Cation‐coupled active transport is an essential cellular process found ubiquitously in all living organisms. Here, we present two novel ligand‐free X‐ray structures of the lactose permease (LacY) of Escherichia coli determined at acidic and neutral pH, and propose a model for the mechanism of coupling between lactose and H+ translocation. No sugar‐binding site is observed in the absence of ligand, and deprotonation of the key residue Glu269 is associated with ligand binding. Thus, substrate induces formation of the sugar‐binding site, as well as the initial step in H+ transduction.
Lactose permease (LacY), a galactoside/H+ symporter, is a well‐studied example of ion electrochemical gradient‐driven symporters and antiporters. LacY carries out the coupled stoichiometric translocation of a d‐galactopyranoside with an H+, utilizing the free energy released from downhill translocation of H+ to drive accumulation of galactopyranosides against a concentration gradient. Notably, in the absence of Δμ̃H+ LacY also catalyzes the converse reaction, utilizing free energy released from downhill translocation of sugar to drive uphill translocation of H+ with generation of Δμ̃H+, the polarity of which depends on the direction of the substrate concentration gradient. In the absence of substrate, LacY does not translocate H+; however, LacY catalyzes exchange or counterflow of sugar without translocation of H+. Since substrate gradients by themselves generate a Δμ̃H+ of either polarity, it is likely that the primary driving force for turnover is binding and dissociation of sugar on either side of the membrane (Kaback et al, 2001; Kaback, 2005).
Recently, an X‐ray structure of C154G LacY, which binds ligand as well as wild‐type LacY but catalyzes little transport, was solved in an inward‐facing conformation (Abramson et al, 2003). LacY consists of 12 transmembrane helices organized in two pseudo‐symmetrical α‐helical bundles. The N‐ and C‐terminal six‐helix domains form a large internal cavity open to the cytoplasm. A single sugar‐binding site is observed at the apex of the cavity near the approximate middle of the molecule. The structure confirms many previous findings obtained by site‐directed mutagenesis in conjunction with biochemical and biophysical studies (Kaback et al, 2001; Guan and Kaback, 2006). Only 6 amino acyl side chains are irreplaceable with respect to ligand binding [Arg144 (helix V) and Glu126 (helix IV)] and/or H+ translocation [Glu269 (helix VIII), Arg302 (helix IX), His322 and Glu325 (helix X)] (reviewed in Frillingos et al, 1998; Kaback et al, 2001; Guan and Kaback, 2005). The structure reveals that Arg144 forms a bi‐dentate H‐bond with the O4 and O3 atoms of the galactopyranosyl ring, confirming the irreplaceable role of this residue in sugar binding and recognition (Venkatesan and Kaback, 1998; Sahin‐Tóth et al, 1999; Abramson et al, 2003). Glu126, another irreplaceable residue, is in close proximity to Arg144 and may interact with the O4, O5 or O6 atoms of the galactopyranosyl ring via water molecules. A direct interaction between Arg144 and Glu126 is not observed in the ligand‐bound structure. Trp151 (helix V), an important residue for substrate binding (Guan et al, 2003; Vazquez‐Ibar et al, 2003, 2004), stacks with the galactopyranosyl ring. Glu269 in the C‐terminal domain, which is critical for sugar binding and H+ translocation, may interact with the O3 atom of the galactopyranosyl ring (Weinglass et al, 2003), forms a salt bridge with Arg144 and is in close proximity to Trp151 (Vazquez‐Ibar et al, 2004). Therefore, it was suggested that contacts between Glu269 in the C‐terminal half and Arg144 and Trp151 in the N‐terminal half may be key for coupling between sugar binding and H+ translocation.
His322, Glu325 and Arg302, as well as Tyr236 (helix VII) which is not irreplaceable with respect to activity (Roepe and Kaback, 1989), are probably directly involved in H+ translocation (Kaback et al, 2001; Guan and Kaback, 2006). His322 may be the immediate H+ donor to Glu325, and Arg302 may interact with Glu325 to drive deprotonation (Kaback et al, 2001; Sahin‐Tóth and Kaback, 2001). However, in the ligand‐bound conformation, there is no direct interaction between Arg302, Glu325 and His322. In addition, Glu269 is far removed from the H+ translocation pathway.
To address ligand binding and the mechanism of coupling between substrate and H+ translocation, we describe here two novel crystal structures from the same mutant (C154G LacY) at pH 5.6 and 6.5 without bound substrate.
The structures were obtained from a new crystal form (tetragonal), which exhibits higher‐resolution limits (2.95 and 3.30 Å for the structures at pH 6.5 and 5.6, respectively) than the orthorhombic form used for the sugar‐bound structure (3.6 Å). The data collection and refinement statistics are summarized in the Table I. The overall fold of the two structures is very similar to the TDG‐bound complex from the orthorhombic crystals with RMSD values for the Cα‐atoms of 1.0 and 1.0 Å, respectively (Figure 1A and B), representing ligand‐free inward‐facing conformations.
Ligand binding and induced fit
The sugar‐binding site in both ligand‐free structures exhibits a similar configuration (Figure 2A and B), which differs from the ligand‐bound configuration, strongly indicating that a rearrangement occurs upon binding of sugar (Figure 2A). In the absence of ligand, Arg144 (helix V) is displaced from Glu269 (helix VIII) and forms a salt bridge with Glu126 (helix IV). Helix VIII is slightly removed from the cavity and the cytoplasmic half appears to undergo a small counterclockwise rotation, consistent with biochemical finding (Frillingos and Kaback, 1997). Glu269 is located away from the hydrophilic cavity in a less‐solvent exposed environment and H‐bonded to Trp151, forming a platform for the initial recognition of the galactopyranosyl ring. From these movements, we conclude that the side chains interacting directly with the hydroxyl groups of the sugar are not in a configuration to bind galactopyranosides in the absence of the sugar. Ligand induces formation of the binding site, mainly through side‐chain movements without a global conformation change.
Critical roles of Glu269 in ligand binding and H+ translocation
Glu269 is surrounded by Cys148, Trp151, Asn272 and Met323 in all structures in the absence and presence of ligand. However, in the ligand‐free conformation, in addition to displacement of Arg144 from Glu269, the carboxyl group is closer to hydrophobic Trp151, Cys148 and Ala273; Asn272 is removed from Glu269, suggesting that Glu269 is in a relatively hydrophobic environment (Figure 2C). Thus, it is likely that Glu269 is protonated in both ligand‐free structures, which is consistent with findings from chemical modification, where Glu269 reacts with hydrophobic rather than hydrophilic carbodiimides (Weinglass et al, 2003). Several lines of evidence indicate that substrate binds to protonated LacY. It has been also postulated that the H+ may be shared between Glu269 and His322 (Kaback, 2005; Guan and Kaback, 2006). Here, we suggest that Glu269 may be the primary protonation site and that protonation precedes ligand binding.
In the ligand‐bound structure, Glu325 is in a hydrophobic environment. A similar arrangement is observed in the ligand‐free tetragonal structure at pH 6.5. However, at pH 5.6, Glu325 moves from the hydrophobic environment and becomes closer to His322, Tyr236 and Arg302, forming a salt bridge with His322 and an H‐bond with Tyr236 (Figure 3A and B). It is likely that His322 is protonated at pH 5.6, and the positively charged imidazole ring causes Glu325 to move into a more hydrophilic environment in the deprotonated form. This H‐bond/salt bridge network suggests that protonated His322 may be able to transfer H+ directly to Glu325. A local acidic pH generated by Δμ̃H+ in the outward‐facing confirmation may facilitate H+ translocation from His322 to Glu325. In contrast, Glu126, Trp151 and Arg302 occupy nearly the same positions in each structure.
Based on the structural observations, we propose an ordered mechanism for the initial step in galactoside/H+ symport. In the ligand‐free conformation, the essential residues for substrate binding (Arg144, Glu126 and Glu269) are not in the correct configuration to bind substrate (Figure 4A). Sugar initially recognizes Trp151 through nonspecific hydrophobic stacking between the galactopyranosyl and indole rings (Guan et al, 2003; Vazquez‐Ibar et al, 2003). This interaction orientates the galactopyranosyl moiety for recognition by Arg144, Glu126 and Glu269 (Figure 4B). Moreover, interaction of sugar with Trp151 disrupts the salt‐bridge between Arg144 and Glu126, and a bi‐dentate H‐bond is formed with Arg144 through the O4 and O3 groups on the galactopyranosyl ring (Figure 4C). Subsequently, protonated Glu269 moves out of a relatively hydrophobic environment, deprotonates, forms a salt bridge with Arg144 and an H‐bond with sugar to complete ligand binding (Figure 4C). In other words, sugar binding drives the deprotonation of Glu269, the initial step in H+ translocation. In the ligand‐bound complex, His322 is far from Glu269, and the H+ released from Glu269 may be delivered to His322 through a tightly bound water molecule(s) either on the O3 group of the galactopyranosyl moiety or bound to His322 (Figure 4D). Although bound water cannot be observed at this resolution, it is common in high‐resolution structures of carbohydrates (Mirza et al, 2001) and membrane proteins (Pebay‐Peyroula et al, 1997).
The postulated scheme for lactose influx based on an alternating access model (Abramson et al, 2003; Guan and Kaback, 2004, 2006) is updated in the following manner (Figure 5). Initially, in the absence of ligand, LacY in the outward‐facing conformation is unstable (Figure 5A), and Glu269 is rapidly protonated (Figure 5B). In this conformation, the side chains involved in direct interaction with sugar are not in a configuration to bind galactopyranosides. However, Trp151 and Glu269 are located in the middle of helices V and VIII in the closely opposed interface between the N‐ and C‐terminal helical bundles at the apex of the water‐filled cavity. Therefore, the recognition platform formed by Trp151 and protonated Glu269 may also be present in the outward‐facing conformation (Figure 5B). Initial recognition of galactopyranoside by Trp151 orientates Glu269 and Arg144 in the correct configuration to bind specifically and induces deprotonation of Glu269 as well (Figure 5C). As a consequence, the salt bridge between Arg144 and Glu126 is disrupted and a new salt‐bridge between Arg144 and Glu269 is formed. This switch is induced by substrate to complete formation of binding site. The H+ released from Glu269 is transferred to His322, and rapid transition to the inward‐facing conformation occurs (Figure 5D). His322 may be the immediate H+ donor to Glu325 through the local H‐bond/charge‐pair network. Substrate is then released into the cavity facing the cytoplasm (Figure 5E), and the salt bridge is broken between Arg144 and Glu269 and re‐established between Arg144 and Glu126. The H+ is then released into the same cavity (Figure 5F) due possibly to a decrease in the pKa of Glu325 caused by either approximation to Arg302 (Kaback et al, 2001; Sahin‐Tóth and Kaback, 2001) or exposure to solvent in the aqueous cavity (cytoplasmic pH is constant at 7.6) or both. After releasing the H+, transition back to the outward‐facing conformation occurs.
In conclusion, the data presented here suggest a simple mechanism for the initial step in coupling. However, it is still not clear how sugar binding correlates with the global conformational change that presumably occurs between the inward‐ and outward‐facing conformations or how the H+ is released from the molecule. Furthermore, it is notable (Guan and Kaback, 2006) that the residues involved in H+ translocation are aligned parallel to the plane of the membrane at a level similar to the sugar‐binding site (Figure 1) and may be exposed to a water‐filled cavity in both the inward‐ and outward‐facing conformations. Therefore, the H+, like the sugar, can be released directly into either cavity. In any event, it is apparent that H+ translocation through LacY cannot involve a pathway through the molecule as observed in bacteriorhodopsin (Pebay‐Peyroula et al, 1997; Lanyi, 2004) or cytochrome c oxidase (Iwata et al, 1995; Tsukihara et al, 1996). These structural features may also help explain how lactose‐driven H+ translocation can occur in both directions across the membrane via the same residues.
Materials and methods
Purification and crystallization
The expression and purification were carried out as described (Abramson et al, 2003). The solubilization step was modified by increasing the detergent:protein ratio to 2.5:1.0 (wt/wt; 10 mg protein/ml) (Guan et al, 2006). Both tetragonal crystals were obtained at the late equilibrium stage under conditions used for the orthorhombic crystals (Abramson et al, 2003), which yielded a pH 6.5. To decrease the pH to 5.6, crystals were soaked in buffer (100 mM MES (pH 5.6)/36% PEG 400/200 mM CaCl2/3% 1,6 hexanediol and 8 mM CHAPS) for 2 min before frozen.
Diffraction and data collection
Diffraction data were collected at 100 K from single crystals on beam‐line X06SA (data from pH 6.5) at the Swiss Light Source (Villagen). Diffraction was anisotropic with both crystals. The crystals at pH 5.6 or 6.5 diffracted to 2.95 or 2.6 Å in one and 4.5 or 3.5 Å in the other dimension, respectively. Due to the relative radiation sensitivity of the crystals, a maximum of 10° was collected at any single position of the crystal. Data processing was performed using Denzo and Scalepack (Otwinowski and Minor, 1997). The crystals belong to the space group P43212. Due to anisotropic diffraction, a sigma cutoff of −1 was used during the Scalepack run, preventing merging of very weak or nonexisting reflections with relatively strong reflections.
The structure at pH 6.5 was solved at a resolution of 2.95 Å by molecular replacement with the C154G mutant monomer by using 1PV6 as template (Table I) (Abramson et al, 2003). The structure at pH 5.6 was solved at a resolution of 3.3 Å by isomorphous molecular replacement using the structure at pH 6.5 (Table I). Subsequent refinement of both structures was carried out by using CNS (Brünger et al, 1998) initially with rigid body minimization with three separate rigid bodies (residues 1–190, 191–210 and 211–417) followed by simulated annealing, individual restrained B‐factor refinement and minimization. Approximately 2.5% of the data were set aside for calculation of R‐free values. Manual rebuilding was done by using O (Jones and Kennedy, 1969) with sigma A‐weighted 2Fo−Fc and Fo−Fc electron density maps (Read, 1986). The stereochemistry of the model was evaluated using Procheck (Laskowski et al, 1996). Due to significant anisotropic diffraction with both crystals, local scaling was applied to the data at the later steps of refinement, calculating individual scale factors for 64 different zones of the reciprocal space. This significantly improved the quality of the maps, and an approximate decrease of 1.5% in R‐factors was observed. All refinement statistics are given in Table I. Five mercury ions were included in the refinement. For the structure at pH 6.5, none of the residues were found in the disallowed, 1.4% in the generously allowed, 28.1% in the additionally allowed, and 70.5% in the most favored regions; for the pH 5.6 structure, the corresponding values are none, 0.8, 32.2 and 66.9%, respectively.
The electron densities throughout the structure were coherent, and most of the amino‐acyl side chains were easily assigned with the exceptions of the central cytoplasmic loop (residues 191–205) and the C‐terminal helix (residues 404–417). All residues forming the sugar‐binding site show clear electron densities, and their positions were determined unambiguously. There is an incoherent electron density in the cavity, stacking on Trp151, which could not be modeled. The coordinates for the pH 6.5 and 5.6 structures have been deposited in the Protein Data Bank (entries 2cfq and 2cfp, respectively).
This work is supported by NIH Grants DK51131‐09, 1U54GM074929 and DK069463‐01 (HRK), EU framework programme 6, E‐MEP and the Biotechnology and Biological Research Councils of the UK (SI), and a fellowship from the Alfred Benzon Foundation (OM). We are also indebted to the Swiss Light Source for use of the synchrotron and to Clements Schulze‐Brieze and Momi Iwata for invaluable help with data collection and Jeff Abramson for sharing his expertise.
- Copyright © 2006 European Molecular Biology Organization