A New Perspective on Response Regulator Activation

Two-component signaling pathways are the most prevalent of strategies used by bacteria for coupling environmental signals to adaptive responses (10). The response regulator component typically functions at the end of the pathway, with the conserved regulatory/receiver domain acting as a phosphorylation-activated switch to control an associated or attached effector domain that elicits the output response (24). In the simplest mechanistic model, these domains are considered to exist in two conformations, an inactive (“off”) state and an active (“on”) state, with phosphorylation at a conserved active-site aspartic acid residue serving to shift the equilibrium toward the active state (22). In this issue of the Journal of Bacteriology, Dyer and Dahlquist provide direct structural evidence that this model is overly simplistic (8). Their crystallographic characterization of the unphosphorylated Escherichia coli chemotaxis response regulator CheY bound to a peptide of its effector protein target, the flagellar switch protein FliM, reveals a conformation that is intermediate between inactive and active states. This finding provides evidence against a strict two-state mechanism in which the orientations of key residues that distinguish the inactive and active states are presumed to be obligatorily coupled. The intermediate state reported in their study not only provides insight into the mechanism of switching between inactive and active states but also has potential physiological relevance for some response regulators.

THE TWO STATES: INACTIVE AND ACTIVE

Structures of several unphosphorylated and phosphorylated, or otherwise activated, receiver domains have provided descriptions of inactive and active conformations (4, 6, 12, 14, 15). The structural changes associated with the active state are relatively subtle, involving backbone displacements of ∼1 Å on the α4-β5-α5 face of the domain, distant from the active-site phosphorylation (18). The magnitudes and extents of these structural perturbations vary substantially in different response regulators and can be assessed only by scrutiny of superpositioned structural images.

A more readily apparent feature of the two states is the distinct orientation of two highly conserved residues with hydroxyl and aromatic side chains, Ser/Thr and Phe/Tyr (Thr⁸⁷ and Tyr¹⁰⁶ in CheY). In the active state, the aromatic side chain is oriented toward the interior of the domain and the hydroxyl-containing side chain is repositioned closer to the active site to allow formation of a hydrogen bond with an oxygen of the active-site phosphoAsp (Fig. 1). As the positions of these two residues are easily recognized in receiver domain structures, they have received much attention. They are commonly referred to as “switch” residues, and their conformations alone have been used to categorize structures as inactive or active. However, it should be noted that features that distinguish the active state, though much less readily discernible than switch residue orientations, are much more extensive and involve a broad surface of the regulatory domain that mediates the macromolecular interactions necessary for the output response of the regulator.

FIG. 1.

CheY activation. Previously determined structures have defined two discrete conformational states: (i) an inactive conformation in which Thr⁸⁷ and the β4-α4 loop are positioned away from the active site and Tyr¹⁰⁶ is solvent exposed and (ii) a phosphorylated or active conformation in which the positions of Thr⁸⁷ and the β4-α4 loop are oriented toward the active site and are stabilized by a hydrogen bond with the phosphate group at Asp⁵⁷ and Tyr¹⁰⁶ is buried, occupying a cavity vacated by the repositioned Thr⁸⁷. The “equilibrium shift” hypothesis postulates a preexisting equilibrium of the two conformations that is shifted toward the active conformation upon phosphorylation of the active-site Asp⁵⁷. The “Y-T” coupling hypothesis postulates a strict correlation between the outward and inward conformations of Tyr¹⁰⁶ and the orientation of Thr⁸⁷, away from and toward the active site, respectively. The existence of intermediate conformations, exemplified by the CheY-FliM structure (8), expands the conformational landscape beyond two discrete states and extends the “Y-T coupling” model to a “T-loop-Y coupling” model that recognizes the importance of the β4-α4 loop in the mechanism of activation and abolishes the strict coupling of the orientations of Thr⁸⁷ and Tyr¹⁰⁶. This schematic representation shows CheY molecules as hexagons with active sites colored yellow and α4-β5-α5 signaling surfaces colored blue. The side chains of the switch residues Thr⁸⁷ and Tyr¹⁰⁶ are depicted as sticks, hydroxyl groups as red circles, and β4-α4 loops as arcs or squiggles.

The observation of active conformations in the absence of phosphorylation or phosphoryl analogs led to the widely accepted “two-state equilibrium” hypothesis that postulates an equilibrium between inactive and active conformations that can be shifted by phosphorylation, binding to targets, or specific mutations (20, 22). There are two basic premises to this model that define the number of states and the mechanism of conversion between them. The hypothesis invokes a preexisting population of domains that exist in an active conformation and are selectively stabilized by phosphorylation or target binding, in distinct contrast to an allosteric mechanism in which phosphorylation or target binding is required to induce an active conformation. The aforementioned data strongly support this notion. The supposition of two discrete states is less solid. Experimentalists, mindful of Occam's razor, frequently discuss two-state mechanisms although they are fully aware of greater complexity in the system. By necessity, the transition between two states requires the existence of intermediate states, and the essential question is not whether intermediate states exist but rather the extent to which they are populated. Proteins are dynamic macromolecules, and there is always some degree of structural heterogeneity for any state. From this perspective, arguments between two-state and multistate models rest on semantics. This caveat aside, all significantly populated states are of potential physiological significance and warrant exploration.

NEW INTERMEDIATES

The structure of unphosphorylated CheY bound to a peptide of the flagellar switch protein FliM reported by Dyer and Dahlquist (8) joins a growing body of evidence that argues convincingly against a simple two-state equilibrium model (7, 11, 20). When bound to the FliM peptide, the conformation of CheY is neither inactive nor active. As in the active state, the side chain of Tyr¹⁰⁶ is completely buried. However, the position of the β4-α4 loop and the orientation of Thr⁸⁷ that resides within it are midway between the conformations observed in the inactive and active states (Fig. 1). Similar, though distinct, intermediate states have also been observed for unphosphorylated CheY-CheZ peptide complexes (11) and for a “meta-active” CheY subpopulation found in the high-resolution crystal structure of unphosphorylated CheY (20). Analogous intermediates have been observed in structures of receiver domains of response regulator transcription factors of the OmpR/PhoB subfamily that were crystallized as homodimers in the absence of phosphoryl analogs (21). Thus, intermediate states induced by target binding appear to be a general feature of receiver domains of response regulator proteins.

The CheY-FliM peptide complex provides a direct structural rationale for previous biochemical observations. Fewer intermolecular contacts between the FliM peptide and unphosphorylated CheY compared to BeF₃⁻-activated CheY likely contribute to the 20-fold-lower affinity of FliM for CheY compared to phosphoCheY (17, 23). The structure also provides insight into the observed 30-fold acceleration of CheY phosphorylation in the presence of FliM peptide (19). The two-state equilibrium model allowed a simplistic interpretation of this conformational coupling. Phosphorylation enhances target binding and target binding enhances phosphorylation, each by shifting the equilibrium toward the active conformation (2, 19). The unphosphorylated CheY-FliM peptide structures clearly indicate that this is not the case, though perhaps only by degree. Phosphorylation and FliM binding do not produce identical conformations of CheY. FliM binding likely lowers the energetic barrier to phosphorylation by stabilizing a conformation of CheY that is closer to the active conformation than to the inactive conformation. Similar interpretations have been made for the enhancement of CheY phosphorylation by the CheZ peptide (11).

PHYSIOLOGICAL SIGNIFICANCE OF INTERMEDIATES

The existence of intermediates observed when unphosphorylated CheY binds to its targets stimulates questions about relevance in vivo. Based on the cytoplasmic concentration of unphosphorylated CheY and the binding affinity to FliM at the flagellar switch (1), it has been estimated that ∼30% of the switch binding sites are occupied by unphosphorylated CheY (19). Further, it has been proposed that the observed low level of activity of unphosphorylated CheY in vitro (3) provides direct evidence for a population of CheY molecules in the active conformation (22). However, in vivo, a basal level of kinase-independent activity might be explained by unphosphorylated CheY binding to the switch in an intermediate conformation and subsequently converting to a fully activated state by accelerated phosphorylation by intracellular phosphodonors such as acetyl phosphate (19), which are estimated to be present at concentrations of ∼1.5 mM (16).

The distinct spatial distributions of CheY when bound to its two targets, FliM and CheZ, suggest different physiological roles for the two intermediates. When bound to CheZ, CheY is localized to the chemoreceptor complex (5), where the source of phosphoryl groups is the kinase CheA. Here, the acceleration of phosphorylation of CheY by binding to the phosphatase CheZ would be predicted to produce a cycle of nonproductive phosphorylation that might fine-tune signaling (11). While the targets of CheY are specific to the chemotaxis system, target-induced acceleration of phosphorylation appears to be a general feature of response regulators (2). Binding of transcription factor response regulators to DNA targets may facilitate phosphorylation by small-molecule metabolites, thus contributing to background levels of activity independent of kinase-mediated phosphorylation.

A NEW PERSPECTIVE ON ACTIVATION

The notable reorientations of Tyr¹⁰⁶ and Thr⁸⁷ in the active states of receiver domains have led to the proposal of a Y-T coupling mechanism for activation (6, 13, 14, 25) (Fig. 1). In fully inactive and active conformations, the orientations of these residues are strictly correlated, and it has been argued that an inward, active orientation of Tyr¹⁰⁶ is sterically incompatible with an active-site-distal, inactive orientation of Thr⁸⁷ (20), although recent structures demonstrate that very minor global adjustments can relieve this constraint (11). The intermediate orientation of Thr⁸⁷ despite the complete burial of Tyr¹⁰⁶ observed in the CheY-FliM peptide structures argues further against strict Y-T coupling.

In further examination of available CheY structures, Dyer and Dahlquist found that the positions of Tyr¹⁰⁶ and Thr⁸⁷ are also dependent on the conformation of the β4-α4 loop, and they consequently propose a “T-loop-Y” coupling model, a modification of the Y-T coupling model that incorporates the conformational restrictions between Tyr¹⁰⁶ and the β4-α4 loop (8). Phosphorylation and the position of Thr⁸⁷ gate the conformation of the β4-α4 loop. Specifically, when phosphate is absent from the active site, Thr⁸⁷ and the β4-α4 loop are in predominantly inactive conformations and there are no restrictions on the orientation of Tyr¹⁰⁶. When phosphate is present, a hydrogen bond dictates an active conformation of Thr⁸⁷, which is coupled to an active conformation of the β4-α4 loop, strongly favoring burial of Tyr¹⁰⁶. Such coupling, collective motions, or gating potentially have mechanistic implications for response regulator activation (9).

The diversity of CheY structures observed in different complexes, and even within a single complex among different molecules within an asymmetric unit, attests to the enormous plasticity of receiver domains. Whether the accumulated structures represent steps within a concerted mechanistic pathway or merely a sampling of numerous allowable states remains an open question. Trying to identify the mechanism of activation from knowledge of inactive and active states and an assortment of additional structures is analogous to trying to fully describe a movie from knowledge of the opening and closing scenes and a random set of still frames from in between. In the case of CheY, this is complicated by a cast of interacting characters that provide the potential for subplots that decorate the main story line. A clearer picture is likely to require either a more complete set of still images or techniques that allow the ordering of individual states into a series of sequential steps.