Specific binding of adamantane drugs and direction of their polar amines in the pore of the influenza M2 transmembrane domain in lipid bilayers and dodecylphosphocholine micelles determined by NMR spectroscopy

Abstract

The transmembrane domain of the influenza M2 protein (M2TM) forms a tetrameric proton channel important for the virus lifecycle. The proton-channel activity is inhibited by amine-containing adamantyl drugs amantadine and rimantadine, which have been shown to bind specifically to the pore of M2TM near Ser31. However, whether the polar amine points to the N- or C-terminus of the channel has not yet been determined. Elucidating the polar group direction will shed light on the mechanism by which drug binding inhibits this proton channel and will facilitate rational design of new inhibitors. In this study, we determine the polar amine direction using M2TM reconstituted in lipid bilayers as well as dodecylphosphocholine (DPC) micelles. (13)C-(2)H rotational-echo double-resonance NMR experiments of (13)C-labeled M2TM and methyl-deuterated rimantadine in lipid bilayers showed that the polar amine pointed to the C-terminus of the channel, with the methyl group close to Gly34. Solution NMR experiments of M2TM in DPC micelles indicate that drug binding causes significant chemical shift perturbations of the protein that are very similar to those seen for M2TM and M2(18-60) bound to lipid bilayers. Specific (2)H-labeling of the drugs permitted the assignment of drug-protein cross peaks, which indicate that amantadine and rimantadine bind to the pore in the same fashion as for bilayer-bound M2TM. These results strongly suggest that adamantyl inhibition of M2TM is achieved not only by direct physical occlusion of the channel, but also by perturbing the equilibrium constant of the proton-sensing residue His37. The reproduction of the pharmacologically relevant specific pore-binding site in DPC micelles, which was not observed with a different detergent, DHPC, underscores the significant influence of the detergent environment on the functional structure of this membrane protein.

Figure 1

2D ¹⁵N TROSY-HSQC spectra of 1 mM (monomer concentration) M2TM(19–49) in the absence (a) and presence (b) of 2.5 mM WJ10. The spectra were measured at 313 K in 100 mM DPC micelle (50 mM sodium phosphate, pH 7.5, in 10% D₂O and 90% H₂O) on a cryoprobe-equipped Varian INOVA 600 MHz NMR. Upon drug binding, the signals became better dispersed in the ¹H dimension, and the improved lineshape and uniformity of the linewidths indicate that the bound protein adopts a well-folded conformation. Assignments are labeled for the bound protein and inset shows an expanded view of part of the spectrum. The chemical structure of spiro-piperidine WJ10 (IC₅₀ = 0.92 μM) is shown in the spectrum on the right. For comparison, Amt has an IC₅₀ = 16 μM.^53.

Figure 2

Titration of 0.94 mM M2(19–49) (monomer concentration) by Rmt. The intensities of the indicated peaks from the ¹⁵N ¹H HSQC spectra are plotted as a function of Rmt concentration. The curve has a well-defined break at a molar ratio of one drug/tetramer shown in blue in panel A. The corresponding titration curve expected for tight binding of the drug in a 4 drugs/tetramer complex is shown in red. Panel B shows a least-squares fit, in which the stoichiometry and K_D were allowed to vary, as described in Methods. The curve was generated using best-fit parameters of 0.88 ± 0.04 drug/tetramer and K_D = 3.9 μM. A sensitivity analysis (Fig. S1) showed that the value of K_D was less than or equal to 5 μM under these conditions, although it was not possible to obtain a precise value for K_D under these conditions. Data were collected at 313 K, pH 7.5, in DPC (protein : DPC ratio 1:50), 50 mM sodium phosphate buffer.

Figure 3

Upfield region of 2D ¹³C-edited ¹H NOESY spectra with 200 ms mixing of ¹³C, ¹⁵N-labeled VAG-M2TM with 2 eq Amt (A) and Rmt (B). The left spectra are those of protonated drugs, and the middle spectra are from samples containing perdeuterated Amt and methyl-deuterated Rmt. The right spectra are the difference between the left and middle spectra. (C) The protons that show NOE cross peaks with M2 are highlighted in red in the Amt and Rmt structures. Spectra in (A) and (B) were recorded on a 500 MHz and a 600 MHz spectrometer, respectively. The concentrations were 2 mM peptide, 100 mM DPC, 1 mM Amt or Rmt, and pH 7.5 phosphate buffer as in Fig. 3. The methyl groups of Val were not stereospecifically assigned. (D) The structure of Amt bound in the channel pore in the crystal structure of M2TM , viewed from the C-terminal end. Shown in balls are Gly34 Cα (orange), Val27 sidechains (green), and Ala30 sidechains (cyan). The hydrophobic adamantyl cage (magenta) interacts extensively with the Val27 sidechains, while the polar group (blue) points to the C-terminus in the crystal structure.

Figure 4

Static ²H quadrupolar echo spectra of d₁₅-Rmt for determining the tilt angle of the adamantyl cage in M2TM (residues 22–46) channels versus lipid bilayers. (a) d₁₅-Rmt bound to DMPC bilayers without the protein. (b) d₁₅-Rmt bound to M2TM in DMPC bilayers with 1 drug/tetramer. (c) d₁₅-Rmt bound to M2TM with 4 drugs/tetramer. (d) One of the two degenerate orientations of d15-Rmt in lipid bilayers at 303 K. (e) Orientation of d₁₅-Rmt in M2TM channels at 303 K.

Figure 5

Simulations of the ²H spectra of d₁₅-Rmt at 303 K. Top row: experimental spectra reproduced from Fig. 5. Bottom row: simulated spectra. (a) d₁₅-Rmt bound to DMPC bilayers without M2. Simulation used a 4 : 1 area ratio of the small and large couplings, consistent with the number of equatorial and axial deuterons in the adamantyl cage. (b) d₁₅-Rmt bound to M2 at 1 drug/tetramer. Simulated spectrum used an area ratio of 63% : 27% : 10% for the 36 kHz, 13.3 kHz, and isotropic components, which represent the pore bound, lipid bound and isotropic drugs. (c) d₁₅-Rmt bound to M2(22–46) at 4 drugs/tetramer. Simulated spectrum used an area ratio of 13% : 83% : 4% for the 36 kHz, 12.5 kHz and isotropic components.

Figure 6

2D ¹⁵N-¹³C correlation spectra of Val27, Ser31, Gly34 and Asp44-labeled M2TM in DMPC bilayers without and with Rmt. (a) The spectrum of drug-free peptide. (b) The spectrum of the 1 drug/tetramer sample. (c) The spectrum of the 5 drugs/tetramer sample. Ser31 and Gly34 ¹⁵N chemical shift increases and Val27 Cα chemical shift decreases upon drug binding.

Figure 7

¹³C{²H} REDOR spectra of DMPC-bound M2TM with CD₃-Rmt at 5 drugs/tetramer. Intensity difference between the control (S₀, black) and dephased spectra (S, red) indicate proximity of the ¹³C-labeled residues to the deuterated methyl group. (a) 16.9 ms REDOR spectra of Val27 Cγ1 showing S/S₀ = 1.02 ± 0.04. (b) 15.1 ms REDOR spectra of Ser31 Cα (S/S₀ = 1.02 ±0.03) and Asp44 Cα (S/S₀ = 0.89 ± 0.03). (c) Gly34 Cα REDOR spectra at 18.8 ms, with S/S₀ = 0.81 ± 0.04. The difference spectrum is shown in blue. (d) Schematic of rimantadine structure in the pore, with the polar amine pointing to the C-terminus and the adamantyl cage tilted by ~13°.

Figure 8

¹³C{²H} REDOR spectra of M2TM in DMPC bilayers with CD₃-Rmt at 1 drug/tetramer. (a) 15.1 ms REDOR spectra. S/S₀ values are 1.2±0.16 for Val27 Cγ1, 1.00 ± 0.19 for Ser31 Cα, and 0.92 ± 0.08 for Asp44 Cα. (b) 16.9 ms REDOR spectra of Gly34 Cα. S/S₀ = 0.94 ± 0.07.

Scheme 1

Synthetic scheme for CD₃-rimantadine and d₁₅-rimantadine.