INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Many species of motile bacteria are propelled by flagella, each consisting of a helical filament that functions as propeller, a rotary motor in the cell membrane, and a flexible coupling that joins the two (for reviews, see references 2, 22, and 28). The source of energy for motor rotation is the membrane gradient of protons or, in some species, sodium ions (14, 16, 19, 23). The flagellar motor thus converts chemical energy into work. Extensive physiological studies of the flagellar motors of Escherichia coli and of a motile Streptococcus species suggest that the efficiency of this energy conversion is very high, probably approaching 100% when the motor is rotating slowly under high load (17, 24-26). The molecular mechanism of torque generation by the flagellar motor is not understood.

MotA and MotB are integral membrane proteins essential for flagellar rotation (5, 8, 9, 32, 39). They are not needed for flagellar assembly, because null mutants of motA and motB produce flagella that, although unable to rotate, appear outwardly normal (11, 15, 22, 31). MotA and MotB bind to each other (36) and function together to conduct protons across the cytoplasmic membrane (4, 5, 12, 13, 33, 38). MotB has a large domain located in the periplasm (8), which includes a sequence motif found in proteins that bind peptidoglycan (10). MotB may therefore serve to anchor the MotA-MotB complexes to the cell body, which would make these complexes the nonrotating part (the stator) of the motor (8, 10). MotB mutants have phenotypes consistent with such an anchoring role (6). Each flagellar motor contains several MotA-MotB complexes that are arranged in a ring at the base of the flagellum (18) and function independently to generate torque (3, 7).

The membrane topologies of MotA and MotB have been deduced by both sequence analysis (9, 32) and experiments (5, 8, 39) and are shown in Fig. 1. MotA has four membrane segments, two small segments in the periplasm, and two sizable segments in the cytoplasm. The conserved proline residues Pro 173 and Pro 222, subjects of this study, are located at the cytoplasmic ends of membrane segments 3 and 4. Two charged residues in the cytoplasmic domain of MotA, Arg 90 and Glu 98, are also important for motor rotation and have been shown to interact with charged residues of the rotor protein FliG (20, 40, 41). MotB has only a single membrane segment, which connects a small N-terminal domain in the cytoplasm to a large C-terminal domain in the periplasm. The conserved residue Asp 32 of MotB is located near the cytoplasmic end of the membrane segment (32). Mutational studies suggest that Asp 32 of MotB functions in proton conduction through the motor and that no other protonatable residues function directly in this role (29, 30, 37, 42). Proton flux through the MotA-MotB channels can be assayed by a growth impairment that results when MotA and an appropriate fragment of MotB (which includes Asp 32 and the membrane segment) are overexpressed (4, 38). Mutations of MotB Asp32 reduce proton conductance in this assay (42).

View larger version (25K): [in this window] [in a new window]   FIG. 1.   Membrane topologies of MotA and MotB. Functionally important residues identified in previous mutational studies (40, 42) are indicated. Residue numbers are for the MotA and MotB proteins from E. coli.

The topology model (Fig. 1) predicts that residues Pro 173, Pro 222, and Asp 32 could be near each other, since all are near the cytoplasmic surface of the membrane. Previously, we proposed that these three residues might form a site that functions to couple proton flow to motor rotation (42). Here, we have sought to test this proposal by examining more closely the motor defects in Pro 173 and Pro 222 mutants. The mutants were characterized by measurements of swarming in soft agar, swimming in liquid medium, torque when tethered to coverslips, dominance, and ability to impair growth when overexpressed. In mutants that swam well enough to allow accurate measurements of speed, the effects of load and solvent isotope (H₂O versus D₂O) were also measured. The results provide clues to the functions of Pro 173 and Pro 222 in flagellar rotation and suggest features of a mechanism for torque generation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids and strains. Plasmids and strains used are listed in Table 1. DNA purification and manipulation were performed by standard procedures. Mutations were made by using the Altered Sites procedure (Promega) on the motA gene in plasmid pRF4. This plasmid expresses MotA at a level comparable to that of the wild-type chromosomally encoded protein.

Measurements of motor performance. Swarming rates were measured at 32°C in tryptone plates containing 0.28% agar. For each measurement, 1 µl of a saturated overnight culture in tryptone broth was spotted onto a swarm plate. Once swarming began, swarm diameter was measured at regular intervals for 4 to 12 h, and plots of diameter versus time were fitted to a line. Reported rates are averages for three to six determinations and are relative to wild-type control strains, which were present on the same plates in most cases. Motor torque under high load was measured by tethering cells to coverslips at 32°C, by using methods described previously (35). Rotating cells were recorded on videotape, and motor torque was computed from rotation speed, cell size, radius of gyration, and solution viscosity, as described elsewhere (35). Reported torque values are averages for at least 90 cells. Swimming speeds were measured at 32°C, by using methods described elsewhere (4), and are averages for at least 50 cells. The medium used for swimming speed measurements was 10 mM potassium phosphate (pH 7.0), 67 mM NaCl, 5 mM sodium lactate, 0.1 mM EDTA, and 1 µM methionine. Solvent isotope effects were measured in the same medium prepared in D₂O, with the pH adjusted to 6.6. Swimming speeds in D₂O were compared to those in H₂O solutions containing 1% Ficoll (average molecular weight, 400,000) to match the viscosity of D₂O. The load dependence of swimming speed was measured in media containing Ficoll at concentrations from 1 to 16%. Viscosity of solutions was measured with a Cannon-Ubbelohde viscometer (International Research Glassware, Kenilworth, N.J.) immersed in a water bath at 32°C.

Growth rate measurements. Growth rates of cells overexpressing MotA, MotB fragments, and various mutant versions of the proteins were measured at 34°C in Luria-Bertani (LB) medium (1% tryptone, 0.5% yeast extract, 0.5% NaCl). Fresh transformants were cultured overnight at 34°C in LB medium containing 100 µg of ampicillin/ml. An aliquot of cells was diluted 100-fold or 200-fold into fresh LB medium (containing ampicillin), and overexpression was induced by adding -indoleacrylic acid to a final concentration of 100 µg/ml from a 30-mg/ml stock in ethanol. Negative controls received just the ethanol. Optical density was measured at intervals of 30 min for a total of 4 to 6 h, and plots of ln (optical density at 600 nm [OD₆₀₀]) versus time were used to determine periods of exponential growth and to estimate the rate.

Estimation of cellular levels of MotA. The relative amounts of mutant MotA proteins in the cells were estimated by immunoblotting, as described elsewhere (35). For level estimation the proteins were expressed from derivatives of pLW3 not induced with indoleacrylic acid, which gives protein levels somewhat (three- to fivefold) higher than expression from the chromosome. Antiserum against MotA was raised in rabbits (Covance, Denver, Pa.), with a purified, soluble fragment of MotA used as the immunogen. A construct was made that overexpresses MotA residues 48 to 174 fused to six His residues, from the T7 promoter. This MotA fragment was overexpressed in strain BL21De3 (34) and accumulated in inclusion bodies. The inclusion bodies were isolated and dissolved in 6 M urea, and the protein was bound to a Ni-ion affinity column (Novagen, Madison, Wis.). The column-bound protein was renatured in binding buffer (5 mM imidazole, 500 mM NaCl, 20 mM Tris [pH 8]). The column was washed with the same buffer containing 100 mM imidazole and then eluted in 400 mM imidazole. The anti-MotA antiserum was purified by passage through a column containing immobilized proteins from a motA motB deletion strain (Table 1), as described elsewhere (35).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Swarming and swimming of Pro 173 and Pro 222 mutants. In a previous mutational study of MotA, we found that the conserved proline residues Pro 173 and Pro 222 are important for motility (40). These residues are located at the cytoplasmic ends of membrane segments 3 and 4 in all MotA sequences reported thus far. Several mutations of Pro 173 and Pro 222 were isolated in the previous study, and their phenotypes were characterized by measurements of swarming rates in soft agar and assays of dominance of plasmid-borne alleles over wild-type motA on the chromosome. Here, we have isolated additional replacements of Pro 173 and Pro 222 and have undertaken more detailed measurements of their effects on motor function.

View larger version (189K): [in this window] [in a new window]   FIG. 2.   Examples of swarming phenotypes of Pro 173 and Pro 222 mutants. The motA-defective strain MS5037 was transformed with plasmids (derivatives of pRF4) expressing the mutant MotA proteins indicated. The plate contained tryptone broth and 0.28% agar. It was inoculated with 1 µl of a saturated culture of each strain and photographed after 8 h at 32°C.

Protein stability. To determine whether the mutations affected the ability of the proteins to accumulate in cells, we used immunoblots to estimate the levels of mutant MotA proteins in cell membranes. An example is shown in Fig. 3. All of the Pro 222 mutant proteins were found at approximately the same level as that of the wild-type protein, except the P222F protein, which was not detected in cell membranes from either mid-log or saturated cultures. Most of the Pro 173 mutant proteins were also found at approximately the same level as that of the wild-type protein. Some (P173D, P173N, P173W, and P173Y) were present at relatively low levels in saturated overnight cultures but at nearly normal levels in mid-log cultures. The P173Q protein was not detected in either mid-log or saturated cultures but must nevertheless be present at some level, because it supports torque generation (at 50% of normal; Fig. 5). Thus, in the majority of cases the defects in motor function are not caused by failure of the proteins to accumulate.

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Immunoblot showing cellular levels of representative mutant MotA proteins. Cell membranes were prepared from mid-log cultures and electrophoresed and immunoblotted as described in Materials and Methods. w.t., wild type.

Load and isotope effects. In free-swimming cells the load on the flagellar motors is relatively light, because the flagellar filaments are thin and several filaments work together in a bundle (21). In this circumstance, the rotation rate of the motors is limited mainly by the rates of internal processes and not by the external load. MotA is not required for flagellar assembly, and MotA mutations are not expected to affect the shape or propulsive efficiency of the flagellar bundle. The slow swimming of the mutants thus suggests that a process occurring in the motor is made slower by the replacements of Pro 173. To learn more about the rate limitations in the mutants, we examined the effects of load and solvent isotope, by measuring swimming speeds in media containing various amounts of Ficoll and in media made with D₂O.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Effects of viscous load on the swimming speeds of wild-type cells and the MotA mutants P173S and P173T. Swimming speeds and medium viscosity were measured at 32°C in media that contained Ficoll (average molecular weight, 400,000) at concentrations ranging from 0 to 16%, as described in Materials and Methods. The reciprocals of swimming speeds are plotted against the relative viscous load, which was obtained as the product of viscosity and swimming speed. Results are shown for three independent experiments with the wild type and two independent experiments with each mutant. Values are the means ± standard errors of the mean for at least 50 cells. Practically all of the uncertainty in the estimates of load comes from the measurements of swimming speed, because measurements of viscosity are much more precise. Errors along the two axes are therefore expected to be correlated, as indicated. Best-fit straight lines are drawn to guide the eye.

Motor torque in Pro 173 and Pro 222 mutants. When cells are tethered to coverslips by individual flagellar filaments, the motors turn the cell bodies at speeds between 1 and 20 Hz. The viscous load resisting rotation is large in this case and determines the rotation rate. When the medium viscosity is increased by a given factor, rotation rate decreases by the same factor, so that torque remains constant (17, 24). Because the torque of a motor turning a tethered cell is just balanced by the viscous drag opposing rotation, motor torque can be estimated from the rotation rate, cell size and radius of gyration, and medium viscosity. Cells of each Pro 173 and Pro 222 mutant were tethered to coverslips, and rotation speed, cell size, and radius of gyration were measured for all of the rotating cells in several microscope fields. Torques were then computed by using equations for the viscous drag on rotating cylinders, as described previously (4, 35). The results are shown in Fig. 5 for all of the mutants that produced measurable torque.

View larger version (48K): [in this window] [in a new window]   FIG. 5.   Motor torques in tethered cells of Pro 173 mutants. Flagellar filaments were sheared and tethered to coverslips on a microscope stage thermostated at 32°C, and cell rotation rate, size, and radius of gyration were measured. Torques were computed as described in Materials and Methods. Values are the means ± standard errors of the mean for the sample sizes indicated.

Function of the mutant proteins in proton conduction. Overexpression of MotA, together with a MotB fusion protein that contains the membrane segment and residue Asp 32, reduces the growth rate of cells by a factor of about 2 (38). This growth impairment has been shown to correlate with an increased permeability of the cytoplasmic membrane to protons (4, 33), so it can be used to assay proton conduction by the MotA-MotB complexes separately from other functions they may have in motor rotation. To determine whether mutations of Pro 173 and Pro 222 affect proton conduction as assayed by growth impairments, the mutations were transferred to plasmid pLW3 (38), which expresses the motA and motB fusion genes from the trp promoter. The resulting plasmids were transformed into wild-type cells, and growth impairments were measured as described previously (4). Examples of growth curves and a summary of growth rates for all of the Pro 173 and Pro 222 mutants are shown in Fig. 6.

View larger version (43K): [in this window] [in a new window]   FIG. 6.   Proton conductance, as assayed by growth impairments, of MotA-MotB channels with mutations in Pro 173 and Pro 222. Wild-type cells were transformed with derivatives of plasmid pLW3 (38), which overexpresses MotA or its mutant variants and also a MotB-Rop fusion protein, both from the inducible trp promoter. (Left) Representative growth curves. Open symbols, uninduced cultures; filled symbols, cultures induced at t = 0 with indoleacrylic acid (final concentration, 100 µg/ml). Circles and triangles represent results of two independent experiments, which are fitted to solid and dashed lines, respectively. (Right) Summary of growth rates of the Pro 173 and Pro 222 mutants, with and without induction by indoleacrylic acid (mean ± standard deviation; n = 3).

View larger version (30K): [in this window] [in a new window]   FIG. 7.   Proton conduction, as assayed by growth impairments, by MotA-MotB channels with mutations in both proteins. The mutations present in MotA and MotB are indicated. Q4Ter signifies a stop codon in place of codon 4 of motB (allele 33 in reference 6). Black bars, uninduced; grey bars, induced with indoleacrylic acid. Values are the means ± standard deviations (n = 5 to 9). w.t., wild type.

Dominance of Pro 222 mutations. The strong dominance of certain Pro 222 mutations suggests that some Pro 222 mutant proteins are capable not only of substituting for wild-type protein in the motors but also of impeding (i.e., jamming) motor rotation (40). A clue to the function of Pro 222 would be provided if mutations of this residue were found to jam the motor more than other mutations of MotA. Mutations isolated by Garza et al. provide a means for testing this hypothesis (12, 13). Garza et al. found that certain mutations of the stator protein MotB are suppressible by certain mutations in the rotor protein FliG. Their results suggested that the MotB mutations cause the stator to become misaligned relative to the rotor and that the FliG mutations cause compensating movements in the rotor that restore a functional alignment. Detailed physiological studies suggested that the FliG mutations increase the clearance between the rotor and the stator (13). To determine whether the dominance of Pro 222 mutations can be modulated by changes in rotor-stator clearance, we measured the dominance of several Pro 222 mutations in strains containing the FliG mutations that are believed to increase clearance. These experiments used a fliG null strain transformed with one plasmid encoding the mutant FliG proteins and another plasmid encoding the dominant nonfunctional MotA proteins (or wild-type MotA as a control). The strains thus contained the wild-type motA gene on the chromosome, allowing dominance of the plasmid-borne motA alleles to be assessed. The results for the strongly dominant MotA mutations P222R and P222M and controls with wild-type MotA are shown in Fig. 8.

View larger version (105K): [in this window] [in a new window]   FIG. 8.   Dominance of Pro 222 mutations of MotA and modulation of this dominance by mutations in fliG. The plates were inoculated with 1 µl of saturated cultures of the indicated strains and incubated at 32°C. Strains that contained only wild-type MotA were incubated for 6 h (top) or 7.5 h (bottom); others were incubated for 7.5 to 8 h. Three isolates of each strain are shown. w.t., wild type.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A previous study showed that residues Pro 173 and Pro 222 of MotA are important for torque generation in the flagellar motor but did not shed light on their specific roles (40). Previous work also showed that Asp 32 of MotB is critical for function, probably because it forms part of the proton path through the motor (42). All three residues are predicted to be near the cytoplasmic ends of membrane segments, so they could be near each other, possibly forming a site that functions in coupling proton flow to motor rotation. The present results provide clues to the functions of Pro 173 and Pro 222 of MotA and generally support the coupling site hypothesis.

Role of proline 173 of MotA. Mutations of Pro 173 gave complex phenotypes, in many cases impairing but not eliminating motor function. Replacements that supported some motor function tended to be uncharged residues of medium size. Motor performance was diminished under both low load and high load in Pro 173 mutants. In some cases (P173A, P173I, and P173Q), motor speed at low load was decreased so much that cells did not swarm at all and were immotile in liquid medium, even though they produced significant torque when tethered. Thus, mutations of Pro 173 can slow some step(s) in the operation of the motor, in some cases severely. In the two mutants that swam steadily enough to allow measurements of speed (P173S and P173T), the speed was more sensitive to viscous load than wild type (Fig. 3) but no more sensitive to solvent isotope (D₂O versus H₂O). Given the limited precision of the swimming speed measurements, we cannot rule out the possibility that Pro 173 mutations affect a proton transfer step(s). Our results do suggest, however, that proton transfer is not the main step affected. Instead, the increased sensitivity of the mutants to viscous load suggests that Pro 173 functions in a step that involves rotor movement. An alternative possibility is that it functions in a step that immediately precedes or follows rotor movement, with the first of the two steps being a rapid pre-equilibrium. In either case, the results imply a link between Pro 173 and events at the rotor-stator interface.

Role of proline 222 of MotA. All replacements of Pro 222 abolished motor rotation in both the swarming and tethered cell experiments and also eliminated or reduced proton flow as assayed by growth impairments. Even residues with side chains comparable in size to proline did not support any motor function, suggesting that the special conformational constraints imposed by a proline residue are essential at this position. We propose that Pro 222 regulates the conformation and/or conformational changes of the MotA-MotB complex. The diminished growth impairments observed in the Pro 222 mutants also implicate this residue in proton conduction. This finding and the location of Pro 222 near the cytoplasmic surface of the membrane suggest that it is in the general vicinity of Asp 32. We cannot conclude that the side chain of Pro 222 is very near Asp 32 or is necessarily inside the channel, however.

Sequence context of the key proline residues of MotA. An alignment of selected MotA sequences (Fig. 9) shows that no other residues are conserved in the segments around Pro 173 and Pro 222. Evidently, no other side chains in these segments of MotA are critical for function. A feature that is conserved is the spacing between the proline residues and some invariant glycine residues in the membrane segments, suggesting that the positions of Pro 173 and Pro 222 relative to the membrane-embedded parts of the channel are an important feature that has been conserved through evolution. The conformation of MotA in the segments around these proline residues is likely to be important for function. Secondary structure predictions suggest that the segments flanking both Pro 173 and Pro 222 are -helices (Fig. 9). The Pro residues would introduce kinks in these helices and, as described below, might mediate conformational changes coupled to the protonation and deprotonation of Asp 32. 

View larger version (33K): [in this window] [in a new window]   FIG. 9.   Alignment of selected MotA sequences in the segments around Pro 173 and Pro 222. M3 and M4 indicate parts of membrane segments 3 and 4. Coils below the sequences indicate segments predicted to be -helical by a neural-net algorithm (27).

A model for motor rotation. Figure 10 illustrates a proposed mechanism of torque generation in the flagellar motor. The model shown is only one of several possibilities, but it is offered as a concrete example of how proton movements might be coupled to rotor movements at the putative Asp 32-Pro 173-Pro 222 site. The model has two main features. First, we suggest that when the rotor and the stator are appropriately aligned, interactions across the rotor-stator interface induce a conformational change in the stator that allows a proton to enter from the periplasm and bind to Asp 32. This channel-gating step involves mainly Pro 173. (Such a role for Pro 173 in gating proton flow, as opposed to a role in forming the proton pathway, would not imply an increased deuterium isotope effect for Pro 173 mutations.) The triggering site that controls the timing of this gating step might consist of the charged residues of MotA and FliG that have been shown to interact at the rotor-stator interface (41). Electrostatic interactions at the rotor-stator interface could also serve to drive some movement of the rotor, and this function is also indicated in the figure. Second, we suggest that the binding of the proton to Asp 32 causes a further conformational change in the stator, which alters the rotor-stator interface to drive and/or restrict the movement of the rotor. In the model drawn, most movement of the rotor is driven actively either by electrostatic interaction or by direct contact with the stator. Alternatively, in a rectified Brownian movement model, rotor movement would be driven mainly by thermal energy, with conformational changes in the stator serving to provide restraints that prevent reverse movements. We suggest that Pro 222 functions in this rotor-driving (or latching) conformational change. To complete the cycle, the rotor moves further forward, allowing the proton to exit to the cytoplasm and restoring the stator to its initial state.

View larger version (14K): [in this window] [in a new window]   FIG. 10.   Proposed mechanism for coupling proton flow to motor rotation at the Asp 32-Pro 173-Pro 222 site. (A) Mechanism of motor rotation. The diagram shows a single stator complex and a small section of the rotor. A series of projections on the edge of the rotor, drawn as white triangles, interact with points projecting from the stator so that relative movement of rotor and stator is restricted. Other sites on the rotor and stator, colored gray and termed trigger sites, interact so that when the rotor and stator are properly aligned, a conformation change is triggered in the stator. The trigger sites might be formed from the charged residues of MotA and FliG that have been shown to interact at the rotor-stator interface (41). Other elements pictured are the functionally important residues Asp 32 of MotB and Pro 173 and Pro 222 of MotA. In an initial state (panel i), Asp 32 is unprotonated, and the trigger sites on the rotor and stator are close to each other but not yet aligned. A small movement of the rotor, which might be accelerated by electrostatic interactions between the rotor and stator, brings the trigger sites into alignment so that they can interact to promote a conformational change in the stator (panel ii). This conformational change, which is suggested to involve Pro 173, opens a gate in a gate in a proton channel to the periplasm. A proton enters and binds to Asp 32 (the filled oval signifies the protonated site), triggering a further conformational change in the stator (panel iii). This conformational change affects contacts at the rotor-stator interface to drive movement of the rotor, toward the right for the geometry pictured. Next, the proton is released from Asp 32 to the cytoplasm, and the stator returns to its starting conformation. The rotor is driven farther to the right at this step because the stator engages the next projecting site on the rotor (panel iv). The net result is transfer of one proton across the membrane and movement of the rotor to the right by one rotor subunit. (B) A mechanism for switching the direction of rotation. Switching could occur by a coordinated movement of the projecting sites on the rotor relative to the trigger sites. The events in the stator would be the same as those pictured in part A, but for the geometry shown at the bottom, the rotor would move to the left.