Stimulation-dependent regulation of assembly and stability of sensory complexes can be important in signalling, and many signal transduction pathways in eukaryotes are regulated on this level . Here we show that protein exchange at the sensory complexes in E. coli chemotaxis is affected by the signalling state of the pathway on many levels. First, stability of the sensory receptor-kinase core is higher for complexes formed by receptors that are in a higher modification state and consequently are more active. Such dependence is generally consistent with previous biochemical experiments [7, 42], with lower structural stability of less modified receptors , and also with higher sensitivity of sensory complexes that are formed in vitro by the less modified receptors to destabilizing factors such as high pH or low ionic strength . Our data also agree with in vivo studies that reported an increase in protein localization to the chemoreceptor clusters [50, 51] at higher levels of receptor modification or activity. However, the effect in vivo is rather modest, and the observed regulation of complex stability dependent on receptor modification is unlikely to be directly involved in signal transduction. Rather, it may play a role in the adjustment of the signalling properties of receptor clusters, and can indeed explain the previously observed increase in the strength of cooperative receptor interactions within clusters upon increase in receptor modification . Since increased methylation results from adaptation to increasing concentration of ambient attractant, higher stability and cooperativity within clusters can enhance the gain of the chemotaxis system at higher levels of ambient ligands, to closely follow physical limits of sensitivity posed by the noise in ligand binding .
The regulation of exchange at the cluster that was observed for the adaptation enzymes may be of even greater physiological significance. When CheR is unable to bind its substrate sites on the receptor, whether due to the mutation in the catalytic site of CheR or lack of unmethylated glutamates, the turnover was greatly accelerated. This suggests that the overall rate of CheR dissociation from receptors (k
) largely depends on its binding to the substrate sites, although such dependence remains to be confirmed by direct biochemical measurements. In principle, the level of receptor modification might also affect this turnover indirectly, through an allosteric regulation of CheR association with the C-terminal pentapeptide sequence of receptors. Regardless of the detailed molecular mechanism of such methylation-dependent acceleration of CheR exchange, we propose that faster turnover can increase the efficiency of adaptation by limiting the amount of time CheR spends in an unproductive association with a receptor molecule that cannot be further modified. This is particularly important for adaptation to high levels of ambient stimulus, when the kinetics and precision of adaptation become severely limited by the shortage of the free methylation sites [15, 52].
Another important effect of the faster turnover of CheR at the cluster may be to specifically reduce the noise in the signalling output at increased levels of receptor methylation. Previous studies suggested that the level of phosphorylated CheY in adapted E. coli cells can vary substantially on the time scale of tens of seconds . This can be explained by stochastic fluctuations in the number of cluster-associated CheR molecules [53–55] that would translate into the variable level of receptor methylation and ultimately into fluctuations of the activity of the pathway. Such fluctuations are expected to result in E. coli cells occasionally undertaking very long runs, enhancing the overall efficiency of the population spread through the environment in the search of chemoattractant gradients [54, 55]. However, fluctuating levels of CheY-P are also predicted to severely impair the ability of bacteria to precisely accumulate at the source of the chemoattractant gradient, posing a trade-off dilemma for the chemotaxis strategy . We propose that the observed increase in the turnover of CheR at the highly methylated receptors will specifically decrease noise in the pathway output for cells that have already reached high attractant concentration along the gradient, enabling them to efficiently accumulate at the source of attractant. The observed regulation of CheR exchange may therefore be an evolutionary selected trait that increases overall chemotaxis efficiency.
An acceleration of exchange was also observed for the catalytic mutant of CheB. This indicates that the CheB exchange is dependent on its binding to substrate sites, similar to CheR, though the molecular details of this effect remain to be clarified. Moreover, CheB exchange was strongly stimulated by mutating the phosphorylation site in the regulatory domain, which prevents CheB activation by phosphorylation. This latter effect confirms that the binding of CheB to receptor clusters is strengthened by phosphorylation, which may provide an additional regulatory feedback to the chemotaxis system (; Markus Kollmann, personal communication).
Finally, we analyzed here the effects of temperature and showed that the thermal stability of the cluster core in the cell, determined by the exchange of CheA, is much higher than that of the biochemically reconstituted complexes . While the factors responsible for the observed differences between stability of the sensory complexes in vivo and in vitro remain to be identified, such differences may be due to the active process of assembly and/or disassembly of the sensory complexes by specialized cellular chaperones as previously proposed . In any case, thermal stability of the cluster core may be an important component of the overall thermal robustness of the chemotaxis pathway . Consistent with that, the deterioration of chemotaxis in some E. coli strains above 37°C is apparently caused by the reduced expression of chemotaxis and flagellar genes rather than by the malfunction of the pathway. Moreover, although the observed effect of temperature on gene expression was not strain-specific, chemotaxis of the wild type strains MG1655 and W3110 was significantly less affected than chemotaxis of RP437. This difference was apparently due to the generally higher expression of chemotaxis proteins in MG1655 or W3110, which enables these strains to maintain expression that is sufficient for chemotaxis up to 42°C. Thus, the ability to maintain chemotaxis at high temperature is likely to be accomplished by a combination of the thermally robust pathway design  with the high thermal stability of chemosensory complexes and high basal expression levels of chemotaxis and flagellar proteins.