Charge clusters and the orientation of membrane proteins

Although hydrophobic forces probably dominate in determining whether or not a protein will insert into a membrane, recent studies in our laboratory suggest that electrostatic forces may influence the final orientation of the inserted protein. A negatively charged hepatic receptor protein was found to respond to trans-positive membrane potentials as though "electrophoresing" into the bilayer. In the presence of ligand, the protein appeared to cross the membrane and expose binding sites on the opposite side. Similarly, a positively charged portion of the peptide melittin crosses a lipid membrane reversibly in response to a trans-negative potential. These findings, and others by Date and co-workers, have led us to postulate that transmembrane proteins would have hydrophobic transmembrane segments bracketed by positively charged residues on the cytoplasmic side and negatively charged residues on the extra-cytoplasmic side. In the thermodynamic sense, these asymmetrically placed charge clusters would create a compelling preference for correct orientation of the protein, given the inside-negative potential of most or all cells. This prediction is borne out by examination of the few transmembrane proteins (glycophorin, M13 coat protein, H-2K^b, HLA-A2, HLA-B7, and mouse Ig μ heavy chain) for which we have sufficient information on both sequence and orientation. In addition to the usual diffusion and pump potentials measurable with electrodes, the "microscopic" membrane potential reflects surface charge effects. Asymmetries in surface charge arising from either ionic or lipid asymmetries would be expected to enhance the bias for correct protein orientation, at least with respect to plasma membranes. We introduce a generalized form of Stern equation to assess surface charge and binding effects quantitatively. In the kinetic sense, dipole potentials within the membrane would tend to prevent positively charged residues from crossing the membrane to leave the cytoplasm. These considerations are consistent with the observed protein orientations. Finally, the electrostatic and hydrophobic factors noted here are combined in two hypothetical models of translocation, the first involving initial interaction of the presumptive transmembrane segment with the membrane; the second assuming initial interaction of a leader sequence.