|Home | About | Journals | Submit | Contact Us | Français|
A remarkably high pKa of approximately 10.5 has been determined for sugar-binding affinity to the lactose permease of Escherichia coli (LacY), indicating that, under physiological conditions, substrate binds to fully protonated LacY. We have now systematically tested site-directed replacements for the residues involved in sugar binding, as well as H+ translocation and coupling, in order to determine which residues may be responsible for this alkaline pKa. Mutations in the sugar-binding site (Glu126, Trp151, Glu269) markedly decrease affinity for sugar but do not alter the pKa for binding. In contrast, replacements for residues involved in H+ translocation (Arg302, Tyr236, His322, Asp240, Glu325, Lys319) exhibit pKa values for sugar binding that are either shifted toward neutral pH or independent of pH. Values for the apparent dissociation constant for sugar binding (Kdapp) increase greatly for all mutants except neutral replacements for Glu325 or Lys319, which are characterized by remarkably high affinity sugar binding (i.e., low Kdapp) from pH 5.5 to pH 11. The pH dependence of the on- and off-rate constants for sugar binding measured directly by stopped-flow fluorometry implicates koff as a major factor for the affinity change at alkaline pH and confirms the effects of pH on Kdapp inferred from steady-state fluorometry. These results indicate that the high pKa for sugar binding by wild-type LacY cannot be ascribed to any single amino acid residue but appears to reside within a complex of residues involved in H+ translocation. There is structural evidence for water bound in this complex, and the water could be the site of protonation responsible for the pH dependence of sugar binding.
The lactose permease of Escherichia coli (LacY)1 is a member of the major facilitator superfamily (MFS) of membrane transport proteins (1, 2) that can utilize the free energy of an electrochemical H+ gradient to drive uphill accumulation of galactopyranosides (3-9). Alternatively, downhill transport of sugar can drive uphill transport of H+ with generation of , the polarity of which depends on the direction of the sugar concentration gradient (10, 11). Sugar/H+ symport is tightly coupled with a stoichiometry approximating unity (4, 12, 13). Cys-scanning and site-directed mutagenesis has shown that a relatively small number of amino acid side chains are irreplaceable with respect to active transport (14). Thus, Glu126 (helix IV), Arg144 (helix V), Glu269 (helix VIII), Arg302 (helix IX), His322, and Glu325 (helix X) have been implicated as key participants in substrate binding and/or H+ translocation (7). Trp151 (helix V) is also critically involved in sugar binding, undergoing aromatic stacking with the galactopyranoside ring and also H-bonding with Glu269 (15, 16).
Glu126, Arg144, Trp151, and Glu269 form the galactopyranoside-binding/recognition site (17-19), which is located in the middle of the molecule and at the apex of a large hydrophilic cavity open to the cytoplasm (Figure 1A,B, green sticks). H+ translocation involves residues primarily in the C-terminal six-helix bundle, which are positioned across the cavity from the sugar-binding site (Figure 1A,B, cyan sticks). Arg302 (helix IX), His322 (helix X), Tyr236 (helix VII), Glu325 (helix X), Asp240 (helix VII), and Lys319 (helix X) form a H-bond/salt bridge network between helices VII, IX, and X (Figure 1C). Arg302, Tyr236, His322, and Asp240 are within 3Å of each other, and Glu325 and Lys319 flank the network from the cytoplasmic and periplasmic sides, respectively. A charge pair between Asp237 and Lys358 connects helices VII and XI and is important for membrane insertion and stability of LacY but not for sugar/H+ symport (20-22).
LacY is a highly dynamic transport protein in which the sugar and H+ binding sites become alternatively accessible to the periplasmic and the cytoplasmic sides of the membrane (9, 23-27). Despite the availability of high-resolution, three-dimensional structures of LacY and a wealth of biochemical data, the mechanism of the coupling of sugar transport to H+ translocation and of H+ translocation itself remains unclear.
The finding that LacY affinity for galactopyranosides decreases at the alkaline pH (28, 29) with a pKa of about 10.5 (30) and numerous other observations (reviewed in refs 7 and 9) indicate that protonation of LacY is required for sugar binding. Because the Kdapp remains constant from pH 5.5 to pH 9.0, it follows that under physiological conditions LacY is fully protonated with respect to sugar binding. Although amino acid side chains with alkaline pKas such as those of Arg, Lys, or Tyr could be responsible for the high pKa of binding, a hydrophobic environment, as well as clustering of amino acyl side chains, can markedly alter the pKas of ionizable groups in proteins (31, 32).
Here we describe the effects of mutations in residues involved in sugar binding and/or H+ translocation on the pKa for sugar binding. Since replacement of essential side chains can dramatically decrease sugar affinity (29, 33), a highly sensitive fluorometric assay based on galactopyranoside-specific conformational change(s) was used to measure Kdapp for galactosides (30). Kinetic parameters of sugar binding were also determined directly by Trp to p-nitrophenyl α-d-galactopyranoside Förster resonance energy transfer (Trp→NPG FRET) combined with stopped flow (34). The pH dependence of the forward and reverse rate constants indicates that the reverse rate constant (koff)is a major determinant of the affinity change at alkaline pH. The findings demonstrate that the high pKa observed for sugar binding is not related to the pKa of a single amino acid side chain. We suggest that structural water coordinated within the H-bond/salt bridge network in the C-terminal six-helix bundle of LacY may participate in H+ translocation as hydronium ion intermediate(s).
Oligonucleotides were synthesized by Integrated DNA Technologies, Inc. (Coralville, IA). Restriction enzymes were purchased from New England Biolabs (Beverly, MA). The QuickChange II kit was from Stratagene (La Jolla, CA), and β-d-galactopyranosyl-1-thio-β-d-galactopyranoside (TDG) and 4-nitrophenyl α-d-galactopyranoside (NPG) were purchased from Sigma (St. Louis, MO). Fluorophores [2-(4′-maleimidylanilino)-naphthalene-6-sulfonic acid (MIANS) and N-(7-dimethylamino-4-methylcoumarin-3-yl)maleimide (DACM)] were obtained from Molecular Probes/Invitrogen (Eugene, OR) and from Anaspec (San Jose, CA). Dodecyl β-d-maltopyranoside (DDM) was obtained from Anatrace (Maumee, OH). Talon superflow resin was purchased from BD Clontech (Palo Alto, CA). All other materials were of reagent grade obtained from commercial sources.
Construction of mutants, expression in E. coli, and purification of LacY were performed as described (26). All mutants contained an additional Cys residue (V331C in helix X) for labeling with maleimide-based fluorophores (MIANS or DACM) and a C-terminal 6-His tag that was used for affinity purification with a Talon resin. Purified proteins (10-15 mg/mL) in 50 mM sodium phosphate (NaPi; pH 7.5)/0.02% DDM were frozen in liquid nitrogen and stored at -80 °C until use.
Purified LacY mutants (40-50 μM) were labeled at Cys331 with an equimolar concen-tration of fluorophore in 50 mM NaPi (pH 7.0)/0.02% DDM in the presence of 15 mM TDG in order to protect Cys148 against alkylation as described (30). Control experiments with wild-type LacY exhibited essentially no labeling by MIANS under the same conditions.
Steady-state fluorescence was monitored at room temperature using an SLM-Aminco 8100 spectrofluorometer (Urbana, IL) modified by OLIS, Inc. (Bogart, GA), and on a SPEX Fluorolog 3 spectrofluorometer (Edison, NY) as described (30) with excitation and emission wavelengths, respectively, of 330 and 415 nm for MIANS or 397 and 440 nm for DACM. Titrations were recorded after sequential addition of 5-10 μL of concentrated TDG to 2 mL of protein solution (0.4 μM) in 0.02% DDM/50 mM buffers with over-lapping pH ranges. The buffers used were citrate-Pi (pH 5.5-6.5), NaPi (pH 6.5-8.0), N-(1,1-dimethyl-2-hydroxyethyl)-3-amino-2-hydroxypropanesulfonic acid (AMPSO) (pH 8.0-9.5), and 3-(cyclohexylamino)-1-propanesulfonic acid (CAPS) (pH 9.5-11.0). Titration data were corrected for dilution and fluorescence drift at high pH. Data fitting was carried out by using SigmaPlot 10 (Systat Software Inc., Richmond, CA).
Stopped-flow measurements of Trp→NPG FRET were performed at 25 °C using an SLM-Aminco 8100 spectrofluorometer (Urbana, IL) modified with a USA stopped-flow unit (OLIS, Inc., Bogart, GA) as described (34). Protein concentrations were 0.5-2 μM. The same buffers were used as described for steadystate measurements. Typically, protein preincubated at an indicated NPG concentration was rapidly mixed with saturating concentrations of TDG (10-20 mM after mixing). Six to twelve stopped-flow traces were recorded for each data point, averaged, and fitted with a single exponential equation with the built-in Olis Globalworks software package or with SigmaPlot 10 (Systat Software Inc., Richmond, CA).
Lactose accumulation by E. coli T184 expressing wild-type LacY or given mutants was performed as described (35).
The best available structure of C154G LacY was used in a search for the presence of structural water molecules (see Supporting Information Refinement of LacY Structure for details). The refinement included all data in the resolution range from 2.81 to 9.99 Å without σ cutoff (28407 reflections) available from the deposited structure factor file (2cfq-sf.cif).
As shown previously (30), MIANS-labeled wild-type LacY binds TDG with a Kdapp near 1 μM in the pH range 5.5-8.0; the Kdapp is about 5-fold greater at pH 9.5 and 30-fold greater at pH 11.0 (Figure 2; Table 1). Although DACM-labeled LacY exhibits lower affinity (i.e., higher Kdapp), a similar dependence on pH is observed (Table 1).
Residues Glu126, Arg144, Trp151, and Glu269 participate directly in sugar binding (Figure 1B, green sticks), and mutation of each drastically decreases TDG-binding affinity (Figure 2, y axes; Table 1; see also Supporting Information Figure 11). Mutations E126A or E126Q completely abolish sugar binding and transport (36, 37), thereby precluding studies on the effect of pH. The conservative mutant E126D exhibits a Kdapp increased more than 500-fold (Table 1); nevertheless, the pH dependence of sugar binding remains unchanged (Figure 2A; Table 1). Only the conservative replacement of Glu269 with Asp maintains activity; the mutants E269A, Q, N, H, K, G, and Y exhibit no sugar binding (data not shown). Although the mutant E269D exhibits a pKa similar to that of wild type, affinity for TDG decreases about 70 times relative to wild type (Figure 2B; Table 1). The observation is consistent with other evidence (38, 39) indicating that Glu269 may directly interact with the galactopyranosyl ring.
Trp151 is important for correctly orientating bound galactopyranosides, and even relatively conservative replacements with Tyr or Phe decrease sugar affinity about 20 or 50 times, respectively (15). Sugar binding by MIANS-labeled mutant W151Y exhibits a pH dependence profile similar to wild-type LacY with a Kdapp that is about 30 times higher (Figure 2C; Table 1). Cys154 is located one turn of helix V from Trp151 toward the periplasmic side, and mutant C154G binds sugar but does almost negligible transport because it is paralyzed on the periplasmic side (25, 26, 40-42). The pH dependence of sugar binding for the C154G mutant is also similar to WT LacY, with an increase in Kdapp of about 20-fold (Figure 2D).
Arg144, which likely forms a bidentate H-bond with the 3- and 4-OH groups in the galactopyranoside ring (17), is strictly conserved in related symporters (43). Even replacement with Lys leads to complete loss of sugar binding (36, 37). Therefore, replacements for Arg144 were precluded from this study.
Arg302, Tyr236, His322, and Asp240 form an H-bond/salt bridge cluster in the middle of the C-terminal six-helix bundle (Figure 1C, cyan sticks), and any single replacement results in a marked decrease in the affinity for sugar and also a concomitant change in the pH dependence of Kdapp (Figure 3; Table 1; see also Supporting Information Figures 12-14). Two distinct effects on the pH profiles are observed: (i) a significant shift of pH dependence toward lower pKa values, or (ii) suppression of the effect of alkaline pH on sugar affinity. An acid shift of 1.5-2.5 pH units is observed for mutant R302K (Figure 3A; Table 1) and for all Tyr236 mutants, except Y236F and Y236K (Figure 3C,D; Table 1). Kdapp values increase sharply between pH 7.5 and pH 9.5 and approach a plateau with at least a 10-fold decrease in TDG affinity in each mutant. The other effect, suppression of the increase in Kdapp at alkaline pH, is observed with mutant R302A (Figure 3B; Table 1) and for mutants D240N or D240A (Figure 3H; Table 1), as well as for all the His322 replacements (H322K, Y, F, A, N, Q, R; Figure 3E,F; Table 1) and mutants Y236F or Y236K (Figure 3C; Table 1). The only conservative replacement in this group, mutant D240E, does not exhibit a change in pKa relative to wild-type LacY (Figure 3G; Table 1). However, sugar affinity in all of these mutants, particularly the replacements for His322, is drastically decreased (Table 1; Figure 3, y axes).
The most striking effect of mutations is found when Lys319 or Glu325 is replaced with neutral residues in the mutants K319L, K319Q, E325A, and E325Q (Figure 4; Table 1; see also Supporting Information Figure 15). The affinities are similar to those observed for wild-type LacY at pH 6-8 and remain high (i.e., low Kdapp) over the entire pH range from 5.5 to 11. Mutant E325Q was also tested with DACM-labeled V331C LacY (Figure 4C) with the same result: the high-affinity TDG binding is independent of pH. It is remarkable that elimination of an ionizable side chain at position 319 or 325 results in high-affinity TDG binding even at extremely low H+ concentrations.
The affinity for TDG does not change significantly with the mutant K319R (Figure 4A; Table 1) but decreases markedly with the mutant E325D (Figure 4B; Table 1), and both mutants exhibit essentially the same pH dependence as wild-type LacY. These observations are consistent with the notion that neutral replacements of Glu325 and Lys319 completely block transport with little effect on sugar binding, but the protein with conservative replacements at these positions retains at least partial transport activity (29, 35, 44-49) (see also Supporting Information Figure 17).
Rates (kobs) of TDG induced conformational change with MIANS-labeled LacY have been shown to increase at alkaline pH. The forward rate (konapp) does not depend on pH, while the reverse rate (koffapp) is clearly pH dependent, resulting in an increase of Kdapp (Kd = koff/kon) at alkaline pH with pKa of about 10.5 (30). Consistently, a similar effect of pH on the kinetic parameters is observed when sugar binding is measured directly by Trp→NPG FRET. Protein samples equilibrated with NPG were mixed rapidly with TDG at saturating concentrations, and reverse rate constants (koff) were measured directly by recovery of Trp fluorescence after displacement of bound NPG with excess TDG (30, 34). Measurement of the reverse rates is convenient for testing the effect of pH since koff is concentration independent and slower than kobs (kobs = koff + kon[NPG]). Stopped-flow traces recorded at given pH values and NPG concentrations for wild-type LacY and mutant K319L are shown in Figure 5. All traces are fitted with a single-exponential equation, which allows estimation of rates and amplitudes of the fluorescence change. The results demonstrate that (i) the rate of NPG displacement (koff) is independent of NPG concentration with either wild-type LacY (Figure 5A; koff = 80 ± 2 s-1 for traces 1-4) or the K319L mutant (Figure 5B; koff = 70 ± 5 s-1 for traces 1-3), (ii) the amplitudes increase with NPG concentration for both the wild-type LacY (Figure 5A; amplitudes are 9%, 16%, 22%, 25% for 12, 50, 100, 300 μM NPG, respectively) and mutant K319L (Figure 5B; amplitudes are 9%, 26%, 33% for 5, 20, 100 μM NPG, respectively), and (iii) the displacement rate (koff) increases at alkaline pH for wild-type LacY (Figure 5C; koff values are 69 ± 2 s-1, 79 ± 2 s-1, 160 ± 7 s-1, 224 ± 10 s-1 at pH 6.0, 9.0, 10.25, 10.5, respectively), while koff hardly changes with mutant K319L (Figure 5D; koff values are 54 ± 2 s-1, 44 ± 4 s-1, 69 ± 10 s-1 at pH 6.0, 9.0, 10.5, respectively).
The concentration dependence of the change in amplitudes with wild-type LacY shows that affinity for NPG decreases at alkaline pH. Thus, Kd values calculated from hyperbolic fits are 43 ± 5 μM, 30 ± 2 μM, and 320 ± 60 μM at pH 7.0, 9.0, and 10.25, respectively (Figure 6A). The pH dependencies of koff as measured by stopped flow and kon calculated from the Kd are presented in Figure 6D. Clearly, kon is pH independent (1.5 ± 0.9 × 106 M-1 s-1), while koff increases dramatically at alkaline pH with pKa at about 10.5. The pH dependencies of the kinetic constants measured directly here for binding of NPG to unlabeled protein and reported earlier for TDG binding to MIANS-labeled LacY (30) are practically identical: the on-rate is pH independent, while the off-rate increases at alkaline pH with pKa around 10.5. It follows that the pKa of ~10.5 observed with wild-type LacY is an intrinsic property of the protein that reflects binding specifically and is not due to a conformational change or to labeling with fluorophores.
Neutral replacements for Lys319 or Glu325 practically abolish the effect of pH on the kinetic parameters for NPG binding. Kd values estimated from the effect of NPG concentration on fluorescence amplitude are similar to the Kd for wild-type LacY at neutral pH but do not change significantly at alkaline pH (Figure 6B,C; Figure 7A-C). Unlike with wild-type LacY, koff is essentially unchanged in the pH range from 6 to 10.8 (compare Figure 6D with Figure 6E,F and Figure 7D-F). The displacement rate measured in each experiment represents a true value of koff since it is not different from the koff value obtained from the dependence of kobs on NPG concentration (see Figure 8 for example). Calculated kon values for all these mutants are also unaffected by pH. These data provide strong support for the remarkable conclusion that neutral replacements for Lys319 or Glu325 in LacY result in high-affinity sugar binding even at pH 11.
All replacements for amino acid side chains involved in the H-bond/salt bridge cluster (Arg302, Tyr236, His322, or Asp240) decrease sugar affinity (Table 1), thereby complicating pre-steady-state measurements. Nevertheless, the pH dependence of kinetic parameters for NPG binding was measured directly for R302A, D240E, and D240A mutants (Figure 9). The pH dependence of koff for each mutant is very similar to the pH dependence of the Kdapp obtained with the corresponding MIANS-labeled protein. R302A and D240A demonstrate little change of koff with pH, while E240E exhibits a sharp increase of the displacement rate in alkaline pH (compare Figure 9D-F, respectively, with Figure 3B, G,H). Clearly, the rate of sugar release (koff) is a key parameter, which is primarily affected by the protonation state of the H-bond/salt bridge cluster that defines the pH dependence of the Kdapp.
LacY binds galactopyranosides specifically in a manner that is pH dependent (28, 29), and protonation is required for high-affinity binding (reviewed in refs 7 and 9). In addition, it has been shown recently (30) that various galactopyranosides bind to wild-type LacY with pKa of about 10.5, indicating that at physiological pH LacY is protonated. There are several irreplaceable amino acid side chains involved in sugar binding and/or H+ translocation that may be responsible for the alkaline pKa. However, as revealed by the studies presented here, the pKa cannot be assigned to a single amino acid side chain in this group. Although conservative replacement of the side chains directly involved in sugar binding greatly decreases affinity, each replacement exhibits the same pH dependence for TDG affinity as wildtype LacY (Figure 2). Furthermore, Arg302, Tyr236, His322, and Asp240, which do not make direct contact with the galactopyranosyl ring (17), are important for affinity nonetheless, because all replacements markedly increase Kd (Figure 3; Table 1). The conservative mutations R302K or Y236W cause acidic shifts in pKa by ca. 2 pH units (Figure 3A,C). Replacements of Tyr236 with nonaromatic amino acids also cause an acid shift in pKa (Figure 3D). However, Y236F or Y236K, as well as uncharged replacements for Arg302 or Asp240 and all replacements for His322, exhibit low-affinity TDG binding that is largely independent of pH (Figure 3; Table 1). These results indicate that protonation of individual amino acid residues does not determine pKa for sugar-binding affinity in LacY, unlike in the case of KcsA channel where a clear effect of selected mutations in the H+ binding cluster on pH sensitivity is observed (50).
Remarkably, neutral replacements for either Glu325 or Lys319 result in high-affinity galactoside binding over the entire pH range tested (Table 1; Figure 4; Figure 6B,C; Figure 7A-C; Figure 8). In contrast, conservative replacements at these positions exhibit decreased affinity, but pH dependencies are similar to wild-type LacY. Uncharged side chain replacements at positions 325 (29, 44, 45, 48, 49) or 319 (46) block all reactions that require net H+ translocation with little or no effect on equilibrium exchange or counterflow. Pre-steady-state kinetic analysis indicates that the rate of sugar release rises sharply at alkaline pH with wild-type LacY, while it remains practically constant in the pH range 6-11 with uncharged mutants of Glu325 or Lys319. Therefore, the pH dependence of sugar affinity is determined by the koff value, since kon remains constant for wild-type LacY and the mutants studied. Notably, similar studies on the H+-coupled multidrug antiporter EmrE indicate that the koff for tetraphenylphosphonium does not depend on pH, while kon increases markedly at alkaline pH so that affinity increases dramatically (Kd decreases) (51). In other words, the effects of pH on substrate binding kinetic parameters of EmrE are exactly opposite from those observed with LacY.
The residues involved in H+ translocation in LacY do not form a pathway through the membrane (Figure 1) (9). Rather Arg302, His322, Tyr236, and Asp240 form a tightly interconnected H-bond/salt bridge cluster in the middle of the molecule with Glu325 and Lys319 within 5 Å on opposite sides (Figure 1C). Any disturbance in the central core of the H+ translocation site decreases sugar affinity by affecting H+ binding, while uncharged side chains at positions 325 or 319 prevent H+ escape from the cluster and maintain coupling with the sugar-binding site that retains high affinity even at very alkaline pH. Since no single amino acid residue in the central cluster of Arg302, His322, Tyr236, and Asp240 can be identified as an individual H+ acceptor, perhaps coordinated water molecules within the H+ binding site form a hydronium ion intermediate that cannot be deprotonated with neutral replacements for Lys319 or Glu325. By this means, water may behave like a cofactor in H+ translocation.
Such a notion may also explain the phenotypes of the mutants in the H+ translocation site. Mutants with neutral replacements for Glu325 or Lys319 catalyze equilibrium exchange and counterflow because LacY remains protonated, while replacement of the other side chains in the H-bond/salt bridge cluster results in loss of coordinated water molecule(s) so that LacY protonation and coupled sugar binding are impaired. Involvement of hydronium ions in H+ binding and translocation was proposed originally by Boyer for F1/Fo-ATPase (52) and used recently by von Ballmoos and Dimroth (53) to explain the unexpectedly high pKa for the conserved carboxylic acid in the C subunit of the enzyme. Structural water molecules coordinating H+ within the hydrophobic core of bacteriorhodopsin have also been identified (54, 55).
The electron density map of the amino acid residues involved in H+ translocation is well-defined in all X-ray structures of LacY (17-19). Therefore, the structure of C154G LacY with the best overall resolution 2.95 Å (PDB ID 2cfq) (18) is reanalyzed here, and the refinements included additional reflections to 2.81 Å that were omitted originally (see Supporting Information Refinement of LacY Structure for details). There are two strong globular densities within the H+ binding site in the composite omit map (over 5σ contour level) consistent with single water molecules (Figure 10). One water molecule forms an H-bond with Asp240 and another with Lys319 and Thr265 (Figure 10, broken lines). These structural water molecules liganded by residues in the H+ binding site may be important as hydronium ion intermediates during sugar/H+ symport. Moreover, Thr265 is located in helix VIII one turn from irreplaceable Glu269, so that the water molecule connecting Lys319 and Thr265 may participate in the coupling between protonation and sugar binding.
We are indebted to Dr. Douglas Rees (California Institute of Technology) for helpful discussions regarding the refined crystal structure of LacY.
†This work was supported by NIH Grants DK51131, DK069463, GM073210, and GM074929, as well as NSF Grant 0450970 to H.R.K.