Mechanisms of selectivity in channels and enzymes studied with interactive molecular dynamics - PubMed (original) (raw)

Comparative Study

Mechanisms of selectivity in channels and enzymes studied with interactive molecular dynamics

Paul Grayson et al. Biophys J. 2003 Jul.

Abstract

Interactive molecular dynamics, a new modeling tool for rapid investigation of the physical mechanisms of biological processes at the atomic level, is applied to study selectivity and regulation of the membrane channel protein GlpF and the enzyme glycerol kinase. These proteins facilitate the first two steps of Escherichia coli glycerol metabolism. Despite their different function and architecture the proteins are found to employ common mechanisms for substrate selectivity: an induced geometrical fit by structurally homologous binding sites and an induced rapid dipole moment reversal. Competition for hydrogen bonding sites with water in both proteins is critical for substrate motion. In glycerol kinase, it is shown that the proposed domain motion prevents competition with water, in turn regulating the binding of glycerol.

PubMed Disclaimer

Figures

FIGURE 1

Our system for IMD. A customizable view of a running simulation is displayed on a desktop PC as it is computed. The PC displays a pointer that corresponds to the position of the haptic device used to interact with the system. A researcher uses this haptic device to apply a downward force to a simulated molecule of glycerol, feeling resistance in his hand as it moves through the channel.

FIGURE 2

Pathway for pentatol conduction through GlpF, revealed through IMD. (Top) Pathway for ribitol (C1 first). The half-helices M3 and M7 are shown as green cylinders, connected through the NPA motifs to the half-membrane-spanning nonhelical sections (red tubes). Water in the periplasm (top) and cytoplasm (bottom) is shown in licorice representation. Snapshots of ribitol are shown in a larger licorice representation at four key locations: entrance from the periplasmic vestibule (top), selectivity filter (_Trp_48, _Phe_200, and _Arg_206 shown in gold), NPA motifs (_Asn_68 and _Asn_203, also in gold); and exit into the cytoplasm. (Bottom) An expanded view of the selectivity filter and NPA motifs, showing one step in the C5-first conduction of arabitol, at which arabitol adopts an alternating conformation. The two neighboring water molecules are visible, and hydrogen bonds are drawn as dotted lines. Bonds are formed between arabitol's hydroxyl groups and from hydroxyl groups to protein. Motion of the sugar up or down would cause arabitol to exchange hydrogen bonds with the water molecules. All figures of steps in conduction in this article use the equilibrated structures from the end of the noninteractive simulations.

FIGURE 3

Pathway of glycerol unbinding from GK, as revealed through IMD. (A) The equilibrated monomer of glycerol kinase used in this study, in its closed (inactive) form. GK is drawn in a cartoon representation, colored according to its division into three domains: the central hinge (red) and the two regions (purple and blue) that were tilted apart into the open (active) form. Ions are drawn as yellow VDW spheres, and glycerol is visible in VDW representation near the center of the protein. All of the water molecules solvating the protein are drawn, using thin sticks. (B) An expanded view of the GK binding pocket in the closed (left) and open (right) conformations. Glycerol is drawn in licorice representation using two snapshots from the interactive simulation, the initial bound state (below) and the final unbound state (above), to illustrate its unbinding pathway. The residues forming the selectivity filter of the GK binding pocket clockwise from the right, in _gold: Trp_103, _Arg_83, _Asp_245, and _Phe_270. All water molecules within 3 Å of the second hydroxyl group of the initial glycerol are drawn in licorice representation. No water molecules are this near the group in the closed conformation, although two are present in the open conformation.

FIGURE 4

Simulation of the opening and closing of GK. (Top) The constraint angles (straight lines) and average relative angles of the constrained C_α_ carbons (rough lines) in GK domains I (above) and II (below) during the simulation. (Bottom) The aligned C_α_-only RMSD between the simulated structure and the closed (black) and open (gray) crystal structures (from 1GLF and 1GLJ). The minima, maxima, and crossings of the two curves can be used to assess the progress of the simulation (see text).

FIGURE 5

Forces applied during IMD. (Left) The component of the force on ribitol and arabitol along the channel as a function of the distance along the channel, averaged in bins of width 1 Å. The scale of the graph corresponds to that in Fig. 2. Significantly more force needed to be applied before passing the selectivity filter (at ∼5 Å) than after. (Middle) The magnitude of the total force applied to the closed (solid) and open (dashed) GK to extract glycerol from its binding site, averaged in bins of 2-ps width. (Right) The integral of the total force shown in the middle graph over time, with the total impulse applied to the system. Significantly more force needed to be applied for a longer time to extract glycerol from closed GK than from open GK.

FIGURE 6

Orientation of the HCOH group dipole moments. The dipole moments were sampled in bins of 2-Å width from 88-ps simulations that started after the equilibration of steps in the pathways. In GlpF (top) one can discern a reversal of the dipole moments at the NPA motifs and a disruption of the alignment in the selectivity filter. In closed GK (bottom, solid line) there is a large dipole moment reversal at 0 Å; this corresponds to the location of the negatively charged residue _Asp_245. In open GK (bottom, dashed line) a dipole moment reversal is also present, but its magnitude is reduced. The distance axes of the graphs correspond to those in Figs. 2 and 3.

FIGURE 7

Orientation of the side groups that form the selectivity filter in GlpF (top) and the binding pocket in the closed form of GK (bottom), with (orange) and without (blue) ribitol and glycerol. Ribitol is shown with C1 entering first; glycerol is shown in its normal bound state, with C3 (above). The figure shows significant motion of the side groups in both proteins: the groups shown generally moved more in response to the motion of glycerol than any other protein side chains. The RMSD of the protein atoms shown was 0.6 Å for GlpF and 0.9 Å for GK. In GlpF, the motion is consistent with a widening necessary to hold ribitol. In GK, the extraction of glycerol causes _Trp_103 to move toward the exit of the binding pocket, _Asp_245 to tilt outward, and _Arg_83 to move inward.

FIGURE 8

Fluctuation of individual HCOH groups perpendicular to the conduction/unbinding pathways. The root-mean-square (RMS) motion was calculated from 88-ps simulations that started after the equilibration of steps in the pathways. A high-pass filter was applied to eliminate motions with a period longer than 8.8 ps. (Top) The combined values for ribitol and arabitol conduction through GlpF, sampled in bins of width 1 Å. (Bottom) The values for the open (dashed lines) and closed (solid lines) forms of GK, sampled in bins of 2-Å width. In GlpF, fluctuations are reduced in the selectivity filter and to a lesser degree near the NPA motif and a third binding site at 11 Å. In GK, fluctuations are smallest in the binding pocket (<0 Å). The open form of GK shows consistently larger fluctuations than the closed form. The overall range of fluctuations in GK is close to the range in GlpF. The distance axes of the graphs correspond to those in Figs. 2 and 3.

FIGURE 9

The hydrogen bonding patterns of sugar molecules in the ar/R regions of GlpF (left, middle) and GK (right). GlpF is shown with ribitol (C1 first, left) and arabitol (C5 first, middle). Ribitol and arabitol both make hydrogen bonds to _Arg_206 and the carbonyl oxygen of _Gly_199. Ribitol, however, also makes a hydrogen bond to the carbonyl oxygen of _Phe_200. GK is shown with glycerol in its natural orientation (C3, above) and reversed (C3, below; glycerol, in orange). The same hydrogen bonds are formed in both orientations. The two HCOH groups at the bottom of the GK binding pocket assume almost the same conformation in both orientations, whereas the HCOH group nearest to the ATP binding site (above the visible region) is rotated slightly in the reversed orientation.

References

1. Ai, Z., and T. Fröhlich. 1998. Molecular dynamics simulation in virtual environments. Comp. Graph. Forum. 17:C267–C273.
1. Borgnia, M., S. Nielsen, A. Engel, and P. Agre. 1999. Cellular and molecular biology of the aquaporin water channels. Annu. Rev. Biochem. 68:425–458. - PubMed
1. Bystrom, C. E., D. W. Pettigrew, B. P. Branchaud, P. O'Brien, and S. J. Remington. 1999. Crystal structures of Escherichia coli glycerol kinase variant S58→W in complex with nonhydrolyzable ATP analogues reveal a putative active conformation of the enzyme as a result of domain motion. Biochemistry. 38:3508–3518. - PubMed
1. de Groot, B. L., A. Engel, and H. Grubmüller. 2001. A refined structure of human aquaporin-1. FEBS Lett. 504:206–211. - PubMed
1. de Groot, B. L., and H. Grubmüller. 2001. Water permeation across biological membranes: mechanism and dynamics of aquaporin-1 and GlpF. Science. 294:2353–2357. - PubMed

Mechanisms of selectivity in channels and enzymes studied with interactive molecular dynamics - PubMed (original) (raw)