The adenomatous polyposis coli protein unambiguously localizes to microtubule plus ends and is involved in establishing parallel arrays of microtubule bundles in highly polarized epithelial cells

Loss of functional adenomatous polyposis coli (APC)^* protein occurs early in the progression of colon cancer (Powell et al., 1992). APC has been characterized most extensively in the context of the Wnt signaling pathway, where it is a crucial component of a protein complex that regulates the degradation of β-catenin (Peifer and Polakis, 2000). Recently, APC has emerged as an important cytoskeletal regulator. It binds to microtubules directly (Munemitsu et al., 1994; Näthke et al., 1996; Zumbrunn et al., 2001) and indirectly via EB1 (Su et al., 1995; Askham et al., 2000) and may also be involved in regulating actin dynamics via its interaction with Asef, a Rac-specific nucleotide exchange factor (Kawasaki et al., 2000). Consistent with the idea that APC has links to both F-actin and microtubules, a number of intracellular locations have been described for APC. In subconfluent cells, APC concentrates in clusters at the dynamic ends of microtubules (Näthke et al., 1996; Mimori-Kiyosue et al., 2000; Rosin-Arbesfeld et al., 2001). In highly confluent cultured cells, APC has been described in two major locations: in clusters near the basal surface and near the lateral plasma membrane (Näthke et al., 1996; Reinacher-Schick and Gumbiner, 2001; Rosin-Arbesfeld et al., 2001). The basal clusters require an intact microtubule network, whereas the localization to the lateral membrane is dependent on a stable actin network (Näthke et al., 1996; Rosin-Arbesfeld et al., 2001). Truncation mutations in APC found in sporadic and familial colon cancer lead to loss of its cytoskeletal association (Polakis, 1995; Polakis, 1997; Rosin-Arbesfeld et al., 2001), suggesting that this function of APC is important for the maintenance of normal epithelial function. Mutations in APC manifest themselves most prominently in polarized epithelial cells of the gut, making the localization of APC in polarized cells important to determine. The detailed analysis of APC localization with respect to cytoskeletal organization has been restricted to cultured cells so far. Because the correct three dimensional organization of epithelial tissues is of particular importance for their function, the information obtained from cultured cells may be incomplete. Therefore, we determined the distribution of APC protein in highly polarized epithelial cells in vivo using supporting cells from the organ of Corti in the inner ear (see Fig. 1 A for schematic). The organ of Corti consists mainly of a strip of neuroepithelial tissue. Sensory hair cells and the adjacent supporting cells are arranged in rows that extend along the length of this strip (Lim, 1986; see Fig. 1 A). This organization together with the cytoskeletal organization of the supporting cells is crucial for efficient transmission of vibrations to the sensory hair cells and thus for auditory perception (Patuzzi, 1996). Supporting cells contain an extremely high number of microtubules organized in an apico-basal array that facilitates the detection of low abundance proteins that specifically associate with microtubule ends (Henderson et al., 1994; Tucker et al., 1992, 1995; Mogensen et al., 1997). In comparison, epithelial cells in the gut contain an order of magnitude fewer microtubules, making their preservation and the detection of microtubule ends significantly more difficult.

We found that APC concentrates near the basal plasma membrane of supporting cells where microtubule plus ends terminate in a dense matrix. Hook decoration was used to confirm that the microtubules in these highly polarized cells are oriented with their plus ends near the base and their minus ends near the apex. Ninein, a microtubule minus end binding and anchoring protein is found near the apex of the cell, further supporting this organization (Bouckson-Castaing et al., 1996; Mogensen et al., 2000; Piel et al., 2000). During the development and assembly of this highly structured, polarized microtubule array, APC associates with microtubules as they extend toward the cellular base. These data suggest that APC may play an important role in the stabilization of the microtubule arrays during their formation. This was further supported by our finding that in the cochlea of Min mice, which are heterozygous for APC and express reduced levels of full-length APC protein, these apico-basal microtubule arrays showed a significant reduction in the number of microtubules present in the parallel bundles when compared with wild-type litter mates.

To confirm the specificity of available antibodies against APC, cell lysates from human colonic tumor cells with wild-type (HCT116) and truncated APC (DLD1) (Rowan et al., 2000) and UE1 cultured mouse inner ear cells expressing full-length APC (Lawlor et al., 1999) were probed with a panel of three different polyclonal APC antibodies (Fig. 1, B and C). Affinity-purified antiserum raised against the middle domain of APC, and crude serum raised against the NH₂-terminal domain (Midgley et al., 1997), detected APC as the major protein in all lysates (Fig. 1, B and C). A commercially available, affinity-purified anti-APC antiserum, N15 (raised against the NH₂-terminal domain of APC) did not detect any APC protein in the lysates from human cells, even after prolonged exposure and only very faintly detected APC in mouse UE1 cells. Instead, proteins with molecular masses of 65–85 kD were detected as major bands by this antibody in human cells lysates, and the 65-kD protein was also detected in the mouse cells. After longer exposure (Fig. 1 C), a band that comigrates with full-length APC appeared in blots exposed to N15 in all samples including DLD1 cell lysates, although these cells do not contain full-length APC. It is important to note that the gels shown in Fig. 1 (B and C) were simply scanned and not processed any further. Additionally, the entire gel lanes are depicted showing all proteins above 35 kD. These data confirm that the N15 antibody is not suitable for detecting endogenous APC. For our studies, we used the affinity-purified anti–M-APC antibody. However, immunofluorescence staining with another anti–C-APC antibody (Midgley et al., 1997) gives identical results in cultured cells, and the staining pattern with either the anti–M-APC or anti–C-APC antiserum is independent of fixation conditions.

The association of APC with microtubules in epithelial cells has been well documented and suggests that APC associates primarily with microtubule ends. However, microtubule polarity has only been inferred and never been directly demonstrated relative to APC accumulations (Näthke et al., 1996; Mimori-Kiyosue et al., 2000). Cultured colonic tumor cells, the subject of previous investigations, polarize their membrane domains when grown to confluency on glass coverslips, however, they rarely polarize their microtubule network under these conditions, making it impossible to identify specific microtubule ends and associated proteins unambiguously (unpublished data). To establish the localization of endogenous APC protein in polarized epithelial cells with a well-defined microtubule organization, we used supporting cells isolated from the organ of Corti (Fig. 1 A, schematic). These epithelial cells contain an apico-basal array of several thousand microtubules that provides a large target for end-associated proteins allowing their unambiguous detection.

We examined microtubule polarity in all three types of supporting cells in the organ of Corti: inner pillar cells, outer pillar cells, and Deiters cells. All three contain large microtubule arrays. Mature supporting pillar or Deiters cells contain two microtubule arrays whose ends are anchored at the apex and base of the cell. The apical end of one of the arrays in each cell is situated near the apical centrosome and its centrioles (Fig. 1 A, dark blue). The apical end of the other array is remotely located with respect to the centrosome (>10 μm distant; Fig. 1 A, light blue). The largest arrays occur in the pillar cells and include several thousand microtubules (Henderson et al., 1995; Tucker et al., 1995). In these cells, the microtubules splay at the cell base to either side of cone-shaped fibrous meshworks rather than terminating within them. The ends of many of these microtubules are situated within 0.5 μm of the basal membrane (see Fig. 4 D).

To determine the orientation of the microtubule array in the supporting cells we performed hook decoration experiments in the organ of Corti from three mature and one 6 d cochlea when assembly of the apico-basal array is still progressing (for a detailed description of these results see

We would like to thank Michel Bornens (Institue Curie, Paris) for the gift of the antininein antibody, Matthew Holley (University of Sheffield) for the UE1 cell line, Charles Patek (Western General Hospital, Edinburgh) for the gift of the Min mice, Birgit Lane for support and use of lab space for M.M. Mogensen, John James and Richard Evans-Gowing for assistance with TEM analysis of the Min mutants. We are grateful to Dina Dikovskaya for valuable comments on the manuscript.

This work was supported by grants from the Medical Research Council (Grant G9326558MB) to J.B. Tucker, J.B. Mackie, and M.M. Mogensen, the Wellcome Trust to A.R. Prescott and M.M. Mogensen (Grant 049616/Z/96/Z/WRE/MK/JAT). I.S. Näthke is supported by a Cancer Research United Kingdom Senior Research Fellowship and a Burroughs Wellcome Fund Career Development Award.