Tasting the Terroir with Tsr

Main Text

Often, a discriminating palate is essential for survival. In bacteria like Escherichia coli, chemical sensors analogous to taste buds cluster at the poles of its rod shape. Four such taste buds—Tsr, Tar, Tap, and Trg, collectively known as methyl-accepting chemotaxis proteins—are expressed in E. coli, each detecting distinct classes of chemical compounds (Tap and Trg are expressed at low levels). Tsr itself detects amino acids such as serine, alanine, glycine, and cysteine. Like the rest of its cohort, its functional form is as a transmembrane α-helical trimer of dimers bound up with CheA and CheW to form an ultrasensitive switch in the inner membrane of the Gram-negative organism expressing it (1). This sensor actuates changes in the direction of flagellar rotation by transmitting a signal to a switch in the flagellar rotary motor. Counterclockwise rotation propels nearly straight “runs,” whereas clockwise rotation causes “tumbles” that randomly change the heading of the swimming cell. Runs are extended in the favorable direction: up a gradient of an attractant chemical and down a gradient of a repellent chemical.

How, then, do E. coli cells actually distinguish useful components in their environment from highly similar, but possibly less useful, forms? In this issue of the Biophysical Journal, Orr et al. show evidence of a compelling structural mechanism for distinguishing subtle enantiomeric differences between processed forms of norepinephrine, a hormone that is released in high quantities in the gastrointestinal tract (2). E. coli converts norepinephrine into 3,4-dihydroxymandelic acid (DHMA) before reexporting it. DHMA is expected to emerge as a racemic mixture of (R) and (S) enantiomers. Orr et al. use an approach that utilizes multiple experimental and computational methodologies, with the scientific goal of exploring the sensitivity of Tsr to each enantiomer of DHMA.

Although comparison of ligands binding to a receptor using computationally costly molecular dynamics simulations is increasingly important (3,4), the high resolution of these powerful techniques raises a key consideration: the resources required to run such algorithms scale with both the size of the system and the amount of time simulated, quickly becoming intractable. However, Orr et al. used a logical, multifaceted pipeline of approaches to ameliorate this problem and develop a set of results that combine into a coherent and well-justified model that could not have been found with a single method alone (2).

They started with ligand docking to predict the molecular pose most suited to be the initial condition for molecular dynamics simulations that demonstrated the difference in effect of the binding of (S)-DHMA and (R)-DHMA to Tsr. Initial docking studies suggested a difference in receptor-ligand interaction and orientation between the two enantiomers. Starting with the docking poses, they found an ensemble of five molecular dynamics runs per system demonstrated the stability of the putative poses and their effect on the system at large. The mechanism of receptor activation, a known piston-like motion by one of the α-helices (5,6), was predicted to be of higher magnitude for (R) versus the (S) enantiomer (Fig. 1). The (R)-enantiomer-bound structure allows the formation of a key (R)-DHMA-threonine interaction, stabilizing the downward push of the helix; this interaction is absent in the (S)-enantiomer-bound structure. One would thus expect a difference in chemotactic response between the enantiomers, which they showed using tests of motility in vivo. In short, the stronger action of (R)-DHMA over (S)-DHMA results from a difference in receptor binding and translates into different behavioral responses in E. coli (2). This prediction justified the effort to experimentally separate the commercially available racemic mixture of (R)-DHMA and (S)-DHMA into the pure enantiomers, which could then be tested for their activity independently.

Figure 1.

The piston motion that stimulates sensing of Tsr ligands. (a) Coarse-grained representation of the four key α-helices in Tsr without ligand. (b) Representation of the downward piston mechanism when Tsr is stimulated by (R)-DHMA, which has a key hydrogen bond that stabilizes downward motion of the helix. (c) (S)-DHMA induces less downward motion of the α-helix because the hydrogen bond is missing. To see this figure in color, go online.

The implications of Orr et al.’s approach are far-reaching. From a methodological perspective, they cleverly avoided the temptation to just run molecular dynamics simulations for a longer time by predicting near-optimal initial poses with docking. In testing an experimentally tractable hypothesis, they were able to confirm the distinction between responses to the enantiomers, lending empirical support to the conclusion of the simulation study (2).

But finally, the biological implications are most fascinating. With just four sensor proteins, how diverse and discriminating is the palate of E. coli chemotaxis receptors? How many different attractants and repellants can be distinguished, and is the mechanism for subtle distinctions in the milieu similar to what Orr et al. found? Is there a unifying principle for how Tsr and the other sensor proteins distinguish molecular structures, or does each ligand have its own unique relationship with each sensor? Like most of the best science, this study has raised more questions than it has answered.

Editor: Julie Biteen.