Gramicidin is a good model for studying the structure and function of membrane channels for several reasons. Because of its relatively small size and simple side chains, it doesn't denature quickly and is relatively simple to synthesize and purify. Although Gramicidin can adopt many different helical conformations depending on solvent conditions(3), it exists in a single conformation with a definite, fairly rigid structure in lipid membranes and micelles. It also selectively transports monovalent cations, such as Li+, Na+ and K+(2), as well as uncharged atoms, such as Xe(6), making it possible to observe changes in the function of the channel as a result of changes in amino acid sequence.
The three-dimensional conformation of Gramicidin A in SDS micelles has already been determined, using 2-D NMR spectroscopy.(4,5) In this experiment, an analog of Gramicidin A was produced, in which the tryptophan at position 11 was replaced with glycine. The effect of that substitution on the three dimensional conformation of the channel was studied by comparing it with the known structure for Gramicidin A.
The structure of this G-11 Gram A analog was calculated using several pieces of information. Many bond lengths and angles that do not change with changes in conformation were known from the primary structure of the peptide. Twenty-six hydrogen bonds per channel were also known from previous structural studies. Most importantly, experimental inter-proton distances for the G-11 analog were determined from NOESY experiments. Energy minimization calculations were done using dspace, and visualization of the channel was performed using Insight II, from Biosym.
Gramicidin A was synthesized by solid state method, using a Rainin Protein Technologies PS3 synthesizer. The resin was suspended in equal volumes of DMF and ethanolamine and heated in a water bath at 65øC for 8 hours, with stirring. The cleaved peptide was dried and purified using a size exclusion LC column and a preparative reverse phase HPLC column.
d25SDS from CIL was recrystallized from 95% ethanol and dissolved in 90% deionized H2O, 10% D2O, phosphate buffered to a pH of 6.5. G-11 Gramicidin A was dissolved in d3 trifluoroethanol, and the aqueous solution was added to it. A 50:1 molar ratio of SDS to gramicidin was used, to help ensure that no more than two molecules (one dimer) of gramicidin were incorporated into each micelle. The final concentration of G-11 was 5 mM. The sample was sonicated for five minutes, to encourage micelle formation and gramicidin incorporation. The resulting transparent solution was then filtered through a sintered glass frit and transferred to a 4 mm NMR sample tube.
All NMR spectra were acquired using at 55øC, without spinning, on a Varian VXR 500 MHz NMR operating at 11.74 Tesla. DQCOSY and TOCSY experiments were performed and 1H chemical shift assignments were made for each proton in G-11 Gram A using coupling information and previously assigned spectra for Gramicidin A(5). NOESY spectra were acquired and 130 unambiguous pairs of peaks were assigned. The volumes of each peak on both sides of the diagonal were measured and the average volumes were converted to inter-proton distances in Angstroms.
Using all of these constraints, a matrix was produced containing the maximum and minimum distance that any atom in the molecule could be from any other atom. Random distances within these bounds were chosen for every pair of atoms, and the matrix was smoothed, so that there were no nonsense distances relationships.
An initial three-dimensional structure was produced from these distances, called an embed. Because each distance was randomly selected, this initial structure was extremely distorted. Using Dspace, each embed was put through a series of annealing and energy minimization steps. The molecule was given the mathematical equivalent of heat (kinetic energy), and then subjected to a series of cooling and annealing cycles until a structure was produced which violated the initial constraints as little as possible. These calculations were repeated many times starting from different randomly generated embeds, and a few calculated structures were averaged to get a final structure for the peptide.
Gly-11 Gramicidin A was successfully synthesized and purified. Most proton chemical shifts of Gly-11 Gram A were very similar to those of samples were stable for two to four days at room temperature and at 55øC. After this time, a white gelatinous precipitate formed that did not dissolve when the sample was sonicated. The largest chemical shift changes observed were those of protons on amino acids adjacent to, or 6 amino acids away from glycine 11. Gly-11 structures calculated with and without hydrogen bond information did not differ significantly (data not shown).
Gly-11 was found to adopt a stable dimeric channel conformation in aqueous SDS micelles. Gly-11 was very similar in secondary structure to Gram A, in that both were dimeric, composed of two identical single stranded, right-handed helices, and both channels had nearly the same backbone conformation (Fig 1). Although the positions of most of the aliphatic side chains in the Gly-11 channel were slightly different from their positions in Gram A, each of the tryptophan rings were oriented very similarly in both channels, with the indole N-H protons pointing toward the ends of the channel. This orientation would maximize the interaction between the N-H groups and the polar head groups on the SDS molecules, helping to stabilize position of the channel in the micelle(Fig. 2).
Because this is a preliminary structure, based on only three calculated structures, little can be said about the meaning of the differences in conformation between Gly-11 Gram A and Gram A. No major structural changes were expected to result from this single amino acid substitution. More interesting than the conformational effects of this substitution, are the functional effects. Future studies will compare the way that Gly-11 Gram A binds to and transports cations to the binding and transport of other Gramicidin analogs. This will allow us to see how each amino acid substitution affects the way the channel actually functions, giving us more insight into properties and dynamics of membrane channels.