An understanding of the features essential to receptor methylation and its role in chemotaxis has developed through a sustained effort that has (i) identified the sites of methylation and the methylation consensus sequence [15–18], (ii) demonstrated that increases in methylation result from attractant binding [26, 27, 29], (iii) established the role of interdimer methylation [30, 37, 38], and (iv) probed the factors influential in methylation site – CheR active site interactions [29, 39, 40]. These studies have been conducted almost exclusively on receptor samples without any of the cytoplasmic signaling proteins present other than CheR. However, it is now appreciated that a substantial fraction of the receptors in the cell are often located in patches at the cell poles [32], and are part of the signaling complexes that form with CheW and CheA [3, 4, 32, 41]. Their influences on receptor methylation have not been systematically studied. A single experiment in the study of Tar transmethylation conducted by LeMoual, Quang and Koshland failed to find an influence of ternary complex formation on Tar methylation, possibly because insufficient amounts of CheA and CheW were present [37]. Levit and Stock [22] made the qualitative observation that the methylation of Tsr proceeded to a greater extent in the presence of CheA and CheW, which is in agreement with the results of the experiments presented here.
Figure 1 maps the locations of the serine and CheW binding sites in relation to the sites of methylation on the dimer model of Tsr [6]. The serine and CheW binding sites are nearly at opposite ends of the dimer, and thus their effects are communicated in opposite directions toward the centrally-located sites of methylation. The panels in Figure 8 provide schematic representations of the methylation region from one subunit, which help to summarize the relative rates of methylation at the four sites without ligand, in the presence of either serine or CheW, and in the presence of both ligands (Figure 8A to 8D, respectively). The values next to the four sites (not in parentheses) are relative to the rate at that site without ligand, while the rate values inside the parentheses are relative to site 4 under the same conditions. The following observations are made from these data: (i) the increases in the methylation rate generated by high concentrations of CheW are larger than the increases generated by serine. (ii) The effects of serine and CheW are synergistic at sites one, two and three. This leads to the additional conclusion that the rate increases produced by serine are significantly larger when Tsr is saturated with CheW (see also Table 3 and Figure 6). (iii) The differences in rate between TsrQQQE and all other single site forms of Tsr lessen as serine and CheW are added, to the extent that TsrEQQQ is methylated more rapidly when both serine and CheW are present.
CheW and serine: different influences on Tsr structure?
Our results demonstrate that serine and CheW binding to Tsr both increase the methylation rate, and that the effects of CheW and serine are synergistic (super additive). These behaviors are consistent with the individual and combined effects of two allosteric activators that have similar effects on receptor structure. Yet a qualitative argument, based on the known influences of serine and CheW on receptor signaling activity, suggests otherwise. To illustrate this point, we assume the simplest situation, one in which Tsr is engaged in a two state equilibrium between kinase activating and kinase inhibiting states. Based on experimental data, serine binding shifts this equilibrium toward the kinase inhibiting state; a state that is also characterized by more rapid methylation. On the other hand, CheW is expected to either shift the equilibrium toward the kinase-activating state or at least not influence it, which according to our argument is characterized by a smaller methylation rate. Clearly, the experimental results are not consistent with this picture (CheW binding increases the methylation rate), which thus suggests that the binding of CheW and serine produce different effects on the structure and arrangement of Tsr in the membrane.
The distinct locations and properties of CheW and serine binding are also consistent with this idea. Serine-induced changes in Tsr structure are expected to be like aspartate-induced changes in Tar, which have been thoroughly investigated (reviewed in ref. [1]). One effect of aspartate binding on Tar is consistent with a propagated piston-like motion of TM2 relative to TM1 within the Tar dimer. Attractant binding shifts TM2 toward the cytoplasm; this motion is correlated with increases in the methylation rate and the inhibition of CheA activity (ref. [1] and references therein, [42]). Sites one, two, and three are clustered on the first methylation helix and joined to TM2 through the linker (Figure 1). In view of this circumstance, it is perhaps significant that the serine-stimulated rates relative to the rates without ligand (r = v+serine/vno ligand) display the trend rEQQQ >rQEQQ >rQQEQ ~ rQQQE (Figure 8B), which is opposite to the trend in separation between the methylation sites and the end of TM2 in the primary structure. The trend is not evident in the increases in rate produced by CheW alone (rEQQQ ≈ rQEQQ ≈ rQQEQ >rQQQE, Figure 8C), but it is evident again when both serine and CheW are present (Figure 8D), although the values of r are larger. It is as though the effect of serine binding on methylation diminishes with increasing distance from TM2, but because serine binding also serves to inhibit the kinase, it must also exert an influence at the CheW (and CheA) binding site(s). This suggests that the nature of the influences produced by serine at the CheW/CheA binding site and methylation sites are distinct, which may reflect the differences in the processes of receptor methylation and kinase regulation. Attractant-induced changes in receptor structure that influence methylation seem to be restricted to the receptor dimer binding attractant [43], while kinase regulation is a cooperative process shared among several receptor dimers [20].
In contrast to the attractant/receptor complexes, high-resolution structural information is not available yet for the CheW-receptor complex; consequently specific models for changes in receptor structure are difficult to develop. Nonetheless, the distinguishing features of serine and CheW binding provide some clues to the basis for different influences. CheW binds to MCPs at a location in the primary structure between methylation sites, i.e. between sites 1–3 and site 4, in contrast to attractants like serine, which have an identifiably more direct connection to the first three sites (Figure 1). As noted above, one of the changes in structure generated by serine may be mediated primarily through one of the two helices in the cytoplasmic domain. In contrast, the methylation data suggests that CheW exerts a more uniform influence on both methylation helices, because CheW (when added alone) produces significant rate increases at all four sites.
The mechanism of the CheW-mediated increase
As it is currently understood, receptor methylation involves two interactions between the MCPs and CheR, a tethering interaction and an active site – methylation site interaction. The tethering interaction involves specific binding of a conserved pentapeptide (NWETF) at the C-terminus of Tar and Tsr to CheR [44], and provides a mechanism to increase the CheR concentration in the vicinity of the methylation sites. The available evidence suggests that the tethering interaction is significantly stronger than active site – methylation site interactions [43, 44], and together the two interactions allow CheR tethered at the C-terminus of either Tar or Tsr to visit the multiple sites of methylation many times through the lower affinity (and presumably more labile) active site – methylation site interactions before a methyl group transfer event takes place. The small turnover number of CheR under saturating substrate concentrations (~7 min-1) [45], is consistent with this scenario. Importantly, this active site – methylation site interaction can, and in some circumstances must, occur in receptor dimers that are different from the receptor dimer involved in the tethering interaction [20, 37, 38, 43].
According to this line of reasoning, the increases in the methylation rates generated by CheW could result from increases in (i) the fraction of CheR tethered to Tsr at the C-terminus, (ii) the number of methylation sites accessible to tethered CheR, and/or (iii) an increase in the probability of methyl group transfer per active site – methylation site encounter. An increase in the recruitment of CheR (possibility i) requires that a substantial fraction of the CheR tethering sites are obscured in the absence of CheW. This is not supported by previous binding experiments involving CheR and Tsr-containing membranes, which demonstrated that ~90% of the CheR tethering sites on Tsr are accessible in the absence of CheW [46]. Moreover, it is estimated that ~70% of CheR is tethered to Tsr in the methylation experiments, based on the measured CheR-Tsr binding affinity (Ka = 4 × 105 M-1) [44, 46] and the concentrations of Tsr and CheR used here. Thus, the increases in rate produced by CheW cannot be accounted for by an increase in the recruitment of CheR to the membrane.
Possibility ii, that tethered CheR could access a larger number of methylation sites when CheW is bound, is consistent with the observations, because as noted above, CheW promotes receptor clustering [32]. Electron microscopic images of cells and isolated membranes with overexpressed Tsr indicate that receptor molecules are packed closely together in the membrane [7, 47]. CheW binding may either promote further clustering or a change in the arrangement of receptors that are already clustered, which serves to increase the number of methylation sites that can be accessed by tethered CheR molecules. To generate the rate increase, this larger number of sites must lead to a larger number of active site – methylation site encounters per unit time. Because of the complexity of these samples we cannot yet be specific about how this might occur, i.e. changes in the interactions between receptor dimers, between CheW and CheA, and between MCPs and CheW (and CheA) could all be involved.
If binding CheW produces a change in the arrangement of Tsr dimers, either via clustering or some other mechanism, without generating a significant shift in the equilibrium between the kinase-activating and kinase-inhibiting receptor states, then the logical inconsistency presented at the beginning of the section is circumvented. In this respect, possibility iii, which postulates an increase in the probability of methyl group transfer, seems less likely, although we cannot formally exclude it. If CheW binding were to produce a change in the conformation and/or dynamics of Tsr, which is a consequence of possibility iii, it seems likely that this would shift the equilibrium between the kinase-activating and kinase-inhibiting states in a manner inconsistent with the experimental results. This line of reasoning is also predicated on the expectation that serine binding shifts the equilibrium toward the kinase-inhibiting/methylation-active state, a state in which the receptor dimer is a better substrate for methylation. Consequently, the effects of CheW and serine are multiplied, because each new site to which CheR has access as a result of CheW binding also has a better chance of being methylated due to serine binding. Overall, an influence of CheW binding that increases the number of methylation sites accessible to CheR provides a plausible explanation of the results.
Implications for adaptation
Adaptation is the ability of a cell to return toward its prestimulus behavior following an increase or decrease in a stimulus. For E. coli adaptation is (near) perfect; that is the return to the prestimulus swimming behavior (tumble frequency) following a persistent attractant stimulus is complete (in most circumstances). This property is robust [48], i.e. the extent of adaptation is not strongly influenced by the relative amounts of receptor and methyltransferase in the cell [49]. Yet the chemotactic ability of E. coli is tolerant to variations in the adapted state; bacteria with significantly different adapted state tumble frequencies can be similarly efficient in chemotaxis assays [50]. Moreover the kinetics of adaptation are not a robust property of the system; the time required to return to the prestimulus behavior is strongly influenced by relative abundances of receptor and methyltransferase [48, 49]. Taken together, these observations suggest that the kinetics of adaptation is more essential than either the specific value of the adapted state tumble frequency or a requirement for perfect adaptation. The large influence that CheW can have on the methylation rate suggests that it plays an important role in determining the kinetics of adaptation in the cell. To achieve adaptation to a specific ligand, it seems necessary for changes in methylation level to occur primarily on the receptors to which the ligand binds. This requires ligand-specific increases in methylation [43], and perhaps also receptor-specific decreases in receptor demethylation. CheW appears to play an important role here, because the absolute difference in rate between serine-stimulated and unstimulated rates are larger in the presence of CheW. Also, the ligand-specific responses observed in transmethylation experiments involving Tar and Tsr [43] are not affected by the presence of CheW (F. Antommattei, unpublished observations). Finally, we noticed that site four differs from the other major sites in two respects: it is the most rapidly methylated site (under most conditions) and it is influenced least by ligand. In contrast, sites one, two and three are methylated more slowly, but are influenced more significantly by serine, especially with CheW present. In E. coli, Tsr is the most abundant receptor, and Tsr together with Tar comprise the large majority of receptors in the cell [34]. Its abundance suggests that Tsr has a strong influence on the kinetics of adaptation and setting the tumble frequency of the adapted state. In this respect, it appears that three of the major sites are involved in serine-specific adaptation, while site four, which is influenced little by ligand, plays a role in buffering the methylation level on all the MCPs in the cell.