A novel type bacterial flagellar motor that can use divalent cations as a coupling ion

The bacterial flagellar motor is a sophisticated nanomachine embedded in the cell envelope and powered by an electrochemical gradient of H+, Na+, or K+across the cytoplasmic membrane. Here we describe a new member of the bacterial flagellar stator channel family (MotAB1 of Paenibacillus sp. TCA20 (TCA-MotAB1)) that is coupled to divalent cations (Ca2+and Mg2+). In the absence of divalent cations of alkaline earth metals, no swimming was observed in Paenibacillus sp. TCA20, which grows optimally in Ca2+-rich environments. This pattern was confirmed by swimming assays of a stator-free Bacillus subtilis mutant expressing TCA-MotAB1. Both a stator-free and major Mg2+uptake system-deleted B. subtilis mutant expressing TCA-MotAB1 complemented both growth and motility deficiency under low Mg2+conditions and exhibited [Mg2+]in identical to that of the wild-type. This is the first report of a flagellar motor that can use Ca2+and Mg2+as coupling ions. These findings will promote the understanding of the operating principles of flagellar motors and molecular mechanisms of ion selectivity.

(Sr 2+ ), for growth ( Fig. 1). On the other hand, E. coli and B. subtilis showed no requirement of such an alkaline earth metal for growth. Adding Ca 2+ and Mg 2+ to the medium moderately enhanced their growth was better than without divalent cations, even though the growth of B. subtilis was inhibited by Sr 2+ . Swimming behavior of this bacterium showed Ca 2+ -and Mg 2+ -dependent motility at pH less than 8.0 and Sr 2+ -dependent motility at pH 8.0 ( Fig. 2A-C and S1). However, swimming (approximately 20 μ m/s) of strain TCA20 was observed in the absence of divalent cations at pH 9.0. Furthermore, no motility was observed when 1 mM ethylenediaminetetraacetic acid (EDTA) was added to the motility assay buffer (pH 9.0) and > 1 mM CaCl 2 in the same buffer was required for Ca 2+ -dependent motility (Fig. 2D). No swimming was observed when up to 100 mM Na + and/or K + was added to the buffer at pH 8.0 (Fig. 3A,B). Additional Na + and K + in the swimming assay buffer did not stimulate Ca 2+ -dependent motility (Fig. 3C). In the presence of elevated divalent cations, no stimulation was observed in the swimming velocity of Escherichia coli, which has a H + -coupled motor at pH 7 and 8, and no swimming was observed at pH 9.0 ( Fig. 2 and S1). Alkaliphilic Bacillus pseudofirmus OF4, which has a Na + -coupled motor, did not exhibit swimming behavior under these conditions. At low concentrations of divalent cations, the Ca 2+ -and Mg 2+ -dependent swimming velocity by the flagellar motor was increased by elevating the pH ( Fig. 2A,B and S1A,B). At pH 9.0, despite the absence of divalent cations, swimming was observed. However, when the chelating reagent EDTA was added, no swimming was observed up to 1 mM CaCl 2 (Fig. 2D). Therefore, we investigated the possibility of carry-over of Ca 2+ in the buffer and found that it contained 0.69 mM, which may explain why motility was observed with presumably no divalent cations at pH 9.0. These results suggest the possibility of using Ca 2+ , Mg 2+ , and Sr 2+ as coupling cations for flagellar rotation of strain TCA20.
In B. subtilis, the Ca 2+ -specific Ca 2+ /H + antiporter ChaA and the P-type Ca 2+ -transporting ATPase YloB were identified 12,13 , suggesting that ChaA and YloB are important for Ca 2+ signaling during sporulation or germination. The annotation of the draft genome sequence showed that Paenibacillus sp. strain TCA20 has a chaA gene (AN: GAK39775) and a gene that encodes a putative P-type Ca 2+ -transporting ATPase (AN: GAK39789) 11 . In Streptococcus pneumonia, the P-type Ca 2+ -transporting ATPase CaxP is used to avoid Ca 2+ accumulation in the eukaryotic host 14 . In addition to Ca 2+ signaling during sporulation or germination these transporters may be important for the growth of strain TCA20 under high Ca 2+ concentrations.
A phylogenetic tree of the stator proteins of the flagellar stator subunits MotB and MotS from Paenibacillus, Oceanobacillus, and Bacillus sp. and E. coli is shown (Fig. 4). The phylogenetic features of two stator proteins suggest that TCA-MotAB2 functions as an H + -type stator. However, TCA-MotAB1 belongs to a different stator cluster from the H + -coupled MotAB or Na + -coupled MotPS complex (Fig. 4). Functional analysis of the cluster motility containing TCA-MotB1 has not yet been experimentally characterized, although each protein was automatically annotated as MotB in the database. Strains belonging to this cluster were isolated from various environments (e.g., soil, oral swab from patient, rhizosphere of plant, and assembly facility), and had no significant commonality [15][16][17][18] Paenibacillus lactis 154 belonging to this cluster was isolated from milk, which is rich in Ca 2+ 19 . Although P. lactis 154 was isolated from a Ca 2+ -abundant environment, the identity and similarity between TCA-MotB1 and MotB from P. lactis 154 were only 61% and 76%, respectively. A protein BLAST search returned no protein with homologies of > 68% and > 63% against TCA-MotA1 and TCA-MotB1, suggesting unique features of the TCA-MotAB1 stator complex.   threonine residue, respectively. Because strain TCA20 was isolated from a unique environment, critical amino acid residues for ion selectivity in both stators may not be conserved. This apparent exception may result in future identification of critical amino acid residues for coupling ion selectivity.
Stator-free B. subtilis mutants expressing TCA-MotAB1 and TCA-MotAB2. Because strain TCA20 was not genetically accessible, we used each single stator gene to directly compare the swimming properties conferred by TCA-MotAB1 and TCA-MotAB2, which were introduced into the lacA locus of B. subtilis strain Δ ABΔ PS from which both native BS-motAB (motAB of B. subtilis) and BS-motPS (motPS of B. subtilis) were deleted.
The resulting B. subtilis mutant strains expressing TCA-MotAB1 or TCA-MotAB2 were named TCA-AB1 and TCA-AB2, respectively. Each gene pair was controlled by a xylose inducible promoter (P xlyA ). Both stators restored motility to the non-motile Δ ABΔ PS strain on soft agar plates containing 1% xylose. Furthermore, we determined the cations that are preferred for flagellar rotation by these mutants (Fig. 6A). The swimming speeds of TCA-AB1 and TCA-AB2 were measured at several Mg 2+ concentrations in 10 mM potassium phosphate buffer (pH 8.0) with 5 mM glucose, 1% xylose, 10 μ g/ml tryptophan and lysine. TCA-AB1 exhibited no swimming ability without added Mg 2+ and exhibited stimulation by Mg 2+ . We performed the same experiment using CaCl 2 . However, the tumbling frequency of all strains was drastically increased and it was very difficult to measure the linear swimming velocity of each. Therefore, we tested the addition of only Mg 2+ and found that the TCA-AB1 rotor prefers using Mg 2+ in the heterologous neutralophilic host. BS-AB swimming, which has a distinct H + -coupled motor from B. subtilis, and TCA-AB2 were observed in the absence of Mg 2+ , and the speed was stimulated by increasing Mg 2+ concentrations up to 10 mM (Fig. 6A). The stimulated phenotype of the swimming speed of a proton coupled motor by adding MgCl 2 is of interest. The details of this mechanism are unknown. However, we speculated that proton entry by flagellar rotation is influence by intracellular pH homeostasis. The requirement of some amount of Mg 2+ may facilitate proton circulation from inside and outside the cells and allow the influx of more protons by flagellar rotation. It is known that intracellular levels of metal ions of bacterium are carefully maintained by sensing of regulatory proteins and an RNA element [20][21][22] . No swimming was observed for strain BS-PS, which has a distinct Na + -coupled motor from B. subtilis, under these conditions. Further, we tested the effects of several inhibitors on the bacterial flagellar motor. The protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP), which dissipates electrochemical H + gradients, did not affect TCA-AB1 swimming containing Mg 2+ at inhibitor concentrations up to 25 μ M at pH 8.0 (Fig. 6B). Conversely, TCA-AB2 swimming and H + -coupled BS-AB were completely inhibited by CCCP addition at 25 μ M with/without Mg 2+ at pH 8.0 (Fig. 6B). The Na + -coupled flagellar stator inhibitor 5-(N-ethyl-N-isopropyl)-amiloride (EIPA) did not affect the swimming of TCA-AB1 and TCA-AB2 containing 10 mM MgCl 2 and 100 mM NaCl at inhibitor concentrations up to 100 μ M (Fig. 6C). These results support that the coupling ions of the TCA-MotAB1 and TCA-MotAB2 stator complex are Mg 2+ and H + , respectively. TCA20 motility was completely dependent on divalent cation concentrations. Therefore, the stator protein TCA-MotAB2 may not be expressed under these culture conditions. Growth and measurement of intracellular Mg 2+ concentrations of a derivative strain of TCA-AB1 which was deleted in the major Mg 2+ uptake system. Strain TCA-AB1 showed Mg 2+ -dependent motility. However, there was no direct evidence of actual inward translocation of Mg 2+ through the stator protein TCA-MotAB1.
There are at least five distinct Mg 2+ transporters [YkoK (also termed MgtE), YloB, YfjQ, YqxL, and CitM] in B. subtilis. Of these, YkoK and YfjQ are major Mg 2+ uptake systems, and if both genes are deleted, the deleted strain requires > 6 mM Mg 2+ for growth (Fig. 7). We constructed strain Δ Δ TCA-AB1, a derivative strain of TCA-AB1, and additionally deleted both ykoK and yfjQ. This strain required Mg 2+ for growth without xylose. We hypothesized that if TCA-MotAB1 is incorporated into a functional flagellar motor and takes up Mg 2+ during flagellar rotation, strain Δ Δ TCA-AB1 will complement the Mg 2+ growth requirement with xylose. Strain Δ ABPSΔ KQ, a derivative strain of Δ ABΔ PS with deletions of ykoK and yfjQ, was used as a negative control.
Growth curves and intracellular Mg 2+ concentrations of strains BR151MA (wild-type, WT), Δ Δ TCA-AB1, and Δ ABPSΔ KQ were measured at several extracellular Mg 2+ conditions (Figs 7 and 8). Strain Δ ABPSΔ KQ required > 6 mM Mg 2+ for growth (Fig. 7), whereas the strain Δ ABPSΔ KQ intracellular Mg 2+ concentration was lower than that of WT under all conditions. However, the growth curve and [Mg 2+ ] in of strain Δ Δ TCA-AB1 was similar to those of WT, as expected. Δ Δ TCA-AB1 motility was observed under all conditions with xylose. Akanuma et al. reported an intercellular total Mg 2+ concentration of B. subtilis WT of approximately 70 mM 23 , similar to our data (Fig. 8). These results suggest that flagellar rotation and Mg 2+ uptake are coupled.
Stator protein TCA-MotAB1 has a universally conserved Asp-33 residue of MotB1 that is critical for motility and is a predicted H + -binding site in E. coli 24 (Fig. 5). The crystal structure of the flagellar stator has not been determined, except for the C-terminal hydrophilic region of MotB and PomB 25 27 . There is no additional negative charged amino acid residue near residue Asp-33 of TCA-MotB1 and the third and fourth transmembrane segments of TCA-MotA1, suggesting that divalent cations work as coupling ions for flagellar rotation of strain TCA20, although the predicted coupling ion-binding site was a single negatively charged side chain of an aspartic acid residue. Utilizing divalent cations for flagellar rotation, the membrane potential would be consumed two times faster than utilizing monovalent cations. Therefore, motor torque coupled with divalent cations compared with monovalent cations is of interest.
The discovery of a novel type flagellar motor that can use divalent cations as coupling ions shows the diversity of flagellar motor proteins. Recently, Takekawa et al reported that one of the earliest flagellar motor proteins in hyperthermophilic bacterium Aquifex aeolicus can be driven by Na + for energy coupling 28 . The path of evolution of the conventional flagellar motor proteins from the earliest ones is still poorly understood. It will be interesting to elucidate the evolution of stator proteins in connection with their habitat environment.
A major finding of our study was that a single bacterial flagellar stator of Paenibacillus sp. TCA20 coupled motility to divalent cations, including Mg 2+ and Ca 2+ . This finding may facilitate the identification of additional examples of divalent cation-coupling capacity among the increasing number of bacteria that exhibit motility in extreme environments. Altering the coupling ion selectivity of the stator in the laboratory is time-consuming. However, if the identification of stator genes of microorganisms growing in extreme environments is possible, application of the flagellar motor as a nanomachine can progress.

Materials and Methods
Bacterial strains, plasmids and growth conditions. The strains and plasmids used in this study are shown in Table S1. To measure Mg 2+ -and Ca 2+ -dependent growth capacities of E. coli, Paenibacillus sp. TCA20, and B. subtilis, we initially attempted to grow E. coli and B. subtilis in Luria broth (LB) medium. Paenibacillus sp. TCA20 cells were aerobically grown on 30 mM Tris medium (pH 7.7) containing 5 mM CaCl 2 overnight at 37 °C. Tris medium contained 30 mM Tris base, 7 mM citric acid monohydrate, 0.05% (w/v) yeast extract, 50 mM  glucose and 1% (v/v) trace elements 29 . E. coli and B. subtilis cells were grown in LB medium under the same conditions. The culture was then inoculated into 30 mM Tris medium (pH 7.7) containing several different MgCl 2 , CaCl 2 , or SrCl 2 concentrations at an absorbance of 0.01 at 600 nm (A 600 ), and grown aerobically at 37 °C for 14 h. Growth was monitored by measuring the absorbance at A 600 .
The effects of increasing MgCl 2 concentrations (i.e., 0, 1, 2.5, 5, or 10 mM) on growth of B. subtilis 168 strains BR151MA (WT), Δ ABPSΔ KQ, and Δ Δ TCA-AB1 were determined in 2 × TY medium with 1% xylose at 37 °C with shaking and measured at A 600 . Plasmid construction. Selected mot genes were integrated into the chromosomal lacA locus of mutant strains under the control of the xylose-inducible P xylA promoter using the plasmid pAX01. For the construction of a plasmid carrying intact TCA-motAB1 and TCA-motAB2 genes under the control of the P xylA promoter, sets of primers were designed with nucleotides encoding SacII sites. All primer sequences are provided in Table S2. Each amplified fragment was cloned into SacII-digested pAX01, yielding PxylA-AB1 and PxylA-AB2, respectively. Each plasmid was integrated into the lacA locus of a B. subtilis Δ ABΔ PS host. Recombinant transformants were selected using conventional techniques to confirm the correct sequence of the insert.
Measurement of swimming speed. For the measurement of swimming speed, Paenibacillus sp. TCA20 cells were aerobically grown on Tris medium (pH 7.7) plus 5 mM CaCl 2 overnight at 37 °C. The culture was then inoculated into 20 ml of fresh Tris medium (pH 7.7) at an A 600 of 0.01 and aerobically grown at 37 °C. E. coli cells were grown in LB medium under the same conditions. B. pseudofirmus OF4 cells were grown in alkaline complex medium under the same conditions 7 . BS-AB, BS-PS, TCA-AB1, and TCA-AB2 mutant strains were grown in Spizizen I medium 30 plus 1 mM MgCl 2 and 1% xylose at pH 8.0 under the same conditions. Spizizen I medium contained Spizizen salts, 0.5% glucose, 0.02% casamino acids, 0.1% yeast extract, 10 μ g/ml tryptophan and lysine. Spizizen salts contained 85 mM K 2 HPO 4 , 40 mM KH 2 PO 4 , 15 mM (NH 4 ) 2 SO 4 , 6 mM sodium citrate and 0.8 mM MgSO 4 . Highly motile cells in the late logarithmic phase were harvested by filtration on OMNIPORE membrane filters (0.45 μ m) and washed three times with 2 ml of 30 mM Tris-HCl buffer that contained 5 mM glucose, and the indicated amounts of NaCl, KCl, CaCl 2 , MgCl 2 , SrCl 2 and/or EDTA were used at the indicated pH values. Cells were suspended in 1 ml of the same buffer and incubated at 30°C for 10 min. For the measurement of swimming speed, B. subtilis and its derivative mutants were grown for 6 h at 37 °C in Spizizen I medium 30 plus 1 mM MgCl 2 and 1% xylose at pH 8.0 with shaking. Cells were suspended in 1 ml of phosphate buffer (pH 8.0) and the indicated amounts of CCCP, EIPA, MgCl 2 and NaCl, and then incubated at 37 °C for 10 min. Phosphate buffer contained 10 mM potassium phosphate (pH 8.0) plus 5 mM glucose, 1% xylose, 10 μ g/ml tryptophan and lysine. Cell motility was observed under a dark-field microscope using a Leica DMRE microscope and recorded in high definition with a digital color camera (model DFC310FX; Leica, Tokyo, Japan). Swimming speed was determined with particle tracking velocimetry software (DigiMo, Tokyo, Japan). The ionic strength of the assay buffers increased with the cation concentration. All results are the averages of three independent experiments in which the speed of 30 different cells was measured.
Construction of ΔykoK and ΔyfjQ mutants. The Δ ykoK and Δ yfjQ mutants were constructed by gene splicing via overlap extension as described previously 31 . For the construction of a fragment upstream and downstream of ykoK, two independent PCR reactions were performed on WT DNA with the primer sets YkoK-cm-1 and YkoK-cm-2 (i) and YkoK-cm-3 and YkoK-cm-4 (ii). The PCR products were used as templates for a second PCR with primers YkoK-cm-1 and YkoK-cm-4. The purified products were cloned into SmaI-digested pUC18Tc, yielding pUC18Tc-Δ ykoK.
For the construction of a fragment upstream and downstream of yfjQ, two independent PCR reactions were performed on WT DNA with the primer sets YfjQ-cm-1 and YfjQ-cm-2 (i) and YfjQ-cm-3 and YfjQ-cm-4 (ii). The PCR products were used as templates for a second PCR with primers YfjQ-cm-1 and YfjQ-cm-4. The purified products were cloned into SmaI-digested pUC18Tc, yielding pUC18Tc-Δ yfjQ.
Strain Δ ABPSΔ KQ was constructed using the gene replacement approach described elsewhere 32 and the ykoK and yfjQ region deletion of the B. subtilis Δ ABΔ PS host was confirmed by PCR.
Measurement of intracellular Mg 2+ concentration. B. subtilis BR151MA, Δ Δ TCA-AB1, and Δ ABPSΔ KQ cells were grown at 37 °C in 2 × TY medium with 1% xylose plus MgCl 2 (0, 1, 2.5, 5, or 10 mM) with shaking. The cells in the late logarithmic growth phase were harvested by centrifugation (8,000 rpm, 10 min, 25°C), such that there were an equal number of bacterial cells, and washed by suspension in 10 ml of 300 mM sucrose solution. Then, the cells were resuspended in 10 ml of 300 mM sucrose solution with 100 mM hydrochloric acid (HCl) solution. After shaking for 10 min, cell debris was removed by centrifugation and Mg 2+ was measured by MG Metallogenics Mg 2+ measurement LS (Metallogenics Co., Ltd., Chiba, Japan) that was calibrated with standard Mg 2+ solutions of known concentrations. Intracellular Mg 2+ was calculated by assuming that a B. subtilis cell is a cylinder (radius, 0.4 μ m; length, 4 μ m) 23 .