Chemotactic Methylesterase Promotes Adaptation to High Concentrations of Attractant in Bacillus subtilis”

The Bacillus subtilis gene encoding CheB (cheBB), the chemotactic methylesterase, has been sequenced. The 39-kDa protein which resulted from the expression of cheBB, using a T7 expression system was con- sistent with the predicted open reading frame. CheBB shares 39.6% identity with Escherichia coli CheBE and can complement a cheBE null mutant. CheBB is re- quired for removal of methyl groups from the receptors upon attractant stimulus and appears to play an impor- tant role in adaptation to the addition of attractants, whereas CheBE plays an important role in adaptation to the addition of repellents. Unlike the cheBE and cheRE mutants of E. coli, which show extreme flagellar rotational biases, the unstimulated cheBB mutant showed a normal (wild type) bias. Upon addition of attractant, the cheBB null mutant showed a counterclockwise bias that was higher than for wild type and demonstrated only partial adaptation. In the capillary assay for the attractant azetidine-2-carboxylic acid, the mutant gave a wild type response at low concentrations but a very reduced response at high eoncentra-tions. We conclude that B. subtilis has an effective methylation-independent adaptation system but must utilize the methylation system for adaptation to high concentrations of attractant.

in unstimulated cells (for a review, see Refs. 2 and 3). The importance of CheB in performing this function is underscored by the fact that mutants in c h B E always tumble (4) due to a strong bias in clockwise flagellar rotation (5).
Like E. coli, Bacillus subtilis has CheAB, CheBB, CheYB, and MCPs. However, null mutants in cheY~ and null mutants in C~A B are very tumbly whereas the corresponding mutants of E. coli are very smooth swimming. Therefore, unlike E. coli, it appears that attractants activate CheAB to lead to smooth swimming behavior. Furthermore, the MCPs of B. subtilk are larger than those of E. coli, having apparent molecular masses greater than 70 kDa,2 compared with 58-60 kDa for E. coli (6). Addition of attractants such as aspartate causes an increased turnover of methyl groups on all three MCPs (7) in a reaction requiring CheAB.3 That is, methyl groups are transferred away and replaced by new methyl groups from S-Adenosylmethionine. The identity of the methyl acceptor is unknown. However, on removal of the attractant, the lost methyl groups may return to the MCPs (8).
In E. coli adaptation to positive and negative stimuli involves changing the degree of methylation of the MCPs. The importance of the degree of methylation of these proteins is clear from the phenotype of strains unable to methylate or demethylate them. C~R E mutants are smooth swimming, and chBE mutants are tumbly. The double mutant is able to move on a swarm plate and shows some degree of chemotaxis in capillary assays. These bacteria do not adapt very well to moderate or large stimuli but do adapt to weak stimuli (9-11). It is thus apparent that, although the methylation system is very important, it is to some extent dispensable.
In this article, we report the sequence of B. subtilis chemotactic methylesterase, cheB~, and characterization of a null mutant in c~B B .
Our results indicate that CheBB plays a significant role in adaptation, but not in excitation, in relation to attractants. There appears to be a very effective methylation-independent adaptation system in B. subtilis that operates for low concentrations of attractant. Indeed, in chemotaxis toward low concentrations of attractant, as measured in the capillary assay, the cheBB mutant was as efficient as wild type.

MATERIALS AND METHODS
Bacterial Strains and Plasmids-The bacterial strains and plasmids used in this investigation are described in Tables I and 11, respectively.
Growth Media-Tryptone broth is 1% Tryptone and 0.5% NaCl.  Mutagenesis-Two mutations in C~B B were created, resulting in strains 012715 and 012836. Strain 012715 was created by digesting pMKl00 at a unique SpeI restriction site, (Fig. 1) located in the first third of che& and filling in the 3' recessed ends with Sequenase enzyme in the presence of 20 p~ deoxynucleotide triphosphates. The resulting blunt ends were subsequently religated, resulting in a nonsense mutation. The mutation was then crossed onto the B. subtilis chromosome by gene conversion (14) yielding a truncated c~B B .
Strain 012836 was created by first introducing two StuI restriction sites, using oligonucleotide-directed in vitro mutagenesis, at each end of c~B B , in pMK100. The resulting 1.1-kb StuI fragment containing C~B B in ita entirety was removed. The vector with the adjacent B. subtilis DNA was religated to create pMK113. The oligonucleotides used in the site-directed mutagenesis were designed so that after religation, the expression of the adjacent genes would not be affected. In order to facilitate the transfer of the mutated DNA onto the B. subtilis chromosome, an EcoRI restriction site was introduced between orf298 and CMB (Fig. l). The resulting 1.1-kb PstI fragment was then subcloned into pMK116, to create pMK117. A promoterless chloramphenicol resistance marker (chloramphenicol acetyltransferase, cat gene) (15) was inserted into the EcoRI restriction site of pMK117, to create pMK118. The mutation was then crossed onto the B. subtilis chromosome by gene conversion (14).
Expression of chBB-The method of Tabor and Richardson (16) was used to express c~B B .
pT7-5 derivatives were introduced into competent E. coli K38 cells. Subcloned genes were expressed after induction at 42 "C in the presence of rifampicin and ~-[~~S ] m e t h i onine. The labeled proteins were electrophoresed on a 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and visualized by autoradiography.
Complementation of C~B B Mutants-Complementation of the mutanta was performed by transforming pMK108, which is an expression vector containing wild type c~B B , into competent 012836. Complementation was assayed on semisolid agar swarm plates containing 0.27% agar (17). Isopropyl-1-thio-8-galactopyranoside (IPTG) was added in varying concentrations to induce the expression of chBB. The swarm plates were incubated for 9 h at 37 'C. The swarm size of the complemented strain was compared to that of the wild type.
In Vivo Methylation-Zn vivo methylation experiments were performed as previously described (18). The in vivo cold chase experiment was performed as previously described (8).
Capillary Assay-Capillary assays have been described (17). Cells were grown overnight on tryptose blood agar base plates, resuspended at l@ cells/ml in tryptone broth, and diluted 50-fold into minimal media. After 4 h, growth cells were supplemented with 0.05% glycerol and 5 mM sodium lactate, grown 15 min longer, and harvested. They were resuspended at ODm = 0.001 and assayed for chemotaxis. Cells accumulated in the capillary tubes were plated out and counted. The assay was performed in triplicate.
Tethering Analysis-The method used to tether the bacteria has been described (19, 20). Cells were diluted 1:100 from an overnight tryptone broth 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 was videotaped on a Panasonic 1960 VCR using a Panasonic BL200 videocamera and a Zeiss standard RA phase-contrast microscope. Each bacterium was then analyzed separately. In order to analyze the behavior, a computer program was written to obtain a time-dependent behavioral profile of a collection of bacteria. Analysis would continue throughout the entire 8-min period during which chemoeffectors or buffer would be passed over the bacteria. Each time a bacterium would change the direction of rotation, a particular key would be pressed on the computer keyboard. Each keystroke would record the time and direction of each change. The behavior was thus digitized by giving a clockwise rotation a value of 1 and a counterclockwise rotation a value of 0. After an entire set (15-30 bacteria) was entered, the program then averaged the behavior of the set at 4-s intervals. The values obtained were then smoothed (width of 5) and plotted using the Cricket Graph graphing program on a Macintosh IIcx computer.
Nucleotide Sequence Accession Number-The nucleotide sequence data reported in this paper have been submitted to GenBank.

Nucleotide Sequence-A
2.1-kb PstI restriction fragment was subcloned from a 10.9-kb EcoRI restriction fragment obtained from a Xcharon4A library ( Fig. 1) (21). Through recombination experiments, it was determined that this 2.1kb PstI fragment contained c~B B (formerly called cheL) (22). This DNA fragment was sequenced in its entirety on both strands and was found to contain an open reading frame encoding a protein of 357 amino acids (Fig. 2). It also contained fragments of two other genes, an upstream gene, orf298, encoding a protein with no known E. coli homolog, and a downstream gene, (23). The orientation of the open reading frame agrees with the direction of transcription of the major operon (22). A potential ribosome-binding site appears 10 base pairs upstream of a UUG start codon. The ribosomebinding site is preceded by the stop codon of the upstream gene, orf298 (Fig. 2). The termination codon of cheBB is downstream from the potential ribosome binding site of cheAB (Fig. 2).
Homology to CheB of E. coli and FrzG of Myxococcus m nthus-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 methylesterase CheBE (24). It was found to share 39.5% identity along its entirety (   Expression of chBB-The 2.1-kb PstI fragment was subcloned in the coding orientation of pT7-5, resulting in pMKllO (Table 11). Expression of this fragment produced a protein having an apparent molecular mass of 40 kDa (Fig.  4). This agrees well with the predicted molecular mass of 39.2 kDa, based on the amino acid sequence, deduced from the DNA sequence. Another expression plasmid was constructed which contained a truncated form of chBB from pMKlll (Table 11). The last 66% of the protein was deleted by the creation of a nonsense mutation. Upon expression of this truncated gene, a protein of approximately 12.5 kDa is visualized (Fig. 4), which corresponded very well with the predicted size of 13.5 kDa. A band migrating at 18 kDa resulted from the expression of the 5' end of CMB fused to the vector. The predicted molecular mass of this fusion protein was 16.4 kDa. Expression of this fusion was unaffected by the C~B B mutation (Fig. 4). No proteins were observed when pT7-5 itself was expressed (data not shown).

. T C A G A A A T T A T C T T C T G~C T T A A T
Complementation of the c h e B~ Mutants-Two C~B B mutants were made. One, strain 012715, allowed the expression of the first 34% of C~B B (Fig. 4). The other, 012836, contained no DNA encoding the methylesterase but left the upstream and downstream genes intact, as confirmed by complementation. Both mutants showed identical in vivo meth-ylation of MCPs (Fig. 5, data not shown), and swarm plate morphologies (Fig. 6, data not shown), indicating that the two mutations caused the same defects. The corresponding mutations were verified by sequence analysis and Southern hybridization (data not shown).
In the absence of CheBE, MCPs cannot undergo enzymatically catalyzed demethylation (26). Thus in an in uiuo methylation experiment, in a chBB mutant, the only way that radiolabeled methyl groups can replace the existing unlabeled methyl groups is by the spontaneous demethylation of the glutamate methylesters: As expected, the level of labeling of MCPs was significantly reduced but not absent in both strain 012836 and strain 012715 (Fig. 5, data not shown).
The complementation as observed by MCP methylation in strain 012837 was 88% of wild type when 1 mM IPTG was added during the growth of the culture. This is a large increase over the amount of methylation in 012836, which was only 17% of the wild type level. (Fig. 5).
To ensure that the mutations created in C~B B were nonpolar on other genes required for chemotaxis, the mutations were complemented with pMK108 (Table 11), which expressed only cheB~. The swarm of the complemented strain (012837) was similar to that of wild type but smaller (Fig. 6). The complementation was also verified by tethering analysis. The complemented strain demonstrated wild type behavior (data ' F. Dahlquist, personal communication.

~I R V L V~D D S A F~R K~I S D F L T E E K~I E V n~~~v~s v~~s~~n~q~n r~~~~s n s~n~n ~~F R V L~V C K C --L~A L~~~C L F~C E~L V P
. * * * * .

V A T A P D P -L V A R D L I K K F N P D V L T L D V E~P I G T A R N G -E E A L K K I E L L K P D V I T L D V E~P V G P A E V D F A G A L V A V Q R H F P D V V -L -V D L S
. . . * * . . . . . . *

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

E C T I N C L E I G A F D F I T K P S G S I S L D L Y K I K
* * * . . . *

P G G K N I S V I K N S E G L Q V V L D N H D T P S R H K P P S G S H L --L V P P E G -R L E L D A G P A L R G F R P
. . *

P G D R H~E L S R S G A N Y~I K I H D G P A V N R H R P
. . * . * . .

A G L K D~L T A G N V K A I A E S E E S C V V Y G~P K A A G~L A~R~A G A -~T L A~N E A S C V V F G~P R E R G L K E I R E R G G -R T I A q D E A S S V V W G N P R E
* t * CheBr, and M. x a n t h s FrzG. Alignment of CheBe with each protein was performed using the AALIGN program from DNASTAR. The two alignments were then combined manually to optimize the apparent homology among the three proteins. Amino acids that appear in all three proteins are designated with a star (*). A dot (.) represents a match between CheBB and CheBE only.

A V K A G L I H E I K H V E D I * A S I T S C V K K E R V
not shown). To further verify the complementation (rather than recombination repair), the strain was cured of the plasmid (resulting in strain 012838) and the mutant phenotype was restored (Figs. 5 and 6). Furthermore, our previous experience with this cat insert indicates that it does not cause polar effects (27, 28). Complementation of the cheBE Mutants-In view of the high amino acid sequence homology between CheBB and CheBE, we wanted to test the functional homology. Previous studies showed that the methylesterase of B. subtilis could remove methyl groups from E. coli MCPs in vitro (29). To determine whether it could function in vivo in E. coli, a C~B E strain, RP4310 was transformed with pMK108. In the presence of IPTG, the defect in c~B E , was complemented, as determined by swarm plate analysis (Fig. 7).
Effect of Attractant on Methyl Group Turnover in cheBB Mutant-To assess whether attractant could cause an increased turnover of methyl groups in the cheBB null mutant, 0.1 M Asp was added to the wild type and to the mutant in a cold chase experiment. Although, the attractant caused a substantial decrease in labeling of the wild type MCPs, it had little effect on the MCPs in the mutant (Fig. 8).
Tethering Analysis-To investigate the role of CheBB in the process of excitation and adaptation to chemotactic stimuli, tethered cells of wild type and strain 012836 were sub-jected to the addition and removal of the attractant azetidine-2-carboxylic acid, a non-metabolizable proline analog (30) and the repellent indole. The behavior of the wild type (011085) had several noteworthy features (Figs. 9 and 10). As expected, positive stimuli (either addition of attractant or removal of repellent) caused an increase in the bias of counterclockwise flagellar rotation. Negative stimuli (either addition of repellent or removal of attractant) resulted in a decrease in the bias of counterclockwise rotation. The adaptation period to positive stimuli was shorter than the adaptation period to negative stimuli by about a factor of four. The B. subtiilis population never reached a 100% counterclockwise bias in response to the addition of attractant. Finally, there appeared to be a slight overshoot, a period of excess clockwise rotation following adaptation.
In the absence of stimuli, the c h e B~ mutant showed a baseline level of counterclockwise rotation similar to the wild type. Addition of attractant caused a higher than wild type level of counterclockwise rotation of the flagella (Fig. 9). After the 95% counterclockwise bias was achieved (compared with 85% for wild type), the bacteria partially adapted. The extent of adaptation depended inversely on stimulus strength. The bacteria adapted much more completely to the lower concentration of attractant than to the higher concentration (Fig.  9). Prolonged exposure to 10 p~ azetidine-2-carboxylic acid showed that a 75% bias was maintained, even after 8 min (data not shown). Upon removal of the attractant, the bacteria showed a counterclockwise bias far below the base-line value. Interestingly, the onset of this low counterclockwise bias was slower and less extreme for removal of the higher concentration of attractant. The chBB mutant was also subjected to repellent stimuli. When 3.15 mM indole was added to the cheBB mutant, it responded like the wild type. However, when indole was removed, the mutant adapted more slowly (Fig.  10). Thus, CheBB appears to be required for normal adaptation to positive stimuli.
Capillary Assay-To further analyze the cheBB mutants ability to perform chemotaxis at different concentrations of attractant, capillary assays were performed. At low concentrations of attractant, the mutant was capable of performing chemotaxis as well as the wild type (011085) (Fig. 11). However, as the concentration of attractant was increased, the efficiency of chemotaxis decreased greatly.

DISCUSSION
Homology of Methylesterases from Several Bacteria-In this article, we report the identification of a gene encoding the chemotactic methylesterase in B. subtilis and show that it is homologous to the E. coli chemotactic methylesterase in both structure and function. The amino acid sequences of CheBB and the E. coli and M. xanthw counterparts were compared (Fig. 3). There are conserved regions among all three of the distantly related proteins. While there are regions that share identity between any two of the proteins, the regions where all three show identity can be indicative of regions that are likely to be essential for proper functioning. These highly conserved regions occur uniformly throughout all three proteins, in both the regulatory (NHz-terminal) and in the enzymatic (COOH-terminal) regions (31).
Requirement of C~B B for Removal of Methyl Groups from MCPs-A C~B B mutant would be expected to show a deficiency in the level of radioactive labeling of the MCPs in an in vivo methylation experiment. In order to radiolabel the MCPs, non-radioactive groups added during growth must first be removed. Thus, if the enzymatic machinery to remove these is disabled, then the MCPs will remain unlabeled fol-  lowing addition of radioactive methionine. As expected, the C~B B mutant showed poor labeling of the MCPs (Fig. 5). The low level (17% of wild type) of methylation that does occur can be attributed to spontaneous demethylation of the MCPs. A 10% level of methylation has been observed in c h e B~ mutants: It is unlikely that there is a "back-up" methylesterase since in a continuous flow assay (32, 33), no increase in methanol production was seen in response to the addition of attractant to the c~B B null mutant strain (012836) as occurs in the wild type strain (data not shown).
CheBB, like its E. coli counterpart, is presumably activated by phosphoryl transfer, for it shares the conserved Asp residue which is phosphorylated in E. coli. In B. subtilis, addition of attractant causes an increase in methyl group turnover on the MCPs. If this turnover occurs due to activation of CheBB by CheAB, then it should not occur in the methylesterase mutant, and indeed it did not (Fig. 8).
Role of ChBB in Adaptation-Tethering analysis indicates that CheBB plays a role in adaptation to attractant (Fig. 9) as the C~B B mutant remained somewhat biased to smooth swimming until attractant was removed. Thus, the removal of methyl groups from the MCPs by CheBB contributes to the adaptive response to addition of attractants, rather than to the addition of repellents as in E. coli (26). The   Azetidine-2-carboxylic acid was added at 1 min and then removed at 4 min. 011085 (exposed to 10 p~) is represented by the heauy solid line, 012836 (exposed to 10 p~) is represented by the thin solid line, and 012836 (exposed to 2.5 p~) is represented by the dotted line. CheYE-P is widely held to cause tumbling (35-37). As in E.
coli, activation of CheAB may cause increased phosphorylation of both CheYB and CheBB. In B. subtilis, however, this may be due to addition of attractant rather than addition of repellent. Adaptation to positive stimuli is more rapid than to negative stimuli in E. subtilis. This probably reflects the activation of CheBB by phosphorylation. By analogy, CheR, which is not known to be activated in E. coli (38), presumably helps bring about the slower adaptation to removal of attractants in B. subtilis.
The capillary assay (Fig. 11) shows that for lower concentrations of azetidine-2-carboxylic acid (0.1 mM and below), CheBB is not needed for chemotaxis. In fact, in this concentration range, chemotaxis is equal in the wild type and in the c h B B mutant. Concomitantly, addition of lower concentrations of azetidine-2-carboxylic acid to tethered cells shows more nearly complete adaptation than does addition of higher concentrations. It would seem that there exists in B. subtilis a methylation-independent adaptation that is very successful at low concentrations of attractant, but not at high concentrations. Indeed, at 0.1 M azetidine-2-carboxylic acid, the accumulation in the capillary assay was 50-fold less than for wild type. It thus appears that the purpose of the adaptation system involving methylation in B. subtilis is to allow the '~m CheBB does not play a direct role in excitation (Fig. 9). Both