Mutations That Simultaneously Alter Both Sugar and Cation Specificity ...

Vol . 263, No. 26. Issue of SeDtember 15,PP. 12909-12915,1988 Printed in U.S. A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY

D 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Mutations That Simultaneously Alter Both Sugar and Cation Specificity in the Melibiose Carrier ofEscherichia coli* (Received for publication, May 9, 1988)

Martyn C. BotfieldS and T. Hastings Wilson From the Department of Cellular and Molecular Physiology, Harvard Medical S c h o o l , Boston, Massachusetts 02115

The isolation and deduced amino acid sequence of 70 melibiose carrier mutants with impaired methyl-8-Dgalactopyranoside (TMG) andcation recognition properties is described. The K , for TMG transport ranged from 1 to >lo0 mM. Amino acid substitutions occurred at 23 unique sites within the protein. These sites were clustered into four distinct regions: Asp-15 through Ile-18 (cluster I), Tyr-116 through Pro-122 (cluster II), Val-342through Ile-348 (cluster 111), and Ala-364 through Gly-374. Only two sites fell outside of these clusters: Ile-61 and Ala-236. In the native conformation, some or all of these clusters may interact to form the substrate recognition site. Impairment ofTMG recognition was accompanied by decreased Li+ inhibition of melibiose transport inall but one mutant. That changes in sugar recognition properties should so frequently accompany changes in cation recognition properties suggests an interaction between the two substrates. A modelfor such interaction is proposed.

The melibiose carrier of Escherichia coli is a cytoplasmic membrane protein which mediates the cotransport of galactosides with monovalent cations. The carrier has a broad substrate specificity which includes both a-galactosides and @-galactosides(see Ref. 1 for a complete review). Although most cotransport systems of bacteria utilize H+ gradients (2), the melibiose carrier is unusual in its ability to use H+, Na+, or even Li+ as the coupled cation (3-7). Direct measurements of cation movement have shown that there is an obligatory coupling between the cation and sugar species (5,8,9). These data suggest that the substrates may share common or overlapping binding sites. However, little isknown about the location or characteristics of these sites. The melB gene, which codes for the melibiose carrier, has been cloned and theentire gene sequenced (10,11).From the nucleotide sequence, the protein is deduced to be composed of469 amino acids and has a molecular weight of 52,029. Based on its strongly hydrophobic profile (-70% nonpolar residues), it has been suggested that the protein crosses the membrane at least 10 times (11).However, little is known about the secondary structure or three-dimensional characteristics of the protein. Although it has been reconstituted into proteoliposomes (12, 13), the carrierhas never been purified to homogeneity, and no direct measurement of physical characteristics have been made. * This work was supported by National Institutes of Health Grant 5-R01-DK05736-27 and National Science Foundation Grant DMB84.14848. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18U.S.C. Section 1734 solelyto indicate this fact. $Supported by National Institutes of Health Grant T32-GM07258.

A useful approach to theinvestigation of protein structure/ function relationships involves the isolation of mutants with known alterations in substrate specificity. Since it is likely that the substitution of amino acids at thesubstrate binding site would affect specificity, such substrate recognition mutants may help to identify the regions of the protein involved in forming this site. This paper describes the isolation and nucleotide sequencing of 70 melibiose carrier mutants which have impaired abilities to transport methyl-0-D-thiogalactopyranoside (TMG)’ and altered Li+ recognition properties. The transportcharacteristics anddeduced amino acid substitutions of these mutants are described. The results provide insights concerning the location of the substrate binding site by identifying four regions of the protein where amino acid substitutions consistently altersubstrate specificity. This strong clustering of mutations suggest that these areas may come together in the tertiary structure to form the substrate recognition site. That changes in sugar recognition properties were almost always accompanied by changes in cation (Li+) recognition properties suggests a direct interaction between the two substrates. A model for substrate-interactive cation binding is proposed. MATERIALS AND METHODS

Reagents-Melibiose andTMG were purchased from Sigma. Methyl-[14C]TMGwas purchased from Du Pont-New England Nuclear and was further purified by paper chromatography on Whatman 3mm chromatography paper using a mixed solvent phase of 3 parts 1-propano1:l part H20. 35S-dATP was purchased from Amersham Corp. Isopropyl-P-D-thiogalactoside and 5-bromo-4-cbloro-3-indolyl/%galactosidewere from Boerhringer Mannheim. DNA polymerase I (Klenow Fragment), T4 DNA ligase, DNA sequencing reagents, and all restriction enzymes were purchased from New England Biolabs. All other chemicals were of reagent grade. Mutagenesis and Isolation of Mutants-Mutants were isolated on the melibiose permease expression plasmids pSTY-37’ and pSTY-91 (10) containing the temperature resistant melB gene (Fig. 1).The plasmids were introduced into the mutagenic E. coli strain ES953 mutD5 (14) by the standard CaClz transformation method (15) and plated on LB agar plus tetracycline. This strainincreases spontaneous mutation rates over 1000-fold (16) by adversely affecting the activity of the editing 3’-exonuclease (17). Individual colonies arising from independent transformation events were grown overnight in 1.5 ml of LB media plus tetracycline and the plasmid DNA isolated by the standard NaOH/sodium dodecyl sulfate lysis method (18).The mutagenized plasmid DNA was transformed into E. coli strain DW2 (melA+AB, Alae ZY) (provided by Dorothy Wilson of this laboratory) and plated onto M63 minimal media (19) agar, pH 5.8, containing tetracycline, 2 mM melibiose as the sole carbon source, and 10 mM TMG as a competitive inhibitor of the melibiose carrier (1).Under these conditions, growth of the cells containing the wild-type melB gene was strongly inhibited. Cells released from TMG competition via mutation of the melB gene grew more rapidly and were clearly The abbreviations used are: TMG, methyl-P-D-thiogalactopyranoside; 35S-dATP,deo~yadenosine-5’-(o-[~~S]thio)-triphosphate. T. Tsuchiya, unpublished.

12909

12910

Mutations in theMelibiose Carrier of E. coli protein secondary structure and hydropathic profiles were performed with MicroGenie Sequence Analysis Software (Beckman Instruments, Palo Alto, CA). The specific algorithm used for the secondary structure analysis was that of Garnier et al. (24). This algorithm is at least as accurate as Chou and Fasman (25) and more adapted to rapid computation. The algorithm used for the prediction of hydropathic character was that of Hopp and Woods (26). Determination of the K,,, for TMG-Transport was performed as described in Fig. 4. Samples were rapidly filtered through a 0.65-pm pore size nitrocellulose membrane filter (Sartorious Filters Inc, Haywood CA). Accumulated label was quantified by liquid scintillation counting using Liquiscint (National Diagnostics, Somerville, NJ). RESULTS

TMG Recognition Mutants-A total of 70 independent mutants exhibiting impaired TMG recognition were isolated and sequenced. The observed nucleotide substitutions and their deduced amino acid substitutions arelisted in Table I. Amino acid substitutions occurred at 23 unique sites within the protein. The most striking observation was that nearly all of these sites were clustered into four distinct stretches: aspartate 15 through isoleucine 18 (cluster I), tyrosine 116 through proline 122 (cluster 11), valine 342 through isoleucine 348 (cluster 111), and alanine 364 through glycine 374(cluster IV) (See Table I). Only two sites fell outside of these clusters: isoleucine 61 andalanine 236. When plotted ona topological model of the p r ~ t e i nclusters ,~ 11,111, and IV occur at corresponding positions of transmembrane segments 3, 9, and 10 (Fig. 2). It is not hard to envision a folding of the protein that would place all of these sites adjacent to each other in the tertiary conformation, thus forming the sugar or sugar/cation binding site. The implications of this suggestion are addressed below. Effect of Substitutions on Hydropathy and Secondary Structure Predictions-In general, the substitutions were conservFIG. 1. Melibiose permease expression plasmids pSTY-37 ative with regard to both charge and hydropathic index. No and pSTY-91. The construction of pSTY-91 is described elsewhere potentially charged residues were replaced by uncharged res(9). Plasmid pSTY-37 was constructed by ligating the PuuII fragment from pSTY-91 to theScaI site of pBR322.' The melB gene of pSTY- idues; only one uncharged residue was replaced by a poten37 and the melA and melB genes of pSTY-91 are constitutively tially charged residue (tyrosine 166 by histidine). While the pK, for the side chain of free histidine in an aqueous environexpressed using the amp promoter. ment at 25 "C is 5.97 (27), itspK, within amembrane-inserted visible after 48 h. One to ten mutant colonies were found per lo6 protein is a complex function of the size and shape of the transformants. To ensure the independence of mutants, no more than protein, location of the residue within the tertiary structure, one mutant from each plasmid preparation was studied. DNA Sequencing-Plasmids containing me@ mutations were iso- and the dielectric environment (28). Given the extreme dielated from overnight cultures of DW2 as described above. Plasmid lectric difference between the periplasm or cytoplasm and the DNA from pSTY-37-derived mutants was double-digested with the interior of a phospholipid membrane, and the lack of inforrestriction endonucleases HincII and EcoRI. Mutants derived from mation concerning the tertiary structure it is impossible to pSTY-91 were double-digested with PstI and EcoRI. The digested predict if this substitution introduced a charge. fragments were electrophoretically separatedon 1% low melting While examination of the hydropathy changes listed in SeaPlaqueO-agarose (FMC Bioproducts, Rockland, ME) in 50 mM Table I1 revealed a number of substitutions that appeared to Tris-acetate, pH 8.2, containing 0.5 pg/ml ethidium bromide. The 1675-base pair melB fragment from pSTY-37 derivatives and 2923- introduce amino acids of disparate hydropathic character, base pair melAB fragment from pSTY-91 derivatives were visualized when these substitutions were permuted over the 17-residue with long wavelength UV light, excised from the gel, and carefully segments used to calculate the net hydropathic character of trimmed of excess agar. The fragments were ligated into appropriately the protein at any given position (26), it became clear that cut M13mp19 directly from the agarose blocks according tothe the overall hydropathic character was conserved. method of Struhl (20). Ligated M13 was used to transform E. coli Although the validity of applying probabilistic methods of strain JMlOl (21), and single-stranded DNA template was produced according to standard methods (22). DNA sequencing was accom- secondary structure prediction to membrane proteinshas plished via the dideoxy chain termination method of Sanger et al. been called into question (29), the effect of the observed (23) using 35S-dATPas the incorporated label. Elongation and chase reactions were carried out at 50 ' C to eliminate sequencing artifacts resulting from temperature-sensitive secondary structure characteristics of the single-stranded DNA template. The me@ template was primed at approximately 200-base pair intervals using synthetic oligonucleotide primers manufactured with a Du Pont Coder 300 DNA Synthesizer (Du Pont Biotechnology Systems, Wilmington, DE) thus allowing the entire gene to be sequenced using a single template species. Sequences read from adjacent primers overlapped by at least 50 nucleotides in all cases. Areas of ambiguous interpretation were repeatedly sequenced until all ambiguities were resolved. Secondary Structure andHydropathy Calculations-Predictions of

Although no formal model has been proposed, Tsuchiya (12)has suggested that the melibiose carrier passes through the membrane a minimum of 10 times. Hydropathy analysis using the algorithm of Hopp and Woods (26) also indicates that theprotein has 10 strongly hydrophobic regions. However, one of these segments, between Arg291 and Ala-346, is 52 residues long. This is more than sufficient to span the membrane twice in helical conformation while leaving an adequate number of residues to join the two helices. Therefore, two transmembrane regions have been assigned to this segment. The resulting model with eleven transmembrane regions is shown in Fig. 2.

de

12911

Mutations in the Melibiose Carrier of E. coli TABLE I Nucleotide substitutions, deduced amino acid substitutions, and kinetic parameters of TMG transport Kinetic data were determined as described in Fig. 4 and under “Materials and Methods.” When mutants were available as both pSTY-37 and pSTY-91 derivatives, kinetic data were determined using the pSTY-37 construct. Kinetic data prefaced by > are minimum estimates based on substrate concentrationranges that did not result in complete saturation of the transportprocess. Control cells containing the parentalplasmids pSTY-37 or pSTY-91 had a K,,, for TMG of 0.8 mM and maximal velocities ( Vma) of 40 and 30 pM TMG/lOg/min, respectively (data not shown). acid change

Amino change

Plasmid designation

ASP-15 + Glu Ala-17 + Thr Ile-18 + Asn Ile-61 + Val Tyr-116 + Phe Tyr-116 + His Met-119 + Val Pro-122 + Ser

T-45 +G G-49 + A T-53 + A A-181 + A A-347 + T T-346 + C G-357 + A C-364 + T

Ala-236 + Thr Ala-236 + Val

G-706 + A C-707 + T

Val-342 + Ala Val-345 + Met Ile-348 + Val Ala-364 + Val Tyr-365 + Phe Gln-368 + Leu Thr-369 + Ala

T-1025 + C G-1033 + A A-1042 + G C-1091- T A-1094 +T A-1103 -+ T A-1105 + G

Thr-369 + Ser Met-370 + Val

C-1106 ”* G A-1108 + G

Met-370 -+ Ile Val-372 + Gly Val-372 -+ Ala Gly-374 -+ Ser

G-1110 + A T-1115 + G T-1115 -+ C T-1120 + C

No. isolated

pSTY37-El5 pSTY91-Tl7 pSTY91-Nl8 25 pSTY91-VGl pSTY91-Fl16 pSTY37-Hl16 pSTY91-Vl19 pSTY37-Sl22 pSTYSl-Sl22 pSTY91-T236 pSTY37-V236 pSTY91-V236 pSTY91-A342 pSTY37-M345 pSTY91-V348 pSTY91-V364 pSTY91-F365 pSTY91-L368 pSTY37-A369 pSTY91-A369 pSTY91-S369 pSTY37-V370 pSTY91-V370 pSTY37-I370 pSTY37-G372 pSTY91-A372 pSTY37-S374 pSTY91-S374

substitutions on the predicted secondary structure was calculated according to Garnier et al. (24) (See Table 11). When permuted over the entire 17-residue span used in the algorithm, the substitutions made little or no difference to the predicted conformation of the region. However, substitutions in cluster IV consistently altered a %residue sequence predicted to form a reverse turn. Hydropathy analysis placed these residues (serine 362, isoleucine 363, and alanine 364) at the extreme end of a strongly hydrophobic segment suggesting that they may form a reverse turn between two transmembrane segments. It is therefore possible that cluster IV substitutions impair transport by perturbing a critical reverse turn rather thanthrough direct modification of the substrate binding site. TMG Transport Churacteristics-The K, and V,,, for TMG transportwere determined to assess the degree to which each substitution affected the ability of the carrier to transport TMG. The results aregiven in TableI and representative data are shown in Fig. 3. In all but one mutant, the K , for TMG was substantially increased, ranging from 1 to >lo0 mM. Both parental constructs exhibited a K , for TMG of approximately 0.8 mM. This suggests that the substitutions mayhave altered the ability of the carrier to bind TMG, rather than affecting its gendral ability to mediate transport. This supports the contention that the substitutions occurred at or near the substratebinding site. While the parentpSTY-37 and pSTY-91 constructsexhibited maximal transport velocities of 40 and 30 pmol/lOg cells/ min, respectively, the mutant Vmsxvalues ranged from 2 to 25 pmol/lOg cells/min. The broad range of values observed for

1 1 4 1 1 1 1 1 3 1 4 4 1 1 6 1 1 1 6 12 1 1 1 6 1 2 2 4

TMG Km

Vl0.X

rnM

pMII0’ cells/rnin

18 4 40

>loo >loo

9 5

8 5 11

>15 >13 17 20 17

6 15

3 11

7 1 4 21 6 7 >50

7 06 25 8 2 14 >25

>50

r13

>loo

>13 18 5 10

16 2 2

the mutants and the discrepancy between the two parent constructs might be attributable to different levels of expression between plasmids and/or impairment of carrier insertion as a result of the mutations. Since transport activity was measured as afunction of cell number and not carrier number, either of these possibilities would dramatically alter the calculated Vmax.Due to the lack of an antibody specific for the melibiose carrier, thesepossibilities could not be investigated. It is also possible that the mutations directly impaired V,.,, although the observed effects on K,,, are sufficient to account for the observed growth characteristics. DISCUSSION

The Sugar Binding Site-Mutations which alter sugar specificity without altering the overall ability of the carrier to transport substrate likely do so through modification of the sugar binding site. These modifications could be the result of local substitutions at the binding site or radical conformational changes induced by substitutions made elsewhere in the protein. Examples where single amino acid substitutions within the putative sugar recognition site have altered specificity for the normal oligosaccharide moiety are known (3033). However, to definitively assign a residue to the substrate binding site, the crystal structure of the protein must be solved. Unfortunately, membrane proteins have proved to be exceedingly difficult to crystalize. To date, only five have been successfully crystallized and resolved three from the photosynthetic reaction center of Rhodopseudomnas uiridis (3436), one from the Photosystem I reaction center of Phormidium laminosum (37),and the plant protein of unknown

12912

Mutations inthe Melibiose Carrier of E. coli

CYTOPLASM

COOH

HIS Phe

PERIPLASM

Ala

IIe

Ala

Val

GI y

FIG. 2. Topological model of the melibiose carrier protein showing mutational clusters. Hydrophobic segments are shown in boxes as transmembrane, a-helices connected by hydrophilicdomains. Two transmembrane segments have been assigned to the 52-residue hydrophobic sequence between Arg-291 and Ala-346.3 Amino acid substitutions resulting in impaired TMG recognition and transport are indicated. The inset illustrates more clearly the distinct clustering of mutations.

'

function crambin (38). None appear to be involved in sugar binding and therefore add little to our knowledge of sugar binding sites. Although none are membrane proteins, more than two dozen carbohydrate binding proteins, including the arabinose, maltose, and galactose binding proteins from E. coli, have been successfully crystallized and studied by x-ray diffraction (see Ref. 39 for a review). These studiesindicate that a general schema for the interactions between proteins and their carbohydrate substrates exist. It is abundantly clear that hydrogen bonds are the dominant form of interaction. In complexing with the protein, the sugars exchange their solvation shell of water for hydrogen bonds with polar groups within the substrate binding site. Concomitantly, the previously bound water is displaced and often excluded from the binding site. With the exception of the anomeric hydroxyl whichonly serves as ahydrogen bond donor, evidence from the arabinose binding protein indicates that all the hydroxyl groups of simple sugars simultaneously serve as hydrogen bond donors and acceptors (40). Taken together, these findings indicate that sugar specificity is determined by the location of hydrogen bond donating and/oraccepting residues within the binding site. Given the importance of hydrogen bonds, it seems logical that removal of a residue participating in a hydrogen bond or introduction of a residue capable of forming a new hydrogen bond, may alter sugar binding and specificity. In addition, substitution by any amino acid that alters the alignment of a residue participatingina critical hydrogen bond may also

affect sugar binding. In the TMG recognition mutants reported on in thispaper, eight substitutions removed potential hydrogen bond formers, six introduced potential hydrogen bond formers, and two substituted one potential hydrogen bond forming residue for another of different side-chain length. The remaining seven substitutions (Ile-61- Val, Ala236 + Val, Val-342 + Ala, Ile-348 + Val, Ala-364 + Val, Val-372 + Ala, and Val-372 + Gly) occurred adjacent to an amino acid capable of participating in one or more hydrogen bonds (see Fig. 2). Although any conclusion concerning the functional involvement of specific residues would be speculative, the pattern of substitution is obviously consistent with the idea of a sugar binding site. This suggests that many of the mutations occurred at the binding site and did not exert their effect through distant conformational changes. When taken asa group, the most remarkable feature of the 23 sugar recognition mutants are theirdistribution: the TMG mutations are predominantly clustered into four discrete regions. This suggests two hypotheses: these clusters are directly involved in forming the sugar recognition site or substitutions a t these sites consistently alter some conformational feature of the protein which effects sugar binding at some distant site. Although substitutionsincluster IV did consistently interfere with an adjacent predicted reverse turn, substitutions at the other siteshadlittleor no influence on the predicted secondary structure. Similarly, the hydropathic profile was not affected by any of the substitutions.Taken together, this distribution of mutations further supports the contention that the substitutions occurred at, or near, the

12913

Mutations in the Melibiose Carrier of E. coli TABLE I1 Effect of mutations on predicted hydropathic and secondary structure characteristics Hydropathy analysis was performed according to Hopp and Woods (26), secondary structure prediction according to Gamier et al. (24). Negative numbers indicate a hydrophobic tendency. The propensity of an amino acid to participate in an a-helix or @-sheet islisted as a-tendency/@-tendency.aa = strong helix former, a = helix former, @@ = strong @ former, B = B former, . = indifferent, - - = strong helix or breaker, - = helix or P breaker, R.T. = reverse turn. characteristics structure Secondary character Hydropathic

uted Mutant

Original Mutation index

------

3.0Glu Asp-15 3 Ala-17 --* Thr Ile-18 Asn Ile-61- Val Tyr-116 -+ Phe Tyr-116 + His Met-119 -+ Val Pro-122 -+ Ser Ala-236 Val Ala-236 Thr Val-342 Ala Val-345 Met Ile-348 + Val Ala-364 Val Tyr-365 + Phe Gln-368 Leu Ser Thr-369 Thr-369 Ala Met-370 -+ Val Met-370 Ile Val-372 + Gly Val-372 + Ala Gly-374 -+ Ser

20

effect

index 3.0 -0.5 -1.8 -1.8 -2.3 -2.3 -1.3 0.0 -0.5 -0.5 -1.5 -1.5 -1.8 -0.5 -2.3 0.2 -0.4 -0.4 -1.3 -1.3 -1.5 -1.5

None None None None None None None None None None None None None None None None None None None None None None None

-0.4 0.2 -1.5 -2.5 -0.5 -1.5 0.3 -1.5 -0.4 -0.5 -1.3 -1.5 -1.5 -2.5 -1.8 0.3 -0.5 -1.5 -1.8 0.3 -0.5 0.3

0.0

,

I

15

v,

10

5

0 0

2

1 vo

3

/[SI

FIG. 3. Eadie-Hofstee plot of TMG transport by representativemutantspSTY9l-V119 (0),pSTY37-H116 (A), and pSTY37-G372 (0).DW2 containing a mutant melB plasmid was grown to early logarithmic phase at 37 "C in LB media containing tetracycline. Due to theposition of the melI3 gene behind the constitutive amp promoter, no induction was necessary. Cells were washed twice and resuspended to 6 X 10' cells/ml in M63 plus 10 mM NaCl, pH 7.2. The addition of Na' was necessary to facilitate TMG cotransport (4). Reactions were initiated by rapidly mixing 100 pl of cells with 50 pl of3X stock TMG solutions containing 5 &i/ml of ["C] TMG and incubating at 25 "C. Both cells and stock TMG solutions were pre-equilibrated to this temperature. After 30 s, 100 pl was rapidly filtered, washed with 5 ml of M63, and assayed as described under "Materials and Methods." Nonspecific adherence of label to the filter and cells was determined by duplicate assays a t 0 "C and zero incubation length. Background values were subtracted and never exceeded more than 5% of total counts. Plots represent linear least squares fitsof the data.

Original index

Mutant

a/B

a/B

Permuted

index

effect

None None None None None None None None None None None None None Interferes with R.T. Interferes with R.T. Interferes with R.T. Interferes with R.T. Interferes with R.T. Interferes with R.T. Interferes with R.T. None None None

sugar recognition site and did not exert their influence by inducing distant conformational changes. Even the cluster IV mutations influence only local conformation, i.e. the immediately adjacent reverse turn, again indicating that theregion forms part of the binding site. We therefore feel that in the native conformation, some or all of these mutational clusters interact to form the sugar recognition site. Cation Specificity-Sugar cotransport in the wild-type melibiose carrier can be driven by H+, Na+, or weakly by Li+ gradients (3,5,6,8). In addition, mutantsthat can participate strongly in Li+-coupled cotransport of melibiose have been isolated by Tsuchiya and co-workers. They found that when proline 122 was replaced by serine, H+ coupling was lost and an absolute requirement for Na' or Li+ was acquired (41,42). Interestingly, this is thesame mutation as found in our TMG mutants pSTY37-S122 and pSTY91-Sl22. Cation specificity is also influenced by substrate concentration. For example, Li' stimulates transport at low melibiose concentrations, but inhibits transport at high melibiose concentrations, Mutants that are resistant to the inhibitory effects of Li+ on melibiose transport have also been isolated by Tsuchiya (Ref. 43 and Footnote4).One of these Li+-resistant mutants, alanine 236 + threonine, is the same substitution as found in our TMG mutantpSTY91-T236. When we tested our remaining TMG recognition mutants, all but one (Ile-61 "-* Val) was also found to have acquired L L resistance ' (Fig. 4). That changes in sugar specificity should so frequently be associated with changes in cation specificity suggests an interaction between the two substrates. Not all cation species can be cotransported with all sugar substrates in the wild type carrier. P-Galactosides are transported only in conjunction with Na+, whereas a-galactosides T. Tsuchiya, personal communication.

Mutations in the Melibiose Carrier of E. coli If both sugar and protein contributed coordination atoms, different sugars might complete different coordination clusters that favor the complexation of different cations. a-Gal10 actoside binding may favor a H' coordination cluster, whereas &galactoside binding may complete a Na+coordination clusl5 ter. Any perturbation in protein structure that altered the position of the sugar within the binding site would influence the completion of these potential coordination clusters and therefore influence the choice of cation. Such modest perturbations occur in our TMG recognition mutants and indeed affect Li' binding. It has been hypothesized by Kaback (44, 48, 49)that pairs of histidine and glutamate (or aspartate) residues form ion pairs which participate in the reversible protonation of a I1 carboxyl-imidazole-hydroxyl"charge relay" chain responsible l5 for H' translocation in bacterial transport systems. Specifically, histidine 357 and glutamate361 of the melibiose carrier 10 were proposed to form an ion pair intrinsic to themechanism I I of the carrier. When glutamate 361 was replaced by glycine or arginine using oligonucleotide-directed site-specific muta5 genesis, H+/melibiose cotransport was inactivated (44). These findings were used to support the charge relay hypothesis. However, Na'-coupled cotransport was also inactivated and 0 B. it seems unlikely that Na' could use a charge relay system of the type proposed for H'. Although these residues are clearly important, the possibility that their removal perturbs transFIG. 4. Growth of TMG recognitionmutants in the presence port not through disruption of a charge relay system but via of Li+. Early log phase DW2 containing the indicated plasmids were some other mechanism should be considered. We find it highly used to inoculate M63 salts supplemented with 10 mM melibiose, B,, suggestive that histidine 357 and glutamate 361 areboth tetracycline, and 2 mM LiCl ( A ) (pSTY-37 and derivatines) or 10 mM potentia1 contributors of coordination atoms and are adjacent LiCl ( B ) (pSTY-91 and derivatives). Cell concentration was deter- to our cluster IV mutations. These residues may be important mined after 8 h growth at 37 'C. When mutants were available as because they participatedirectly as cationcoordination groups both pSTY-37 and pSTY-91 derivatives, the pSTY-91 constructs were used. Li+ concentrations were established by determining the or indirectly by altering the position of the sugar within the binding site and disrupting both Na' and H+ coordination minimum concentration required to completely inhibitparental growth (data not shown). Due to the constitutive expression of the clusters.

1

1

i rl

melA gene (a-galactosidase) in pSTY-91 and derivatives, but not in pSTY-37 constructs,a higher Li' concentration was required to inhibit growth on melibiose.

Acknowledgments-We would like to thankOleg Jardetzky, Elkan Blout, Dorothy Wilson, and Steven King for their helpful discussions and Tomofuso Tsyuchia for the original plasmid constructions.

are coupled to Na' or H' (9). This obligatory coupling supREFERENCES ports the idea that sugar conformation dictates the cation species used for transport. That thereis competition between 1. Wilson, T. H., Ottina, K., and Wilson, D. M. (1982)in Membranes and Transport (Martonosi, A. N., ed) pp. Na' and H' during a-galactoside transport further suggests .. 33-39, Plenum Publishing Cob., New York that both H' and metallic cations share common a recognition 2. Rosen. B. P.. and Kashket. E. R. (1978) in Bacterial Transport site. Is it possible that theshape of the sugar and its position (Rosen, B. 'P., ed) pp. 559-620, Marcel Dekker, New York within the binding site directly determines cation compatibil3. Tsuchiya, T., Raven, J. and Wilson, T. H. (1977) Biochem. ity? Biophys. Res. Commun. 7 6 , 26-31 It has been suggested that proteins may recognize and 4. Tsuchiya, T., Lopilato, J., and Wilson, T. H. (1978) J. Membr. Biol. 42,45-59 interact with cations through the formation of coordination 5. Tsuchiya, T., and Wilson, T. H. (1978) Membr. Biochem. 2, 63complexes using appropriately placed oxygens and/or nitro79 gens similar to that found with crown ethers (45). Members 6. Lodato. J.. Tsuchiva. - . T.,. and Wilson. T. H. (1978) J. Bacteriol. of this family of cyclic polyethers are known to form stable i34,i47-156 complexes with H30' (46), Na', and K+ (47). Crystal struc7. Cohn, D. E., and Kaback, H. R. (1980) Biochemistry 19, 4237tures show that the complexed ion is located on the central 4243 8. Tsuchiya, T., Oho, M., and Shiota-Niiya, S. (1983)J.Biol. Chem. axis of the ring and anchored in place by hydrogen bonds. 258,12765-12767 H30+is complexed by three hydrogen bonds, Na' and K' by 9. Wilson, D. M., and Wilson, T. H.(1986) in Zon Gradient-Coupled six. This haslead Boyer (45) to make the insightful suggestion Transport (Alvarado, F., and van Os, C. H., eds) pp. 97-103, that proton translocation may occur by the initial complexaElsevier Science Publishers, New York tion of H30+ (ratherthanH+) with appropriately placed 10. Hanatani, M., Yazyu, H., Shiota-Niiya, S., Moriyama, Y., Kanoxygen and/or nitrogen atoms. The translocation of Na', K', azawa, H., Futai, M., and Tsuchiya, T. (1984) J. Biol. Chem. 259,1807-1812 or other metallic cations would be similar, but utilize a different cluster of coordination atoms. Thus, modest re- 11. Yazyu, H.,Shiota-Niiya, S., Shimamoto, T., Kanazawa, H., Futai, M., and Tsuchiya, T. (1984) J. Biol. Chem. 259,4320-4326 arrangements in binding site organization might confer 12. Wilson, D. M., Ottina, K., Newman, M. J., Tsuchiya, T., Ito, S., unique cation binding properties. and Wilson, T. H. (1985) Membr. Biochem. 5,269-290 By virtue of its two free pairs of electrons, the ring oxygen 13. Tsuchiya, T., Ottina, K., Moriyama, Y., Newman, M. J., and of simple sugars is available to act as acoordination atom and Wilson, T. H.(1982) J. Biol. Chem. 257,5125-5128 participate in the type of cation complexing discussed above. 14. Vaccaro, K. K., and Siegel, E. C. (1977) Mutat. Res. 42,443-446

Mutations the in

Melibiose Carrier of E. coli

15. Mandel, M., and Higa, A. (1970) J. Mol. Biol. 59, 154-162 16. Degnen, G. E.,and Cox, E. C. (1974) J. Bacteriol. 117,477-487 17. DiFrancesco, R., Bhatnagar, S. K., Brown, A., and Bessman, M. J. (1984) J. Biol. Chem. 269,5567-5573 18, Birnboim, H. e.,and Doly, J. (1979) Acids Res. 7, 15131523 19. Cohen, G. N., and Rickenberg, H, v. (1956) Inst, posteur (Paris)91,683-691 20. Struhl, K. (1983) Gene (Amst.) 26,231-242 21. Messing, J. (1979) Recomb. DNA Tech. Bull. 2, 43-48 22. Schreier, P. H., and Cortese, R. (1979) J. Mol. Biol. 129, 169172 23. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. S. A. 74,5463-5467 24. Gamier, J., Osguthorpe, D. J., and Robson, B. (1978) J. Mol. BWl. 120,97-120 25. Chou, P. Y., and Fasman, G . D. (1974) Biochemistry 13,211-222 26. Hopp, T. P., and Woods, K. R. (1981) Proc. Natl. Acad. Sci. U. S. A. 78,3824-3828 27. Winaolz, M., Budavari, S., Stroumtsos, L. Y., and Noether Fertig, M.(eds) (1976) The Merck Index, p. 619, Merck and Co., Inc., Rahway, NJ 28. Fersht, A. (1977) Enzyme Structure and Mechanism, W. H. Freeman and Co., San Francisco 29. Wallace, B. A., Cascio, M., and Mielke, M. L. (19%) PrOC. NatL Acad. Sci. U. S. A. 83,9423-9427 30. Brooker, R. T.9 and Wilson, T. H. (1985) PrOC. Acad. SCi U. S. A. 82,3959-3963 31. Markgraf, M., B o c k k e , H., and Miiller-Hill, B.(1985) Mol. Gen. Genet. 198,473-475

12915

32. Jacobs, A. A. C., Simons, B. H., and de Graaf, F. K. (1987)EMBO J. 6,1805-1808 33. Rogers, G. N., Paulson, J. c..Daniels, R. s.,Skehel, J. J., Wilson, I. A., and Wiley, C.D. (1983) Nature 304,76-78 34. Deisenhofer, J., EPP, 0.7 Mikip K., Huber, R.9 and Michel, H. (1984) J. Mol. Biol. 180, 385-398 35. Deisenhofer, J.,EPP, 0.9 Miki, K., Hub-, R.9 and Michel, H. (1985) Nature 318,618-624 36. Michel, H., Epp, O., and Deisenhofer, J. (1986) EMBO J. 5 , 2445-2451 37. Ford, R. C., Picot, D., and Garavito, R. M. (1987) EMBO J. 6, 1581-1586 38. Hendrickson, W. A., and Teeter, M.M. (1981) Nature 290,107113 39. Quiocho, F. A. (1986) Annu. Rev. Biochem. 66,287-315 40. Quiocho, F. A., and Vyas, N. K. (1984) Nature 310,381-386 41. Niiya, S., Yamasaki, K., Wilson, T . H., and Tsuchiya, T. (1982) J. Biol. Chem. 257,8902-8906 42. Yazyu,H., Shiota, S., Futai, M., and Tsuchiya, T. (1985) J. Bacteriol. 162,933-937 43. Shiota, % Yaman% y . 7 Futai, M.7 and TsuchiYa, T. (1985) J. Bacteriol. 162,106-109 44. Kaback, H. R. (1987) Biochemistry 26, 2071-2076

:::~ ~ ~ ~ ~ ~ + D , :rLm. (~~~!

~ $ d ~ ~ ~ ~ ~Chem, s sot, ; ~ ~ 104,4540-4543 47. Dobler, M., Dunitz, J. D., and seiler, p. (1974) Acta C r y s t a ~ ~ g r , Sect. B Struct. Sci. B30, 2741-2743 48. Carrasco, N., Antes, L. M., Poonian, M. S., and Kaback, H. R. (1986) Biochemistry 25,4486-4488 49. Menick, D.R., Carrasco, N., Antes, L., Patel, L., and Kaback, H. R. (1987) Biochemistry 26,6638-6644

~