(The FASEB Journal. 1998;12:1281-1299.)
© 1998 FASEB
Cys-scanning mutagenesis: a novel approach to structure–function relationships in polytopic membrane proteins
Stathis Frillingos1,a,
Miklós Sahin-tótha,
Jianhua Wua,
and H. Ronald Kabacka,1
a Howard Hughes Medical Institute, Departments of Physiology and Microbiology and Molecular Genetics, Molecular Biology Institute, University of California Los Angeles, Los Angeles, California 90024–1570
 |
ABSTRACT
|
|---|
The entire lactose permease of Escherichia coli, a polytopic membrane transport protein that catalyzes ß-galactoside/H+ symport, has been subjected to Cys-scanning mutagenesis in order to determine which residues play an obligatory role in the mechanism and to create a library of mutants with a single-Cys residue at each position of the molecule for structure/function studies. Analysis of the mutants has led to the following: 1) only six amino acid side chains play an irreplaceable role in the transport mechanism; 2) positions where the reactivity of the Cys replacement is increased upon ligand binding are identified; 3) positions where the reactivity of the Cys replacement is decreased by ligand binding are identified; 4) helix packing, helix tilt, and ligand-induced conformational changes are determined by using the library of mutants in conjunction with a battery of site-directed techniques; 5) the permease is a highly flexible molecule; and 6) a working model that explains coupling between ß-galactoside and H+ translocation.—Frillingos, S., Sahin-Tóth, M., Wu, J., Kabac, H. R. Cys-scanning mutagenesis: a novel approach to structure-function relationships in polytopic membrane proteins. FASEB J. 12, 1281–1299 (1998)
Key Words: active transport • bioenergetics • membrane protein structure • site-directed mutagenesis • ligand binding • conformational change
|
INTRODUCTION
|
|---|
DIFFICULTIES INHERENT in crystallization represent a major obstacle to the study of most polytopic membrane proteins and account for the availability of only a handful of high-resolution structures for this important class of proteins (1–5). Nonetheless, the use of molecular biological techniques to engineer membrane proteins for various site-directed techniques is providing detailed information about the structure and function of the lactose permease (lac permease)3 of Escherichia coli, a paradigm for polytopic membrane proteins (see refs 6–9). By using Cys, which is average in bulk, relatively hydrophobic, and amenable to highly specific modification, site-directed mutagenesis can be used in conjunction with biochemical and biophysical techniques to elucidate membrane topology (10–15) and accessibility of intramembrane residues to the aqueous or lipid phase of the membrane (16–25), as well as spatial proximity between transmembrane domains (23, 26–33). Cys-scanning mutagenesis has now been applied systematically to essentially every residue in lac permease in order to provide insight into the structure and mechanism of this membrane transport protein.
Lac permease, the product of the lacY gene, catalyzes the coupled stoichiometric translocation of ß-galactosides and H+, and is representative of membrane proteins from Archaea to the mammalian central nervous system that transduce free energy stored in electrochemical ion gradients into solute concentration gradients (secondary active transport). The permease has been solubilized from the membrane, purified to homogeneity in a completely functional state (reviewed in ref 34), and shown to act as a monomer (see ref 35). All available evidence indicates that the permease contains 12 
-helices that traverse the membrane in a zigzag fashion connected by relatively hydrophilic loops with both amino and carboxyl termini on the cytoplasmic face (
Fig. 1).
A functional permease molecule devoid of eight native Cys residues has been constructed (C-less permease) (36) and used as a background for Cys-scanning mutagenesis (10, 13, 25, 37–46). Placement of single-Cys residues at every position of the molecule except the carboxyl-terminal 15 amino acid residues, which can be deleted with no effect (47–49), has yielded a library of mutants that represents a unique experimental tool for structure/function studies.
View larger version (39K):
[in this window]
[in a new window]
|
Figure 1. Secondary structure model of lac permease. The one-letter amino acid code is used, and putative transmembrane helices are shown in boxes. Residues irreplaceable with respect to active transport are enlarged: those involved in substrate translocation are black [Glu126 (helix IV) and Arg144 (helix V)]; those involved in H+ translocation and coupling are in green [Glu269 (helix VIII), Arg302 (helix IX), His322 (helix X), and Glu325 (helix X)]. Asp237 (helix VII), and Lys358 (helix XI), which are charge-paired, are in red; Asp240 (helix VII) and Lys319 (helix X), also charge-paired, are in brown. Nonessential residues thought to be involved in substrate translocation [Met145 and Cys148 (helix V); Val264, Gly268, and Asn272(helix VIII)] are encircled. Blue arrowheads represent sites where discontinuities in the permease can be introduced and functional complementation is observed; purple arrowheads indicate sites where functional complementation is not observed. Helices IV and V have been modified based on deletion analysis (C. Wolin and H. R. Kaback, unpublished observations).
|
|
|
CONSTRUCTION OF FUNCTIONAL LAC PERMEASE DEVOID OF CYS RESIDUES
|
|---|
A principal difficulty with Cys-scanning mutagenesis is the complexity resulting from the presence of multiple native Cys residues in most proteins, eight in the case of E. coli lac permease. However, extensive site-specific mutagenesis of the eight residues (50–54) shows that the permease is relatively unaffected by replacement of Cys117, Cys176, Cys234, Cys333, Cys353, or Cys355 with Ser, whereas Cys154 tolerates replacement with Val. Furthermore, although Cys148 interacts directly with the galactosyl moiety of the substrate (see below); it can be replaced with Ala, for example, with complete retention of activity (55). C-less permease is expressed normally in the membrane and catalyzes lactose accumulation to about 60% of the wild-type level with an unperturbed Km (ca. 0.3 mM) (36). In addition, site-directed fluorescence studies indicate that several mutants constructed in C-less permease bind substrate with normal affinity (37, 41, 45, 55–59) Therefore, C-less permease presents an appropriate molecular background for systematic Cys-scanning mutagenesis.
|
401 UNIQUE SINGLE-CYS MUTANTS
|
|---|
By replacing each amino acid residue from position 2 to 402 in C-less permease one-by-one with Cys, 401 unique single-Cys permease mutants have been constructed (
Fig. 1and Fig. 2).
With the exception of residues 402–417 (47–49), the sequence mutagenized corresponds to the entire length of the protein. The great majority of the 401 mutants are expressed normally in the membrane and catalyze accumulation of lactose against a significant concentration gradient, thereby demonstrating that Cys replacement for most residues does not induce severe perturbations in the structure of the permease or in the mechanism of ß-galactoside/H+ symport. Conversely, it is clear that very few residues are irreplaceable. Only 18 single-Cys mutants are unable to catalyze lactose accumulation, and an additional 10 mutants transport to marginally significant levels ( Table 1).
View larger version (58K):
[in this window]
[in a new window]
|
Figure 2. Active lactose transport by single-Cys replacement mutants and the effect of NEM. Upper panel: steady-state levels of lactose accumulation by each mutant expressed as a percentage of C-less permease. Lower panel: rates of lactose transport observed after treatment with NEM expressed as a percentage of the rate in the absence of NEM. The horizontal axes show schematically the 417 residue sequence of E. coli lac permease with the 12 putative -helical transmembrane domains as boxes and the intervening hydrophilic segments as connecting lines. Red bars on the upper horizontal axis identify positions of the six irreplaceable residues (Glu126, Arg144, Glu269, Arg302, His322, Glu325). Blue bars along the lower horizontal axis identify positions of single-Cys LacY mutants that cannot be tested for NEM because of low activity. Finally, the dark red vertical bars in the upper panel identify positions where single-Cys mutants exhibit little or no activity, but other replacements in C-less or wild-type permease support significant levels of accumulation [Pro31 Gly (wt), His35Arg/His39Arg (wt), Gly64Ala (wt), Asp68Ser/Thr45Arg (wt), Leu76Cys (wt), Gly115Ala, Lys131Cys (wt), Phe140Cys (wt), Gly147Ala, Trp151Phe (wt), Ala177Cys (wt), Leu184Cys (wt), Asp237Cys/Lys358Cys, Asp240Cys/Lys319Cys, Pro327Ala (wt), Phe334Leu, Ser346Cys (wt), Thr348Ser, Tyr350Phe (wt), Leu400Cys (wt)]. For clarity, levels of accumulation by mutants constructed in wild-type permease (wt) are given as percentages of the level observed with wild-type permease.
|
|
The 28 mutants can be classified into four subsets with respect to their properties. 1) Those mutants in which Cys replacement inactivates, but other substitutions are well tolerated (e.g., Ala or Gly for Pro31, Ala for Gly64 or Gly147, Leu or Trp for Phe334), suggesting that there is a requirement for appropriate side chain volume and/or hydrophobicity at these positions. 2) Those mutants in which Cys or other replacements inactivate C-less permease, but the same mutations are tolerated in wild-type permease, suggesting interaction of the altered side chain with one or more of the eight native Cys residues in wild-type permease. Thus, Ala177 and/or Leu184 in helix VI may interact with Cys154 in helix V (42), Gly64 in helix II with Cys234 in helix VII (31, 38, 60, 61), and Thr348 and/or Ser346 in helix XI with Cys355 in the same helix and/or Cys333 in helix X (S. F., J. Sun and H. R. Kaback, unpublished observations). 3) Mutants that can be rescued by second-site suppressor mutations. Permease inactivated by neutral replacement of Asp237 or Asp240 in helix VII is rescued respectively by neutral replacement of Lys358 in helix XI or Lys319 in helix X and vice versa, indicating that Asp237 is ion-paired with Lys358 and Asp240 is ion-paired with Lys319 (62–67). Other mutations in residues such as Asp68 or Gly64, which are thought to be near the cytoplasmic end of helix II ( Fig. 1), are suppressed by second-site mutations on one face of helix VII, but also by a number of mutations on the opposite side of the membrane (68, 69). In addition to providing preliminary evidence that helices II and VII are in close proximity, the behavior of the suppressor mutations complements a body of evidence indicating that lac permease is a highly flexible molecule (see below). 4) Only six positions in the permease are clearly irreplaceable, as judged by the finding that none of the strategies outlined above yield mutants that catalyze active lactose transport. The residues are Glu126 (helix IV), Arg144 (helix V), Glu269 (helix VIII), Arg302 (helix IX), His322, and Glu325 (helix X) ( Figs. 1 and 2; Table 1).
|
NONCRYSTALLOGRAPHIC APPROACHES TO HELIX PACKING
|
|---|
The approaches used involve site-directed mutagenesis—Cys-scanning mutagenesis, in particular—with the express purpose of engineering lac permease for biochemical and biophysical techniques that yield both static and dynamic information. In most instances, construction of the mutants begins with a cassette lacY gene (36) (EMBL X-56095) containing a unique engineered restriction site about every 100 bp that encodes lac permease with a single-Cys residue at a specified position. With a library of mutants containing a single-Cys residue at each position in the permease encoded by a cassette gene, it is a simple operation to construct paired-Cys replacement mutants by restriction fragment replacement. Furthermore, by inserting the biotin acceptor domain from a Klebsiella pneumoniae oxaloacetate decarboxylase in either the middle cytoplasmic loop or at the carboxyl terminus, the mutant proteins are readily purified in a single step by monovalent avidin affinity chromatography (70). By using this library of mutants in conjunction with a battery of biochemical, biophysical, and immunological techniques that include second-site suppressor analysis and site-directed mutagenesis, excimer fluorescence, engineered divalent metal binding sites, chemical cleavage, electron paramagnetic resonance, thiol cross-linking, and identification of discontinuous monoclonal antibody epitopes, a helix packing model has been formulated ( Fig. 3;
see refs 8, 9 for a more detailed description of the techniques).
View larger version (36K):
[in this window]
[in a new window]
|
Figure 3. Helix packing of the lac permease viewed from the cytoplasmic surface. The positions of the four residues irreplaceable for coupling [Glu269 (helix VIII), Arg302 (helix IX), His322 (helix X), and Glu325 (helix X)] are enlarged and in blue. Positions of the two interacting pairs of Asp-Lys residues [Asp237 (helix VII)/Lys358 (helix XI), and Asp240 (helix VII)/Lys319 (helix X)] are enlarged and in red. Positions of NEM-sensitive Cys replacements are indicated with a small green dot. Positions of residues involved in substrate recognition and/or translocation [Glu126 (enlarged), Arg144 (enlarged), Met145, Cys148, Val264, Gly268, and Asn272] are in yellow. Positions shown in purple are involved in thiol cross-linking. Placement of helices III and IV is based on recent thiol cross-linking experiments between Cys residues in periplasmic loops (33) and/or transmembrane helices (J. Wu and H. R. Kaback, unpublished observations).
|
|
Although the two pairs of interacting Asp and Lys residues [Asp237 (helix VII)-Lys358 (helix XI) and Asp240 (helix VII)-Lys319 (helix X)] are not essential for activity, the interactions demonstrate that helix VII is close to helices X and XI. Evidence for the Asp-Lys pairs is derived primarily from second-site suppressor analysis (62, 64), site-directed mutagenesis, and chemical rescue experiments (63, 65–67, 71) demonstrating that neutral replacements for either Asp or Lys residue lead to inactivation; but double neutral replacement of Asp237 and Lys358 or Asp240 and Lys319 or neutralization of the unpaired charge in the single mutants yields permease with highly significant activity. Moreover, Asp237 can be interchanged with Lys358 without loss of activity, whereas reversal of Asp240 and Lys319 inactivates (63). The Asp237-Lys358 (or Lys237-Asp358) pair is needed for optimal insertion of the permease into the membrane, indicating that the carboxyl-terminal half of the permease is inserted into the membrane posttranslationally (65). Thus, permease with double neutral replacements for Asp237 and Lys358 exhibits high activity, but immunoblots reveal low membrane levels of the mutants; pulse-chase and other studies demonstrate that the mutants are stable once inserted into the membrane. In contrast, the Asp240-Lys319 pair is not obligatory for active transport or for folding and insertion, but is needed for maximal activity (63). Furthermore, site-directed excimer fluorescence shows that helix VIII (Glu269) is close to helix X (His322), helix IX (Arg302) is close to helix X (Glu325), helix X is in -helical conformation (His322/Glu325) (72), and helix VIII (Ala273) is close to helix IX (Met299) (73).
Many of the spatial relationships have been confirmed by engineering divalent metal binding sites (bis-His residues) within the permease. These studies demonstrate that permease with bis-His residues at positions 269 and 322, 302 and 325 or 269, or 237 and 358 (but not at positions 240 and 319) binds Mn(II) with a stoichiometry of unity, a Kd in the micromolar range, and an apparent pKa of about 6.3 (74–77). Therefore, although these positions are presumably buried in the membrane, the sites are readily accessible to water. In addition, although D240H/K319H permease does not bind Mn(II), thiol cross-linking of Cys residues at the two positions has been demonstrated (J. Wu and H. R. Kaback, unpublished observations). Site-directed chemical cleavage confirms the positioning of helix X next to helices VII and XI and indicates that helix V is close to helices VII and VIII (60). The relationship between helices V, VII, and VIII has been firmly documented by site-directed spin labeling and thiol cross-linking experiments (61). Similarly, double nitroxide-labeled A273C/M299C permease exhibits spin–spin interaction, confirming the proximity between helices VIII and IX, as demonstrated by excimer fluorescence (73).
Completely independent support for close proximity between helices VIII to XI is provided by the demonstration (13) that monoclonal antibody (mAb) 4B11 binds to an epitope comprised of the last two cytoplasmic loops in the permease. In addition, a portion of the helix packing model has been confirmed by distance measurements between an engineered Cu(II) binding site and spin-labeled single-Cys residues in this region of the permease (78–80).
Most recently, site-directed thiol cross-linking has been used to demonstrate that helix I is close to helix VII, helix II is close to helices VII and XI (31), and helix VI is close to helices V and VIII (J. Wu and H. R. Kaback, unpublished observations). In addition, the alternative method involving cross-linking across an engineered factor Xa protease site (61) was used between Cys residues in periplasmic loops of the permease and has led to the placement of helices III, IV, and XII to complete the helix packing model (33).
Site-directed thiol cross-linking studies with coexpressed amino- and carboxyl-terminal halves of the permease (split N6/C6 permease) have also provided dynamic information. By using iodine-catalyzed disulfide formation and homobifunctional thiol cross-linking agents (32), it has been shown that position 245 (helix VII) is up to 6 Å from positions 52 and 53 (helix II) and that ligand binding results in a symmetrical increase in distance to up to 10 Å ( Fig. 3). Thus, ligand binding causes a translational or scissors-like movement between helices VII and II with little or no rotation. On the other hand, little information regarding the important matter of helix tilt is available. Recent site-directed thiol cross-linking (81) demonstrates clearly that helix II is tilted in such a manner that the periplasmic end of helix II is close to the periplasmic end of helix VII, whereas the cytoplasmic end is close to the cytoplasmic ends of helices XI and XII.
|
SCANNING SINGLE-CYS MUTANTS FOR SENSITIVITY TO ALKYLATION
|
|---|
Resistance to inactivation by membrane-permeant, thiol-specific alkylating agents such as N-ethylmaleimide (NEM) distinguishes C-less permease from wild-type permease and provides the basis for systematic scanning of single-Cys permease mutants for NEM inactivation (36). Apart from the inactive mutants that are untestable ( Table 1), inhibition is greater than 70% for 43 mutants ( Fig. 2and Fig. 4)
and complete for 16 ( Table 2).
Furthermore, formation of the S-(1-ethyl-2,5-dioxopyrrolidin-3-yl)-Cys derivative has been demonstrated in situ for many of the single-Cys mutants examined thus far whether they are in transmembrane helices or in loops (24, 82). Therefore, inactivation of a given single-Cys mutant or the lack thereof is probably not due solely to differences in reactivity, but rather reflects the relative importance of a given Cys residue with respect to the transport mechanism. On the other hand, Cys replacement mutants in helix XII do not react with NEM (P. Gao and H. R. Kaback, unpublished observations), as has been reported for certain transmembrane domains in the metal-tetracycline/H+ antiporter (14, 83, 84). Thus, the reactivity or the lack thereof of a given Cys replacement mutant with NEM appears to be highly dependent on the local environment of the particular Cys side chain.
View larger version (41K):
[in this window]
[in a new window]
|
Figure 4. Secondary structure model of E. coli lac permease showing positions of NEM sensitive single-Cys replacement mutants. Positions permease where the initial rate of transport is inactivated by >60% after treatment with NEM are enclosed in filled circles; positions where the initial rate is inactivated by 40–60% are enclosed in open circles.
|
|
In any event, the argument that NEM inactivation in most cases reflects the relative importance of a given Cys residue with respect to the transport mechanism rather than reactivity vs. lack of reactivity is supported by numerous observations. From a survey of NEM-sensitive single-Cys mutants, the bulk or hydrophobicity of the side chain in wild-type permease is not correlated with the NEM sensitivity of a specific Cys replacement mutant. For example, the positions of 34% of the NEM-sensitive single-Cys mutants are normally occupied by the six smallest amino acid side chains (Gly, Ala, Ser, Cys, Thr, or Asp), and 38% are occupied by the seven bulkiest (Phe, Ile, Leu, Tyr, Trp, Arg, or Lys) ( Fig. 4). Similarly, the six smallest residues occupy 28% and the seven bulkiest 44% of the 401 positions mutagenized. Therefore, a more subtle change than an increase in side chain volume of the modified Cys residues must account for the inhibition observed.
The distribution of NEM-sensitive single-Cys mutants in the sequence of the permease is shown in Fig. 4. Essentially none are found in helices III, VI, IX, or XII, and the mutants are located primarily in helices I (periplasmic half), II, V, VII (periplasmic half), VIII, X (cytoplasmic half), and XI. In addition, the alkylation-sensitive mutants in each transmembrane domain cluster on one face of a transmembrane helix (10, 25, 40, 41, 43, 44, 55, 57). The arrangement of these helical faces in the tertiary structure is such that they appear to line a cavity or pathway within the permease ( Fig. 3). Thus, the interface between transmembrane helices V, VIII, X, XI, and VII is lined by residues where single-Cys replacement mutants are inactivated by NEM. Ten of the 16 mutants that are completely inactivated ( Table 2) are distributed at opposite ends of helices VII and VIII ( Fig. 3).
|
SUBSTRATE PROTECTION AGAINST ALKYLATION BY NEM
|
|---|
To elucidate the role of helical interfaces where single-Cys replacement mutants are inactivated by NEM, the effect of substrate on inactivation of transport, as well as the reactivity of many mutants with radioactive NEM in situ, was studied (24, 25, 44, 82). Whereas reactivity of many mutants increases in the presence of ligand, indicating widespread conformational changes, the Cys replacement mutants that are protected are few and they cluster on adjacent faces of helices V (Cys148, Met145) and VIII (Val264, Gly268, Asn272) ( Fig. 3). Of these residues, native Cys148 is completely protected against alkylation, interacts weakly and hydrophobically with the galactosyl moiety of the substrate (41, 55, 57), and is accessible to solvent from both sides of the membrane (24, 82). Met145 appears to interact with the glucosyl moiety of lactose, and the interaction is even weaker than that of Cys148. Thus, M145C permease transports lactose well, NEM inactivates by only about 50% (41), and reactivity with 2-(4'-maleimidylanilino)naphthalene-6-sulfonic acid (MIANS) is blocked weakly by substrate (57). On the other hand, ligand binding significantly decreases iodide quenching of MIANS at position 145 (57). Cys replacement mutants at positions 264, 268, or 272, which are on one face of helix VIII ( Fig. 3), exhibit significant transport activity that is inactivated by NEM, and both inactivation and reactivity with NEM are decreased in the presence of substrate (25). In contrast, a Cys in place of Thr265 on the same face of helix VIII as positions 264, 268, and 272 exhibits a dramatic increase in reactivity with NEM in the presence of ligand, and the increase is blocked by an impermeant thiol reagent. Therefore, the substrate-induced decrease in NEM reactivity exhibited by V264C, G268C, or N272C permease may reflect a conformational change involving one face of helix VIII rather than direct steric interaction of the three residues with substrate. In any case, the five residues clearly play a relatively minor role in substrate recognition and/or translocation.
|
GLU126 AND ARG144
|
|---|
Glu126 and Arg144, which initially were placed by hydropathy profiling (85) at the membrane–water interface at the cytoplasmic ends of helices IV and V, respectively ( Fig. 1), were among the last 18 residues in the permease to be mutagenized (40) because two or six contiguous His residues can be inserted into loop IV/V with little or no effect on activity (86). Thus, it was surprising when Glu126 and Arg144 were found to be irreplaceable. Furthermore, recent studies using single amino acid deletions (C. Wolin and H. R. Kaback, unpublished observations), nitroxide-scanning, accessibility measurements (M. Zhao, J. Hernandez-Borrell, W. L. Hubbell and H. R. Kaback, unpublished observations), and lac permease fusions with the NG domain of FtsY (E. Bibi, personal communication) indicate that loop IV/V is much smaller than indicated by hydropathy profiling, extending only from about Val132 to Phe138 ( Fig. 1).
Replacement of either Glu126 or Arg144 with various neutral amino acids in either the wild-type or C-less background completely abolishes active transport (40). Moreover, rescue is not observed with double neutral substitutions nor is active transport observed when Glu126 and Arg144 are interchanged. The only mutations that exhibit significant transport activity are Asp in place of Glu126, which exhibits wild-type steady-state levels of accumulation but a Km that is at least sixfold higher than wild-type with a similar Vmax (M. Sahin-Tóth and H. R. Kaback, unpublished observations), or Lys in place of Arg144, which transports lactose at a very poor rate to a steady state that is only about 25% of wild-type. It is also significant that lactose-induced H+ translocation is observed at a slow rate with E126D permease, but not with E126A or E126Q or with any of the replacement mutants for Arg144 (R144A or R144K) (J. le Coutre, J. Lee, and H. R. Kaback, unpublished observations). Finally, none of the neutral replacement mutants for either Glu126 or Arg144 catalyze efflux, equilibrium exchange, or counterflow.
Recent experiments (86a) provide evidence that Glu126 and Arg144 play a direct role in substrate binding and recognition. Replacement of either Glu126 or Arg144 with Ala in permease with a single-Cys residue at position 148 markedly decreases the reactivity of Cys148 with NEM, indicating that interaction between Glu126 and Arg144 confers a conformation that leads to enhanced reactivity of Cys148, and no substrate protection is observed. Furthermore, normal reactivity of Cys148 is observed when Glu126 and Arg144 are interchanged, suggesting that the two residues interact directly (e.g., via a salt bridge), and no substrate protection against NEM labeling of Cys148 is observed. With Lys in place of Arg144, normal reactivity of Cys148 with NEM is also observed, but, significantly, substrate elicits no protection against NEM labeling. Therefore, although a Lys residue at position 144 can interact with Glu126 in such a fashion that reactivity of Cys148 is normal, the mutant recognizes and binds substrate poorly. In addition, K131A or F140A permease, each of which exhibits significantly decreased activity (see ref 40), has little or no effect on the reactivity of Cys148 with NEM or protection by substrate in the single-Cys148 background, thereby highlighting the specificity of the mutations in Glu126 and Arg144 (K. C. Zen and H. R. Kaback, unpublished observations).
Taken together with observations regarding the roles of Cys148 and Met145 in substrate binding (55, 57), the results are consistent with the model shown in Fig. 5.
The major points are as follows. 1) One of the guanidino N atoms of Arg144 H-bonds to the hydroxyl at the fourth and possibly the third position of the galactosyl moiety of the substrate, an interaction that plays a key role in the substrate specificity of the permease. Galactose binds with low affinity, but has all the properties of any substrate of the permease; glucose, which differs only in the orientation of the hydroxyl group at the fourth position of the pyranose ring, has no affinity whatsoever (57). 2) The other guanidino N is salt bridged with Glu126, and the interaction holds Arg144 and Cys148 in the proper conformation to interact with the galactosyl moiety. One of the oxygen atoms of the carboxylate could also act as an H-bond acceptor from the hydroxyl at the sixth position of the galactosyl moiety. Thus, when Arg144 is replaced with Lys, interaction with Glu126 is still possible and Cys148 reacts normally with NEM, but Lys is unable to H-bond simultaneously to the galactosyl moiety of the substrate. 3) Cys148 interacts weakly and hydrophobically with the galactosyl moiety of lactose and other galactosides (55, 57). Small hydrophobic side chains at position 148 (Ala, Val) generally increase apparent affinity for substrate, whereas hydrophilic side chains (Ser, Thr, Asp) decrease apparent affinity and bulky or positively charged side chains (Phe, Lys) virtually abolish activity. In addition, hydrophilic substitutions (Ser, Thr, Asp) decrease the activity of the permease with respect to monosaccharides (e.g., galactose) relative to disaccharides (e.g., lactose). These and other observations (57) demonstrate that Cys148 is located in the binding site of the permease, interacting weakly and hydrophobically with the galactosyl moiety. 4) As discussed above (cf. previous section), Met145 interacts even more weakly with the glucosyl moiety of lactose.
View larger version (109K):
[in this window]
[in a new window]
|
Figure 5. Putative substrate binding site in lac permease. Helices IV (white) and V (green) are shown with Glu126 (green), Arg144 (white), Met145 (orange) and Cys148 (yellow). The galactosyl and glucosyl moieties of lactose are in light and dark blue, respectively. As indicated, one of the guanidino NH2 groups of Arg144 H-bonds with the hydroxyl group at the C4 and/or C3 positions of the galactosyl moiety, and the other guanidino NH2 group interacts electrostatically with Glu126 which may also act as an H-bond acceptor from the C6-OH of the galactosyl moiety. Cys148 interacts weakly and hydrophobically with the galactosyl moeity, and Met145 interacts even more weakly with the glucosyl part of lactose.
|
|
|
CONSERVATION OF RESIDUES AT POSITIONS WHERE SINGLE-CYS MUTANTS REACT WITH NEM
|
|---|
Although E. coli lac permease shares common structural motifs with a large family of related membrane proteins, it displays strong sequence homology with only two other oligosaccharide/H+ symporters from E. coli and two from other enteric bacteria (87) Conservation of residues among the four homologous transporters (excluding the sucrose transporter CscB) is much greater within the small set of NEM-sensitive positions identified in E. coli lac permease than within the entire sequence ( Table 3).
The differences cannot be explained simply by the preferential localization of NEM-sensitive positions in transmembrane domains. For example, identical residues between E. coli and K. pneumoniae lac permease account for 60% of the entire sequence (242 of 401), 72% of the membrane-spanning segments (179 of 249), and 90% of the positions where transport by the single-Cys replacement mutants is inhibited by NEM (39 of 43). On the basis of this information, site-directed mutagenesis can be used to rationally design permease molecules with altered substrate specificity and identify specific residues involved in substrate recognition. In this respect, two lac permeases that transport lactose much less efficiently than E. coli lac permease (88, 89) differ in only 4 of 43 of the NEM-sensitive positions ( Table 3). More strikingly, the raffinose permease, which displays increased affinity for raffinose and melibiose and no affinity for methyl-ß,D-thiogalactopyranoside (90), differs in only one of the seven residues that may be involved in substrate translocation in E. coli lac permease with Ala in place of Val264.
View this table:
[in this window]
[in a new window]
|
Table 3. Conservation of residues in the family of oligosaccharide/H+ symporters homologous to E. coli lac permeasea
|
|
|
DYNAMICS AND FLEXIBILITY
|
|---|
Site-directed excimer fluorescence can be used to study dynamic aspects of permease folding (91). Excimers formed between transmembrane domains are markedly diminished by denaturants, whereas the excimer observed within helix X is unaffected, indicating that tertiary interactions are disrupted with little effect on secondary structure. One of the excimers described also exhibits ligand-induced alterations. Excimer fluorescence due to the interaction between helices VIII (E269C) and X (H322C) is quenched by Tl(I), and the effect is markedly and specifically attenuated by permease substrates, although at relatively high concentrations.
The reactivity of single-Cys residues placed in many transmembrane domains is also dramatically altered in the presence of ligand, implying that transport involves widespread changes in tertiary structure (45, 55–59, 82). With V331C permease, the effect of ligand on reactivity is mimicked by imposition of 
µH+ (24). However, the effect of ligand on the reactivity of single-Cys mutants appears to be considerably more widespread than the effect of µH+ (82).
As discussed above, ligand binding increases the distance between position 245 (helix VII) and positions 52 and 53 (helix II) in a symmetrical fashion by up to 4 Å (32). Furthermore, it has been shown recently (92) that proximity relationships between certain positions in loops I/II and VII/VIII are altered dramatically in the presence of ligand.
Since many Cys replacement mutants that are inactivated by alkylation cluster on helical faces ( Figs. 3 and 4), it seems likely that surface contours of the helices may be important for sliding or tilting movements between transmembrane helices that may occur during turnover. This surmise coupled with the indication that very few residues are essential in a mechanism that involves widespread conformational changes is encouraging, as it suggests that a relatively low-resolution structure (i.e., helix packing) and localization of the translocation pathway(s) can provide important insights.
The notion that the permease, a 12-helix membrane protein without prosthetic groups or metals, is a highly flexible, metastable molecule may help explain the general difficulty in crystallizing this type of membrane protein. Recent experiments (93) measuring the average helix tilt of the permease as a function of phospholipid to protein ratio by polarized attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR) yield an average tilt angle of 33° for the helices relative to the bilayer normal at a lipid-to-protein ratio of about 800:1 (mol/mol). However, upon decreasing the lipid-to-protein ratio, the average tilt angle increases to as much as 51° in a manner that correlates with a decrease in both activity and the lipid order parameter; the effects are reversible when a high lipid to protein ratio is reestablished. A rough estimate of the lipid requirement for a single permease molecule on the basis of an average protein diameter of 40 Å (94, 95) and a cross-sectional area of 75 Å2 per membrane lipid head group (96) yields about 25 lipid molecules in direct contact with the permease. Taking this calculation to the extreme, `native' permease requires an intact bilayer area with a radius of approximately 100 Å.
In addition, ATR-FTIR measurements demonstrate that the permease exhibits unusually fast H/D exchange to 90–95% completion (93, 97). This unexpected accessibility of the protein backbone to bulk water, which is also observed with the human erythrocyte glucose transporter (98), is consistent with the ability of intramembrane bis- or tris-His residues between helices to bind divalent metal with a pKa similar to that of unperturbed imidazole (74–77). This high degree of hydration suggests a common feature for transporters of hydrophilic substrates. In contrast, other membrane proteins such as EmrE, a multidrug transporter for hydrophobic substrates (99), bacteriorhodopsin, a light-driven H+ pump (100), or the prokaryotic K+ channel SliK (97) exhibit much lower rates and extents of amide H/D exchange.
|
PROPERTIES OF MUTANTS IN RESIDUES IRREPLACEABLE FOR H+ TRANSLOCATION AND COUPLING
|
|---|
Differences in the transport properties of mutants indicate that Glu269, Arg302, His322, and Glu325 are critical for H+ translocation and coupling to substrate translocation. Because individual steps in the overall translocation cycle cannot be delineated by studying µH+-driven active transport, carrier-mediated efflux down a chemical gradient, equilibrium exchange, and entrance counterflow are used to probe the mechanism (101, 102). Efflux, exchange, and counterflow with wild-type permease are explained by a simple kinetic scheme ( Fig. 6).
Efflux consists of five steps: 1) binding of H+ and lactose to the permease at the inner surface of the membrane; 2) a conformational change in the permease that results in translocation of lactose and H+ to the outer surface of the membrane; 3) release of substrate; 4) release of H+; and 5) a conformational change corresponding to return of the unloaded permease to the inner surface of the membrane. Exchange and counterflow involve steps 1–3 only. Efflux, exchange, and counterflow are blocked in His322 and Arg302 mutants, although the mutants catalyze lactose influx down a concentration gradient. In contrast, Glu325 mutants are specifically defective in all steps involving net H+ translocation but catalyze exchange and counterflow as well or better than wild-type (103). Since Glu325 appears to be the only residue in the permease that is directly involved in H+ translocation (see below), it seems highly likely that it is required for step 4.
View larger version (20K):
[in this window]
[in a new window]
|
Figure 6. Schematic representation of reactions involved in lactose efflux, exchange and counterflow. C represents lac permease; S is substrate (lactose).
|
|
Glu325 mutations are mimicked by D2O (104) or mAb 4B1 (105), and substrate affinity is unaffected by neutral replacements, D2O or mAb 4B1. Replacement of Arg302 with Lys or His322 with Arg yields permease that does not catalyze active transport, exchange, or counterflow even though the replacements would be expected to preserve putative charge pairs. Therefore, it is unlikely that disruption of charge pairs between the irreplaceable residues per se leads fortuitously to the properties described.
Permease with Asp in place of Glu325 exhibits about 20% of wild-type activity and is markedly defective with respect to efflux down a concentration gradient. Remarkably, exchange of lactose across the membrane is pH dependent: below pH 7.5, exchange is rapid and the rate is comparable to wild-type; above 7.5, the rate decreases and is nil at pH 9.5 with a midpoint at about 8.5; inhibition at alkaline pH is completely reversible (106). In contrast, wild-type exchange is only mildly inhibited above pH 9.5, and exchange by E325A permease is comparable to wild-type and unaffected by pH.
Although none of the His322 mutants catalyze active transport, permease with Tyr or Phe in place of His322 (107–109) exhibits sugar-dependent H+ influx with low efficiency and melibiose efflux remains coupled to H+ translocation. In addition, the reactions involving exchange are limiting for lactose but not melibiose efflux, and a double mutant with Val in place of Ala177 and Asn in place of His322 catalyzes lactose-dependent H+ influx with a stoichiometry close to unity (110). Therefore, although His322 is irreplaceable with respect to active transport, it does not appear to play a direct role in H+ translocation. Similarly, mutants in Arg302 catalyze little or no lactose accumulation, but mutant R302S exhibits lactose-dependent H+ influx (111, 112).
Mutants with Asp, Gln, or Cys in place of Glu269 do not catalyze active transport of lactose, exchange, counterflow, influx, or efflux down a concentration gradient (113). Remarkably, however, E269D permease accumulates the high-affinity analog ß,D-galactopyranosyl 1-thio-ß,D-galactopyranoside (TDG) in a partially uncoupled fashion with an increase in H+/TDG stoichiometry; with the exception of galactose (S. Frillingos and H. R. Kaback, unpublished observations), other substrates tested are not accumulated. These findings and others (114) indicate that Glu269, like His322 and Arg302, plays an essential role in the mechanism, but is not directly involved in H+ translocation.
|
MONOCLONAL ANTIBODY 4B1 ALTERS THE pKa OF A CARBOXYLIC ACID AT POSITION 325
|
|---|
mAb 4B1 or the Fab fragment binds to an epitope (Phe247, Phe250, and Gly254) in the periplasmic domain between helices VII and VIII ( Figs. 1 and 3), uncoupling lactose from H+ translocation in much the same manner as Glu325 mutants or D2O (i.e., only those reactions involving net H+ translocation are inhibited, with little or no effect on translocation of the ternary complex; Fig. 6) (12, 105). Since no residue in this domain is important for transport and avidin (ca. 70 kDa) binding to a biotinylated Cys residue in the domain has no effect on activity, it was suggested that 4B1 binding exerts a torsional effect that perturbs helix VII and/or VIII and alters the pKa(s) of a residue(s) critical for coupling. Consistent with this idea, 4B1 binding shifts the midpoint of the pH profile for exchange by mutant E325D from pH 8.5 to pH 7.5 (106). Since the midpoint presumably reflects the pKa of Asp325 and mAb binding is unaffected at the pHs tested, 4B1 probably causes an acidic shift in the pKa of Asp325.
|
DOES PERMEASE TURNOVER BEGIN WITH GLU325 PROTONATED OR UNPROTONATED?
|
|---|
In a mechanism proposed (7) for coupling lactose and H+ translocation in lac permease, the catalytic cycle was initiated with Glu325 (helix X) salt-bridged to Arg302 (helix IX) and His322 (helix X) H-bonded with Glu269 (helix VIII) (i.e., with Glu325 unprotonated). However, recent experiments (P. Venkatesan and H. R. Kaback, unpublished observations) indicate that the carboxylic acid at position 325 must be protonated in order for the permease to bind substrate effectively. Although single-Cys148 permease with a wild-type Glu residue at position 325 exhibits complete protection by TDG against NEM labeling at neutral or high pH (see ref 24), when Glu325 is replaced with Asp, protection similar to that observed with wild-type is observed at pH 5.5 and pH 7.5, but there is no protection at pH 9.5. The most reasonable explanation is that the pKa of a Glu residue at position 325 is very high (>pH 10) due to the low dielectric of the phospholipid bilayer in which the carboxylic acid is embedded (see below). With an Asp at the same position, the side chain is one methylene group shorter and the carboxylic acid is more accessible to water, which lowers the pKa to about pH 8.5. Therefore, at pH 5.5 and pH 7.5, Asp325 is protonated and the permease binds ligand, but at pH 9.0 the permease is predominantly unprotonated and negatively charged, conditions that prohibit binding of substrate. This explanation also applies to the exchange titration profile exhibited by E325D permease discussed above.
|
A MECHANISM FOR COUPLING LACTOSE AND H+ TRANSLOCATION
|
|---|
The findings lead to a model for coupling that is significantly different from that proposed originally (7), although the basic principles remain the same. The model is based on four propositions: 1) in the absence of substrate, the permease does not catalyze significant H+ translocation; 2) a substrate concentration gradient generates µH+, the polarity of which depends on the direction of the substrate concentration gradient; 3) Glu325 is the primary and possibly the only residue directly involved in H+ translocation; and 4) the catalytic cycle starts with Glu325 protonated.
A cytoplasmic view of the six helices thought to play a central role in the mechanism is shown in Fig. 7.
Cys148, which makes direct contact with the galactosyl moiety of the substrate, is at the top of helix V on the same face as Met145 (not shown), which also interacts weakly with substrate; on the adjoining face of helix VIII are Val264, Gly268, and Asn272, where the reactivity of single-Cys replacements with NEM is decreased by ligand, and Thr265 (not shown), where Cys reactivity is increased due to increased solvent accessibility. Most important for substrate binding are Arg144, one turn above Cys148 in helix V, and Glu126 in helix IV (see Fig. 5). Thus, the interface between helices IV and V plays a major role in substrate recognition and translocation, but it is likely that the face of helix VIII with Asn272, Gly268, Thr265, and Val264 is also important for substrate translocation, probably by coupling conformational changes at the interface between helices IV and V to the interface between helices IX and X, and vice versa. In the upper left are helices VIII, IX, and X with Glu269, Arg302, and His322, which are postulated to interact, forming a stable, neutral triad. In this configuration, Glu325 in helix X faces the low dielectric interior of the membrane and must, as a result, be protonated. Finally, Asp240 (helix VII) is charge-paired with Lys319 (helix X), and neither residue nor the salt bridge is directly involved in the mechanism. On the basis of the relationships described, substrate-induced structural changes at the interfaces between helices IV, V, and VIII are transmitted through the network of interacting residues to the interfaces between helices VIII, IX, and X. Conversely, changes between helices VIII, IX, and X are transmitted to the interfaces between helices IV, V, and VIII.
View larger version (25K):
[in this window]
[in a new window]
|
Figure 7. Proposed mechanism for energy coupling in lac permease. 1) In the outward-facing conformation (lower right), Glu325 (helix X) is protonated and Glu269 (helix VIII), Arg302 (helix IX), and His322 (helix X) form a triad. 2) Ligand binding at the interface between helices IV and V induces a conformational change that disrupts the triad (upper right). With saturating substrate concentrations at both surfaces of the membrane, the protonated form of Glu325 is stabilized and the permease can oscillate between outward- and inward-facing conformations, thereby catalyzing equilibrium exchange and counterflow with no H+ translocation. Moreover, mAb 4B1 stabilizes the permease in this configuration, as does replacement of H2O with D2O or neutral amino acid replacements for Glu325. 3) In the presence of a substrate concentration gradient (µlac) or µH+, the changes associated with substrate binding lead to a marked decrease in the pKa of Glu325 and, ultimately, its deprotonation by bringing Arg302 into proximity with the carboxylic acid (upper left). When H+ is released from Glu325 between helices IX and X, it can be acted upon equally by either the electrical potential or the pH gradient across the membrane, particularly if the changes described are accompanied by appropriate changes in helix tilt, resulting in the opening and closing of crevices on respective sides of the membrane with transient accessibility to both sides of the membrane. The stable, uncharged triad between Glu269, His322, and Arg302 is reformed in the outward-facing conformation, but Glu325 is negatively charged and embedded in the low dielectric of the membrane, which is thermodynamically unfavorable (lower left). 4) Glu325 is reprotonated and the cycle can be repeated.
|
|
The postulated mechanism for influx is as follows ( Fig. 7). 1) In the `outward-facing' conformation (lower right), Glu325 (helix X) is protonated and Glu269 (helix VIII), Arg302 (helix IX), and His322 (helix X) form a triad. 2) Ligand binding at the interface between helices IV and V induces a conformational change that disrupts the triad (upper right). With saturating substrate concentrations at both surfaces of the membrane, the protonated form of Glu325 is stabilized and the permease can oscillate between outward- and inward-facing conformations, thereby catalyzing exchange and counterflow with no H+ translocation. Moreover, mAb 4B1 stabilizes the permease in this configuration, as does replacement of H2O with D2O or neutral amino acid replacements for Glu325. 3) In the presence of a substrate concentration gradient (µlac) or µH+, the changes associated with substrate binding lead to a marked decrease in the pKa of Glu325 and, ultimately, its deprotonation by bringing Arg302 into proximity with the carboxylic acid (upper left). Since bis-His residues at positions 302 and 325 form a divalent metal binding site with an apparent pKa that approximates an unperturbed imidazole, it seems reasonable to assume there is a water-filled crevice between helices IX and X. Thus, when the H+ is released from Glu325 between helices IX and X, it can be acted upon equally by either the electrical potential or the pH gradient across the membrane, particularly if the changes described are accompanied by appropriate changes in helix tilt resulting in the opening and closing of crevices on respective sides of the membrane with transient accessibility to both sides of the membrane. The order of release is always sugar first and H+ second; however, in the presence of µlac, deprotonation of Glu325 is rate-limiting, whereas in the presence of µH+, dissociation of sugar is limiting (see ref 104). In any case, the stable, uncharged triad between Glu269, His322, and Arg302 is reformed in the outward-facing conformation, but Glu325 is negatively charged and embedded in the low dielectric of the membrane, which is thermodynamically unfavorable (lower left). (4) Glu325 is reprotonated and the cycle can be repeated.
In addition to providing a rationale for coupling H+ and substrate translocation and an explanation for why the electrical potential and the pH gradient across the membrane have the same kinetic as well as thermodynamic effect on transport, the model explains other important observations. 1) Although
TOP
ABSTRACT
INTRODUCTION
