Oligomerization of the phosphatase CheZ upon interaction with the phosphorylated form of CheY. The signal protein of bacterial chemotaxis.

Earlier studies have suggested that CheZ, the phosphatase of the signaling protein CheY in bacterial chemotaxis, may be in an oligomeric state both when bound to phosphorylated CheY (CheYP) (Blat, Y., and Eisenbach, M.(1994) Biochemistry 33, 902-906) or free (Stock, A., and Stock, J. B.(1987) J. Bacteriol. 169, 3301-3311). The purpose of the current study was to determine the oligomeric state of free CheZ and to investigate whether it changes upon binding to CheYP. By using either one of two different sets of cross-linking agents, free CheZ was found to be a dimer. The formation of the dimer was specific, as it was prevented by SDS which does not interfere with cross-linking mediated by random collisions. The dimeric form of CheZ was confirmed by sedimentation analysis, a cross-linking-free technique. In the presence of CheYP (but not in the presence of non-phosphorylated CheY), a high molecular size cross-linked complex (90-200 kDa) was formed, in which the CheZ:CheY ratio was 2:1. The size of the oligomeric complex was estimated by fluorescence depolarization to be 4-5-fold larger than the dimer, suggesting that its size is in the order of 200 kDa. These results indicate that CheZ oligomerizes upon interaction with CheYP. This phosphorylation-dependent oligomerization may be a mechanism for regulating CheZ activity.

Bacteria such as Escherichia coli or Salmonella typhimurium use chemotaxis to navigate toward favorable environments and retreat from non-favorable ones (1). The sensory information from the receptors is integrated by a cytoplasmic signal transduction network of chemotaxis proteins (see for review, Refs. [2][3][4][5] and transmitted to the flagella by the signaling molecule CheY (6,7). CheY interacts with the switchmotor complex at the base of the flagellum (6 -10) and changes the direction of rotation from the default direction, counterclockwise (9 -17) to clockwise (12,15,17), and thereby causes the cell to reorient (18). The clockwise causing activity of CheY is regulated by phosphorylation (19). The phosphorylation level is determined by the kinase CheA and the phosphatase CheZ (20 -22). The activity of the kinase, CheA, is modulated by chemotactic stimuli via the membrane chemotaxis receptors and the chemotaxis protein CheW (23)(24)(25). On the other hand, regulation of CheZ activity by chemotactic stimuli has not been demonstrated.
In a previous study we found that the binding of CheZ to phosphorylated CheY (CheYϳP) is 2 orders of magnitude higher than to non-phosphorylated CheY, and that several molecules of CheZ can bind to a single CheYϳP molecule (26). Earlier observations indicated that CheZ can be in two oligomeric forms, 115 and Ͼ500 kDa, as estimated by size-exclusion chromatography (27) (the molecular size of the monomer is 23.9 kDa (27)). The observations of both studies taken together suggest that the oligomeric state of CheZ is modulatable. Here we examine this possibility and demonstrate that CheZ is a dimer which oligomerizes upon interaction with CheYϳP. The possibility that this phosphorylation-dependent oligomerization is a regulation mechanism for CheZ activity is investigated in a subsequent work (28).
Protein Radiolabeling-Radiolabeling of CheY E and CheZ E was carried out by inducing the expression of CheY and CheZ in the presence of L-[ 14 C]leucine as described previously (26).
Protein Purification-The purification of CheY E (nonlabeled) and CheZ E (nonlabeled and radiolabeled) was carried out on a Cibacron column followed by a G-50 (for CheY) or Sepharose CL-6B (for CheZ) column as described previously (26). Radiolabeled CheY E was purified as the non-labeled CheY E , only that smaller-scale columns (4 ml of Cibacron and 65 ϫ 1 cm G-50) were used. CheY S was purified as CheY E * This study was supported by Grant 93-00211 from the United States-Israel Binational Science Foundation (BSF), Jerusalem, Israel. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This paper is dedicated to Julius Adler on the occasion of his sixtyfifth birthday.
‡ Incumbent of Jack and Simon Djanogly Professorial Chair in Biochemistry. To whom correspondence should be addressed: Dept. of Membrane Research and Biophysics, The Weizmann Institute of Science, 76100 Rehovot, Israel. Fax: 972-8-344112; Tel: 972-8-343923; E-mail: bmeisen@weizmann.weizmann.ac.il. (26), except that the bacteria were grown on Luria broth and induced with 0.66 mM isopropyl-␤-D-thiogalactopyranoside. CheZ214FC was purified as follows. RP1091 cells containing pEWC1 were grown at 35°C in 1.5 liter of Luria broth containing 100 g/ml ampicillin. At OD 590 ϭ 0.4, overexpression was induced by the addition of 1 mM isopropyl-␤-Dthiogalactopyranoside. After 4 h, the cells were harvested by centrifugation, washed once by buffer A (20 mM HEPES, pH 7.4, 2 mM dithiothreitol, and 0.2 mM phenylmethylsulfonyl fluoride), resuspended in 35 ml of buffer A, and sonicated. Membranes and unlysed cells were removed by centrifugation at 165,000 ϫ g for 60 min. The lysate was loaded onto a 40-ml Sepharose CL-6B column pre-equilibrated with buffer A. The column was washed with 100 ml of buffer A, followed by 170 ml of buffer A containing 275 mM NaCl. CheZ was eluted from the column by a 250-ml linear gradient of 275-450 mM NaCl in buffer A. The CheZ-containing fractions were pooled and concentrated by ultrafiltration through a 5-kDa cut-off membrane, using an Amicon chamber (model 52). The concentrated fractions were loaded onto a 20-ml hydroxylapatite column, pre-equilibrated with buffer B (10 mM NaP i , pH 7.0, 2 mM dithiothreitol, and 0.2 mM phenylmethylsulfonyl fluoride). The column was washed with 60 ml of buffer B, and CheZ was eluted with a 200-ml linear gradient of buffer B containing 10 -200 mM NaP i . The CheZ-containing fractions were concentrated to a volume of 2 ml and loaded onto a Sephadex G-150 column (50 ϫ 1.5 cm), pre-equilibrated with a solution of 50 mM Tris-HCl, pH 7.9, 200 mM NaCl, 2 mM dithiothreitol, and 0.2 mM phenylmethylsulfonyl fluoride. The CheZcontaining fractions were concentrated and stored in 10% glycerol at Ϫ80°C.
Size Estimation of CheZ by Zonal Centrifugation-CheZ E , CheY E (14 kDa), CheW (18 kDa), and ovalbumin (43 kDa) samples (150 l of 2 mg/ml) were loaded onto a 5-20% isokinetic sucrose gradient (5 ml) in Tris-HCl (50 mM, pH 7.9), KCl (100 mM), and MgCl 2 (5 mM). The samples were spun at 290,000 ϫ g for 24 h (20°C) in a Beckman SW50.1 rotor. After centrifugation, 130-l fractions were collected from the top of the gradient and analyzed for protein concentration by the Bradford technique (33). The size of CheZ E was estimated by plotting a calibration line of (molecular size) 0.67 against the distance traveled by the protein peak (34).
Fluorescence Depolarization of CheZ-Fluorescence depolarization studies with fluorescein-labeled CheZ214FC were carried out with a Perkin Elmer LS 50 B luminescence spectrometer equipped with a fluorescence polarization accessory. The excitation and emission wavelengths were 490 and 520 nm, respectively (5-nm slit width).
Phosphatase Activity of CheZ-The phosphatase activity of CheZ was assayed by monitoring the steady-state level of CheY phosphorylation in the presence of [ 32 P]acetyl phosphate (AcP, synthesized as described (35)) and varying concentrations of CheZ. CheY (20 M) was incubated in Tris-HCl (50 mM, pH 7.9) and MgCl 2 (5 mM) with [ 32 P]AcP (20 mM, 200 -400 cpm/pmol) and varying concentrations of CheZ. The reaction (20 l) was quenched after 10 min of incubation at room temperature (25°C) by 100 l of 10% ice-cold trichloroacetic acid and 20 g of ovalbumin. (We verified that trichloroacetic acid does not hydrolyze CheY˜P under our experimental conditions.) The level of CheY phosphorylation was determined as described earlier (26).

CheY Phosphorylation Leads to Formation of a High Molecular Size CheZ-CheY
Complex-In order to examine the possibility of phosphorylation-dependent oligomerization of CheZ, we monitored the oligomeric state of CheZ by cross-linking in the presence or absence of CheYϳP. When cross-linked by DMS which cross-links amines (36), some of the CheZ molecules appeared to form two bands migrating on SDS-PAGE as 43and 49-kDa proteins (Fig. 1, lane 2). These bands probably correspond to CheZ dimers (the monomeric size of CheZ is 23.9 kDa). The two distinct bands of the dimer possibly reflect two different positions of DMS links between the CheZ monomers. The addition of CheY (14 kDa) did not cause any change in the cross-linking products observed (lane 4). However, the addition of CheY together with the phosphodonor AcP (37) caused the appearance of two additional cross-linking products of 38 and Ͼ94 kDa (lane 5). (A more accurate determination of the size of Ͼ94 kDa product is made below.) AcP in the absence of CheY had no effect (lane 3). The 38-kDa band probably corresponds to a 1:1 CheY-CheZ complex which reflects the increase in binding of CheZ to CheY upon phosphorylation (26). The Ͼ94 kDa band may be an oligomeric complex of CheZ which possibly contains also CheY (see below). In order to verify that the CheZ dimer and the oligomeric complex are not cross-linking artifacts, we used an additional cross-linker and cross-linking-free techniques.
A mixture of EDC and NHS mediates the cross-linking of amines to carboxyl groups (38). CheZ exposed to this mixture formed a single defined band at 46 kDa (Fig. 2A, lane 2). In the presence of non-phosphorylated CheY, most of CheZ remained as a dimer (lane 3). A sharp change in the oligomeric state of CheZ was observed when CheY was added together with AcP. Under these conditions all the CheZ formed a Ͼ94 kDa crosslinking product (lane 4), which appeared on a 10% SDS-PAGE as a 90 -200-kDa smear and a distinct band at 145 kDa (Fig.  2B). As in the case of DMS, AcP in the absence of CheY did not form the Ͼ94-kDa complex (Fig. 3A, lane 3). These results confirm that an oligomeric complex of CheZ is formed in the presence of CheYϳP.
To confirm that the cross-linking products observed in this study resulted from specific interactions, we repeated all the above mentioned experiments in the presence of SDS (0.5%). (SDS does not interfere with cross-linking mediated by random collisions (39).) Only the monomeric form of CheZ was observed; the formation of all the cross-linking products was prevented by SDS (data not shown), indicating that the interactions were specific.
The EDC-NHS-mediated cross-linking of CheZ (Fig. 2) was much more efficient than the DMS-mediated cross-linking (Fig.  1). This may be the consequence of the small number of lysine residues and the high number of aspartate and glutamate residues in CheZ (6 residues versus 37, respectively (40)). For this reason the subsequent cross-linking studies were carried out with EDC-NHS.
If the phosphorylation-dependent oligomerization of CheZ is physiologically significant, it should depend on the concentration of CheYϳP. As shown in Fig. 3, A and B, the amount of the oligomer was indeed dependent on the concentration of CheY added in the presence of access AcP. Furthermore, CheY57DE which cannot be phosphorylated due to the substitution of Glu for the phosphorylation site, Asp 57 (26,41), did not promote the oligomer formation even in the presence of AcP (Fig. 3C, lane  7). Similarly, depletion of Mg 2ϩ (necessary for CheY phosphorylation (42)) also prevented the formation of the oligomer (lane 5).
The Amount of CheY in the Oligomer-In order to estimate the relative content, if any, of CheY in the oligomer, we carried out the cross-linking experiment with radiolabeled CheY or radiolabeled CheZ in parallel (Fig. 4). [ 14 C]CheY cross-linked in the presence of CheZ and AcP (Fig. 4, lane 4) indeed formed a band migrating at the same position as the CheZ oligomer (lane 1). This band was not formed in the absence of CheZ and AcP (lane 3; the faint high molecular size band seen in this lane represents aggregates too large to enter the gel). This result indicates that the CheZ oligomer contains also some CheY cross-linked to it. We quantified the amount of CheY crosslinked to the CheZ oligomer by a PhosphoImager. The result was 0.48 Ϯ 0.18 (mean Ϯ S.D., three independent determinations) molecules of CheY per monomer of CheZ, suggesting a CheZ:CheY ratio of 2:1.
Estimation of the Oligomeric Status of CheZ in the Absence of CheYϳP by Sedimentation Analysis-Our observation that, in the absence of CheYϳP, CheZ appears as a dimer in crosslinking experiments, is in apparent conflict with Stock and Stock's (27) observation, reproduced by us, that CheZ behaves as a tetramer when run on an HPLC size-exclusion column. In order to solve this discrepancy we estimated the size of CheZ by zonal centrifugation on a sucrose gradient. As shown in Fig. 5, the CheZ peak appeared between the peaks of CheW (18 kDa) and ovalbumin (43 kDa) at a location which, according to the formula described under "Experimental Procedures," is the site of a 35 Ϯ 4 kDa (mean Ϯ S.D., two determinations) globular protein. This estimation is closer to the cross-linking results, which indicated that CheZ is a dimer (47.8 kDa), than to the HPLC results which indicated a tetramer. The difference between the size estimations of the different approaches suggests that CheZ is non-globular (see "Discussion").
Estimation of the Oligomeric Status of CheZ in the Presence of CheYϳP by Fluorescence Depolarization-To confirm the oligomerization of CheZ in the presence of CheYϳP, we initially attempted to use size-exclusion chromatography. Running a mixture of CheZ, CheY, and AcP through the HPLC column yielded only two peaks, eluted at retention times identical to those obtained when CheZ and CheY were run separately. This could mean either that, in contrast to the crosslinking results, the CheYϳP-induced oligomerization does not occur at all, or that the oligomer is unstable and readily disso- ciates when CheY and/or AcP are separated from CheZ on the column. To distinguish between these possibilities, we used fluorescence depolarization, which, unlike size-exclusion chromatography and sedimentation analysis, allows interaction between CheZ and CheYϳP throughout the experiment.
To enable the measurement of the rotational diffusion of CheZ by fluorescence depolarization, we had first to fluorescently label CheZ. For this purpose we replaced phenylalanine 214 of CheZ with cysteine by using a mismatch primer and PCR, as described under "Experimental Procedures." The mutated protein, CheZ214FC, retained normal CheY phosphatase activity (Fig. 6) and therefore could serve as a model for wildtype CheZ. CheZ214FC contains a single cysteine residue and therefore could be labeled at a specific site by fluorescein-5maleimide. The anisotropy, which reflects the rotational diffusion and therefore also the size of the fluorescein-labeled CheZ214FC (43,44), was dependent on the concentration of added CheY (Fig. 7A). In the absence of AcP, the anisotropy increased only moderately with the concentration of CheY and was not saturated within the concentration range of the experiment (0 -45 M CheY). This moderate change in anisotropy was probably the result of the low-level binding of non-phosphorylated CheY to CheZ (26). In the presence of AcP, the anisotropy was increased over 2-fold, indicating a large decrease in the rotational diffusion of CheZ upon CheY phosphorylation. The increase in anisotropy was saturated already at 4 M CheY. The anisotropy change was rapid; it was completed prior to the first measurable time point (ϳ30 s). In order to estimate the magnitude of the size change, the anisotropy of CheZ was measured at varying viscosity values in the presence or absence of CheYϳP. The data, presented in the form of a Perrin plot (43,44), are shown in Fig. 7B. In the Perrin plot the slopes of the curves are inversely related to the molecular dimension and therefore to the molecular size (43,44). The slope was 4.8-fold smaller in the presence of CheYϳP than in its absence, suggesting that upon CheY phosphorylation CheZ formed a complex which is approximately 4 -5-fold larger in volume than the CheZ dimer. The fact that we did not detect CheYϳP-induced oligomerization of CheZ in size-exclusion chromatography, but did observe it using fluorescence depolarization, suggests that the oligomer is unstable and dissociates in the absence of CheYϳP.

DISCUSSION
In this study we have shown that CheZ is a dimer which further oligomerizes upon interaction with CheYϳP. The significance of these findings is discussed below.
The different approaches used in this study to probe the molecular size of CheZ have yielded conflicting results. When run on a size-exclusion column, CheZ appears as a tetramer (Ref. 27 and this study). In the cross-linking experiments, CheZ appeared as a dimeric protein (Figs. 1-3). In zonal centrifugation, the estimated size of CheZ was between a monomer and a dimer (Fig. 5). Since it is well known, both experimentally (45-47) and theoretically (48), that the molecular size of nonspherical proteins can be overestimated severalfolds by sizeexclusion chromatography (45- 48), and that the size can be underestimated in zonal centrifugation (47), it is reasonable to assume that CheZ is a non-spherical dimer.
The major finding of this study is that, in the presence of CheYϳP, there is further oligomerization of the CheZ dimer. However, it is not possible to determine, on the basis of the results, whether CheYϳP is an integral part of the oligomer at a CheZ:CheYϳP ratio of 2:1, or whether it is just bound to an oligomer comprising CheZ alone. The results of the fluorescence depolarization implied that the oligomer is about 4 -5 times larger than the CheZ dimer formed in the absence of CheYϳP. This suggests that the size of the oligomer is in the order of 200 kDa. This size is about at the top of the size range observed in the cross-linking experiments (Fig. 2B). The size estimation of the oligomer from the fluorescence depolarization is only a first degree approximation, because it was based on the assumption that both the oligomer and the CheZ dimer are spherical (43,44), an assumption which, according to our own results, is probably incorrect. Determination of the exact size or shape is beyond the resolution of the applied methods. A better estimation of the size of the oligomer is rather difficult by currently available methods because, at this stage, there is no obvious way to separate the oligomer from CheYϳP and maintain it in a stable form. For example, techniques in which the shape contribution can be estimated (e.g. light scattering) cannot distinguish between the oligomer and CheYϳP in the mixture. Nevertheless, these difficulties in size and shape estimation do not affect the conclusions reached in this study, as neither the exact size nor the shape of the oligomer are necessary for the conclusions. It should be noted that the CheZ oligomer, observed in this study, is different from the CheZ homopolymer (27) and the CheZ-CheA S multimeric complex (49), observed earlier, in the sense that the oligomer of this study is not stable and it readily dissociates in the absence of CheYϳP.
In the fluorescence depolarization experiments, the anisotropy of fluorescein-labeled CheZ was increased to a large extent in the presence of CheYϳP (Fig. 7A). This observation could, in principle, be attributed either to a significant increase in the molecular volume of CheZ, or to a large conformational change at the vicinity of the fluorescein moiety that restricts its free rotation. The following observations strongly suggest that a significant increase in the molecular volume of CheZ, i.e. CheZ oligomerization, is the mechanism responsible for the anisotropy change. (i) To a first approximation, the Perrin plot (Fig. 7B) is composed of two distinct rotations: a fast rotation of the probe, and a slow rotation of the whole protein. To determine the rotational freedom of the probe, we extrapolated the straight line of the Perrin plot to 1/anisotropy ϭ 0. Since the straight line shown in the figure represents the rotation of the protein, the extrapolated value represents the hypothetical case in which the protein rotation is frozen but the probe rotates freely. As shown in Fig. 7B, both the dimeric and the oligomeric forms of CheZ fall in the same anisotropy range (1/anisotropy values of 4.0 and 4.8 for the dimer and oligomer, respectively). This indicates that the rotation of the probe itself is not significantly affected by the oligomerization, and therefore that the anisotropy change is not the result of a change in the probe rotation. (ii) An increase in anisotropy could, in principle, be due to a decrease in the lifetime of the excited state of the probe, reflected in a reduced efficiency of the fluorescence. However, the large oligomerization-dependent change in anisotropy was accompanied by only a minor reduction (7%) in the fluorescence efficiency. The lack of substantial changes in the fluorescence efficiency and in the motional freedom of the probe, is supported by the observation that both the 214F 3 C substitution (Fig. 6) and the conjugation of this cysteine with fluorescein maleimide 2 did not affect the activity of CheZ.
It is well known in a variety of systems, including bacterial signal transduction systems (50,51), that oligomerization regulates protein activity. Accordingly, it is conceivable that the oligomerization may either activate or inhibit the phosphatase activity of CheZ. In the first possibility, a burst of CheY phosphorylation will activate CheZ and will thereby promote faster deactivation of CheY. Fast deactivation of CheY will prevent non-beneficial too long periods of tumbling. If the other possibility is correct and oligomerization inhibits the phosphatase activity of CheZ, the oligomerization may serve as an amplification step in which phosphorylation of CheY leads to oligomerization of CheZ, inhibition of its phosphatase activity, and, consequently, further increase in the level of CheYϳP. The results described in the subsequent paper (28) suggest that the first possibility is the correct one. . When present, the concentration of CheY S was 20 M. The medium viscosity was increased by sucrose up to 4.6 cPoise. In this viscosity range the change in fluorescence anisotropy reflects mainly the rotation of the whole CheZ-fluorescein conjugate, and to a much lesser extent the free rotation of the probe (fluorescein). The latter rotation is not expected to be affected by the presence of CheY and AcP. Therefore, to a first approximation, the slopes of these lines, can be taken as the inverse of the respective molecular volumes (43).