JOURNAL OF BACTERIOLOGY, Aug. 1986, p. 666-6730021-9193/86/080666-08$02.00/0Copyright © 1986, American Society for Microbiology

Recognition and Transport of Ferric Enterobactin inEscherichia colit

Vol. 167, No. 2

DAVID J. ECKER, BERTHOLD F. MATZANKE, AND KENNETH N. RAYMOND*Department of Chemistry, University of California, Berkeley, California

Received 19 November 1985/Accepted 9 May 1986

The specificity of the outer membrane protein receptor for ferric enterobactin transport in Escherichia coliand the mechanism of enterobactin-mediated transport of ferric ions across the outer membrane have beenstudied. Transport kinetic and inhibition studies with ferric enterobactin and synthetic structural analogs havemapped the parts of the molecule important for receptor binding. The ferric complex of the synthetic structuralanalog of enterobactin, 1,3,5-N,N',N"-tris-(2,3-dihydroxybenzoyl)triaminomethylbenzene (MECAM), wastransported with the same maximum velocity as was ferric enterobactin. A double-label transport assay with[59Fe, 3H]MECAM showed that the ligand and the metal are transported across the outer membrane at anidentical rate. Under the growth conditions used, large fractions of the transported complexes were availablefor exchange across the outer membrane when a large excess of extracellular complex was added to the cellsuspension; at least 60% of the internalized [59Felenterobactin exchanged with extracellular [55Fe]enterobactin.Internalized [59Fe, 3H1MECAM was released from the cell as the intact complex when either unlabeledFe-MECAM or Fe-enterobactin was added extracellularly. The results suggest a mechanism of active transportof unmodified coordination complex across the outer membrane with possible accumulation in the periplasm.

The mobilization and uptake of iron by microbes is medi-ated by low-molecular-weight complexing agents(siderophores) (22, 34). A siderophore produced by Esche-richia coli, enterochelin (27) (here called enterobactin) (31),is the most powerful ferric ion complexing agent known andhas been among the most thoroughly studied of thesiderophores (34). However, the detailed mechanism(s) bywhich iron is delivered to E. coli is unknown. Since E. coli isencapsulated by two membranes, the cytoplasmic mem-brane and the outer membrane, these barriers and theintervening periplasmic space potentially all play a role inthe uptake process. The outer membrane contains about 105molecules per cell of the protein porin (25). The porinproteins span the membrane and associate into trimers,which form water-filled channels that allow nonselectivepassage of solutes measuring up to about 600 daltons. Theinner membrane, however, is a more selective barrier.Virtually everything that is known about active transport inE. coli has come from studies of inner membrane transportsystems.For substrates larger than 600 daltons, E. coli expresses

specific outer-membrane receptors (recently reviewed) (1,23). Outer membrane receptors have been identified forvitamin B12, nucleosides, maltodextrans, certain colicins,phages, and the iron chelates ferrichrome, coprogen, ferriccitrate, and ferric enterobactin.

In the present paper, we report studies on the bindingspecificity and mechanism of receptor-mediated ferricenterobactin transport in E. coli. It has been previouslyreported that a number of synthetic structural analogs ofenterobactin (Fig. 1) serve as a source of iron for E. colisiderophore auxotrophs in growth promotion tests (13). Ofthese compounds, 1,3,5-N,N',N"-tris-(2,3-dihydroxyben-zoyl)-triaminomethylbenzene (MECAM) is also active intransport assays. We extended this work with detailed

* Corresponding author.t No. 35 in the series Coordination Chemistry of Microbial Iron

Transport Compounds.

kinetic and inhibition studies to probe the binding specificityof the outer membrane ferric enterobactin (fepA) receptor.Of all the substrates that are transported via a specific

outer membrane receptor, only in the case of vitamin B12 hasactive transport across the outer membrane into the periplas-mic space been demonstrated (36). In the present paper, wereport evidence that ferric enterobactin is actively accumu-lated across the outer membrane as an undissociated, metal-ligand complex.

Enterobactin possesses a cyclic backbone consisting ofthree ester-linked serine residues (Fig. 1). O'Brien et al. (26)observed that, after 45 min of incubation of cells withFe(ent), the absorbed complex was hydrolyzed because ofthe activity of an intracellular esterase. An iron releasemechanism involving this esterase was proposed, and sup-port for this model came from the observation that thereduction potential of ferric enterobactin at pH 10 is -1.0 V,with an estimated potential at pH 7 of -0.75 V, which wouldseem to preclude physiological reduction (5, 17). However, aquestion arose of whether the substrate for the esterase isFe(ent)3- (2) or the decomplexed free ligand (9). Greenwoodand Luke (9) re-examined the esterase activity and foundthat the esterase acts on both the free ligand and the ironcomplex, but with a 2.5-fold-higher rate for the free ligand.One major argument against the esterase iron release

mechanism is that iron transport and growth promotion havebeen observed by using enterobactin analogs which are notsusceptible to hydrolysis. The synthetic analog MECAMhas, in place of the enterobactin triester ring, a skeleton of atriamine derivative of mesitylene (Fig. 1) that is not suscep-tible to hydrolysis. Growth stimulation tests with severalanalogs, including MECAM (13, 16, 24, 40), showed growthpromotion. Furthermore, nonhydrolytic iron removal fromMECAM by extracts of Bacillus subtilis has been reported(19). Thus, the esterase mechanism for in vivo iron releasefrom ferric enterobactin has been questioned.An alternative mechanism has emerged from investiga-

tions of the coordination chemistry of Fe(ent) at low pH (14,15, 28, 29). In addition to the early electrochemical studies of

666

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

FERRIC ENTEROBACTIN TRANSPORT IN E. COLI 667

ENTEROBACTIN

OH

H

MECAM

TRIMCAM

OHR

R ROH OH

R = CH2-NH-C-

OH H

0

R = C-NH--CH2i&,

NH-CH2-CH2-CH2--N-CH2-CH2-CH2-CH2-NH

C=O C=0 C=0S03O0H OAOH S OH

SO- OH so; OH so-

H3 3 ~~~~~~~~3

3,4-LICAMSFIG. 1. The structures of enterobactin and several synthetic

enterobactin analogs: MECAM [1,3 ,5-N,N' ,N"-tris-(2,3-dihydroxybenzoyl)-triaminomethylbenzene]; TRIMCAM [1,3,5-tris-(2,3-dihydroxybenzoylcarbamido)benzene]; and 3,4-LICAMS[1,5,10-N,N',N"-tris-(5-sulfo-2,3-dihydroxybenzoyl)triazadecane].

Fe(ent) (5, 11, 17), it has been found that reductive removalof Fe from enterobactin is achieved by glutathione at pHvalues lower than 6 (10). The observed protonation con-stants of ferric enterobactin (11, 12) led to our suggestion(28) that the sequential protonation of the Fe(ent) complexwith decreasing pH increases the potential sufficiently that areductive mechanism or iron removal might be possible in alow-pH environment of the cell. This proposal has beencriticized by Hider et al. (14), who stated that "Bacteria,unlike eukaryotes, possess a single intracellular compart-ment, and it is extremely unlikely that a bacterium wouldpermit the pH of the cytoplasm to fall to these acidic pHvalues..." However, E. coli, like other gram-negativebacteria, has an inner and an outer membrane. The interven-ing periplasmic space is an acidic compartment of the cell (6,

38). It constitutes up to 40% of the cell volume. From recentobservations (20), including those reported here, this regionof the cell seems to be the principal region of ferricenterobactin accumulation in its early uptake by E. coli. Wehave recently measured the reduction potential of ferricenterobactin in the pH range 6 to 11 (17). The potential at pH6 is -0.56 V. Thus, reduction in an acidic environment at thesurface of the cell remains a plausible mechanism for ironremoval from E. coli.

In this and the following paper (20), we will consider threepossible mechanisms for iron removal from ferricenterobactin. (i) The first step in the metabolism of Fe(ent)after transport into the cytoplasm is hydrolysis of the esterbonds in the backbone by a specific esterase, the fes geneproduct (9). (ii) The iron is removed from Fe(ent) by areductase. (iii) A local low pH environment of the cell allowsthe reductive removal of iron.

MATERIALS AND METHODS

Bacterial growth. For all transport assays, E. coli K-12strain RW193 (ATCC 33475) was used (9). This enterobactinsynthesis-deficient mutant (pro leu trp thi purE entA) waskindly provided by Phil E. Klebba. For enterobactin isola-tion, the strain AN311 was used (43). It is a strain proficientin enterobactin synthesis but deficient in enterobactin-mediated iron uptake (F- pro leu thi fep) and was kindlyprovided by Carmen Sciortino. Cultures were maintained onagar slants containing YM broth (Difco Laboratories, De-troit, Mich.; 21 g/liter, pH 7.4). Cultures were grown in 100ml of iron-deficient medium in baffled 500-ml Erlenmeyerflasks in shaking water baths (New Brunswick Scientific Co.,Inc., Edison, N.J.; model G-76) at 37°C with the platformrotating at approximately 240 rpm. All culture glassware wasacid washed and soaked overnight in a 1 mM EDTA solu-tion, followed by extensive rinsing with deionized, double-distilled water.

Iron-deficient medium was prepared with deionized, dou-ble-distilled water as described by Neidhardt et al. (21) butleaving out the iron and tricine, and supplementing with 4mM citric acid, 50 mg each of L-proline, L-leucine, andL-tryptophan per liter and 10 mg of thiamine per liter. Thecalculated iron contamination from the reagent chemicalsused in this medium is approximately 0.2 ,uM, and no furthersteps were taken to remove it.

Cell growth was followed by measurement of the opticaldensity at 650 nm in an HP-8450A spectrophotometer (Hew-lett-Packard Co., Palo Alto, Calif.). A standard curve ofoptical density at 650 nm versus cell dry weight was linear toan optical density at 650 nm of 1.8 (1.0 mg of cell dry weightper ml) on this instrument. Cell dry weight was determinedas previously described (21). The E. coli RW193 grown iniron-deficient medium attained a maximum density of ap-proximately 0.5 mg of dry weight per ml after 24 h ofincubation. Cultures supplemented with 5 ,uM ferricenterobactin attained more than double this density. Nogrowth was obtained if the citrate was left out of themedium. For transport assays, cells were grown for 24 h iniron-deficient medium to a density of approximately 0.5mg/ml. Cells were washed twice with iron-deficient mediumwithout phosphate, citric acid, glucose, or amino acid sup-plements at pH 7.4, and were resuspended in that medium.When glucose was added, the concentration was 1 g/liter.The cell concentration was adjusted to 1.0 to 1.5 mg of celldry weight per ml unless otherwise noted, and 10-ml aliquotswere pipetted into a 125-ml baffled Erlenmeyer flask, cov-

VOL. 167, 1986

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

668 ECKER ET AL.

ered with Parafilm, and stored on ice until use. At 15 minbefore each experiment, the flasks were placed in the waterbath at 37°C and shaken at 240 rpm.

Preparation of metal complexes. Enterobactin was isolatedfrom cultures of E. coli AN311 as previously described (43).Syntheses of catechoylamide enterobactin analogsMECAM, TRIMCAM, and 3,4-LICAMS have been de-scribed previously (41, 42). Enterobactin, MECAM, andTRIMCAM were dissolved in methanol, and 3,4-LICAMSwas dissolved in methanol and water (1:1). Standardized55Fe, 59Fe, or 67Ga (New England Nuclear Corp., Boston,Mass.; specific activities approximately 60 to 80 dpm/pmol)solutions were prepared as the nitrilotriacetic acid (NTA)complexes at a ratio of metal to NTA of 1:2 in 0.1 M MOPS(morpholinepropanesulfonic acid) buffer at pH 7.4. Metal-siderophore complexes were formed by adding the M(NTA)2solution to the siderophore solution under argon. Themetal/siderophore ratio was 1:1.1 for most measurements,except inhibition studies, where it was exactly 1:1. Themethanol was removed under vacuum, and the volume wasmade up with MOPS buffer. The final metal concentrationwas 1 mM. For [59Fe, 3H]MECAM, the complex was puri-fied by high-voltage paper electrophoresis as previouslydescribed (7), but with 0.1 HEPES (N-2-hydroxy-ethylpiperazine-N'-2-ethanesulfonic acid) buffer at pH 7.8.The metal complex was eluted from the paper and filtered,and the concentration was determined from the extinctioncoefficient (at pH > 8, 490 = 4,900 M-1 cm-1) (11). Chemicaland radiopurity of all ligands and metal-ligand complexeswere determined by analytical high-pressure liquid chroma-tography and thin-layer chromatography (ligands) and elec-trophoresis and autoradiography (metal complexes). Underthe electrophoresis conditions described above, the[67Ga]enterobactin complex migrated as a trianion, identicalto [59Fe]enterobactin. Greater than 95% of the 67Ga countswere associated with the band identified as Ga(ent), asdetermined by an overexposed autoradiogram.

Tritium labeling of MECAM. All labeling and purificationprocedures were carried out at the Lawrence BerkeleyLaboratory, National Institutes of Health tritiation facility.To a Pyrex tube (0.25-in. [0.63 cm] outside diameter) sealedon one end was added 10 mg of MECAM, 61 ,lI of aceticanhydride, 15 ,lI of trifluoroacetic acid, and 20 RI of tritiatedwater (0.8 Ci). The tube was frozen in liquid nitrogen,evacuated to 65 ,um, sealed with a torch, and placed in an oilbath at 140°C for 24 h. The tube was then frozen and opened,and the contents were rinsed into a lyophilization flask.Excess water and methanol were added, and the labeledcompound was lyophilized five times to remove exchange-able tritium. The compound was purified by preparativereverse-phase high-pressure liquid chromatography. Themobile phase was methanol and water (60/40), each contain-ing 5 mM acetic acid. Greater than 90% of the radioactivitywas eluted with the major peak. This procedure resulted inisolation of radiopure tritiated MECAM, as determined byanalytical high-pressure liquid chromatography and thin-layer chromatography with autoradiography (specific activ-ity, 1.2 mCi/p.mol). This was diluted 1 to 5 with nonradioac-tive MECAM for transport experiments.

Transport assays. In each of the transport experiments,time-dependent accumulation was linear for at least 6 min at0.5 and 6 ,uM complex concentration. For kinetic measure-ments, 0.8 ml of cell suspension was added to test tubescontaining concentrations of [55Fe]siderophore in a volumeof 0.2 ml, thus initiating uptake. After 2 (or 5) min, 5 ml of0.9% LiCl was added to each tube, and the cells were

filtered, washed, and counted as described. The uptake rateswere calculated from the interval between min 2 and 5.

Inhibition studies were performed in the same way, butwith a fixed [55Fe]enterobactin concentration and increasingconcentrations of inhibitor (ferric MECAM and ferric 3,4-LICAMS). Uptake of [Fe(ent)]3- at 0.5 p.M was taken asreference (100%). Ligand exchange cannot affect theradiodistribution. At neutral pH, the half-life of ferric ionexchange between the fully formed hexadentate siderophorecomplexes is at least many hours (35, 39). For complexconcentrations of 50 puM, we estimate the half-life forexchange to be about 800 min. Thus, in the course of theinhibition experiment (typically 10 min), no interference dueto ligand exchange is expected.

Radioactively labeled substrate in a volume less than 50 pulwas added to a 10-ml cell suspension (1.0 to 1.5 mg [dryweight] per ml) (final concentration of metal chelate, usually2 ,uM). When lower substrate concentrations were used inthe Km and V,a, determinations, the cell concentration waslowered proportionately to maintain a large excess of sub-strate relative to transport enzyme(s). Samples (0.4 ml) wereremoved at regular intervals, filtered through 0.45-p.m-poremembrane filters, and immediately rinsed twice with 5-mlportions of the same medium, without the glucose andsupplemented with 100 p.M NTA at pH 7.4. The backgroundcounts for iron adsorbed to filters without cells were deter-mined and subtracted from the values obtained with cells. Agrowth medium buffered with MOPS rather than with highconcentrations of phosphate was used (since MOPS buffer issubstantially free of contaminating iron, and the phosphatesalts are usually the largest source of contaminating iron[and other metals] in growth media). This departure fromprevious methods (21) eliminated the need for pretreating themedium by extraction or column chromatography to removecontaminating iron. More important, unlike phosphate,MOPS is a very weak metal chelator. For the same reason,we caution against the routine addition of NTA or citric acidin transport assay media and suggest that the use of thesechelating agents be eliminated or restricted to the solutionused to rinse the cells after filtration to lower the backgroundon the membrane filter.To the membrane filters in glass scintillation vials was

added 0.5 ml of 0.5 M HCl followed by 1 ml of ethyl acetate.Vials were capped and incubated for 1 h, after which 10 mlof Aquasol 2 (New England Nuclear) was added. Vials werevigorously shaken, stored overnight, shaken again, andallowed to equilibrate for several more hours at 12°C. Thisprocedure resulted in complete solubilization of cell-containing filters and in a homogeneous distribution ofradioactivity in the scintillation vial as determined by thedouble-channels method (4). Quench correction and calcu-lations of disintegrations per minute for double-label sampleswere made (3).

RESULTS

Transport assays with whole cells showed variability inthe rate of substrate uptake between batches of cells derivedfrom different cultures, probably caused by variability in thedegree of expression of outer membrane receptor or in thephysiological condition of the cells. However, with a singlebatch of cells, transport assays were repeatable withinapproximately 5%. Therefore, all direct comparisons of theuptake rates of different substrates were made within a singlebatch of cells.The requirement for metabolic energy in ferric entero-

J. BACTERIOL.

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

FERRIC ENTEROBACTIN TRANSPORT IN E. COLI 669

400

E

0E0

ECLa.

aL.

300

200 -

100 '

0 L I I I

0 4 8 12 16 20

minutesFIG. 2. Uptake of [59Fe]enterobactin and [59Fe]MECAM in E.

coli RW193 (pH 7.4). The concentration of each ferric complex was2 ,uM. The cell concentration was 1.2 mg of dry weight per ml.Glucose-starved cells were washed twice, resuspended in uptakemedium without any carbon source, and placed at 37°C with shakingfor 2 h before the experiment to delete endogenous glucose. Sym-bols: 0, [59Fe]enterobactin in glucose-supplemented cells; A,[59Fe]enterobactin in glucose-starved cells; 0, [59Fe]enterobactin inglucose-starved cells with glucose (4 g/liter) added after the 8-minsample was removed (arrow); *, [59Fe]MECAM in glucose-supplemented cells.

bactin uptake is shown in Fig. 2. Cells starved for an energysource took up ferric enterobactin at a greatly reduced ratecompared with glucose-supplemented cells. However, bothglucose-starved and glucose-supplemented cells repeatedlybound approximately 30 pmol/mg (dry weight) of ferricenterobactin within 15 s of addition, which probably repre-sents energy-independent binding to the outer membranereceptor. Similar energy-independent binding has been ob-served with cells poisoned by dinitrophenol (33). Transportactivity was immediately restored by the addition of glucoseto a glucose-starved culture (Fig. 2).Under rate-saturating substrate concentrations (2 ,uM or

more), [55Fe]enterobactin and [55Fe]MECAM were taken upat virtually identical rates (Fig. 3). The Vma and apparent Kmvalues for ferric enterobactin were 73 pmol/min per mg and0.23 ,uM, respectively, in reasonable agreement with apreviously reported value, Km = 0.1 ,uM (8). The values forferric MECAM in the same cell culture were Vma = 68pmol/min per mg and an apparent Km = 0.69 ,uM. The higherKm value for ferric MECAM is not surprising, since only afraction of the ferric MECAM exists as the same coordina-tion isomer as ferric enterobactin at the pH of the transportassay (see below).The rate of uptake in E. coli from the enterobactin

structural analogs TRIMCAM and 3,4-LICAMS was morethan 10 times lower than that of ferric enterobactin, consis-tent with previous growth studies (13). The ferric complexesof these two compounds did not compete with[55Fe]enterobactin for the receptor; ferric MECAM didinhibit [55Fe]enterobactin uptake (Fig. 4). These resultssuggest that ferric TRIMCAM and ferric 3,4-LICAMS donot bind to the ferric enterobactin (fep) receptor. The cell

may have some other mechanism to remove small amountsof iron from these compounds, which is adequate for growthpromotion. Thus, growth promotion assays are not reliableindicators of the functional equivalence of enterobactin andsynthetic analogs with respect to active transport.The results of the double-label transport assay with [59Fe,

3H]MECAM showed that the ligand and metal were takeninto the cell at an identical rate (Fig. 5). Glucose-starvedcells initially bound the same amount of [59Fe, 3H]MECAMas did glucose-sufficient cells, but subsequent transport wasinhibited. Adding glucose immediately restored the ability ofthe cells to transport the intact complex. A 10-fold excess ofunlabeled ferric MECAM added during the transport assayresulted in the release of intact [59Fe, 3H]MECAM from thecell.The uptake of 2 ,uM [59Fe]enterobactin showed biphasic

transport kinetics when monitored for a long period. A rapidrate of uptake over the first 20 min was followed by a muchlower rate of uptake that continued for more than 90 min(Fig. 6). The break in the curve appeared after approxi-mately 400 pmol/mg was accumulated by the cell. The breakdid not occur because of diminishing extracellular concen-trations of substrate. In the experiment shown in Fig. 5,100% uptake corresponded to 1,640 pmollmg. Thus, after thecell accumulated 400 pmol/mg, only 25% of the substratewas consumed, and the external concentration of substratewas 1.5 ,uM. Uptake was increased if, in the middle ofuptake, the ferric enterobactin concentration was increased.

Double-labeling experiments were performed. When a

60

IO

i 0

0

c

Ec0E-

Ea.

50

40

30

20

-5 | II (M 5

S (,a)M

FIG. 3. Substrate concentration (Lineweaver-Burk) plot of con-centration-dependent uptake rates (pH 7.3). Symbols: *,[55Fe]enterobactin; *, [55Fe]MECAM uptake.

VOL. 167, 1986

_ I Zi

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

670 ECKER ET AL.

15-fold excess (30 ,uM) of substrate labeled with 55Fe wasadded, rapid exchange between cellular and extracellularferric enterobactin resulted, with an overall net accumula-tion (Fig. 6). Approximately 60% of the accumulated[59Fe]enterobactin was exchanged with external compoundwhen the experiment was terminated at 40 min after additionof excess substrate. Identical results were obtained when[55Fe]enterobactin uptake was followed with the addition of[59Fe]enterobactin in excess of 50 min (data not shown).These results suggest that a major portion of absorbed ferricenterobactin is not shunted into metabolism immediatelyafter transport. This conclusion is also supported by exper-iments with Mossbauer spectroscopy (20).To examine the possible role of redox processes in metal

ion release from ferric enterobactin, [67Ga]enterobactin up-take was monitored. Over a short time (10 min)[67Ga]enterobactin was taken up by E. coli at about 25% therate of [59Fe]enterobactin uptake (Fig. 7). The initial rate for[67Ga]enterobactin uptake was about 3 pmol/mg per min;cells derived from the same culture took up [59Fe]en-terobactin at approximately 12 pmol/mg per min (data notshown). The initial binding of [67Ga]enterobactin to the cellsin the first 15 s was only 5 pmol/mg, compared with 30pmol/mg for [59Fe]enterobactin. Glucose-starved cells stillbound 5, pmol of [67Ga]enterobactin per mg initially, butsubsequent transport was greatly inhibited compared withthat of glucose-sufficient cells. The addition of glucoseimmediately restored the rate of transport to the glucose-sufficient level. Over a longer time, [67Ga]enterobactin up-take stopped when the cells accumulated approximately 70pmol/mg (Fig. 7). In another set of experiments (data notshown), [67Ga]enterobactin uptake stopped when the cellsaccumulated approximately 100 pmollmg. Adjusting the pH(within the range of 6.0 to 8.0) of the uptake medium showedno effect on the accumulation of [67Ga]enterobactin. Addi-tion of an eightfold excess of unlabeled ferric enterobactin orgallium enterobactin resulted in quantitative release of cel-lular [67Ga]enterobactin. Uptake of 1 ,uM [59Fe]enterobactinwas not inhibited by 30 puM Ga-enterobactin.

0

6,.a

6,

II

InlFIG. 4. Inhibition of

analogs. The [55Fe]enteroQ.5 ,uM complex concent100lo. Compounds added(0); Fe-TRIMCAM (U);

400

3001

200

E

0

E6

ECk.

%ob0aca.

100

0

300

200

100

020

minutesFIG. 5. Transport of [59Fe, 3H]MECAM. Conditions were as

described in the legend to Fig. 2. The final complex concentrationwas 2 ,uM. The fate of 59Fe (0, *) and 3H (0, O) during thetransport process was monitored in the same assay by two-channelanalysis. Symbols: upper panel, (0, *), glucose-supplementedcells; (l, *), glucose-starved cells. Lower panel: (0, 0), glucose-supplemented cells with a 10-fold excess (20 ,tM) of unlabeledFe-MECAM added immediately after the 6-min sample was re-moved (arrow); (l, *) glucose-starved cells with glucose (4 g/liter)added immediately after the 6-min sample was removed.

DISCUSSION

Of the enterobactin analogs studied, only ferric MECAMinhibited uptake of [55Fe]enterobactin (Fig. 4). Kinetic anal-ysis of the concentration-dependent uptake rate of[55Fe]MECAM shows saturation kinetics with a maximumvelocity identical to [55Fe]enterobactin (Fig. 3). The appar-ent Km value for ferric MECAM, 0.69 ,uM, compared with0.2 ,uM for ferric enterobactin, suggests a lower affinity forthe receptor. However, both (Fe.enterobactin)3- and (Fe-

-v_ I MECAM)3- are protonated in a one-proton-per-metal stoi-I 2 3 4 5 chiometry; for (Fe-MECAM)3- KMHL = 7.08, whereas (Fe-

enterobactin)3- has a KMHL of 4.89 (11, 12). Thus, at pH 7.3,hibitor Concentration (,u M) ferric MECAM is a mixture of tris-catecholate and bis-[55Fe]enterobactin uptake by structural catecholate monosalicylate complexes, but ferrictbactin uptake rate (36 pmol/mg per min at enterobactin exists almost exclusively as the tris-catecholateration) without any inhibitor is defined as complex. Furthermore, ferric MECAM is a racemic mixtureI as potential inhibitors were Fe-MECAM of Acis and Acis optical isomers, but ferric enterobactin isand Fe-3,4-LICAMS (A). present only as the Acis isomer. These factors combine to

J. BACTERIOL.

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

FERRIC ENTEROBACTIN TRANSPORT IN E. COLI 671

"10400E

E 300 59FFe0

~'200

55Fe500

00 20 40 60 80

minutesFIG. 6. Exchange of external and cellular ferric enterobactin.

The initial concentration of [59Fe]enterobactin was 2 ,uM. Symbols:[59Fe]enterobactin uptake with no additions; A, (control)[59Fe]enterobactin uptake with the addition (at 51 min, arrow) of thesame substrate at 30 ,uM; *, [59Fe]enterobactin uptake with addition(at 51 min, arrow) of [55Fe]enterobactin at 30 ,uM; A,[55Fe]enterobactin accumulation in the same experiment; O, sum of"Fe (A) and 59Fe (G) in the same experiment.

decrease by 70% the concentration of ferric MECAM that isstructurally identical with ferric enterobactin at the metalcenter. Correction for this gives an apparent Km of 0.21 p.Mand a Vmax of 66 pmol/mg per min, identical to the values forferric enterobactin.The results of these concentration-dependent kinetic and

inhibition studies identify more closely the fepA receptorspecificity. It is the metal-catechol end of the moleculewhich is recognized by the receptor; the ligand backbone isimportant only in that it allows the catechols to coordinatethe iron in a structure similar to ferric enterobactin. Sulfon-ate groups on the catechol ring (as in 3,4-LICAMS) result inloss of recognition, as does removal of the adjacent carbonyl(as in TRIMCAM). Only ferric MECAM (with unsubstitutedcatechol rings and adjacent carbonyl groups) competedeffectively with [55Fe]enterobactin for the receptor and wastransported into the cell with a Vmax identical with that of[55Fe]enterobactin. These results corroborate the observa-tion that ferric MECAM, but not the other analogs, protectsthe cells from killing by colicin B, which enters cells via theferric enterobactin receptor (13).

Iron-starved E. coli cells showed energy-independentbinding of 30 pmol/mg of both [59Fe]enterobactin and[59Fe]MECAM under saturating substrate concentrations. Ifthe observed energy-independent binding event representsreceptor binding in a one-complex-per-receptor stoichiom-etry, 30 pmol/mg of cell dry weight would correspond toabout 7,500 receptors per cell (2.4 x 109 cells per mg [dryweight]), compared with about 200 outer 'membrane recep-

tors per cell reported for vitamin B12 and 105 copies per cellof porin (25, 36). The results in Fig. 6 show cellular accu-mulation of [59Fe]enterobactin to about 450 pmollmg. With acell water volume of 2.4 ,u/mg (dry weight), this correspondsto an average cell concentration of 186 puM (38). The externalconcentration at this point was about 1.5 ,uM (25% of thesubstrate was transported).Two results in this paper strongly suggest that transported

400

300

200

100 -

E

0E%o0

SataOm

0

50

40

30

20

10

0

minutesFIG. 7. Metal ion uptake from [67Ga]enterobactin and

[59Fe]enterobactin. Conditions were as described in the legend toFig. 2. Upper panel: uptake was monitored over a longer time for[59Fe]enterobactin (0) and [67Ga]enterobactin (0). Immediatelyafter the 60-min samples were removed, unlabeled Fe enterobactinwas added to each flask at 20 puM (10-fold excess over the initialconcentration of each labeled substrate). Lower panel: uptake wasmonitored over a shorter time (with a different batch of cells fromthose used in the experiment shown in the upper panel). Symbols forlower panel: [6'Ga]enterobactin uptake in glucose-supplementedcells; O, [67Ga]enterobactin uptake in glucose-starved cells, glucose(4 g/liter) was added immediately after the 8-min sample wasremoved; U, [67Ga]enterobactin uptake with a 10-fold excess ofunlabeled gallium enterobactin added immediately after the 8-mimsample was removed; 0, [67Ga]enterobactin uptake with a 10foldexcess of unlabeled Fe-enterobactin added after the 8-min samplewas removed.

VOL. 167, 1986

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

672 ECKER ET AL.

[59Fe]enterobactin is accumulated in an unmodified form.First, at least 60% of the accumulated [59Fe]enterobactinexchanged with extracellular [55Fe]enterobactin added inlarge excess (Fig. 6). Although there is no proof that the 59Fecounts released from the cell are unmodified[59Felenterobactin, it seems unlikely that iron removed from[59Fe]enterobactin for metabolism would be exchangeablewith extracellular [55Fe]enterobactin. Second, [59Fe,3H]MECAM was taken into the cell intact (counts from themetal and ligand were assimilated at an identical rate [Fig.5]). Addition of a large excess of unlabeled ferric MECAM(Fig. 5) or ferric enterobactin (data not shown) resulted inrelease of a significant fraction of undissociated [59Fe,3H]MECAM (counts from the metal and ligand were re-leased at an identical rate).Two other groups have previously reported experiments

monitoring ferric ["4C]-enterobactin uptake in E. coli. Re-cently, Plaha and Rogers (30) reported no accumnulation offerric ['4C]enterobactin in E. coli and only a trace of accu-mulation in fes mutants. The authors pointed out that thedata for the fes mutant are difficult to explain and suggesteda rapid intracellular reductive removal of the iron and releaseof the ligand to the exterior. We suggest that the problemmay be experimental. The [14C]enterobactin used in thoseexperiments was obtained by incubating cells with[14C]glucose for cellular incorporation and subsequent isola-tion of the natural product with a reported specific activity of1.16 x 106 cpm/Ipmol. This is a very low specific activity. Inthe other report, contained in a review article (37), noaccumulation of ferric [l4C]enterobactin was found in E.coli, but accumulation in a fes strain was observed. Datawere not available on the specific activity of the['4C]enterobactin used by these workers (the data reportedin the review [37] are cited from a paper [8] which does notinclude this information).Our data were obtained with very high specific activities of

[3H]MECAM, counted as a homogeneous sample. Unfortu-nately, enterobactin is too hydrolytically labile to be tritiatedby the same procedure that we used to label MECAM.Although MECAM does not have the same ester-linkedserine backbone as enterobactin, the evidence suggests thatat least some fraction of the iron from ferric MECAM ismetabolized. This is consistent with growth promotion stud-ies with other synthetic analogs which, although not ab-sorbed via the enterobactin receptor, do promote some ironuptake (13, 19).The conclusion that ferric enterobactin is accumulated by

active transport, independent of subsequent metabolism, issupported by the requirement of outer membrane transportprocesses for the cell to carry a normal tonB gene (1). Thisgene, which has recently been sequenced (32), codes for a26,000-dalton protein that is required for all outer mem-brane, receptor-mediated transport processes including fer-ric enterobactin, ferric citrate, ferrichrome, vitamin B12, theenergy-dependent irreversible step of +80 and Tibacteriophage infection, and the action of the B groupcolicins. The common property of all these substrates is thatthey contain a coordinated metal ion or exploit a metal ionreceptor to gain entry into the cell. Of these substrates, theonly example for which active transport across the outermembrane into the periplasm has been demonstrated isvitamin B12. The evidence suggests that vitamin B12 andferric enerobactin are taken up by very similar mechanisms(23).

It is remarkable that [67Ga]enterobactin was a poor trans-port substrate. Gallium enterobactin has been shown to be

isostructural with ferric enterobactin by nuclear magneticresonance spectroscopy (18), and [67Ga]enterobactin mi-grates at a rate identical with [59Fe]enterobactin on high-voltage cellulose electrophoresis at pH 7 (data not shown).There are now several examples where 67Ga complexed tohydroxamate siderophores is transported into the cell at thesame rate as is the iron complex (35). Yet E. coli bound onlyone-sixth as much [67Ga]enterobactin as iron complex, andthe subsequent transport rate was only one-fourth of the rateof [59Fe]enterobactin uptake for the first 10 min. Totaluptake of [67Ga]enterobactin was 70 to 100 pmol/mg, whichis only two to three times the amount of [59Fe]enterobactinthat binds to the cell in the first 15 s, independent of energy.The lack of inhibition of Fe(ent) uptake by Ga(ent) wouldseem to indicate that the gallium complex does not bind tothe receptor. The apparent energy dependence of the proc-ess (Fig. 7) supports this. Plaha and Rogers (30) haverecently reported that [67Ga]enterobactin uptake occurs at3.5% the rate of [59Fe]enterobactin uptake. They also re-ported that 46Sc enterobactin is a better transport substratethan [67Ga]enterobactin, but still is absorbed at only 20% ofthe rate of [59Fe]enterobactin uptake. The differences cannotbe due to the reducibility of the femrc complex, since neitherGa or Sc has a stable +2 oxidation state, and so they wouldbe expected to behave identically.These data and those presented in the following paper (20)

support a mechanism of enterobactin-mediated iron uptakein E. coli in which rapid transport across the outer mem-brane results in an accumulation and intertnediate storage offerric enterobactin before its release of iron to the cell.

ACKNOWLEDGMENTS

We thank Gertraud Muller and Larry Loomis for helpful discus-sions. We gratefully acknowledge the help of B. E. Gordan and thestaff of the Lawrence Berkeley Laboratory, National Institutes ofHealth tritiation facility.

This research was supported by Public Health Service grant AI11744 from the National Institutes of Health, and by an AmericanCancer Society fellowship (to D.J.E.).

LITERATURE CITED1. Braun, V., and K. Hantke. 1981. Bacterial cell surface recep-

tors, p. 1-73. In B. K. Ghosh (ed.), Organization of prokaryoticcell membranes, vol. 2. CRC Press, Boca Raton, Inc., Fla.

2. Bryce, G. F., and N. Brot. 1972. Studies on the enzymaticsynthesis of the cyclic trimer of 2,3-dihydroxy-N-benzoyl-L-serine in Escherichia coli. Biochemistry 11:1708-1715.

3. Bush, E. T. 1964. Liquid scintillation counting of doubly-labeledsamples: choice of counting conditions for best precision intwo-channel counting. Anal. Chem. 36:1082-1089.

4. Bush, E. T. 1968. A double ratio technique as an aid to selectionof sample preparation procedures in liquid scintillation count-ing. Int. J. Appl. Radiat. Isot. 19:447-452.

5. Cooper, S. R., J. V. McArdle, and K. N. Raymond. 1978.Siderophore electrochemistry: relation to intracellular iron re-lease mechanism. Proc. Natl. Acad. Sci. USA 75:3551-3554.

6. Cramer, W. A., J. R. Dankert, and Y. Uratoni. 1983. Themembrane channel-forming antibiotic protein, colicin El.Biochim. Biophys. Acta 737:173-193.

7. Ecker, D. J., C. W. Passavant, and T. Emery. 1982. Role of twosiderophores in Ustilago sphaerogena: regulation of biosynthe-sis and uptake mechanisms. Biochim. Biophys. Acta720:242-249.

8. Frost, G. E., and H. Rosenberg. 1973. The inducible citrate-dependent iron transport system in Escherichia coli K-12.Biochim. Biophys. Acta 330:90-101.

9. Greenwood, K. T., and R. K. J. Luke. 1978. Enzymatic hydro-lysis of enterochelin and its iron complex in Escherichia coli

J. BACTERIOL.

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

FERRIC ENTEROBACTIN TRANSPORT IN E. COLI 673

K-12: properties of enterochelin esterase. Biochim. Biophys.Acta 525:209-218.

10. Hamed, M. Y., R. C. Hider, and J. Silver. 1982. The competi-tion between enterobactin and glutathione for iron. Inorg.Chem. Acta 66:13-18.

11. Harris, W. R., C. J. Carrano, S. R. Cooper, S. R. Sofen, A.Avdeef, J. V. McArdle, and K. N. Raymond. 1979. Coordinationchemistry of microbial iron transport compounds. 19. Stabilityconstants and electrochemical behavior of fer-ric enterobactinand model complexes. J. Am. Chem. Soc. 101:6097-6104.

12. Harris, W. R., C. J. Carrano, and K. N. Raymond. 1979. Spec-trophotometric determination of the proton-dependent stabilityconstant of ferric enterobactin. J. Am. Chem. Soc. 101:2213-2214.

13. Heidinger, S., V. Braun, V. L. Pecoraro, and K. N. Raymond.1983. Iron supply to Escherichia coli by synthetic analogs ofenterochelin. J. Bacteriol. 153:109-115.

14. Hider, R. C., D. Bickar, I. E. G. Morrison, and J. Silver. 1984.Siderophore iron-release mechanisms. J. Am. Chem. Soc.106:6983-6987.

15. Hider, R. C., A. R. Mohd-Nor, J. Silver, I. E. G. Morrison, andL. V. C. Rees. 1981. Model compounds for microbial iron-transport compounds. Part 1. Solution chemistry and Moss-bauer of iron(II) and iron(III) complexes from phenolic andcatecholic systems. J. Chem. Soc. Dalton Trans. 609-622.

16. Hollifield, W. C., and J. B. Neilands. 1978. Ferric enterobactintransport system in Escherichia coli K-12: extraction, assay,and specificity of the outer membrane receptor. Biochemistry17:1922-1928.

17. Lee, C.-W., D. J. Ecker, and K. N. Raymond. 1985. The pH-dependent reduction of ferric enterobactin probed by electro-chemical methods and its implications for microbial iron trans-port. J. Am. Chem. Soc. 107:6920-6923.

18. Llinas, M., D. M. Wilson, and J. B. Neilands. 1973. Effect ofmetal binding on the conformation of enterobactin: a proton andC-13 nuclear magnetic resonance study. Biochemistry12:3836-3842.

19. Lodge, J. S., C. G. Gaines, J. E. L. Arceneaux, and B. R. Byers.1980. Non-hydrolytic release of iron from ferrienterobactinanalogs by extracts ofBacillus subtilis. Biochem. Biophys. Res.Commun. 97:1291-1295.

20. Matzanke, B. F., D. J. Ecker, T.-S. Yang, B. H. Huynh, G. I.Muiiler, and K. N. Raymond. Escherichia coli iron enterobactinuptake monitored by Mossbauer spectroscopy. J. Bacteriol.167:674-680.

21. Neidhardt, F. C., P. L. Bloch, and D. F. Smith. 1974. Culturemedium for enterobacteria. J. Bacteriol. 119:736-747.

22. Neilands, J. B. 1981. Iron absorption and transport in microor-ganisms. Annu. Rev. Nutr. 1:27-46.

23. Neilands, J. B. 1982. Microbial envelope proteins related toiron. Annu. Rev. Microbial. 36:285-309.

24. Neilands, J. B., T. J. Erickson, and W. H. Rastetter. 1981.Stereospecificity of the ferric-enterobactin receptor of Esche-richia coli K-12. J. Biol. Chem. 256:3831-3832.

25. Nikaido, H., and T. Nakae. 1979. The outer membrane ofgram-negative bacteria. Adv. Microb. Physiol. 20:163-250.

26. O'Brien, G., G. B. Cox, and F. Gibson. 1971. Biologically activecompounds containing 2,3-dihydroxybenzoic acid and serineformed by Escherichia coli. Biochim. Biophys. Acta 237:537-549.

27. O'Brien, I. G., and F. Gibson. 1970. The structure ofenterochelin and related 2,3-dihydroxy-N-benzoylserine conju-gates from Escherichia coli. Biochim. Biophys. Acta215:393-402.

28. Pecoraro, V. L., W. R. Harris, G. B. Wong, C. J. Carrano, andK. N. Raymond. 1983. Coordination chemistry of microbial irontransport compounds. 23. Fourier transform infrared spectros-copy of ferric catechoylamide analogues of enterobactin. J. Am.Chem. Soc. 105:4623-4633.

29. Pecoraro, V. L., G. B. Wong, T. A. Kent, and K. N. Raymond.1983. Coordination chemistry of microbial iron transport com-pounds. 22. pH-Dependent Mossbauer spectroscopy of ferricenterobactin and synthetic analogues. J. Am. Chem. Soc.105:4617-4623.

30. Plaha, D. S., and H. J. Rogers. 1983. Antibacterial effect of thescandium complex of enterochelin: studies of the mechanism ofaction. Biochim. Biophys. Acta 760:246-255.

31. Poilack, J. R., and J. B. Neilands. 1970. Enterobactin, an irontransport compound from Salmonella typhimurium. Biochem.Biophys. Res. Commun. 38:989-992.

32. Postle, K., and R. F. Good. 1983. DNA sequence of the Esche-richia coli tonB gene. Proc. Natl. Acad. Sci. USA 80:5235-5239.

33. Pugsley, A. P., and P. Reeves. 1977. Uptake of ferrienterochelinby Escherichia coli: energy-dependent stage of uptake. J. Bac-teriol. 130:26-36.

34. Raymond, K. N., G. Muller, and B. F. Matzanke. 1984.Complexation of iron by siderophores. A review of their solu-tion and structural chemistry and biological function. Top. Curr.Chem. 123:49-102.

35. Raymond, K. N., and T. P. Tufano. 1982. Coordination chem-istry of the siderophores and recent studies of synthetic analogs,p. 85-105. In H. B. Dunford, D. Dolphin, K. N. Raymond, andL. Sieker (ed.), The biological chemistry of iron. D. ReidelPublishing Co., Dordrecht, Holland.

36. Reynolds, P. R., G. P. Mottur, and C. Bradbeer. 1980. Trans-port of vitamin B12 in Escherichia coli. J. Biol. Chem.255:4313-4319.

37. Rosenberg, H., and I. Y. Young. 1974. Iron transport in theenteric bacteria, p. 67-82. In J. B. Neilands (ed.), Microbial ironmetabolism. Academic Press, Inc., New York.

38. Stock, J. B., B. Rauch, and S. Roseman. 1977. Periplasmic spacein Salmonella typhimurium and Escherichia coli. J. Biol. Chem.252:7850-7861.

39. Tufano, T. P., and K. N. Raymond. 1981. Coordination chem-istry of microbial iron transport compounds. 21. Kinetics andmechanisms of iron exchange in hydroxamate siderophore com-plexes. J. Am. Chem. Soc. 103:6617-6624.

40. Venuti, M. C., W. H. Rastetter, and J. B. Neilands. 1979. 1,3,5-Tris(N,N',N"-2,3-dihydroxybenzoyl)-aminomethylbenzene, asynthetic iron chelator related to enterobactin. J. Med. Chem.22:123-124.

41. Weitl, F. L., W. R. Harris, and K. N. Raymond. 1979. Sulfon-ated catecholamide analogues of enterobactin as iron sequester-ing agents. J. Med. Chem. 22:1281-1283.

42. Weiti, F. L., and K. N. Raymond. 1979. Ferric ion sequesteringagents. 1. Hexadentate o-binding N,N',N"'-tris(2,3-dihydroxy-benzoyl) derivatives of 1,5,9-triazacyclotridecane and 1,3,5-triaminomethylbenzene. J. Am. Chem. Soc. 101:2728-2731.

43. Young, I. G., and F. Gibson. 1979. Isolation of enterochelinfrom Escherichia coli. Methods Enzymol. 56:394-398.

VOL. 167, 1986

on February 7, 2020 by guest

http://jb.asm.org/

Dow

nloaded from