The Mechanism of Glucose 6-Phosphate Transport by Escherichia coli*

THE

JOURNALOF BIOLOGICAL CHEMISTRY

Vol. 263, No. 14, Issue of May 15, pp.Printed 6625-6630,1988 in U.S.A.

0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

The Mechanism of Glucose 6-Phosphate Transportby Escherichia coli* (Received for publication, November 23,1987)

Larry A. Sonna, SureshV. Ambudkar, andPeter C. Maloney From the Department of Physiology, The Johns Hopkins Uniuersity School of Medicine, Baltimore,Maryland 21205

To evaluate anion exchange as the mechanistic basis behaves as a chemiosmotic carrier that mediates nH+/anion of sugar phosphate transport, natural and artificial symport (11-13). membraneswere used in studies of glucose 6-phosphate The idea of nH+/anion symport, while it has been useful to (Glc-6-P) and inorganic phosphate (Pi) accumulation general interpretations of UhpT function in E. coli, is in by the uhpT-encoded protein (UhpT) of Escherichia conflict with conclusions drawn from studies of sugar phoscoli. Experiments with intact cells demonstrated that phate transport in other bacteria. In particular, analysis of UhpT catalyzed the neutral exchangeof internal and this reaction in Gram-positive cells (14-16) points t o a neutral external Pi, and work with everted as well as right- anion exchange as the mechanistic basis of sugar phosphate side-out membranevesicles showed further thatUhpT transport (reviewed in Ref. 17). In such cases, Glc-6-P may mediated the heterologous exchange Piofand Glc-6-P. be taken up by direct exchange for internal Pi (14, E ) , or, in When loaded with Pi, but not when loaded with mor- a somewhat more complex process, the 2 for 1 antiport of pholinopropanesulfonate (MOPS), everted vesicles took up Glc-6-P to levels 100-fold above medium con- mono- and divalent sugar phosphate anionsmay lead to a net centration in a reaction unaffectedby the ionophores Glc-6-P accumulation in the presence of a pH gradient (16). valinomycin, valinomycin plus nigericin, and carbonyl While it is not necessary that sugar phosphate transport cyanide p-trifluoromethoxyphenylhydrazone.Simi- proceed by the same mechanism in all bacteria, anion exlarly, right-side-out vesicles were capable of Glc-6-P change of the sort described in Streptococcus lactis (14-17) and Staphylococcus aureus (17) is consistent with data gathtransport, but only if a suitable internal countersubstrate was available. Thus, in MOPS-loaded vesicles, ered so far regarding Glc-6-P transportby E. coli. Moreover, the feasibility of antiport as amechanism for UhpT function oxidative metabolism established a proton-motive force that supported proline or Pi accumulation, but in E. coli has been established by preliminary work showing transport of Glc-6-P was found only if vesicles could that cells grown with Glc-6-P acquire the ability to carry out accumulate Pi during a preincubation. After reconsti- certain Pi-linked exchanges (18), including the antiportof Pi tution of UhpT into proteoliposomesit was possible to against Glc-6-P. show as well that the level of accumulation ofGlc-6-P The work reported here has allowed a clear distinction (17 to 560 nmol/mg of protein)was related directly to between these two possible mechanisms of sugar phosphate the internal concentration Pi. of These results are most transport in E. coli. Experiments using both membrane vesieasily understood if the transport of glucose 6-phoscles and reconstituted proteinargue against the participation phate in E. coli occurs by anion exchange rather than of nH’/anion symport and offer direct evidence favoring by nH+/anion symport. anion exchange as the molecular basis of UhpT function. MATERIALS ANDMETHODS

In Escherichia coli the regulation and expression of Glc-6-

P’ transport are directed by a set of four genes clustered in the uhp region (1,2). The uhpT gene specifies the transport protein itself, while the uhpA, uhpB, and uhpC genes encode regulatory elements (1, 3) that ensure expression of sugar phosphate transport only when external substrate is available (3-6). Early descriptions by Pogell et al. (7), Winkler (8, 9 ) , and Dietz (10) showed that sugar phosphate substrates of UhpT were transportedagainstaconcentration gradient, without prior hydrolysis, and that the source of energy for such accumulation correlated with the “energized membrane.” Subsequent studies provided evidence that the UhpTprotein

* This work was supported by United States Public Health Service Grant GM 24195 from the National Institutes of Health and Grant DMB 8609845 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “uduertisement” in accordance with 18U.S.C. Section 1734 solelyto indicate this fact. The abbreviations used are: Glc-6-P, glucose 6-phosphate; octylglucoside, octyl-P-D-glucopyranoside;MOPS, 3-(N-morpholino) propanesulfonic acid; MES, 2-(N-morpholino)ethanesulfonic acid FCCP, carbonyl cyanide p-trifluoromethoxyphenylhydrazone;PMS, phenazine metbosulfate.

Bacterial Strains andGrowth Conditions-All strains were derivatives of E. coli K12. Strain MC4100 (F- araD139 A(argF-hc)U169 relAZ rpsL150 thi) (19) is wild type for regulation and expression of Glc-6-P transport.The derivative RK5000 carries a deletion covering uhpA-T as well as several other markers (A(iluB-uhpABCT’)2056 gyrA219 non metE70 recA). In strainRK5000pRK10, RK5000 serves as host to the multicopy plasmid pRK10, an ampicillin-resistant derivative of pBR322 with a 6.5-kilobase HindIII-BamHI chromosomal insert containing the uhpABCT region (3). Some experiments also used strain RK6814, a derivative of RK5000pRK10 in which the plasmid was modified (3) by a TnlOAAKm insertion into the uhpT gene. With the exception of strain MC4100 (from the Coli Genetic Stock Center), these strains were kindly provided by R. J. Kadner (University of Virginia, Charlottesville, VA). Cells were grown at pH 7 in medium M63 (18) supplemented by 0.1% acid-hydrolyzed casamino acids (Difco-Bacto), 40 yg/ml uracil, 1 pg/ml thiamine hydrochloride, and 0.5% potassium gluconate, and with 25 pg/ml of ampicillin or kanamycin for growth of RK5000pRK10 or RK6814, respectively. After overnight growth at 35 ”C,UhpT was induced by addition of 5 mM Glc-6-P(strain MC4100) or 0 to 0.5 mM (strains RK5000pRK10 and RK6814) during the 2 h preceding harvesting. For transport experiments with intact cells, buffer A (20 mM MOPS/ K, 250 mM KC1, 1 mM MgSO, (pH 7)) was used for washing and resuspension (18).For other work, cells were processed as indicated below. Membrane Vesicles-To prepare everted membrane vesicles, cells were washedand resuspended using 100 mM KPi, 1mM dithiothreitol,

6625

6626

Pi-LinkedPhosphate Sugar

Antiport in E. coli

0.5 mM phenylmethylsulfonyl fluoride (pH 7) before passage through A a French pressure cell at 4000 p.s.i. (20). Membranes were then isolated by centrifugation, washed twice, and finally stored (-80 "C) in this buffer with 5 mM added MgSO,. For assays of transport, thawed membranes were washed by centrifugation and resuspended at 2 mgof protein/ml using buffer B (20 mM MOPS/K, 125 mM KzSO4 (pH 7)). Right-side-out membrane vesicles were prepared by osmotic lysis at pH 7, using the methods described by Kaback (21), with the following modifications: (i) the 100 mM KP, (pH6.6) buffer was replaced with either 50 mM KP, (pH 7) or 30 mM MOPS/K, 25 mM K2SO4(pH 7); (ii) lysozyme digestion was in the presence of 0.5 mM phenylmethylsulfonyl fluoride; and (iii) vesicleswere washed once rather than several times with Pi or MOPS-based buffers conMinutes Seconds taining 10 mM EDTA. Vesicles were finally resuspended and stored FIG. 1. Phosphate exchange mediated byUhpT in intact (at -80 "C) in these same buffers with MgSO, added to 10 mM. Prior to assays of transport, Pi-loaded vesicles were washed with and then cells. A, washed cells were placed in buffer A at 3.9 pl of cell water/ resuspended in buffer B at 0.5 to 2 mg of protein/ml. MOPS-loaded ml of suspension, followed 5 min later by 0.1 mM 3zPi;at the arrow vesicles werethawed and adjusted directly to 1 mg of protein/ml with each suspension received 30 mM KP, (pH 7). B, in a parallel tube, 32Pi.M, 0, buffer B (pH 7) or with the equivalent buffer prepared with MES/K cells received 0.1 mM ["C]Glc-6-P ratherthan RK5000pRK10 (UhpT overproducer); A, A, MC4100 (UhpT wild at pH6 and containing 0.25 mM Na3V04. RK6814 (UhpT-negative). Reconstitution-Everted membrane vesicles (1mg of protein) were type); 0,0, washed and resuspended using 100 mM MOPS/K (pH 7). Protein was then solubilized (22) on ice with 1.1% octylglucoside, 0.37% acetone/ether-washed E. coli phospholipid, 20% glycerol (v/v), in 50 mM MOPS/K (pH 7). After incubation and centrifugation as described (22), the clarified extract was mixed with excess sonicated E. coli phospholipid, and proteoliposomes (or liposomes) were formed by 25-fold dilution into a buffer (pH 7) containing 100 mM (MOPS/K plus KPi or KAsO,) and 1 mM dithiothreitol; final levels of KPi and KAsOl ranged from 0.75 to 30 mM, as indicated in the legend to Fig. 5. Proteoliposomes (or liposomes) were washed and resuspended (22) in buffer C (20 mM MOPS/K, 75 mM K,SOI, 2.5 mM MgSO, (pH 7)) before assays of transport. Transport Assays-Aliquots of cells, vesicles, or proteoliposomes were preincubated 3-5 min at room temperature in the appropriate assay buffers before addition of labeled substrates. Intact cells were placed in buffer A at a density of 3.9pl of cell water/ml of suspension. Everted membranes and right-side-out membrane vesicleswere 5-0' placed a t 100 to 200 pg of protein/ml in either MOPS-based (pH 7) or MES-based (pH 6) buffer B. In most cases the suspension medium Minutes Minutes also contained 0.25 mM Na3V0, to reduce sugar phosphatase activity FIG. 2. Homologous andheterologous PI-linkedexchange in (vanadate did not affect the activity of Pi-linked exchanges), and where indicated the medium also contained 33 mM potassium ascor- membrane vesicles. A, right-side-out vesicles derived from strain bate and 0.12 mM reduced PMS to establish a proton-motive force RK5000pRK10 were loaded with 50 mM KPi (pH7) during preparaby oxidative reactions (23). Transport by proteoliposomes was tested tion ("Materials and Methods"). Washed vesicles at 70 pg of protein/ in buffer C with protein a t about 15 pg/ml suspension. Transport ml in buffer B were given 0.5 mM 32Pito start the reaction, followed at the arrow by either 5 mM KPi or 5 mM KAsO, (pH 7). In parallel assays were terminated by membrane filtration (15, 22). Other Procedures-Total sugar phosphatase activity was measured trials (not shown) external substrate was varied to study the dependby anion exchange chromatography, as described (15). Protein was ence of steady state 32Pilevels on external Pi. That work showed a determined by the method of Lowry et al. (24) (membrane vesicles) linear relationship between external Pi and the fraction of total "Pi incorporated at steady state, as expected if 32Piwere distributed or Schaffner and Weissmann (25) (proteoliposomes). This linear Chernicak-D-[l-'4C]Glucose 6-phosphate (50 Ci/mol), ~ - [ 2 , 3 - ~ H ]passively between outer and inner Pi pools(15,22). proline (28 Ci/mmol), and KH,3'P04 (1 Ci/mol) were obtained from relationship was used (15, 22) to derive external (517 nmol/ml) and New England Nuclear. With theexception of octylglucoside (Behring internal (1.35 nmol/ml) Pi spaces for the experiment shown. Taking Pias 50 mM,the 32Pi-accessibleinternal volume corresponded Diagnostics) all other materialswere purchased from Sigma or Fisher. internal to 0.39 pl/mg of protein.* B, In a different experiment Pi-loaded everted vesicles of RK5000pRK10 (UhpT overproducer) were susRESULTS pended a t 100 pg of protein/ml inbuffer B containing either no other Pi:Pi Exchange by Intact Cells-Analysis of intact cells of additions (ethanol control) (O),1 p~ valinomycin ( V A L ) (M), or 1 strains that differedin the expression of UhpT demonstrated p~ valinomycin plus 1p~ nigericin (NZG) (A).Transport was measured after addition of 0.1 mM ["C]Glc-6-P. At the arrow the control clearly that UhpT could accept 32Pias a substrate for both tube was divided into threeportions, which were giveneither 30 mM transportandexchange. There was,forexample, a pro- KP,(pH 7) (O), 1 p~ valinomycin plus 1 pM nigericin (A), or nounced elevation in the rate of 32P;or [14C]G1c-6-P entry additional buffer (0).Transport by strain RK5000 (UhpT-negative) when UhpT was overexpressed(Fig. 1) Internal 32Piwas not ('(7) was also measured in buffer B. availableforexchangewithexternal Pi unless UhpT was induced, and because this capacity for exchange was elimiHomologous and Heterologous Exchanges in Membrane Vesnated by TnlO insertion intouhpT and because external Glc- icles-Work withmembranevesicleshasconfirmedthat 6-P also prevented 32Pientry (not shown; see Ref. 18), we UhpT mediates the homologous32Pi:Pi exchange identified in concluded that the UhpT protein itself mediated homologous intact cells. This energy-independent exchange, as it appears 32Pi:Pi antiport. We presumed as well that 32Piincorporation inright-side-outvesicles, is shownby Fig. 2 A . Pi-Loaded 32Pi(to 20 nmol of Pi/mg of protein) by the UhpT-negative strain RK6814 reflected the operation vesicles took up external by a process that reflected the passive distribution of isotope of Pit, thenH+/Picotransport system expressed constitutively by K12 strains; it is known (18, 26) that P i t does not carry betweeninside and outside Pi pools. Thisconclusionwas based on four observations. First, "Pi was not taken up by o u t an exchange of internal for external Pi in such experiMOPS-loaded vesicles(unless a n oxidizablesubstratewas ments.

Pi-Linked Sugar Phosphate Antiport in E. coli

6627

present; see below). Second, incorporation of 32Piby Pi-loaded vesicles was unaffected by 5 p M FCCP, 0.5 p M valinomycin, or 0.5 p~ valinomycin plus 0.1 PM nigericin (data not given; but see Fig. 2B). Third, there was rapid chase of internal 32pi when unlabeled Pi or AsO, was added in excess (Fig. 2 A ) . And finally, the quantity of 32Piassociated with vesicles was determined directly by the relative masses of internal and external Pi pools, and expansion of the external pool by added Pi led to predictable decreases of isotope incorporation (Fig. 2 A , legend). Withthe assumption of passive exchange, these variable levels of isotope incorporation were also used to calculate a 32Pi-accessiblespace of 19 nmol of Pi/mg of protein (Fig. 2A, legend) (15, 22). Similar values for a 32Pi-accessible space were found in three other instances, from which we derived an equivalent internal volume of 0.39 f 0.06 pl/mg of protein (mean k S.E.) for right-side-out vesicles? Other work pointed to UhpT function as essential to this distribution of 32Pi,since PJoaded vesicles derived from UhpT-negative strains did not take up32Piin this manner (see below). Taken exchange identogether, such findings verify that the 32Pi:Pi tified in intact cells (Fig. 1)is preserved in membrane vesicles. The movements of Glc-6-P were studied in similar experiments (Fig. 2B), and such work suggested that UhpT also mediates the heterologous exchange of Pi and Glc-6-P. Thus, accumulation of Glc-6-P was found using Pi-loaded everted vesicles of RK5000pRK10, but not for similar vesicles of the UhpT-negative host RK5000 (Fig. 2B), nor was Glc-6-P transport found if vesicles of RK5000pRK10 had been loaded with MOPS ratherthan Pi (not shown; see below). Assuming an internal volume of 0.5 pl/mg of protein for these everted vesicles (20), the observed accumulation of Glc-6-P (Fig. 2B) corresponded to an inside concentration considerably higher than in the external medium (11 uers'sus 0.1 mM), but maintenance of this gradient could not be attributed to nH+/anion symport. Not only was the reaction unaffected by ionophores that collapse ion-motive gradients (Fig. 2B), but, andequally important, the experiment had used everted rather than rightside-out vesicles, so that the proton-motive gradient arising from residual metabolic activity should have been directed outward rather than inward. (Other studies (not given) documented an ionophore-insensitive Glc-6-P transport by Piloaded vesicles of normal polarity.) It is also worth noting that the Glc-6-P taken up by Pi-loaded vesicles was stable with time, even in the presence of ionophores, but was promptly released on addition of unlabeled Pi (Fig. 2B), and this finding, along with the other information given here, strongly suggests that Glc-6-P accumulation (Fig. 2B) derived from an exchange with internal Pi. For these reasons, we concluded that UhpT mediates heterologous Pi:Glc-6-P antiport and that the direction of exchange depends on the direction of the driving (Pi) gradient ratherthan on the sidedness of membrane StNCtUre. In addition, because ionophores had not alteredthe rateor extentof this heterologous exchange, in right-side-out (not shown) or everted (Fig. 2 B )

vesicles, we also concluded that theheterologous reaction was a neutral antiport. Role of the Proton-motive Force in Sugar Phosphate Transport-The finding that UhpTmediates both homologous Pi:Pi and heterologous Pi:Glc-6-P exchange is compatible with anion exchange as the underlying mechanism of Glc-6-P transport, but this observation alone is not enough to reject the idea of nH+/anion symport-for example, it is quite clear that symporters can display substrate exchange (27), often at rates more rapid than the reaction leading to net flux (28). Therefore, to distinguish the two modes of UhpT function (H+/ anion symport, Pi-linked antiport), itwas necessary to establish directly whether transport of sugar phosphate required a suitable countersubstrate in the presence of a driving ionmotive gradient. To examine the dependence of Glc-6-P transport on the presence or absence of trans substrates, MOPS-loaded, rightside-out vesicles wereprepared from both UhpT-positive and UhpT-negative strains. Studies of proline transport were then used to evaluate the effectiveness of substratetransport driven by oxidative metabolism. As shown in Fig. 3A, MOPSloaded vesicles fromboth strains hadsimilar capacities in this regard, and in both instances the 200-fold accumulation of proline indicated a net ion-motive driving force near 140 mV for these conditions (taking internal volume as 0.39 Fl/mg of protein and stoichiometry as 1 H+(orNa+)/proline (29)).The same kinds of experiments (Fig. 3B) showed also that MOPSloaded vesicles of UhpT-positive and UhpT-negative strains were equally effective in the transport of 32Pi(steady state levels of 4.8 +- 0.3 and 3.6 f 0.5 nmol/mg of protein for strains RK5000pRK10 and RK6814,respectively; mean k S.E. of eight and five experiments as in Fig. 3B) and that such 3zPi accumulation was sensitive to FCCP (0.7 f 0.1 and 0.6 +. 0.1 nmol/mg of protein, respectively). This behavior, which was unlike the passive 32Pidistribution in Pi-loaded vesicles (Fig. 2 A ) , was presumed to reflect the energy-dependent 3zPiaccumulation by Pit, a system that carries out nH+/Pi symport (18, 30). Although MOPS-loaded vesicles couldtake up both proline and Pi (Fig. 3), significant accumulation of Glc-6-P was not

This value, which assumes an internal concentration of 50 mM Pi, is lower than the dextran-inaccessible volume (2.2 pl/mg of protein) reported for vesicles of E. coli strain ML308-225 (31). This difference is probably due to straindifferences or to our modifications of the original method of vesicle preparation (see "Materials and Methods") rather than to loss of internal Pi during washing and resuspension in Pi-free media. The latter explanation was ruled out because this experiment (Fig. 2 . 4 ) had included a parallel trial with unwashed vesicles diluted 100-fold to achieve 0.5 mM external Pi;the washed and unwashed vesicles had identical 32Pi-accessiblespaces. ible Size considerations also lead one to expect that a 3 2 P i - a ~ ~ e ~ sspace may be smaller than a dextran-impenetrable space.

FIG. 3. Proline and Pi transport by MOPS-loaded vesicles. A , MOPS-loaded right-side-out vesicles were placed at 98 (RK5000pRK10) or 111 (RK6814) pg of protein/ml in MES-based buffer B (pH 6). Vesiclesweregiven0.12 mM PMS and 33 mM ascorbate/K (pH 6) 3 min before adding 50 p~ ["Clproline, in the presence or absence of 5 p M FCCP asindicated; 5 pM FCCP or control amounts of ethanol were added at the arroms. B, in a separate experiment, 0.1 mM 32Piwas given to MOPS-loaded vesicles placed at 141 (RK5000pRK10) or 198 pg (RK6814) of protein/ml in MESbased buffer B (pH 6); other conditions were as in A. Solid symbols, RK5000pRK10 (UhpT overproducer); open symbols, RK6814 (UhpTnegative).

VhpT'

2 3

UhpT-

1

OO

Minutes

p

.FCCP

4 0 Minutes

12

E, coli

Pi-Linked Sugar Phosphate Antiport in

6628

observedfor thesesameconditions,Instead, vesicles from UhpT-positive and UhpT-negative strains behaved in much the sameway, and in the experiment shown (Fig. 4A) neither took up more than 0.3 to 0.6 nmol Glc-g-P/mg of protein, even after extended incubation;a similar result was found in two other cases inwhich, after 15 minof incubation, MOPSloaded vesicles of the UhpT-positive straingave an apparent net accumulation of 0.5 nmol of Glc-g-P/mg of protein over FCCP-treated controls (see also Table I). Because such negativefindings were madeboth at p H 6 (Fig. 4A), where membrane potential and pH gradient each contribute to the proton-motive force (29), as well as at pH 7 (not shown), where the proton-motive force is mainly a membrane potential (29), it seemed unlikely that Glc-6-P transport was mediated by either electroneutral or electrogenic nH+/Glc-6-P symport. For these same conditions (pH 6 and 7 ) , however, when vesicles of the UhpT-positive strain were first incubated with Piplus PMS/ascorbate, so that accumulationof Pi could occur (cf. Fig. 3B), the subsequent addition of Glc-6-P was followed immediately by its rapid incorporation to3.5 nmol/ mg of protein (Fig. 4B).(Anion exchange chromatography of vesicle contents confirmed that(5% of this label was present as the free sugar.) This response was not observed when Pi

and PMS/ascorbate were added separately, when FCCP was also present, or when vesicles of the UhpT-negative strain were used (Fig. 4B).These same observationswere made in a number of other instances, as summarized by Table I, andin those cases, too,added Pi had a pronounced andpositive effect that elevated steady state Glc-6-P accumulation toa level at least 10-fold above FCCP-treated or UhpT-negative controls (3.7 uersus 0.1 to 0.4 nmol/mg of protein, respectively). Considered against thisresponse, a slow (cf. Fig. 4B)but eventual Glc-6-P incorporation without added Pi (to 1.1 nmol/mg of protein) was of uncertain significance, sincePi may have been present inadvertently, as a result of residual phosphatase activity, from a n endogenous source (phospholipids, etc.), or even as a contaminant of Glc-6-P. We did not pursue this issue further, but concluded instead that overall this work gave strong indicationof a requirement for internal Pi for the for the accumulation of Glc-6-P (Fig. 4,Table I), but not accumulation of proline orPi (Fig. 3 ) . This important finding is not easily interpreted if UhpT mediates nH+/anion symport, but is predictable if UhpT operates by Pi-linked exchange. Accumulation of Glc-6-P by Reconstituted UhpT-The ex60C

UhpT' UhpT-

=

3

No Additions rn 'PMSIA5C

3 0

Minutes

FIG. 4. G6P accumulation requires internal Pi. A , MOPSloaded vesicles were placed at 119 pg (RK5000pRK10, UhpT') or 160 pg (RK6814 UhpT-) of protein/ml in MES-based buffer B (pH 6) in the absence or presence of 0.12 mM PMS plus 33 mM ascorbate (Asc) (pH 6), as indicated, followed several minutes later by addition of 0.1 mM ["C]Glc-6-P. B , in parallel tubes the vesicles fromUhpT-positive and UhpT-negative strains were preincubated for 15 min with 0.1 mM KP,, and with other supplements as noted in A , before adding 0.1 mM ["C]Glc-6-P; 5 p~ FCCP was used where indicated.

A

TABLEI Effect of phosphate on sugar phosphate transport MOPS-loaded vesicles were preincubated for 15 min with 0.12 mM PMS and 33 mM ascorbate/K with supplements as indicated, and steady state Glc-6-P incorporation was measured 12 min after the subsequent addition of 0.1 mM labeled Glc-6-P. Data are shown as mean k S.E. of nine (RK5000pRK10) and four (RK6814) experiments. The table summarizes data from tests at both pH 6 and pH 7, since results did not depend on assay pH within this range. Strain

Additions to assay

Glc-6-P incorporation nrnollmg of protein

RK5000pRK10 None + 0.1 mM KPi + 0.1 mM KP, + 5 p M FCCP

1.1 +. 0.2 3.7 -t 0.6 0.4 +. 0.1

RK6814

0.2 +. 0.02 0.4 k 0.04 0.1 f 0.04

None + 0.1 mM KPi + 0.1 mM KP; + 5 uM FCCP

9

-

J

~

50

100

Minutes FIG. 5 . Reconstitution of UhpT-mediated Pi:Glc-6-P exchange. Membrane protein was reconstituted as described under "Materials and Methods." Proteoliposomes with protein from the UhpT overproducer RK5000pRK10 were prepared to contain KPi at 0 mM (o), 0.75 mM (A), 1.5 mM (v),3 mM (o), 10 D M (a), and 30 mM Pi (A)or 30 mM KAsO, (m);proteoliposomes using protein from the UhpT-negative strain RK6814 were loaded with 30 mM KPi (*). Sugar phosphate transport was measured after addition of 0.035 mM [14C]Glc-6-P.Inset: steady state Glc-6-P (G6P) incorporation as a function of initial internal Pi. Parallel trials (not shown) monitored steady state incorporation of 0.1 mM 32Pito estimate the internal Pi pool available to the exchange reaction. Taking internal volume as 1 pl/mg of phospholipid (22) and internal KPi as stated above, such assays gave zero time Pi contents as 54, 92, 118,510, and 1360 nmol of PJmg of protein, respectively, for proteoliposomes with final levels of Glc-6-P of 17, 34, 63, 215,and 520 nmol/mg of protein.

Pi-Linked Phosphate Sugar Antiport

in E. coli

6629

represented a reasonable interpretation of data. In particular, earlier work had often used functional tests in the presence of external Pi and had always employed either intact cells or Pi-loaded membrane vesicles. But since even low levels of Pi can play a catalytic role in Glc-6-P transport via nH+/Pi symport (Fig. 4B), these conditions would inevitably cause an indirect coupling between the proton-motive force and Glc-6P accumulation. Clearly, interactions among Pi and Glc-6-P transport systems make it difficult to assess the individual contributions of each unless precautions are taken toexclude Pi from the analysis. It also seems likely that the exchange mediated by UhpT in E. coli can function, as in S. &tis, in a manner that more directly mimics a neutral 2H'/Glc-6-P2symport. Quantitative analysis of exchange in S. lac& (16) suggests a bifunctional active site that binds either a pair of monovalent anions or a single divalent species, and for that case the neutral self-exchange involving sugar phosphate (2[HGlc-6-P"]:Glc-6-P2-) leads to net entry of 2Hf and Glc6-P2- in the presence of a pH gradient. In many respects, the exchanges mediated by UhpT in E. coli resemble those associated with sugar phosphate transport in Gram-positive cells such as S.lactis (14-17) and S. aureus (17, 32). Indeed, similarities among these sugar 6-phosphate transport systems seem to extend beyond a superficial classification as anionexchange. For example, in each case,sugar 6-phosphate is the preferred substrate ( K t values of 20-40 ~ L M (15,16, 18, 32)),but inorganic phosphate or arsenate can also take part in exchange, albeit with a somewhat lower affinity ( K t values of 0.3 to 3 mM during homologous Pi:Pi antiport (14,18,32)).For this reason, and also because the homologous and heterologous exchanges involving Pi are convenient experimental tools, such antiport systems have been classified as "Pi-linked" (16, 17). This does not imply that P, must always be a substrate during exchange, for in some cases Pi may play a minor role in the physiological reaction (16, 17). The Pi-linked exchange mediated by UhpT is clearly an electroneutral event, whether one considers the homologous (Pi:Pi)or heterologous (Pi:Glc-6-P) reaction (Figs. 1 and 2 DISCUSSION and text).These exchanges are also neutral events in S. lactis Three kinds of experiments support the idea that anion (15,16) andS. aureus (32), and in those cases the participation Pi appears to be restricted to the monovalent anion (14, exchange forms the mechanistic basis for UhpT function in of 33). If one presumes this is also true in E. coli, then the E. coli. First, work reported here and elsewhere (18) establishes that UhpT mediates rapid transmembrane 32Pimove- neutral antiportof two monovalent Pi anions for onedivalent Pi flux-that is, UhpT carries Glc-6-P (pK= 6.1) offers a simple explanation for the apparment without a concomitant net out a "tightly coupled 32Pi:Pi antiport. Second, it is now also ent stoichiometry (2.5:l) found during heterologous exchange clear that UhpT catalyzes a similarly tight exchange of Pi for in proteoliposomes (Fig. 5). It is otherwise not easy to underGlc-6-P, so that no external source of energy is required for stand why stoichiometry did not fall below 2:1 in that experGlc-6-P accumulation by Pi-loaded vesicles (Fig. 2B) or pro- iment (Fig. 5, inset), since sampling times extended well into teoliposomes (Fig. 5). In these cases Glc-6-P is taken up at the steady state. This line of reasoning, although indirect, rates and to extents determined by internal (trans) Pi, in a seems worth pursuing in light of the similarities of Pi-linked neutral event unaffected by ionophores that collapse cation- exchange in E. coli and S. lactis (above) and also because of motive gradients. Third, and perhaps most important, even the direct measurement of a 2:1 stoichiometry for heterologous though oxidative metabolism drives both proline and Pi ac- exchange in S. lactis. The finding of anion exchange in both Gram-positive and cumulation in MOPS-loaded vesicles, the transport of GIc-6Gram-negative cells indicates a mechanism widely spread P is impaired (Fig. 4A);it recovers only if a suitable trans substrate is available (Fig. 4B, Table I). The requirement for among bacteria and suggests one might anticipate examples a suitable trans substrate (in this case, Pi) is not expected of in addition to those dedicated to sugar 6-phosphate transport. nH+/Glc-B-P symport, but itis clearly predicted by the model In this context, isit worth noting that circumstantial evidence implicates anion exchange as the basis of sn-glycerol 3-phoSof anion exchange. Earlier interpretations of UhpT function had suggested the phate movement by the GlpT protein of E. coli (18, 26) and idea of nH'/Glc-6-P symport (11-13), and the work summa- that preliminary work4 points to the same conclusion for rized here also allows one to understand why that model accumulation of phosphoenolpyruvate by the PgpTtransport protein of Salmonella typhimurium (34). These examples em3For these conditions of reconstitution, internal volume should phasize that Pi-linked exchange serves as a general mechacorrespond to about 40 d/mg of protein (18, 22). The levels of 32Pi nism for the transport of organic phosphates (17), a conclu-

periments outlined above demonstrated that Pi can play a catalytic role in Glc-6-P transport, but to assess the quantitative nature of this relationship it was necessary to examine the exchange reaction in theabsence of an external source of energy. For this purpose we used the technique of reconstitution in studiesof Glc-6-P accumulation by proteoliposomes containing varying levels of internal Pi (0.75 to 30 mM) or arsenate (30 mM). Several important findings were made in this work. For example, elevations of internal Pi were accompanied by increases in both the initial rates and final extents of Glc-6-P transport (Fig. 5), and while rates of Glc-6-P movements were too rapid for quantitative analysis, steady state levels could be directly related to the initial contents of Pi (Fig. 5, inset). It also appeared that the initial load of P; was roughly twice the Glc-6-P content at steady state (aratio of 2.5 & 0.2, mean +- S.E.; Fig. 5, legend), a value consistent with an exchange stoichiometry of 2:l (phosphate:sugar phosphate), as found in the Gram-positive S. lactis (15, 16). This experiment showed also that Pi and arsenate were equally effective in supporting the heterologous exchange (Fig. 5, solid triangles versus squares), and verified the specific role of UhpT by demonstrating that protein from the UhpT-negative strain RK6814 failed to incorporate labeled Glc-6-P (Fig. 5, stars) or 32Pi(not shown). These last findings also excluded possible contributions from Pit (cf. Fig. 3B) orother Pi transport systems. Finally, it was clear that reconstitution led to significantly increased rates of Glc-6-P transport compared to those found in membrane vesicles. This probably reflected an underestimation of rate in vesicles (and perhaps cells), where Glc-6-P movement during heterologous exchange would be limited by both the need to replenish internal Pi and by an increasingly prominent and silent back reaction (Glc-6-P:Glc-6-P antiport)as Glc-6-P (high affinity (18)) builds up at theexpense of Pi (low affinity (18)).These factors should less adversely affect initial rate measurement in the artificial system, since proteoliposomes enclose a much larger volume per unit protein.3

established by Pi:Pi exchange (legend, Fig. 5) suggest this entire volume was accessible to the exchange reaction (18).

S. V. Ambudkar and P. C. Maloney, unpublished data.

6630

Pi-Linked Sugar Phosphate Antiport in E. coli

sion also supportedbysequencehomologiesamong these proteins (1, 35). It is likely that anion exchange will prove useful in other settings as well,because the transport of materials that participatedirectlyinmetabolism is better accommodated by antiport thanby symport. Acknowledgment-We thank Dr. R. J. Kadner for his generous gift of the bacterial strains used in this work.

REFERENCES 1. Friedrich, M. J., and Kadner, R. J. (1987) J. Bacterioll69,35563563 2. Kornberg, H. L., and Smith, J. (1969) Nature 2 2 4 , 1261-1262 3. Weston, L. A., and Kadner, R. J. (1987) J. Bacterwl 1 6 9 , 35463555 4. Winkler, H. H.(1970) J. Bacteriol 101,470-475 5. Shattuck-Eidens, D.M., and Kadner, R. J. (1983) J. Bacterwl 148,203-209 6. Dietz, G.W., and Heppel, L. A. (1971) J. Bwl. Chem.246,28852890 7. Pogell, B. M., Maity, B. R., Frumkin, S., and Shapiro, S. (1966) Arch. Bwchem. Biophys. 116,406-415 8. Winkler, H. H. (1966) Biochim. Biophys. Acta 117, 231-240 9. Winkler, H. H. (1973) J. Bacteriol 1 1 6 , 203-209 10. Dietz, G. W. (1972) J. Bwl. Chem. 2 4 7 , 4561-4565 11. Essenberg, R. C., and Kornberg, H. L. (1975) J.Biol. Chem. 250, 939-945 12. Ramos, S., and Kaback, H. R. (1977) Biochemistry 16, 42714275 13. LeBlanc, G.. Rimon,. G.,. and Kaback, H. R. (1980) Biochemistry 19,2522-2528 14. Malonev. P. C.. Ambudkar. S. V.. Thomas. J.. and Schiller. L. (1984) 2. Bacteriol158,238-245 15. Ambudkar, S. V., and Maloney, P. C. (1984) J. Biol. Chen. 2 5 9 , 12576-12585 16. Ambudkar, S. V., Sonna, L. A., and Maloney, P. C. (1986) Proc. Natl. Acad. Sci. U. S. A . 83,280-284 17. Maloney, P. C., Ambudkar, S. V., and Sonna, L. A. (1987) in I

_

Phosphate Metabolism and Cellular Regulation in Microorganisms (Torrianni-Gorini, A., Rothman, F. G., Silver, S., Wright, A., and Yagil,E., eds) pp. 191-196, American Society for Microbiology, Wash., D. C. 18. Ambudkar, S. V., Larson, T. J., and Maloney, P. C. (1986) J. Biol. Chem. 261,9083-9086 19. Casadaban, M. J. (1976) J. Mol. Biol. 104,541-555 20. Ambudkar, S. V., Zlotnick, G.W., and Rosen, B. P. (1984) J. Bwl. Chem. 259,6142-6146 21. Kaback, H. R. (1971) Methods EnzymoL 22,99-120 22. Ambudkar, S. V., and Maloney, P. C. (1986) J. Biol. Chem. 2 6 1 , 10079-10086 23. Konings, W. N., Barnes, E. M., Jr., and Kaback, H. R. (1971) J. Biol. Chem. 246,5857-5861 24. Lowry, 0.H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Bid. Chem. 193,265-275 25. Schaffner, W., and Weissmann, C. (1973) Anal. Biochem. 56, 502-514 26. Elvin, C. M., Hardy, C. M., and Rosenberg, H. (1985) J. Bacteriol. 161,1054-1058 27. Wong, P. T. S., and Wilson, T. H. (1970) Biochim. Biophys. Acta 196,336-350 28. Kaczorowski, G. J., and Kaback, H. R. (1979) Biochemistry 18, 3691-3697 29. Maloney, P.C. (1987) in Escherichia coli and Salmonella &phimurium. Cellular and Molecular Biology (Neidhart, F., Ingraham, J. L.,Low, K. B., Magasanik, B., Schaechter, M., and Unbarger, H. E., eds) pp. 222-243, American Society for Microbiology, Wash., D. C. 30. Rosenbern. H., Gerdes. R. G., and Harold, F. M. (1979) Biochem. J. 178,-i33-137 31. Kaback, H. R., and Barnes, E. M., Jr. (1971) J. Biol. Chem. 246, 5523-5531 32. Sonna. L.A.. and Malonev. " , P. C. (1988) . , J. Membr. Biol. 101. 267-274 ' 33. Mitchell, P.(1954) J. Gen. Microbiol. 1 1 , 73-82 34. Saier, M. H., Jr., Wentzel, D. L., Feucht, B. U., and Judice, J. J. (1975) J. Biol. Chem. 2 5 0 , 5089-5096 35. Eiglmeier, K., Boos, W., and Cole, S. T. (1987) Mol. Microbwl., in press