Conserved Residues Ser16 and His20 and Their Relative Positioning Are Essential for TonB Activity, Cross-linking of TonB with ExbB, and the Ability of TonB to Respond to Proton Motive Force*

The cytoplasmic membrane protein TonB couples the proton electrochemical potential of the cytoplasmic membrane to transport events at the outer membrane of Gram-negative bacteria. The amino-terminal signal anchor of TonB and its interaction with the cytoplasmic membrane protein ExbB are essential to this process. The TonB signal anchor is predicted to form an α-helix, with a conserved face comprised of residues Ser16, His20, Leu27, and Ser31. Deletion of either Ser16 or His20 or of individual intervening but not flanking residues rendered TonB inactive and unable to assume a proton motive force-dependent conformation. In vivo formaldehyde cross-linking experiments revealed that the ability of this subset of mutants to form a characteristic heterodimer with ExbB was greatly diminished. Replacement of residues 17–19 by three consecutive alanines produced a wild type TonB allele, indicating that the intervening residues (Val, Cys, and Ile) contributed only to spacing. These data indicated that the spatial relationship of Ser16 to His20 was essential to function and suggested that the motif HXXXS defines the minimal requirement for the coupling of TonB to the cytoplasmic membrane electrochemical gradient. Deletion of Trp11 resulted in a TonB that remained active yet was unable to cross-link with ExbB. Because Trp11 was demonstrably not involved in the actual cross-linking, these results suggest that the TonB/ExbB interaction detected by cross-linking occurred at a step in the energy transduction cycle distinct from the coupling of TonB to the electrochemical gradient.

The outer membrane of Gram-negative bacteria serves as a diffusion barrier, that, by virtue of the lipopolysaccharide in the outer leaflet, presents a polar, negatively charged surface to the external surroundings, hindering the passage of detergents, hydrophobic antibiotics, and other toxic agents soluble in standard phospholipid bilayers. This barrier contains aqueous channels that are formed by general and substrate-specific porin proteins that enable diffusion of small, hydrophilic nutrients (Ͻ600 Da in Escherichia coli), across the outer membrane.
Simple diffusion is not sufficient for the acquisition of either the nutrient cobalamin or Fe(III)-siderophore complexes. Instead, these nutrients are efficiently harvested at the outer membrane surface by high affinity receptors, which catalyze the active transport of bound ligand into the periplasmic space (1)(2)(3). The release of cobalamin and Fe(III)-siderophores into the periplasmic space is energy-dependent, requiring an intact proton gradient at the cytoplasmic membrane (4 -6). The separation of transport events from their energy source suggests the need for an energy transducer; this need is met by the protein TonB (5,(7)(8)(9). TonB is anchored to the cytoplasmic membrane presumably by a single transmembrane domain (amino acids , derived from an uncleaved signal sequence (10), with the bulk of the protein occupying the periplasmic space (11,12). TonB spans the periplasm and directly contacts outer membrane active transport proteins (13,14). Roughly one-third to one-half of the total cellular TonB partitions exclusively with the outer membrane upon fractionation, suggesting that TonB functions as an intermembrane energy shuttle (15). The outer membrane components and specific features of TonB that support shuttling to the outer membrane are only defined in part, whereas the requisites for TonB interaction with the cytoplasmic membrane are clear: an intact, wild type signal anchor and the presence of two proteins, ExbB and ExbD (15,16). Both of these integral cytoplasmic membrane proteins physically interact with TonB, as evident by in vivo cross-linking (17), which, in the case of ExbB, appears to involve the TonB signal anchor (18).
The TonB residues Ser 16 and His 20 occur on a conserved face of the predicted ␣-helical TonB transmembrane domain (19). We previously noted that the deletion of a residue (Val 17 ) occurring between the two conserved residues rendered TonB inactive (18), suggesting either that residue Val 17 made a side chain-specific contribution to energy transduction or conversely that the spatial relationship between Ser 16 and His 20 was essential to function. To resolve this dichotomy, we constructed a series of single residue deletions that spanned the amino-terminal two-thirds of the TonB transmembrane domain. Characterization of the resultant deletion mutants and of several substitution mutants suggested by our initial findings indicated that the positional relationship of Ser 16 to His 20 was an essential feature for the apparent coupling of TonB to the proton motive force (pmf). 1

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-All bacteria used in this study were derived from the E. coli K-12 strain W3110 (20). The majority of experiments used the ampicillin-resistant tonB deletion strain KP1344 and its exbB::Tn10 derivative, KP1347 (16). Transport assays used an aroB derivative of KP1344, KP1351 (16). When necessitated by the use of ampicillin-selected plasmids, experiments were performed with the ampicillin-sensitive tonB deletion strains KP1229 (21) and KP1304 (con-* This work was supported by Grant GM42146 from the National Institutes of General Medical Sciences (to K. P.). 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 18 U.S.C. Section 1734 solely to indicate this fact.
A pACYC184 derivative (pKP325) carrying araC and the tonB gene under control of the arabinose promoter and an isogenic construct encoding TonB ⌬V17 (pKP362) have been described (24). In this study, a compatible pBR322 derivative carrying a similarly regulated exb operon was constructed by ligation of a 5Ј-phosphorylated polymerase chain reaction amplimer (generated from a pKP298 (13) template and corresponding to residues 585-1748 of the exb operon (25)) into a pBAD18 vector (26) that had been restricted at the unique SmaI site and dephosphorylated. Constructs were mapped for orientation, and inserts were sequenced in entirety to verify identity. The resultant construct (pKP390) used the vector-provided start site to encode an ExbB with an amino terminus containing a valyl and a prolyl residue between the methionine initiation and what would normally be the second residue of ExbB. The second gene in the operon, encoding ExbD, was unaltered by the cloning strategy, retaining its native initiation and termination codons.
Site-directed mutagenesis was performed in pKP325 and pKP390 by a method adapted from that described by Michael (27), wherein a third, 5Ј-phosphorylated oligonucleotide primer bearing the desired base changes is introduced into a modified polymerase chain reaction reaction that includes a thermostabile ligase, such that the mutagenic primer is incorporated in the extending strand. In this study, the primers used for amplification flanked two BamHI sites, thus the resultant amplimers could be restricted with BamHI and then reintroduced to the parent vectors cut at these same sites. Because the mutagenic primers were designed to also introduce silent mutations that altered characteristic restriction patterns, products bearing desired mutations could be detected by restriction mapping. Candidates thus identified were further mapped to determine insert orientation, with the inserts ultimately verified by sequencing. Plasmids generated in this fashion are described in Table I.
Media and Culture Conditions-Tryptone plates and top agar, and LB broth and agar were made as described (22). Liquid cultures were grown with aeration at 37°C in M9 minimal salts; supplemented with 0.4% w/v glycerol, 0.2% vitamin-free casamino acids, 40 g⅐ml Ϫ1 tryptophan, 0.4 g⅐ml Ϫ1 thiamin, 1.0 mM MgSO 4 , 0.5 mM CaCl 2 , 1.85 M iron (provided as FeCl 3 ⅐6H 2 O), and L-arabinose as indicated below. For transport assays, vitamin-free casamino acids were replaced by a defined amino acid mix (40 g⅐ml Ϫ1 each, except aspartic acid and glutamic acid (30 g⅐ml Ϫ1 each), and tyrosine (10 g⅐ml Ϫ1 )) Antibiotics were used where required at a concentration of 100 g⅐ml Ϫ1 for ampicillin and 34 g⅐ml Ϫ1 for chloramphenicol. In initial cross-linking and proteinase K accessibility experiments, all strains were grown in supplemented M9 containing 67 M L-arabinose to induce tonB expression, which resulted in steady state TonB levels near that of chromosomal controls (i.e. 6 -8-fold greater than the noninduced) for all constructs except those encoding deletions at positions 14/15-20, where levels were greater than noninduced but less than the chromosomal control (data not shown). In subsequent experiments ⌬16 -20 were induced with 130 M L-arabinose, and ⌬14/15 with 260 M L-arabinose to attain steady state levels equivalent to or slightly greater than that of chromosomally encoded TonB.
Chemicals and Reagents-␣-TonB mAb were prepared as previously described (29). The ␣-FhuA mAbs 6.9 and 6.14 were the generous gift of James Coulton. Media components were purchased from Difco Laboratories (Detroit, MI). L-Arabinose was purchased from Pfanstiehl Laboratories, Inc. (Waukegan, IL). Monomeric formaldehyde was purchased from Electron Microscopy Sciences (Fort Washington, PA). ECL immunoblot kits and [ 55 Fe], provided as iron chloride, were purchased from PerkinElmer Life Sciences. Horseradish peroxidase-conjugated goat-␣mouse IgG was purchased from Caltag Laboratories (Burlingame, CA). Polyvinylidiene fluoride membranes (Immobilon-P) were purchased from Millipore Corp. (Bedford, MA). Oligonucleotides were synthesized by Life Technologies, Inc. Thermolabile alkaline phosphatase was purchased from Epicentre Technologies (Madison, WI). All other DNAmodifying enzymes were purchased from either New England Biolabs (Beverly, MA) or Life Technologies, Inc. Proteinase K, lysozyme, and phenymethylsulfonyl fluoride were purchased from Roche Molecular Biochemicals. All other reagents were purchased from Sigma.
Spot Titer Assay-Cells were grown in supplemented M9 to an A 550 of 0.4 (as determined with a Spectronic 20 spectrophotometer; path length, 1.5 cm), harvested in 100-l aliquots, suspended in 2 ml of molten top agar (60°C) containing 1.30 mM L-arabinose and antibiotics as appropriate, and overlaid onto room temperature T plates. Serial 10-fold dilutions of bacteriophage 80 vir (initial concentration, 5.5 ϫ 10 10 plaque-forming units⅐ml Ϫ1 ) were applied to the plate surface in 5-l aliquots and then incubated for 16 h at 37°C. Results were recorded as the reciprocal of the highest dilution at which clearing of the lawn was evident.
Proteinase K Accessibility Assays-Cells were grown in supplemented M9 (containing chloramphenicol and arabinose as indicated) to an A 550 of 0.4 and spheroplasts prepared as previously described (18). The effects of the protonophore CCCP was examined as previously described (16). Briefly, CCCP (or an equal volume of ethanol carrier) was added to spheroplast suspensions to a final concentration of 50 M, with samples then incubated with or without proteinase K (25 g⅐ml Ϫ1 ) for 15 min at 4°C, treated for 2 min with 1.0 mM phenymethylsulfonyl fluoride to inactivate the proteinase, and then precipitated with trichloroacetic acid and processed as described (16).
In Vivo Chemical Cross-linking-Cells were grown in supplemented M9 (containing antibiotics and arabinose as indicated) to an A 550 of 0.4, harvested in 1.0-ml aliquots, centrifuged, and suspended in 938 l of room temperature 100 mM phosphate buffer (pH 6.8), followed by the addition of 62 l of fresh 16% formaldehyde monomer. Suspensions were incubated for 15 min at room temperature, centrifuged, suspended in 50 l of Laemmli sample buffer, incubated for 5 min at 60°C, and then stored at Ϫ20°C until analysis. Electrophoretic Analysis of TonB Protein-Samples prepared as above were subjected to electrophoresis on SDS 11% polyacrylamide gels, and the resolved proteins were electrotransferred to polyvinylidiene fluoride membranes. Immunoblot analyses were subsequently performed using TonB-specific mAbs and ECL, as previously described (30).
Transport of [ 55 Fe]Ferrichrome-Ferrichrome transport assays were adapted from a study by Köster and Braun (31). The strain KP1351 (⌬tonB, aroB), carrying either a control plasmid (pACYC184), or plasmids encoding either wild type TonB (pKP325) or TonB VCI17-19AAA (pKP441) was grown to an A 550 of 0.5 in supplemented M9 salts with chloramphenicol, defined amino acids, and 67 M L-arabinose. Cells were centrifuged and then suspended to 2 ϫ 10 8 colony-forming units ⅐ml Ϫ1 in M9 salts containing 0.1 mM nitrilotriacetate, followed by incubation with shaking for 5 min. at 30°C. Samples of 0.5 ml were removed and precipitated with trichloroacetic acid preparative to electrophoresis. Transport was initiated by the addition of 150 pmol of [ 55 Fe]ferrichrome (generated by preincubation of deferriferrichrome with [ 55 Fe]Cl 3 at a 6.7:1 molar ratio in 10 mM HCl at 37°C for 15 min) to 2 ϫ 10 8 colony-forming units, with incubation continued by shaking at 30°C. Samples of 0.5 ml were harvested at the indicated time points by filtration onto glass filters, which were then washed three times with 5 ml of 0.1 M LiCl and dried. Incorporated [ 55 Fe] was determined by liquid scintillation counting.  Result in the Loss of TonB Activity-Certain bacteriophage and colicins have evolved to exploit the TonB-dependent transport system to gain entry into E. coli (32,33), and their toxicity provides indirect assays for TonB function. In this study, the activity of TonB proteins carrying single residue deletions in the transmembrane domain was assayed by spot titer against the TonB-dependent bacteriophage 80. Every attempt was made to express the plasmid-encoded proteins at chromosomal levels. TonB-encoding plasmids and controls were carried in the tonB deletion strain KP1344, with TonB expression induced by the presence of 1.3 mM L-arabinose in the top agar. All assays were performed in triplicate. Because this is a solid phase assay, direct determination of expression levels is problematic; however, this concentration of L-arabinose resulted in a level of activity for wild type plasmid-encoded TonB (pKP325) that was indistinguishable from that obtained with the chromosomally encoded TonB of W3110 (in each case clearing/plaque formation was evident at the 10 Ϫ8 dilution; data not shown). In the absence of induction, clearing/plaque formation with plasmidencoded wild type TonB was reduced 10-fold (data not shown). In a prior study (16), induction with higher levels of L-arabinose resulted in a similar reduction of activity for wild type TonB, presumably a dominant negative gene dosage effect. Thus, although we do not presume that the levels of TonB achieved in this assay precisely match that encoded by the chromosome, any activity differences caused by variation in expression level were less than the resolving power of this assay, with one possible exception as described below. This noted, the effects of single residue deletion, when present, were unambiguous (Fig.  1). Deletion of single residues from both positions 10 -14/15 and positions 21-23/24 of TonB had no detectable impact on sensitivity to 80. Conversely, single residue deletions in the region of positions 16 -20 completely eliminated TonB activity, with the exception of ⌬V17, where slight clearing with undiluted phage (ϳ2.75 ϫ 10 8 plaque-forming units) was evident in two of three cases (Fig. 1). We have previously found chromosomally encoded TonB ⌬V17 to be completely inactive in this assay, with traces of activity evident upon overexpression (16). Thus it is likely that in the present assay TonB ⌬V17 was somewhat over-expressed relative to chromosomally encoded TonB.

Single Residue Deletions at
Single Residue Deletions at  Result in the Loss of TonB pmf Responsiveness-We had previously demonstrated that the conformation of TonB varies with the presence or absence of pmf (16). Comparison of spheroplasts in the presence and absence of a protonophore (CCCP or dinitrophenol) revealed that upon collapse of the cytoplasmic membrane proton gradient, wild type TonB acquires a distinct conformation wherein the amino-terminal 152-156 residues exhibit an enhanced resistance to digestion by proteinase K. This ability to respond to pmf changes is dependent upon the presence of ExbB/D and is lost upon mutation of either TonB residues Ser 16 or His 20 or the deletion of residue Val 17 (16). Here, identical assays of pmf responsiveness were performed on the present deletion series (Fig. 2). The results obtained paralleled those of the 80 spot titer assays (Fig. 1), with proteins bearing single residue deletions involving positions 16 -20 rendered unresponsive to pmf, whereas proteins with deletions on either flank of this region (residues 10 -14/15 and 21-23/24) retained pmf responsiveness (Fig. 2). In addition, each TonB derivative examined was susceptible to proteinase K in spheroplasts, indicating that negative phenotypes did not merely reflect improper secretion.
Some Single Residue Deletions Alter the in Vivo Chemical Cross-linking Profile of TonB-Wild type TonB can be crosslinked in vivo with formaldehyde to form a set of characteristic higher molecular mass complexes that includes a TonB⅐FepA complex that migrates with an apparent mass of 195 kDa, a cluster of uncharacterized complexes that migrate at 77 kDa, an additional unidentified complex that migrates at 43.5 kDa (13), and a TonB⅐ExbB heterodimer with an apparent molecular mass of 59 kDa (13,17). Initial studies with TonB ⌬V17 indicated that the mutant could not form a detectable crosslinked complex with ExbB upon formaldehyde treatment yet retained the ability to form the other characteristic complexes (18). These experiments were extended here to include the entire set of single residue deletions (Fig. 3). Several minor deviations in the cross-linking profile were evident, including variations in the signal intensity of the several bands comprising the 77-kDa cluster of the ⌬14/15-⌬20 mutants, increased signal intensity of the 43.5-kDa complex for the ⌬13-⌬20 mutants, and the appearance of a novel 66-kDa complex (indicated by the arrow in the right panel) for the ⌬17-⌬19 mutants. Because the non-TonB components of these complexes are not as yet determined, the implications of these differences were unclear. More readily interpretable were the effects of the mutations on the ability of TonB to form the 59-kDa heterodimer with ExbB. These results paralleled the results of the activity and pmf responsiveness assays, with the ability of the ⌬16-⌬20 mutants to form the characteristic cross-linked complex with ExbB greatly diminished or absent. This phenotype All strains were assayed in triplicate, with identical values obtained within each set, except for TonB ⌬V17 , where two plates indicated sensitivity to undiluted phage and one plate was scored as fully resistant.
was also evident with the ⌬14/15 mutant and, surprisingly, with the ⌬11 mutant. Preliminary studies 2 suggest the inability to detect cross-linked complexes of the ⌬14/15 mutant with ExbB may be related to its apparent instability (see "Experimental Procedures"). Because tryptophan 11 is theoretically formaldehyde-reactive, the inability to detect cross-linked complexes of the ⌬11 mutant with ExbB suggested the possibility that the actual cross-linking occurred through this residue. To test this possibility, two additional mutations were generated. First, the Trp 11 residue of TonB was conservatively replaced by phenylalanine, a residue nonreactive with formaldehyde. Second, Trp 38 of ExbB was replaced with a phenylalanine, because topological comparisons suggested that this was the nearest formaldehyde-reactive residue with which TonB Trp 11 could interact. The TonB activity of strains carrying one or both of these alleles, as determined by spot titer assays, was indistinguishable from wild type (data not shown). Alleles were paired and evaluated by in vivo formaldehyde cross-linking (Fig. 4). Unexpectedly, replacement of TonB Trp 11 with a nonreactive residue did not result in the loss of formaldehyde-mediated cross-linking of TonB to ExbB. Similarly, ExbB W38F did not lose the ability to cross-link to TonB; however, the relative amount of complex formed appeared to be diminished with both wild type TonB and TonB W11F .
Alanine Replacement of  Does Not Alter the TonB Phenotype-The deletion of either residue 16 or 20 or of the individual amino acids that reside between these positions resulted in essentially the same phenotype; inability to support 80 infection (Fig. 1), inability to assume a pmf-dependent conformation (Fig. 2), and greatly decreased ability to form formaldehyde-mediated cross-links with ExbB in vivo (Fig. 3). Although these results appeared to exclude the possibility that residues flanking residues 16 -20 made direct contributions to energy transduction, they did not distinguish whether any of the residues between positions 16 and 20 were of specific consequence or merely functioned as spacers to provide the proper degree of helix twist between Ser 16 and His 20 . To address this issue, we replaced these intervening residues (Val 17 , Cys 18 , and Ile 19 ) with three alanines. Initial characterization of the resultant TonB VCI17-19AAA found it to be indistinguishable from wild type TonB on the basis of 80 spot titer-determined activity, ability to assume a pmf-dependent conformation, and in vivo formaldehyde cross-linking profile (data not shown). To more rigorously address function, transport assays using 55 Fe-loaded ferrichrome were performed. The ability of TonB  to transport radiolabeled ferrichrome did not significantly differ from that of wild type TonB (Fig. 5). Transport was not evident for cells carrying a control plasmid (pACYC184) alone or when pmf was collapsed in cells encoding wild type TonB by the addition of CCCP, with counts remaining consistently below 1,000 cpm/1 ϫ 10 8 cells throughout the experiment (data not shown). Immunoblot analysis of TonB and the ferrichrome receptor FhuA levels in the assayed cultures indicated that the levels of both were essentially identical in cells expressing either wild type TonB or TonB VCI17-19AAA (data not shown).

DISCUSSION
TonB functions as an energy transducer, apparently harvesting the pmf of the cytoplasmic membrane to drive active transport at the outer membrane of Gram-negative bacteria (2, 34 -36). TonB-dependent processes can be blocked by protonophores (4 -6) and, in unc strains (which lack membrane-bound ATP synthase), by cyanide (6). Together these results suggest pmf is necessary, but whether it is the direct, or even the sole energy source, is unresolved. Regardless of the energy source, recent observations suggest a dynamic model of energy transduction, wherein TonB transits through a set of conformations that alternately store and release potential energy, with transition to a higher energy isomer coupled to the energy source and transition to a lower energy isomer coincident to energy release to a ligand-occupied outer membrane receptor (16). The means by which TonB is coupled to the pmf of the cytoplasmic membrane is unknown, but it is clear that the transmembrane signal anchor of TonB (residues 12-32) and its ability to interact with ExbB are required. Beyond targeting and tethering TonB to the energy source (10), the TonB signal anchor is essential for both activity (37,38) and formation of the formaldehyde cross-linked TonB⅐ExbB heterodimer (38). More re-2 R. A. Larsen and K. Postle, unpublished observations.

FIG. 2. Identification of a pmf-responsive TonB conformation in spheroplasts.
Immunoblots are shown of spheroplasts (Sph) and CCCP-treated spheroplasts (Sph/CCCP), generated from cells in which tonB genes were induced with L-arabinose to levels approximating that of chromosomally encoded TonB. Samples were resolved on SDS-11% polyacrylamide gels, transferred to polyvinylidiene fluoride membranes, and probed with 1:5,000 mAb 4H4 (specific for TonB residues 79 -84; Ref. 29). The positions and apparent molecular masses of intact TonB and the proteinase K-resistant product are indicated on the left. The presence of full-length TonB in the proteinase-treated lanes is most likely due to incomplete spheroplast formation. Panels representing ⌬S16 through ⌬H20 were overexposed relative to the other panels to more stringently establish absence of the pmf-dependent band. The relative amount of proteinase K-resistant product varies among the positively scored strains. Whether or not this variation correlates to subtle differences in TonB activity remains to be established.
cently, we noted that TonB could assume a pmf-sensitive conformation. This required both a competent signal anchor and ExbB/D, with mutants such as TonB S16L and TonB H20Y unable to achieve the conformation unless an ExbB suppressor allele was present (16).
Further evidence for the importance of the TonB signal anchor stems from the ability of TolQ and TolR to substitute in part for ExbB and ExbD (39 -41). These respective protein sets are similar in both primary amino acid sequence (25,42) and topology (43)(44)(45)(46), with the greatest identity evident in their transmembrane domains. Although TonB and TolA also have a common topology (12,24), the similarity does not extend to the primary amino acid sequences, which are highly dissimilar (28,47), except for their signal anchors, which share a conserved set of residues comprising one face of the putative transmembrane ␣-helix (19).
The conserved face common to the signal anchors of TonB and TolA consists of the E. coli TonB residues Ser 16 , His 20 , Leu 27 , and Ser 31 (depicted in red, Fig. 6). Of these residues, mutation of Ser 31 did not appear to overtly affect TonB activity (49). In our hands, mutations involving either Leu 27 or Ser 31 have not been recovered in extensive mutant hunts using both high and low stringency selections for TonB activity (16,18,50). Similar mutations of the corresponding residues in TolA do not disrupt function (51). Thus, residues Leu 27 and Ser 31 do not appear to be essential for energy transduction. Conversely, mutations involving either Ser 16 or His 20 have a profound impact on TonB activity. When expressed at wild type levels, TonB S16L , TonB H20Y (16), and TonB H20A 2 lack demonstrable activity, whereas for TonB H20R (assayed under conditions of overexpression) activity was greatly diminished (49). Similarly, substitutions at the corresponding S and H residues of TolA render it inactive (51).
We previously isolated a tonB mutation where loss of function resulted from the deletion of a single residue (Val 17 ) in the region separating Ser 16 and His 20 (18). The phenotype of this mutant was similar to that of the later isolated Ser 16 and His 20 mutations, to the extent that ExbB suppressors isolated for Ton ⌬V17 and TonB S16L were essentially interchangeable (16,18). When considered together, these and the above-cited results led us to consider whether or not transmembrane domain residues exclusive of positions 16 -20 were of direct consequence to the coupling of TonB to the cytoplasmic membrane proton gradient. That hypothesis was examined using a set of single residue deletions, the phenotypes of which are summarized in Fig. 6.
Not surprisingly, deletion of either Ser 16 or His 20 resulted in a negative phenotype by all measures, similar to the results obtained with point mutations involving these residues (16). Deletion of any single intervening residue also produced a negative phenotype, consistent with the previous ⌬V17 results (18). Deletion of single residues to the carboxyl-terminal flank of His 20 generated a wild type TonB, indicating that amino acids 21-24 have no role in energy transduction and suggesting that the position of Ser 16 and His 20 relative to the other conserved residues (Leu 27 and Ser 31 ) was also not essential to function. Deletion of single residues to the amino-terminal flank of Ser 16 also generated essentially wild type TonB (except for formaldehyde-mediated cross-link formation with ExbB by TonB ⌬L14/15 (which may be unstable) and TonB ⌬W11 , discussed below), indicating that amino acids 10 -15 have no role in energy transduction and suggesting that the position of Ser 16 and His 20 relative to the TonB cytoplasmic domain and the first several residues of the signal anchor was also not essential to function.
These results suggested two alternatives, either the position of Ser 16 and His 20 relative to each other was sufficient for effective coupling of TonB to the energy gradient or the side chains of one or more intervening residues contributed to the energization mechanism. This was resolved by the substitution of residues 17-19 by three consecutive alanines. The resultant TonB VCI17-19AAA was wild type by all measures, indicating that the intervening residues did not make side chain-specific contributions essential for the energization of TonB. Beyond supporting the hypothesis that the positional relationship of Ser 16 to His 20 is necessary to function, these results suggest that the motif SXXXH (where residues S and H occupy a common face of the ␣-helix) may define the minimal structural requirement for the coupling of TonB to the cytoplasmic membrane proton electrochemical gradient. It should be noted that the initial suspicion that Ser 16 and His 20 were of consequence came from their conservation in TolA and the suggestion that these residues would occupy the same face of an ␣-helix (19). In the absence of a solved crystal structure, the present data represent the strongest advocacy to date for such a structure.
The inability of TonB ⌬W11 to form formaldehyde-mediated cross-links with ExbB, while otherwise remaining ostensibly wild type, is intriguing. Our initial suspicion that the explana-tion for this observation would prove to be trivial (i.e. the deleted tryptophan was the site of cross-link formation) was not borne out. Thus, we are left with the more exciting interpretation that this lack of detectable cross-linking could be indicative of an altered TonB conformation wherein the actual cross-linking site on TonB is no longer proximal to a formaldehyde-reactive site on ExbB. This suggests that the TonB/ExbB interaction detected by cross-linking is not essential to energy transduction. One possibility is that cross-linking detects an interaction that defines the recycling process by which spent TonB is shuttled back to a form competent to store energy. Thus, the reduction or absence of cross-linking to ExbB by TonB mutants unable to support energy transduction might reflect the fact that they are blocked in a portion of the cycle that occurs prior to the stage marked by cross-linking to ExbB.
Studies concerning the contribution of the TonB signal anchor to the recycling phase of the shuttling cycle are ongoing. The present study has addressed the role of the TonB signal anchor in the energization step of the cycle and has defined what appears to be the minimal structural requirement of TonB for efficient coupling to the cytoplasmic membrane proton gradient. It is evident that this process is facilitated by ExbB and ExbD; however, the means by which this occurs remain obscure. ExbB and ExbD can each form, at minimum, homotrimers (17). Because ExbB and ExbD can interact in vitro (52), they may occur as a heterohexamer, structurally consistent with the idea that they could be a proton translocator. They are also required for TonB to respond to the presence or absence of pmf (16). Understanding ExbB and ExbD will be essential for elucidating the mechanism by which TonB be-  (48) and topology (12) to represent the transmembrane domain. The orientation of the structure relative to the cytoplasm and the periplasmic space is indicated. Residues comprising the conserved face are indicated in red and identified by label. The data presented in Figs. 2-4 are summarized at the right of the figure at individual positions corresponding to the position of the involved residue depicted in the modeled structure. Activity is scored as the highest log dilution of 80 vir at which clearing or plaque formation was evident, Ϫ indicates no clearing with undiluted 80 vir . For pmf sensitivity, the ability to form the pmf-dependent proteinase K-resistant product is indicated by ϩ, whereas the apparent absence of such a product is indicated by Ϫ. For in vivo cross-linking to ExbB, the ability to form an identifiable TonB/ExbB heterodimer is indicated by ϩ, whereas the apparent inability to efficiently form such a product is indicated by (Ϫ). comes energized. A likely first step would involve determining whether or not ExbB (or ExbD) directly interacts with the TonB SXXXH motif and, if so, which features of ExbB participate in such an interaction.