Furthermore, we have examined how these domains are involved in the assembly of ZO into the tight junction complex.Our results suggest that ZO may serve as a link between the proteins of the tight junction and the actin cytoskeleton.Exact details of plasmid construction are available from the authors by request.Expression constructs were introduced into cells using a variation on standard calciumphosphate mediated transfection. Constructs are designated by the amino acids included. Black lines denote the extent of the sequences encoded by each construct.The results of the ZO and occludinbinding studies are indicated at left.The vertical dark shaded box denotes the boundaries of PDZ.Each of these halves contains Oleic acid several previously identified protein domains whose specific roles are unknown in ZO. To address the functional role of these domains in protein binding and tight junction assembly, we created a panel of myc epitopetagged expression constructs which encode overlapping fragments of human ZO and introduced these constructs into cultured MDCK epithelial cells.These proteins can be coimmunoprecipitated from epithelial cell extracts under relatively stringent conditions, suggesting a tight and possibly direct interaction. However, if ZO does bind directly to ZO, then the striking homology between these two proteins suggests that these proteins might also form homodimers in vivo.To address these questions and further elucidate the interaction between ZO and ZO, the epitopetagged fusion proteins were immunoprecipitated from lysates of transfected MDCK cells.Analysis of the epitope tag immunoprecipitates with an antibody, and not the transfected human polypeptide, indicates that the endogenous protein does not coimmunoprecipitate with the fulllength epitopetagged ZO construct under these conditions, nor does it interact with any of the deletion constructs.This result strongly suggests that ZO does not form homodimers in vivo, but instead forms a simple ab heterodimer with ZO.When the same blots were probed with a polyclonal antisera against ZO, we found that only deletion constructs which still contained the second PDZ domain of ZO. Furthermore, a construct in which only PDZ was specifically deleted also fails to coprecipitate ZO, indicate that the PDZ domain mediates the interaction with ZO.To identify the reciprocal occludin binding site on ZO, we performed binding assays between the epitopetagged ZO constructs and a GST fusion protein encoding this aa domain.Specifically bound proteins were detected by immunoblotting for the myc epitope.Panel B shows the bound proteins from a survey of all of the epitopetagged ZO constructs.For example, no binding is detected with the constructs z, z, and z.These observations suggest that sequences within the aa domain between aa and mediate a strong interaction with occludin. Interactions with PDZ do not appear to be required for binding to occludin. These results suggest that the primary occludin binding site resides within the aa domain between aa and. Stable MDCK lines expressing the fulllength of ZO were plated on collagen treated filter inserts and cultured for days before fixation and staining with antibodies against the myc epitope. MDCK cells Clindamycin HCl stably transfected with each construct were dissociated by treatment in PBSEDTA, trypsinized, and replated on glass coverslips at a confluent density.