Selective anchoring in the specific plasma membrane domain: a role in epithelial cell polarity

From the two apical markers studied, one (A1) behaved like the basolateral markers in that it displayed large unextractable fractions in the correct (apical) membrane and was almost completely extractable from the incorrect (basolateral) membrane. A main difference with the basal antigens, however, was that the insoluble pool of A1 did not show any dependence on cell-cell contacts. The nature of the apical membrane-specific cytoskeletal structures that bind A1 is still unclear; it may be speculated that they form part of the terminal web or the microvillar core structures (46) . The second apical marker, A2, was completely extractable under the conditions used in this work. Strikingly, it was also the most stringently polarized and it shared with A1 the ability to polarize in low calcium medium (although A1 is less polarized than A2), in clear contradistinction with the basolateral markers B1 and B2 ( Fig. 5 ; see also reference 82). With these data, we cannot exclude the possibility that A2 is anchored via a "labile" cytoskeletal anchorage that may be broken under our experimental manipulations. On the other hand, the maintenance of A2 polarization may be dependent on a fence mechanism different from the tight junction, or on some other restriction mechanism, perhaps complex aggregation or active exclusion from the basolateral domain. We have calculated that even high aggregation numbers would have a low effect on D (Edidin, unpublished observations). Aggregation of membrane proteins might be a significant factor if "filter"-like fences existed in the bilayer. Such a structure, perhaps incomplete tight junctional elements, might permit the passage of single molecules (as would be the case for the mobile fraction of A1), but restrict the diffusion of aggregates (as in the case of fully polarized antigens like A2). Some indirect evidence suggests that filters may be operational even in confluent monolayers: A2 is partially excluded from the boundaries of the apical domain (83, 84) , while A1 is not. If patching of A1 is induced by addition of a second complete antibody on living cells, on the other hand, the patches are excluded from the same boundary area (Salas, E J. I., unpublished observations). Such an incomplete fence or filter may be present in cells that lack complete tight junctions.