Abstract
The membrane assembly of polytopic membrane proteins is a complicated process. Using Chinese hamster P-glycoprotein (Pgp) as a model protein, we investigated this process previously and found that Pgp expresses more than one topology. One of the variations occurs at the transmembrane (TM) domain including TM3 and TM4: TM4 inserts into membranes in an N(in)-C(out) rather than the predicted N(out)-C(in) orientation, and TM3 is in cytoplasm rather than the predicted N(in)-C(out) orientation in the membrane. It is possible that TM4 has a strong activity to initiate the N(in)-C(out) membrane insertion, leaving TM3 out of the membrane. Here, we tested this hypothesis by expressing TM3 and TM4 in isolated conditions. Our results show that TM3 of Pgp does not have de novo N(in)-C(out) membrane insertion activity whereas TM4 initiates the N(in)-C(out) membrane insertion regardless of the presence of TM3. In contrast, TM3 and TM4 of another polytopic membrane protein, cystic fibrosis transmembrane conductance regulator (CFTR), have a similar level of de novo Nin-Cout membrane insertion activity and TM4 of CFTR functions only as a stop-transfer sequence in the presence of TM3. Based on these findings, we propose that 1) the membrane insertion of TM3 and TM4 of Pgp does not follow the sequential model, which predicts that TM3 initiates N(in)-C(out) membrane insertion whereas TM4 stops the insertion event; and 2) "leaving one TM segment out of the membrane" may be an important folding mechanism for polytopic membrane proteins, and it is regulated by the N(in)-C(out) membrane insertion activities of the TM segments.