experiments that showed the intracellular localization of the APC protein (for review see Polakis, 1997). The best studied interaction between APC and another protein is its binding to β-catenin (Rubinfeld et al., 1993; Su et al., 1993). As a component of the Wnt signaling pathway this interaction is crucial for the control of signal transduction from the cytosol into the nucleus. The formation of a multiprotein complex of β-catenin, the APC protein and several other proteins leads to an increase of the turnover rate of β-catenin and thereby to a decrease of the intracellular β-catenin concentration (Behrens et al., 1998; Hart et al., 1998; Rubinfeld et al., 1996). The tumor specific truncated APC protein does not lower the β-catenin level (Munemitsu et al., 1995; Rubinfeld et al., 1997). As a consequence, β-catenin enters the nucleus in complex with a transcription factor and activates transcription (Behrens et al., 1996; Molenaar et al., 1996).
Recent studies described different intracellular localizations of the APC protein depending on the analysed organism or cell line. Näthke et al. (1996) found the APC protein localized at the margins of microtubules in the lamellipodia of epithelial cells. In cells of the murine intestine, the APC protein was found in the lateral and apical cytoplasm and the nucleus (Miyashiro et al., 1995). High cytosolic and nuclear APC levels were described in cells at the upper villus as well as in differentiated Paneth cells at the basal crypt of murine intestinal epithelial tissue (Wong et al., 1996). More precisely, the colocalization of the nuclear APC with rRNA was described (Neufeld and White, 1997). Until now, nothing is known about a possible function of the nuclear APC.
To further characterize the intracellular localization of the APC protein we analysed tissue sections by immunostaining and cell lysates by Western blot. To get a first idea of a possible role of the nuclear APC we looked for the direct interaction with genomic DNA. Using recombinant APC protein fragments we mapped the DNA binding domain and screened for DNA sequences that are preferentially bound by the APC protein.
Results and discussion
The nuclear level of the APC protein is high in cells in the basal crypt region of the normal colorectal epithelium
The intracellular distribution of the APC protein was evaluated in sections from normal colorectal mucosa by immunostaining with an antibody directed against an aminoterminal epitope of the APC protein. In addition to a faint cytosolic staining, we observed clear nuclear staining in cells located at and close to the basal tubular gland regions (Figure 1a). To verify the nuclear localization cell nuclei were counterstained with DAPI (Figure 1b). Strikingly, the nuclear APC level decreased in apical direction of the tubular glands. The human colon epithelium regenerates from the bottom of the tubular glands by proliferation of stem cells located at the gland bases (Stevens and Lowe, 1992). To detect cells in the epithelium that are in the active phases of the cell cycle, we immunostained for Ki-67, a nuclear proliferation marker (Gerdes et al., 1983) (Figure 1c). Interestingly, Ki-67 and APC showed very similar staining patterns which might indicate a connection between the high nuclear APC level and proliferation.
A 180 kD nuclear isoform of the APC protein binds to DNA
Recently, it was shown that several different isoforms of the APC protein can be detected in cell lysates (Pyles et al., 1998). Therefore, we first looked what isoform of the APC protein is present in cell nuclei. We fractionated cells of the human embryonic kidney cell line (HEK-293) and analysed the subcellular fractions by Western blot developed with an antibody against the aminoterminus of the APC protein (Neufeld and White, 1997). In total lysates of HEK-293 cells we detected four prominent bands of approximately 310, 200, 180 and 150 kD (Figure 2a). Interestingly, in the cytosolic fraction only the 310, 200 and 150 kD bands were detectable, whereas the 180 kD band was specific for the soluble nuclear fraction. All four isoforms and their respective subcellular distribution were also detected with antibodies against carboxyterminal epitopes of the APC protein indicating that the isoforms all include the amino- and the carboxyterminal parts of the APC protein. Similar results have been described recently (Pyles et al., 1998). We obtained comparable results with cells of the human colorectal tumor cell line HCT116 that show high levels of nuclear wild-type APC protein (data not shown) (Heinen et al., 1995; Neufeld and White, 1997).
To investigate whether the endogenous nuclear APC isoform binds to DNA, we incubated the soluble nuclear lystate of HEK-293 cells with matrix-immobilized DNA and pelleted the beads by centrifugation. The flowthrough, wash and pellet fractions were analysed by Western blot. In the pellet fraction we again detected the 180 kD band indicating the precipitation of the nuclear APC isoform by the DNA matrix (Figure 2b).
Immunoprecipitation of nuclear APC co-precipitates DNA
To confirm the interaction of the endogenous APC protein with DNA we attempted to co-precipitate 3H-thymidine labeled genomic DNA by immunoprecipitating APC from nuclear extracts of HEK-293 or HCT116 cells. After isolation and homogenization of the cell nuclei the labeled DNA was fragmented by ultrasonic treatment. The nuclear APC protein was immunoprecipitated from the soluble nuclear lysate with a carboxyterminus specific antibody and the [3H]-specific radioactivity of the washed immunoprecipitates was measured. We found that the anti APC precipitated samples contained at least twice as much [3H]-radioactivity compared to the control without anti APC antibody (Figure 3). Since the precipitated radioactivity can be regarded as a parameter for the amount of APC-DNA complex, the results confirmed that the endogenous APC protein binds to genomic DNA. To verify that the precipitated pellets contained APC protein, the same pellets were analysed by Western blot. As expected, the nuclear 180 kD isoform of the APC protein was detected in the pellets (Figure 3). In control gel shift assays we excluded the possibility that the antibody used for the immunoprecipitation, binds to DNA (data not shown).
Recombinant APC protein fragments interact with DNA
The DNA binding site on the APC protein was mapped using several APC protein fragments that were subcloned, expressed and purified from bacteria (Figure 4). As a first simple screening experiment we tested the electrophoretic mobility of plasmid DNA in an agarose gel in the presence of the purified APC protein fragments (Figure 5) (Wong et al., 1992). In this assay the DNA showed retardation only in presence of three APC fragments: APC(1340 – 1901) that is located within the β-catenin binding region, and APC(2219 – 2580) and APC(2581 – 2843), the two carboxyterminal fragments.
APC(1340 – 1901) binds with lower affinity to DNA than APC(2219 – 2580)
Additional DNA binding assays showed that the three DNA binding APC fragments also bind to other DNA species like PCR fragments and single and double strand oligonucleotides of different lengths and sequences (data not shown). To see which region of the APC protein plays the major role in the interaction with DNA we measured the affinities of APC(1340 – 1901) and APC(2219 – 2580) to the random sequence oligonucleotide ONC1 (Figure 6). The gel shift assay revealed clear differences in the binding affinities. APC(1340 – 1901) binds to DNA with 20 times lower affinity than APC(2219 – 2580).
The DNA binding domains of the APC protein contain clusters of S(T)PXX motifs
To detect putative DNA binding motifs we compared the sequences of the DNA binding domains APC(1340 – 1901) and APC(2219 – 2580). A segment of 29 amino acids with 45% identity was identififed and used to screen for homologous sequences in the databases. Significant homology to several DNA binding proteins was found, with the highest homology to the DNA binding region of histone H1 (Suzuki, 1989a). Histone H1 mediates DNA interaction by the repetitive sequence motifs serine-proline-X-X and threonine-proline-X-X (Suzuki, 1989b). In histone H1 at least one X is lysine or arginine. Clusters of the so-called S(T)PXX motifs were also found in several other DNA binding proteins like RNA-polymerase (at least one X is tyrosine) and gene regulatory proteins (at least one X is serine, threonine, alanine, leucine or proline) (Suzuki, 1989a).
The full length APC protein sequence contains a total of 33 SPXX and eight TPXX motifs, where at least one X is K, R, Y, S, T, A, L or P (Figure 4). Most of these S(T)PXX motifs are located close to the carboxyterminal end of the APC protein. For DNA binding, at least two proximate repetitive S(T)PXX motifs are necessary (Churchill and Suzuki, 1989; Suzuki, 1989a). Thus, the APC protein comprises five clusters of repetitive S(T)PXX motifs that might be able to interact with DNA. Remarkably, these clusters are located in the three recombinant APC protein fragments that we found to interact with DNA in the agarose gel shift assay (Figure 5). APC(1340 – 1901) includes one S(T)PXX cluster consisting of four proximate S(T)PXX motifs between amino acids 1360 and 1383. APC(2219 – 2580) covers three clusters each consisting of five S(T)PXX motifs: between amino acids 2244 and 2286, between 2323 and 2371, and between 2449 and 2488. The carboxyterminal fragment APC(2581 – 2843) includes one S(T)PXX cluster consisting of three S(T)PXX motifs between amino acids 2760 and 2792. Strikingly, the number of S(T)PXX clusters correlated with the different affinities of the protein fragments APC(1340 – 1901) and APC(2219 – 2580) (Figure 6). These findings provided first indications that the S(T)PXX clusters might mediate the interaction of the APC protein with DNA.
APC binds preferentially to A/T rich DNA sequences
To address the question whether the APC protein specifically interacts with a DNA consensus sequence motif we performed a random binding site selection assay (Blackwell and Weintraub, 1990). Starting with a double stranded oligonucleotide that contains a stretch of 20 random nucleotides, we screened for DNA sequences preferentially bound by APC(1340 – 1901) or APC(2219 – 2580) by repeated preparative mobility shifts. After six rounds of selection, the pools of the shifted oligonucleotides were cloned, and the inserts of 67 individual clones of the APC(1340 – 1901) shifted oligonucleotide pool and of 113 clones of the APC(2219 – 2580) shifted pool were sequenced. The statistical analysis of the sequences did not reveal a single DNA consensus motif that is recognized by either of the two APC fragments. However, we found a striking preference of both DNA binding fragments for A/T rich sequences: 65% of all variable bases in the APC(1340 – 1901) shifted oligonucleotide pool were A or T. Totally, the 1340 variable bases in this oligonucleotide pool contained 457 A, 413 T, 259 C and 211 G. 62% of all variable bases in the APC(2219 – 2580) shifted oligonucleotide pool were A or T. Totally, the 2260 variable bases of this pool included 736 A, 678 T, 421 C and 425 G.
These findings were affirmed in gel shift competition assays by comparing the affinity of APC(2219 – 2580) to the random sequence oligonucleotide ONC2 with its affinity to the A/T rich oligonucleotide ONAT. The APC(2219 – 2580) induced mobility shifts of labeled ONC2 were competed by increasing stoichiometric amounts of unlabeled ONC2 or of unlabeled ONAT (Figure 7). ONAT inhibited the APC induced mobility shift more effectively than ONC2 which indicates a higher affinity of the APC protein to A/T rich DNA stretches than to random sequences. These results are in good agreement with results described for other S(T)PXX proteins. Churchill and Suzuki (1989) showed that S(T)PXX peptides interact preferentially with A/T rich sites rather than with a single specific DNA consensus motif.
Hoechst 33258 and APC compete for DNA binding
The interaction between DNA and S(T)PXX peptides can be inhibited by the DNA binding dye Hoechst 33258 (Suzuki, 1989b). This competition was explained by the similar binding modes. Both Hoechst 33258 and S(T)PXX peptides, bind into the minor groove of the DNA helix (Suzuki, 1989b). In analogy, we could show that Hoechst 33258 and the APC protein compete for DNA binding. Hoechst 33258 inhibited the electrophoretic mobility shifts of the A/T rich oligonucleotide ONAT caused by APC(1340 – 1901) or APC(2219 – 2580) (Figure 8).
All three DNA binding regions are found downstream to the corresponding gene region where most of the inherited and sporadic APC gene mutations were identified (Polakis, 1997). Since the majority of APC gene mutations are nonsense or frameshift mutations, the DNA binding domains are lost in most tumor cells carrying APC mutations.
Regarding the two known main functions of the APC protein, the interaction with β-catenin and the binding to microtubules, different roles of the APC/DNA interaction can be assumed. In a nuclear multiprotein complex consisting of APC, β-catenin and a transcription factor of the Lef/TCF family, the APC protein might control the DNA structure or the binding of the complex to the DNA. On the basis of the presented results, it is also tempting to speculate that the APC protein itself acts as a direct transcription regulator that receives its DNA sequence binding specificity by another yet unknown protein. The APC protein binds to the ends of microtubules (Näthke et al., 1996). Provided that APC interacts with microtubules and DNA simultaneously, APC might function as a linker between the microtubule spindles and the chromosomal centromers during mitosis. Thus, the APC protein might play a structural role and might be directly involved in cell proliferation.
Materials and methods
Antibodies and oligonucleotides
Sections from surgically removed macroscopically and microscopically regular colonic mucosa were analysed. Sample preparation and immunohistochemistry were essentially performed according to standard protocols. After formalin-fixation and dehydration samples were paraffin embedded. Free aldehyde and unspecific antibody binding sites were blocked with glycine and with goat serum, respectively. Sections were incubated with primary antibody and developed with Cy 3-coupled anti-mouse IgG. In control experiments the primary antibody was omitted. Cell nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI; Sigma).
Preparation of nuclear extracts
The human cell lines HEK-293 from human embryonic kidney, and HCT116 from human colorectal tumors were cultured under conditions recommended by the supplier (ATCC). For preparation of nuclear lysates, cells in the exponential growth phase were washed twice with cold PBS and lysed in BL (PBS, 0.1% Triton X-100, 0.1% Nonidet P40, 1 mM of each aprotinin, leupeptin, pepstatin). The nuclei were pelleted by centrifugation at 1000 g, resuspended in BL and centrifuged through a cushion of 0.85 M sucrose in BI1 (PBS, 1 mM of each aprotinin, leupeptin, pepstatin) at 16 000 g for 15 min in order to minimize membrane contaminants. The pellet was washed in BI1, and the nuclei were lysed by sonification in 850 mM NaCl. Next, the insoluble nuclear matrix was pelleted by centrifugation at 100 000 g for 20 min. The resulting supernatant containing typically 7.5 mg/ml total protein was defined as soluble nuclear lysate.
APC precipitation with DNA-cellulose
DNA binding of the APC protein present in the nuclear extracts was tested in a DNA-cellulose precipitation assay (Wong et al., 1992). 2.5 – 5 mg cellulose carrying immobilized DNA (Pharmacia/Amersham/Biotec) in BI2 (50 mM Tris/HCl (pH 7.6), 50 mM NaCl, 2.5 mM EDTA, 5% glycerol) was added to the nuclear lysate. After incubation for 60 min at 4°C the matrix was pelleted by centrifugation. The supernatant was defined as flowthrough. After washing the DNA-cellulose twice with BI2, the matrix was pelleted by centrifugation and eluted with SDS sample buffer for 5 min at 95°C. The fractions were analysed by Western blot. As a control, the same experient was performed using standard cellulose (Sigma).
Immunoprecipitation of APC
To test the binding of endogenous nuclear APC proteins to DNA we immunoprecipitated an endogenous APC-DNA complex. HEK-293 or HCT116 cells in the exponential growth phase were labeled with 3H-thymidine (1.3 TBq/mmol) at a concentration of 0.1 μM for 48 h. Cells were harvested in PBS, 1 mM EDTA, 5% glycerol, 0.1% Triton X-100, 0.1% NP40, 1 mM of each PMSF, leupeptin, aprotinin and nuclear extracts were prepared. We received between 80 and 240 μg DNA out of 2 to 6×106 cells labeled with 1 to 2.5×104 c.p.m. per μg DNA (HEK-293 cells) or 4 to 9×104 c.p.m. per μg DNA (HCT116 cells). The average DNA fragment length was reduced and the total yield of soluble DNA was increased by ultrasonic treatment. APC was immunoprecipitated with Ab-6 and the immunocomplexes were pelleted with protein G-Sepharose (Pharmacia/Amersham/Biotec). The [3H] specific radioactivity in the pellets was measured in a scintillation counter. The same precipitates were analysed by Western blot developed with Ab-1.
Electrophoretic mobility shift assays
Gel retardation assays were used for mapping the DNA binding domain, for determining the affinity of the APC protein fragments to DNA, for the binding site selection assay and for competition assays (Wong et al., 1992; Berman et al., 1987). For agarose gel shift experiments 1 μg plasmid DNA (pGEX4T3) was incubated with 100 ng of one of the bacterially expressed and purified APC protein fragments for 30 min, electrophoresed in 0.5 – 0.7% agarose and visualized by ethidium bromide staining. The affinities of APC(1340 – 1901) and APC(2219 – 2580) to DNA were quantified in polyacrylamide gel shift assays. Oligonucleotide ONC1+ was 5′ end labeled using γ32P-ATP polynucleotide kinase and annealed to ONC1−. Varying APC protein concentrations were incubated with 2 nM labeled double stranded ONC1 for 30 min. After electrophoresis in a 10% native polyacrylamide gel, the DNA was detected by autoradiography. The apparent dissociation constant KDapp was defined as the APC protein concentration that is necessary for the shift of 50% ONC1 and calculated as described (Rebar and Pabo, 1994). For competition gel shift assays ONAT or ONC2 was labeled by filling in the 5′ overhanging ends of the double strands with α32P-dCTP. For oligonucleotide competition gel shift assays 2 nM labeled ONC2 were incubated with 12 nM APC(2219 – 2580). Unlabeled ONAT or ONC2 was added at concentrations of relative stoichiometric surplus compared to the concentration of labeled ONC2. For dye competition assays 2 nM labeled ONAT were incubated with 250 nM APC(1340 – 1901) or 12 nM APC(2219 – 2580) for 30 min. Hoechst 33258 (Sigma) was added at concentrations of relative stoichiometric surplus compared to the protein concentration. After electrophoresis in 10% native polyacrylamide gels, the DNA was visualized by autoradiography.
Western blot analysis
The endogenous APC protein in the fractionated cell lysates and the nuclear APC protein precipitated by DNA-cellulose or immunoprecipitated from the nuclear extracts were localized by Western blot analysis. Samples were separated on 5% SDS polyacrylamide gels and electroblotted onto nitrocellulose membranes. Each primary antibody was used at 1 : 200 dilution in TPBS (PBS, 0.5% Tween 20). After incubation with the secondary antibody at 1 : 20 000 dilution, bands were visualized by enhanced chemoluminescence and fluorography (Pharmacia/Amersham/Biotec).
Recombinant APC protein fragments
The full length cDNA of the human APC gene was provided by Dr Paul Polakis (Onyx Pharmaceuticals, Richmond, VA, USA). Figure 4 shows schematically the positions of all APC protein fragments used in this study. Fragments of the cDNA were PCR-subcloned in the bacterial expression vectors pGEX (Pharmacia/Amersham/Biotec), pET (Novagen) or pQE (Qiagen) and expressed in E. coli as fusion proteins. Purification by affinity chromatography and a second chromatography step were performed as described (Deka et al., 1998).
Random binding site selection assay
In a random binding selection assay we screened for DNA sequences that are preferentially bound by the APC protein (Blackwell and Weintraub, 1990). Complementary strand synthesis and 32P labeling of ONBSS was performed by PCR using 5′- and 3′-linker (de Pater et al., 1996). The oligonucleotide pool selected by six rounds of APC induced preparative gel shift was cloned into the pCRII TOPO cloning vector (Invitrogen). Plasmid inserts of 67 different clones of the APC(1340 – 1901) shifted pool and 113 plasmids of the APC(2219 – 2580) shifted pool were sequenced.
Behrens J, Jerchow BA, Würtele M, Grimm J, Asbrand C, Wirtz R, Kühl M, Wedlich D and Birchmeier W. . 1998 Science 280: 596–599.
Behrens J, von Kries JP, Kuhl M, Bruhn M, Wedlich D, Grossschedl R and Birchmeier W. . 1996 Nature 382: 638–642.
Berman J, Eisenberg S and Tye BK. . 1987 Meth. Enzymol. 155: 528–537.
Blackwell TK and Weintraub H. . 1990 Science 258: 1104–1109.
Churchill ME and Suzuki M. . 1989 EMBO J. 8: 4189–4195.
Deka J, Kuhlmann J and Müller O. . 1998 Eur. J. Biochem. 253: 519–597.
de Pater S, Greco V, Pham K, Memelink J and Kijne J. . 1996 Nucl. Acid Res. 24: 4624–4631.
Gerdes J, Schwab U, Lemke H and Stein H. . 1983 Int. J. Cancer 31: 13–20.
Hart MJ, de los Santos R, Albert IN, Rubinfeld B and Polakis P. . 1998 Curr. Biol. 8: 573–581.
Heinen CD, Richardson D, White R and Groden J. . 1995 Cancer Res. 55: 4797–4799.
Kinzler KW and Vogelstein B. . 1996 Cell 87: 159–170.
Miyashiro I, Senda T, Matsumine A, Baeg GH, Kuroda T, Shimano T, Miura S, Noda T, Kobayashi S, Monden M, Toyoshima K and Akiyama T. . 1995 Oncogene 11: 89–96.
Molenaar M, van de Wetering M, Oosterwegel M, Petersen-Maduro J, Godsave S, Korinek V, Roose J, Destree O and Clevers H. . 1996 Cell 86: 391–399.
Munemitsu S, Albert I, Souza B, Rubinfeld B and Polakis P. . 1995 Proc. Natl. Acad. Sci. USA 92: 3046–3050.
Näthke IS, Adams CL, Polakis P, Sellin JH and Nelson WJ. . 1996 J. Cell Biol. 134: 165–179.
Neufeld KL and White RL. . 1997 Proc. Natl. Acad. Sci. USA 94: 3034–3039.
Polakis P. . 1997 Biochim. Biophys. Acta. 1332: 127–147.
Pyles RB, Santoro IM, Groden J and Paryseck LM. . 1998 Oncogene 16: 77–82.
Rebar EJ and Pabo CO. . 1994 Science 263: 671–673.
Rubinfeld B, Albert I, Porfiri E, Fiol C, Munemitsu S and Polakis P. . 1996 Science 272: 1023–1026.
Rubinfeld B, Robbins P, El-Gamil M, Albert I, Porfiri E and Polakis P. . 1997 Science 275: 1752–1753.
Rubinfeld B, Souza B, Albert I, Müller O, Chamberlain S, Masiarz FR, Munemitsu S and Polakis P. . 1993 Science 262: 1731–1734.
Stevens A and Lowe J (eds). . (1992). In: Histology. Gower Medical Publishing, London.
Su LK, Vogelstein B and Kinzler KW. . 1993 Science 262: 1734–1737.
Suzuki M. . 1998a J. Mol. Biol. 207: 61–84.
Suzuki M. . 1998b EMBO J. 8: 797–804.
Wong G, Müller O, Clark R, Conroy L, Moran MF, Polakis P and McCormick F. . 1992 Cell 69: 551–558.
Wong MH, Hermiston ML, Syder AJ and Gordon JI. . 1996 Proc. Natl. Acad. Sci. USA 93: 9588–9593.
We are indebted to Alfred Wittinghofer for fruitful discussions and continuous support. We thank Paul Polakis for providing the cDNA clone of the human full length APC gene. This work was supported by the Association for International Cancer Research, London (Grant 96-28). The work in the laboratory is supported by the Deutsche Forschungsgemeinschaft, the Bundesministerium für Bildung und Forschung, the Heinrich Hertz-Stiftung and Qiagen GmbH.
About this article
Evidence for the Nucleo-Apical Shuttling of a Beta-Catenin Like Plasmodium falciparum Armadillo Repeat Containing Protein
PLOS ONE (2016)
DNA Repair (2016)
FEBS Journal (2015)
Molecular mechanism of adenomatous polyposis coli-induced blockade of base excision repair pathway in colorectal carcinogenesis
Life Sciences (2015)
APC Is an RNA-Binding Protein, and Its Interactome Provides a Link to Neural Development and Microtubule Assembly