Chemotactic Methyltransferase Promotes Adaptation to Repellents in Bacillus subtilis”

Bacillus subtilis cheRE, which encodes the chemotactic methyltransferase, has been cloned and sequenced. CheRB is a polypeptide of 256 amino acids, with a pre- dicted molecular mass of 28 kDa. A comparison of the predicted amino acid sequence of B. subtilis CheRB with that of Escherichia coli CheRE demonstrates that the two enzymes share 31% amino acid identity. The homology was functional in that the expression of che& in an E. coli cheRE null mutant made the bacteria Che’. In contrast to cheRE null mutants which show a strong smooth swimming bias, che& null mutants were pre- dominantly tumbly. They respond to the addition and subsequent removal of attractant. They also respond to the addition of repellent but do not adapt; they resume prestimulus bias on removal of repellent. Tethering analysis of a culture of a cheRE null mutant revealed two distinct subpopulations, each demonstrating unique behaviors. One showed a strong clockwise flagellar rota- tion bias, whereas the other was more random. The lat-ter phenotype may be due to a deficiency of CheB and may reflect an interaction of CheB and CheR. Measure-menta

are able to methylate the heterologous MCPs i n vitro . In contrast to most of the chemotaxis genes of B. subtilis, cheRB is located near aroF, a locus distinct from the major chelfta operon, which is betweenpyrD and thyA .
CheRB is known to be required for chemotaxis inasmuch as a point mutant, 011100, showed poor chemotaxis toward attractants as measured in a capillary assay. The specific role that cheRB plays has been unclear. The behavior of 011100 as determined by microscopic observation is markedly different than cheRE mutants from E. coli; 011100 appears to swim randomly, whereas the cheRE mutants are smooth swimming.
In E. coli, repellent stimuli act through the MCPs to activate CheAE, which then donates phosphate groups to CheYE, the tumble regulator, and CheBE, the methylesterase. Phosphorylated CheBE is activated to remove methyl groups from the MCPs which brings these receptors back to a prestimulus signaling state. Upon the removal of repellents, CheAE kinase activity decreases which leads to lower levels of phosphorylated CheBE. Consequently, CheBE activity decreases so that CheRE can then replace the methyl groups removed by CheBE. Thus CheBE is the enzyme responsible for the adaptation to negative stimuli and CheRE is responsible for the adaptation to positive stimuli.
The study of chemotaxis in B. subtilis has unveiled several striking differences with E, coli. In both organisms CheA is believed to phosphorylate CheY (Hess et al., 1988;Fuhrer and Ordal, 1991), which in turn regulates the direction of flagellar rotation. However, null mutants in cheA and in cheY are tumbly in B. subtilis but smooth swimming in E. coli (Fuhrer and Ordal, 1991;Bischoff and Ordal, 1991;Oosawa et al., 1988;Parkinson, 1978). In both organisms, methylation of the MCPs causes adaptation to chemoeffectors. However, removal of methyl groups from the MCPs by the methylesterase, CheB, results in adaptation to attractant in B. subtilis, but adaptation to repellent in E. coli. (Kirsch et al., 1993;Hiroyuki et al., 1983).
Thus, in these respects, chemotaxis in B. subtilis is "opposite" to that in E. coli. In this article, we report the cloning and sequencing of cheRB and demonstrate that CheRB is the enzyme responsible for the adaptation to repellents in B. subtilis chemotaxis.  Ordal, 1981 Burgess-Cassler et al. (1982) This work This work This work This work Kirsch et al. (1993) This work This work This work Leonhardt andAlonson (1988) Ferrari et al. (1983) Tabor and Richardson (1985) Yanisch-Perron et al. ( 1985) This work This work This work This work Kirsch et al. (1993) EXPERIMENTAL PROCEDURES Bacterial Strains and Plasmids-The bacterial strains and plasmids used in this investigation are described in Table I. Growth Media-Tryptone broth (TBr) is 1% Tryptone and 0.5% NaCl. LBr is 1% Tryptone, 0.5% yeast extract, and 0.5% NaCI. Minimal medium is 50 rm potassium phosphate buffer pH 7.0, 1 w MgCI,, 1 r m (NHJSO,, 0.14 rm CaCl,, 0.01 m~ MnCI,, 50 pg/d required amino acids (His, Met, Trp), and 20 r m sorbitol.
DNA Sequence and Analysis-DNA sequencing was done by the dideoxynucleotide-chain termination method (Sanger et al., 1977) with the Sequenase kit (U. S. Biochemical Corp.) and [36SldATP (Amersham Corp.). Analyses of DNA sequences and homology alignments were performed using DNASTAR. Since we were able to recover clones with cheR, inserted into M13 and pUC vectors in only one orientation, oligonucleotide primers were designed from the sequence derived from the one strand to facilitate double stranded sequencing of the opposite strand.
Mutagenesis and Analysis-Integration analysis (Piggot et al., 1984) was performed by subcloning DNA fragments containing incomplete regions of chRB into integration plasmids (Table I) and subsequently transforming wild type B. subtilis. Ache& mutant strain, 012680, was created by the integration of pAZ286 (Table I) into 011085. Southern hybridization analysis was performed to verib the integration of each plasmid. Another c~R B mutant (012652) was created by the replacement of the internal DraI fragment with a promoterless chloramphenicol resistance marker (chloramphenicol acetyl transferase, cat gene) (Corarmblas and Bolivar, 1982). The ability to perform normal chemotactic functions was determined by capillary assays (Ordal and Goldman, 1975) and in vivo methylation (Ullah and Ordal, 1981).
Complementation of cheR Mutants-Complementation of the mutants was performed by transforming competent cells with the appropriate plasmids. Complementation was assayed on semisolid agar swarm plates containing 0.27% agar. 1 m~ isopropylthiogalactoside was added to induce the expression of cheBB from pMKlO8. The swarm plates were incubated for 6 h at 37 "C for B. subtilis cells and for 9 h at 30 "C for E. coli cells. The swarm size of the complemented strain was compared to that of the wild type.
Capillary Assay-Capillary assays have been described (Ordal and Goldman, 1975). Cells were grown overnight on tryptose blood agar plates. IO8 bacteridml were then inoculated into 1 ml of TBr, diluted 150 into minimal media, and grown for 4 h at 37 "C. Cells were then supplemented with 0.05% glycerol and 5 l l l~ sodium lactate, grown 15 min longer, and harvested. They were resuspended at AGm = 0.001 and assayed for chemotaxis using azetidine-2-carboxylic acid as the attractant. The contents of the capillary tube were plated on TBr plates to determine accumulation of colony forming units.
Tethering Analysis-The method used to tether the bacteria has been described (Berg and Tedesco, 1975;Berg and Block, 1984). Cells were diluted 1:lOO from an overnight TBr culture into minimal medium and grown for 4.5 h. Each tethered bacterium was subjected to the addition and removal of chemoeffectors over a period of 8 min. The behavior analyzed as previously described (Kirsch et al., 1993).
In Vitro Methylesteruse Assay-The methylesterase activity of cellular extracts was determined by the method of Stock and Koshland (1978) with some modifications. Stationary phase cells were diluted 1:lOO into 6 liters of LBr and were harvested afker 10 h of growth at 37 "C. Cell pellets were washed twice in 1 M KCI, then twice in MT buffer (10 m~ potassium phosphate, pH 7.0, 1 rm MgCl,, 0.1 rm EDTA, 1 rm 2-mercaptoethanol, 0.1 m~ phenylmethylsulfonyl fluoride, and 0.02% sodium azide) and resuspended in a final volume of 30 ml of MT buffer. Cells were incubated with 3 mg/ml lysozyme for 1 h at room temperature and lysed by sonication. Cell debris was removed by centrifugation at 12,000 x g for 30 min. Extracts were then assayed for methylesterase activity. Protein concentration was determined using the Coomassie protein assay reagent (Pierce Chemical Co.). Radioactive-methylated membranes from a cheR mutant (012680) were prepared to be used as the substrate for the methylesterase assay. These membrane preparations were labeled with [methyl-3HlS-adenosyImethionine using soluble extracts of 012714 which overexpresses CheRB in an E. coli strain lacking all other chemotaxis proteins. Labeled membranes were washed twice in 1 M KC1 and once in MT buffer and resuspended in MT buffer at a concentration of 40 mg/ml membrane protein.
The methylesterase assay was performed by adding cellular extracts to the labeled membranes. Reactions contained labeled membranes diluted to a final protein concentration of 0.4 mg/ml, 20% glycerol, and 37.7 mg/ml whole cell extract protein. The reaction was incubated at 28 "C and 100-pl aliquots were taken at different time intervals up to 90 min and added to 900 pl of cold ethanol. The resultant precipitate was  binding sites are underlined and in boldface. Inverted repeat regions suggesting potential pindependent terminators are denoted with arrows.

CheR, F T T K W K Q L T G V D L T L Y K E A Q M K R R L T S L Y 35 CheR, I S Q L I Y Q R A G I V L A D H K R D M V Y N R L V R R L FrzF L A A L L L E R A G L K I T P D G F H S L R L A L S
. * * * * * * * *

" I K T S R P L K I W S A A C S T G E E P Y T L A M L L 115 R S R R E M R K V S I W S A G C A T G E E P Y S L A M V L --R R G S G E Y R V W S A A A S T G E E P Y S I A M T L
.. * . . . * . *

FrzF R F T S R R A I S I N Q A R L T R F F K P V E E G Y E A L
* * * * . .

CheR, -E V K T E I K K N I T F K K H N L L A D R Y E ----Q 193 CheR, V R V R Q E L A N Y V D F A P L N L L A K Q Y T V P --G FrzF P A L R -E ---Y I R F D G Q N L A V P V F D K V A L S
. * * * * * * * * *

FrzF S L D L I L C R N V I I Y F D L P T I R G L M D R F L A A C h e G P F D A I F C R N V M I Y F D Q T T Q Q E I L R R F V P L
* .
* * * * * * * ing frame encoding a 256-amino acid protein with a predicted molecular mass of 29,954 daltons (Fig. 1). The predicted amino acid sequence of this open reading frame was compared to proteins in GenBank and was found to have the highest percentage identity with the chemotactic methyltransferase CheRE (Mutoh and Simon, 1986). It was found to have 29% amino acid identity (Fig. 2). Based on the homology this open reading frame was named cheRB. CheRB is also homologous to FrzF, a methyltransferase of Myrococcus zanthus (McCleary et al., 1990), with 27% identity.

CheR, L K K N G V L F V G -S T E Q I F N -P E K -F G ---L 245 CheRE L K P D G L L F A G H S -E N -F S H L E R R F T ---
A putative &-promoter (Helmann and Chamberlin, 1987) was found just upstream of this open reading frame (Fig. 1  methyltransferase activity (Burgess-Cassler and , was complemented by the plasmid pAZ285, which expresses cheRB ( Table I). The complemented strain formed a wild type swarm on a TBr swarm plate (data not shown).

Directed Mutagenesis and Analysis of cheR-As
stated above, sequence analysis suggests that cheRR is monocistronic. This hypothesis is consistent with the location of cheRs between two genes whose products carry out functions entirely unrelated to chemotaxis (ndk and aroF). Furthermore, integration plasmids formed using DNA spanning the promoter proximal end of the gene, DNA from the middle, and DNA spanning the promoter distal end of the gene (Fig. 3) were subcloned into plasmids which cannot replicate in Bsubtilis. These plasmids were subsequently introduced into wild type B. subtilis to form 012682,012680, and 012681, respectively. In this type of analysis, one expects that if a promoter or terminator is contained in the DNA which is cloned in the integration vector, upon recombination, an uninterrupted operon will remain. However, if only internal DNA included, the operon will be disrupted (Piggot et al., 1984). As expected, only 012680, which was created using pAZ286 (Table I), showed a deficiency in CheR activity (see below) and no metabolic requirement for aromatic amino acids was created.
Analysis of cheRB Mutants-The integration strains 012680, 012681, and 012682 were tested for chemotaxis by the capillary assay and only 012680, having the integration plasmid with only the middle of the gene, was defective. Chemotaxis to the non-metabolizable proline analog, azetidine-2-carboxylic acid, was only 4% that of wild type (Table 11). This is similar to the taxis shown by the cheRR strain 011100, toward isoleucine (Ullah and Ordal, 1981). A null mutant containing a cat gene replacement of cheRR, showed the same taxis as 012680 (data not shown).
To confirm that strain 012680 was defective in methyltransferase activity, in uivo methylation experiments were carried out. Here, radiolabeled methionine is taken up by the cells, converted into S-adenosylmethionine, and used to methylesterify the MCPs in a reaction catalyzed by the CheR methyltransferase . As expected, only 012680 was defective in methylation. The wild type strain (011085) as well as strains 012681 and 012682 were normal for methylation (Fig. 4). MCPs (Burgess-Cassler and Ordal, 1982). Expression of cheRR from pAZ283 in an E. coli cheR null mutant, complemented the mutant as seen on a TBr swarm plate (Fig. 5). CheRB also methylated the MCPs in vivo (Fig. 6).  Behavioral Analysis Using Tethered Cells-Tethering analysis of wild type cells gives a very reproducible behavioral profile upon repetition under identical conditions. Upon tethering cheRB null mutant cells (0126801, quite variable profiles were obtained when randomly selected bacteria were analyzed. When the bacteria were categorized and analyzed as two separate behavioral subpopulations, all the bacteria unambiguously fell into one of the two categories. One subpopulation had a prestimulus flagellar rotation bias that was extremely CW, and the other had a more random bias (Fig. 7). To rule out the possibility of contamination, the cell cultures that were analyzed were derived from a single colony which itself was derived from several cycles of colony purification. Each colony always led to the development of the two subpopulations. The distribution of these cells in each subpopulation was determined by counting a large number of tethered cells (100-300) and categorizing them into one of the two subpopulations based on the observed prestimulus behavior. It was found that 33% of the cheRB population showed the extreme CCW bias while 66% showed the more random behavior. The time dependent behavioral profile of the two subpopulations appear to be superimposable, after normalizing for the initial bias (Fig. 71, indicating that the response to the addition and removal of attractant is independent of the prestimulus bias. Tethering analysis of other cheRB mutants (strains 011100 and 012652) also showed two subpopulations in approximately the same proportions as 012680 (data not shown).

Complementation of E. coli cheR Mutant-Zn vitro experiments have indicated that B. subtilis CheR can substitute for E. coli CheR in methylation of E. coli
AcheRB/cheBB double null mutant (013017) was analyzed by the tethering assay. Its behavioral profile was strikingly similar to that of the more random cheRB mutant subpopulation (data not shown). This suggested that the more random behavior of this subpopulation could be the result of a deficiency in CheBB activity. To test this hypothesis, wild type cheBB, on an expression plasmid (pMK108), was introduced into the cheRB mutant (012680) generating strain 013041. The distribution of the two subpopulations changed from 33% with an extreme CW bias to 66% of the population with this bias.
The behavior of strain 012680 was analyzed when subjected to the addition and removal of the repellent indole at a concentration of 0.82 m~ (Fig. 8). Only the more random subpopulation was analyzed since the more tumbly subpopulation had a prestimulus bias too low for a repellent response to be detected. The cells showed a rapid decrease in the CCW bias upon the addition of repellent but showed no adaptation. Upon removal of the repellent the CCW bias returned to the prestimulus level in the cheR mutant whereas the wild type showed a large increase in CCW bias, above the prestimulus level, upon removal of repellent (Fig. 8).

In Vitro Methylesteruse Activity in Whole Cell Extracts-
The results of the behavioral assays of the cheR mutant suggested that there may be a deficiency in the methylesterase activity due to the absence of the methyltransferase. To test this hypothesis, in vitro methylesterase assays were performed on whole cell extracts from wild type (011085), cheB mutant (0128361, and cheR mutant (012680). It was demonstrated that the radiolabel introduced into the 012680 membranes could be entirely released by incubation for 1 h in 0.1 M NaOH at 28 "C. The reaction containing the extract from the strain lacking any CheB (012836) (Kirsch et al., 1993) showed some release of methanol which may be attributed to spontaneous hydrolysis and nonspecific enzymatic release of methanol. After subtract- ing this non-specific methanol production from the rate of production determined for the wild type and cheR mutant it was found that the wild type cells showed a CheB-specific activity of 0.28% of total methyl group released per mdml of protein in whole cell extract&. The cheR mutant showed a CheB specific activity of 0.06%. Therefore the CheB activity in the extract from the CheR mutant is only 20% that of the wild type.

DISCUSSION
In this study we report the identification of a gene encoding the chemotactic methyltransferase in B. subtilis and show that it is homologous to the E. coli chemotactic methyltransferase (CheR) and to FrzF from M. ranthus. The DNA sequence suggests that cheRB is monocistronic and under the control of the vegetative sigma factor, OA. In addition, it has been shown that while the MCPs are under the control of 8, che& is not (Marquez et al., 1990). This is further strengthened by studies with la&-fusions with cheRB demonstrating that cheR is expressed vegetatively and not under the control of alternate sigma factors.2 Most of the other flagella and chemotaxis genes are located in a large operon which is also under the control of o"' (Zuberi et al., 1990;Bischoff and Ordal, 1992). This is in contrast to the chemotaxis and flagellar genes of E. coli, which are regulated by an alternate sigma factor (Helmann and Chamberlin, 1987;Arnosti and Chamberlin, 1988) The main areas of amino acid similarity are near the middle and near the C-terminal end. CheRB can replace CheRE in allowing chemotaxis on a TBr swarm plate (Fig. 5). The similarity was somewhat imperfect as methylation of the E. coli

MCPs by CheRB only approximated that in wild type E. coli.
Several features of chemotaxis in B. subtilis appear to be opposite to those of E. coli. Null mutants in cheYB and cheAB are tumbly (Bischoff and Ordal, 1991;Fuhrer and Ordal, 19911, whereas those in cheYB and che& are smooth swimming (Oosawa et al., 1988;Parkinson, 1978). It is likely that CheY-P causes smooth swimming in B. subtilis, whereas CheY-P causes tumbling in E.
In the same vein, a null mutant in cheRE is smooth swimming (Chen, 19921, whereas a null mutant in cheRB shows a more tumbly bias. It is believed that unmethylated MCPs in E. coli generate smaller CW signals because the "default" condition in E. coli (when there is no CheY-P) is CCW. By the same line of argument, we infer that unmethylated MCPs of B. subtilis generate weak CCW signals because in B. subtilis the "default" condition is CW rotation (Bischoff and Ordal, 1991).
Adaptation in E. coli is generally thought to be due to changes in the methylation level of MCPs in E. coli (Goy et al., 1977) and the same may be true in B. subtilis. CheB, which catalyzes removal of methyl groups, helps bring about adaptation to positive stimuli in B. subtilis (Kirsch et al., 1993) and to negative stimuli in E. coli (Hiroyuki et al., 1983). Methylation of the MCPs, by CheR,, should then be responsible for the adaptation to negative stimuli in B. subtilis. To test this, the cheR mutant 012680 was subjected to the addition and removal of the repellent indole. It showed no adaptation to the negative stimulus and returned to its prestimulus level only upon the removal the repellent (Fig. 8). Furthermore, in contrast to the wild type (011085), the cheR mutant showed no significant increase above the prestimulus CCW bias upon the removal of repellent. The wild type response to the positive stimulus of removal of repellent must therefore result from highly methylated MCPs. This is the opposite of the E. coli system where the   Ordal (1993) increase in the CCW bias that is seen upon the removal of repellent results from poorly methylated MCPs. Since, in the cheRs null mutant, there is no methylation of the MCPs (Fig. 4), there is no opportunity for CheB to remove methyl groups. However, partial adaptation still occurs in response to the addition of attractant stimuli (Fig. 7) as well as in the cheBB null mutant (Kirsch et al., 1993). Thus, the partial adaptation that does occur must be due to another mechanism which is independent of methylation (Kirsch et al., 1993). This partial adaptation, however, seems only to be involved in the adaptation to attractant stimuli as seen by the behavior of 012680 (Figs. 7 and 8).
A unique phenomenon has been observed, in that two different behaviors are demonstrated by the cheRB mutant. The fundamental difference between the two is that the prestimulus flagellar rotation bias is extremely CW in one of the subpopulations and more random in the other (Fig. 7). The behavior of the more random subpopulation is almost identical to that of a cheBBlcheRB double mutant (013017) (data not shown). This suggests that there may be a deficiency in CheBB activity in the cheRB mutants. Introduction of a n expression plasmid containing wild-type cheBB shifts the distribution of the subpopulations towards a larger proportion with the extremely CW rotating flagella. This implies that CheB activity may be diminished in the cheRB null mutant, and the introduction of pMK108 somewhat compensates for this deficiency.
To further test this hypothesis, methylesterase assays were performed on whole cell extracts. The cheR mutant showed an 80% reduction in CheB activity as compared to the wild type. It was found that cheR mutants of E. coli and Salmonella typhimurium were also deficient in methylesterase activity, showing a 45% decrease and a 55% decrease, respectively (Stock and Koshland, 1978). Perhaps CheBB and CheRB form a complex that stabilizes CheBB from degradation. Experiments are now being done to further study possible interactions between CheBB and CheRB.
It appears that in B. subtilis, the higher the level of methylation of the MCPs, the higher the prestimulus CCW flagellar rotation bias.2 This is the opposite of the E. coli system (Borkovich et al., 1992). It is therefore likely that the true behavior of a strain lacking only CheR activity would be represented by the more tumbly subpopulation. The behavior in the more random subpopulation may result from a deficiency in CheB activity which would limit or prevent the deamidation of glutamine residues that is believed to be performed by CheBB as in E. coli (Sherris and Parkinson, 1981;Kehry et al., 1983;Terwilliger and Koshland, 1984). Since the glutamine residues function similarly to methylated glutamates (Dunten and Koshland, 1991), the signal sent by unstimulated MCPs would promote higher CheAB activity than the unmethylated, deamidated MCPs of the more tumbly cheRB mutant subpopulation. cloned DNA containing cheRB. We thank Dr. J. S. Parkinson for several