A question of fundamental importance concerning the biosynthesis of integral membrane proteins is whether transmembrane secondary structure can insert spontaneously into a lipid bilayer. It has proven to be difficult to address this issue experimentally because of the poor solubility in aqueous solution of peptides and proteins containing these extremely hydrophobic sequences. We have identified a system in which the kinetics and thermodynamics of alpha-helix insertion into lipid bilayers can be studied systematically and quantitatively using simple spectroscopic assays. Specifically, we have discovered that a 36-residue polypeptide containing the sequence of the C-helix of the integral membrane protein bacteriorhodopsin exhibits significant solubility in aqueous buffers free of both detergents and denaturants. This helix contains two aspartic acid residues in the membrane-spanning region. At neutral pH, the peptide associates with lipid bilayers in a nonhelical and presumably peripheral conformation. With a pKa of 6.0, the peptide inserts into the bilayer as a transbilayer alpha-helix. The insertion reaction proceeds rapidly at room temperature and is fully reversible.

Fourier transform infrared (FTIR) difference spectroscopy and site-directed mutagenesis have been used to investigate structural changes which occur during rhodopsin photoactivation at the level of individual amino acid residues. The rhodopsin-->bathorhodopsin FTIR difference spectra of the mutants Asp-83-->Asn (D83N) and Glu-134-->Asp (E134D) incorporated into membranes are similar to that of native rhodopsin in the photoreceptor membrane, demonstrating that the retinal chromophores of these mutants undergo a normal 11-cis to all-trans photoisomerization. Two bands assigned to the C = O stretching mode of Asp and/or Glu carboxylic acid groups are absent in the D83N rhodopsin-->metarhodopsin II FTIR difference spectrum. Corresponding changes are not observed in the carboxylate C = O stretching region. The most straightforward explanation is that the carboxylic acid group of Asp-83 remains protonated in rhodopsin and its bleaching intermediates but undergoes an increase in its hydrogen bonding during the metarhodopsin I-->metarhodopsin II transition. The mutant E134D produced a normal rhodopsin-->bathorhodopsin and rhodopsin-->metarhodopsin II difference spectrum, but a fraction of misfolded protein was observed, supporting earlier evidence that Glu-134 plays a role in proper protein insertion and/or folding in the membrane.

We report that the replacement of Leu-93 in bacteriorhodopsin by Ala (L93A) or Thr (L93T) slows down the photocycle by -100-fold relative to wild-type bacteriorhodopsin. Time-resolved visible absorption spectroscopy and resonance Raman experimnents, respectively, show the presence of long-lived O4ike and N-like intermediates in the photocycles of the above mutants. We infer the existence of an equilibrium between the N and 0 intermediates in the photocycles of these mutants. The L93A and L93T mutants exhibit normal proton pumping under continuous illumination, suggesting that the decay of the N and/or 0 intermediate, and consequently, proton translocation, can be accelerated by the absorption of a second photon. Since the 13-cis --all-trans reisomerization of retinal is completed during the decay of the N and 0 intermediates, we conclude that the interaction of Leu-93 with retinal is important in this phase of the photocycle. This conclusion is supported by a recent structural model of bacteriorhodopsin that suggests that Leu-93 is near the C-13 methyl group of retinal. intermediate (17). Both isomerization steps are generally believed to be associated with changes in protein conformation. To understand how changes in retinal geometry are coupled to conformational changes in the protein, we are carrying out systematic replacements of amino acids that line the retinal binding pocket. We report here that the interaction of Leu-93, a residue in helix C (Fig. 1), with retinal has an important role in the photocycle and in light/dark adaptation. Time-resolved visible spectroscopic studies show that the half-time for the decay of the 0 intermediate of the photocycle is increased to -800 ms in the mutants where Leu-93 was replaced by Ala (L93A) or Thr (L93T) as compared to a rate of <5 ms for wild-type bR. Resonance Raman (RR) experiments provide evidence for the presence ofa long-lived N intermediate in the photocycles of these mutants. We conclude that Leu-93 in bR is involved in coupling protein conformational changes that occur during the decay of the N and 0 intermediates of the photocycle to 13-cis -+ all-trans reisomerization of the chromophore.

Phospholamban is a 52-amino acid residue membrane protein that regulates Ca(2+)-ATPase activity in the sarcoplasmic reticulum of cardiac muscle cells. The hydrophobic C-terminal 28 amino acid fragment of phospholamban (hPLB) anchors the protein in the membrane and may form part of a Ca(2+)-selective ion channel. We have used polarized attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy along with site-directed isotope labeling to probe the local structure of hPLB. The frequency and dichroism of the amide I and II bands appearing at 1658 cm-1 and 1544 cm-1, respectively, show that dehydrated and hydrated hPLB reconstituted into dimyristoylphosphatidycholine bilayer membranes is predominantly alpha-helical and has a net transmembrane orientation. Specific local secondary structure of hPLB was probed by incorporating 13C at two positions in the protein backbone. A small band seen near 1614 cm-1 is assigned to the amide I mode of the 13C-labeled amide carbonyl group(s). The frequency and dichroism of this band indicate that residues 39 and 46 are alpha-helical, with an axial orientation that is approximately 30 degrees relative to the membrane normal. Upon exposure to 2H2O (D2O), 30% of the peptide amide groups in hPLB undergo a slow deuterium/hydrogen exchange. The remainder of the protein, including the peptide groups of Leu-39 and Leu-42, appear inaccessible to exchange, indicating that most of the hPLB fragment is embedded in the lipid bilayer. By extending spectroscopic characterization of PLB to include hydrated, deuterated as well as site-directed isotope-labeled hPLB films, our results strongly support models of PLB that predict the existence of an alpha-helical hydrophobic region spanning the membrane domain.