Report

Two-State Allosteric Behavior in a Single-Domain Signaling Protein

See allHide authors and affiliations

Science  23 Mar 2001:
Vol. 291, Issue 5512, pp. 2429-2433
DOI: 10.1126/science.291.5512.2429

Abstract

Protein actions are usually discussed in terms of static structures, but function requires motion. We find a strong correlation between phosphorylation-driven activation of the signaling protein NtrC and microsecond time-scale backbone dynamics. Using nuclear magnetic resonance relaxation, we characterized the motions of NtrC in three functional states: unphosphorylated (inactive), phosphorylated (active), and a partially active mutant. These dynamics are indicative of exchange between inactive and active conformations. Both states are populated in unphosphorylated NtrC, and phosphorylation shifts the equilibrium toward the active species. These results support a dynamic population shift between two preexisting conformations as the underlying mechanism of activation.

The heart of signal transduction is the switching of proteins between inactive and active states. This process can be promoted by other proteins, domains, ligands, or covalent modifications such as phosphorylation. A central question is how these effectors lead to a change in activity. Two models have classically been discussed: (i) the effector induces a new structure or (ii) shifts a preexisting equilibrium. The second model is also known as allosteric activation, a shift between relaxed (R) and tense (T) states (1, 2), a model also known as allosteric activation. Although allostery is well accepted for multidomain proteins, for regulation within a single domain, the induced-fit mechanism is still widely held. In particular, phosphorylation that regulates signal transduction pathways is usually assumed to trigger a conformational switch. The idea of inducing a new structure is based on the traditional assumption of unique folds dictated by protein sequence. Modern concepts of protein-folding funnels and energy landscapes (3,4), which describe folded proteins as an ensemble of conformational substates, challenge traditional concepts of the control of protein activity (5, 6). According to this statistical viewpoint, protein function is not determined purely by the static structure but rather through a redistribution of already existing populations in response to changes in environment. There is little experimental data that discriminates between induced-fit and population-shift models because the conformational changes are fast and the populations are often strongly skewed. Our data on the nitrogen regulatory protein C (NtrC) provide direct experimental evidence for a population-shift mechanism in this phosphorylation-regulated signaling protein.

NtrC is a member of the “two-component system” signaling family (7). Two highly conserved components, histidine kinases and response regulators, control gene expression, chemotaxis, antibiotic resistance, and many other bacterial processes. The “receiver domain” of the response regulators is the molecular switch, which is controlled by phosphorylation of an active-site aspartate. In NtrC, phosphorylation of the receiver domain results in large structural changes (8). The active conformation of the receiver domain is essential for oligomerization of full-length NtrC, which then promotes transcription by the σ54-holoenzyme form of RNA polymerase (9). NtrC variants with partial activity in the absence of phosphorylation show nuclear magnetic resonance (NMR) chemical shift differences (10) in the same area that changes structure upon phosphorylation. Each mutant displayed evidence of conformational heterogeneity rather than adopting a unique, partially active conformation. To shed light on the mechanism of activation, we characterized at atomic resolution the molecular motion of the NtrC receiver domain in three different functional states.

NMR relaxation of backbone amides provides a powerful experimental approach for detecting conformational dynamics of proteins at atomic resolution (11). Three relaxation parameters were measured for all backbone amide nitrogens of the regulatory domain of NtrC (NtrCr): 1H–15N nuclear Overhauser effect (NOE), rate constants for spin-lattice relaxation (R 1), and spin-spin relaxation (R 2) (Fig. 1) (12). The experiments were performed for NtrCr in three different functional states: the inactive unphosphorylated form NtrCr, a partially active mutant form NtrCr [Asp86 → Asn86/Ala89 → Thr89(D86N/A89T)], and the fully active phosphorylated form P-NtrCr (13). The NMR spectra for P-NtrCr were acquired during steady-state autophosphorylation/dephosphorylation using carbamoylphosphate and Mg2+ as substrates for the enzyme turnover (8), because of the short lifetime of the phosphorylated form.

Figure 1

Backbone 15N relaxation parameters for NtrCr, NtrCr(D86N/A89T), and P-NtrCr are shown in blue, green, and red, respectively. Results of (A) {15N–1H} NOE, (B) 15N R 1, and (C) 15N R 2 measurements at 750 MHz 1H frequency and (D) 15NR 2 at 600 MHz are plotted as a function of residue number. The approximate locations of secondary structure elements are indicated above. Values for backbone amide signals are unavailable for six prolines at positions 48, 58, 74, 77, 103, and 105, as well as the two NH2-terminal residues. Additional residues were excluded from the relaxation and model-free analysis because of spectral overlap or extreme line-broadening. The significantly increased R 1 and decreasedR 2 values for P-NtrCr compared to the unphosphorylated forms is caused by the high-salt conditions required to maintain phosphorylation (8).

To extract internal backbone dynamics, relaxation parameters (Fig. 1) were analyzed using an extended Lipari-Szabo “model-free” approach (14, 15), which allows the characterization of complex internal dynamics over a broad range of time scales (16). Fast (pico- to nanosecond) motions of the N–H bond vector, expressed in terms of an order parameter (S 2) ranging from 0 (unrestricted) to 1 (completely restricted), are comparable for all three forms of NtrCr (Fig. 2A). The key for monitoring the activation process is the micro- to millisecond time-scale motions (Figs. 2B and 3, A through C) identified by the exchange term R ex, which indicates conformational exchange between conformations that sense different chemical environments. The value of R ex is directly proportional to the difference in chemical shift between the exchanging species and not necessarily to the magnitude of structural change. Therefore, those amides with R ex values beyond detection reflect a large difference in chemical shift between the exchanging species.

Figure 2

Correlation between backbone dynamics parameters and conformational switch upon activation. Order parametersS 2 characterizing motions in the pico- to nanosecond time scale (A) and exchange valuesR ex indicating motions in the micro- to millisecond time regime (B) deduced from model-free analysis of 15N relaxation data (Fig. 1) are plotted as a function of residue number. Values for NtrCr, NtrCr(D86N/A89T), and P-NtrCr are shown in blue, green, and red, respectively. R exvalues of 20 s−1 are shown when extremely broad signals precluded the determination of relaxation parameters. (C) Backbone atomic RMS deviations (solid line) and backbone and15N chemical shift between NtrCr and P-NtrCr (purple dots). Displacements were measured using the NMR structures of NtrCr and P-NtrCr(8) superimposing residues 4 to 9, 14 to 53, and 108 to 121. A purple “X” reflects a 15N chemical shift that was not determined in either NtrCr or P-NtrCr, where the missing resonance is presumed to be broadened beyond detection as the result of chemical exchange between two signals with large shift differences. Proline residues are indicated by “P” on thex axis.

Figure 3

Color-coded representation of the conformational exchange dynamics, structural rearrangement, and chemical shift perturbations related to NtrCr activation. Values of R ex (Fig. 2B) are indicated on a ribbon diagram with a continuous color scale for (A) NtrCr, (B) NtrCr(D86N/A89T), and (C) P-NtrCr. (D) The NMR structures of NtrCr (cyan/blue) and P-NtrCr(yellow/orange) are superimposed using residues 4 to 9, 14 to 53, and 108 to 121, indicated in darker colors (orange and blue), with the structural differences highlighted in lighter colors (yellow and cyan) (8). (E) This structural rearrangement upon phosphorylation is presented as backbone RMS deviations (Fig. 2C) in a continuous color scale. (F) Chemical shift perturbations due to phosphorylation of NtrCr (Fig. 2C) are displayed in a similar manner. All structural representations were generated in MOLMOL using Protein Data Bank entries 1DC7 for NtrCr (A, B, and D through F) and 1DC8 for P-NtrCr (C and D).

The two central results are illustrated in Fig. 3: First, many residues of unphosphorylated wild-type (Fig. 3A) and active mutant NtrCr (D86N/A89T) (Fig. 3B) show prominent micro- to millisecond time-scale motions that disappear in the phosphorylated form (Fig. 3C). In addition to the fitted values for R ex, several residues are indicated with R ex values ≥20 s−1 (Figs. 2B and 3, A through C). Those residues could not be detected because of large line-widths from exchange broadening. Second, dynamic regions in both unphosphorylated forms (Fig. 3, A and B) map precisely to the region that experiences a structural change upon activation (Fig. 3, D and E). This suggests that the conformational exchange that is observed be between the inactive and active conformations (17).

How are these microsecond time-scale motions related to function? Our dynamic data are not consistent with a model for activation where NtrCr exists in the inactive conformation (I) and phosphorylation induces a change in structure to the active conformation (A) (Fig. 3A). Previously, we showed that the size of chemical shift differences between a series of mutants relative to wild-type NtrCrstrongly correlated with the level of transcriptional activity (Fig. 4) (10). We proposed that the activating mutations may cause a shift in population toward the active state, a model that has been suggested for response regulators (18–20). This model is buttressed by the dynamics here: resonances showing large shift differences also show microsecond time-scale motions and map exactly to the switch region (Fig. 3). In contrast, amides in the other parts of the molecule, which do not undergo structural changes upon activation, show no motion on the slow time scale. Thus, the observed linear shifts (Fig. 4A) are due to a conformational exchange between I and A, which is fast on the NMR time scale. This results in an average signal with the peak position determined by the relative populations of the exchanging conformers; larger chemical shift changes occur asA becomes more populated (Fig. 4).

Figure 4

Relationship between chemical shift changes and activity of the following NtrCr forms (36): V115I (gray), D54E (red), D86N (green), D86N/A89T (gold), D86N/A89T/V115I (blue), and P-NtrCr (cyan) in respect to wild-type NtrCr (black). (A) Signals for D88 in1H–15N HSQC spectra of five NtrC variants are superimposed. Larger chemical shift changes (shift of the signal to the lower right corner) coincide with increased activity of the corresponding mutants (10). This collinear shift behavior is observed for a number of residues, as shown in (B): D88 (□), D11 (○), D10 (×), M81 (◊), and V91 (▵). Chemical shifts are shown relative to P-NtrCr and therefore reflect the relative population between inactive and active conformers for each mutant.

The key question remains: how does wild-type NtrCrget to the activated state? Surprisingly, wild-type NtrCrundergoes a conformational exchange process in exactly the same area as the mutants (Fig. 3, A and B). In addition, the shifts of residues showing exchange in wild-type NtrCr are collinear with those detected for the mutants (Fig. 4A). Consequently, Iand A already coexist in unphosphorylated NtrC, with the equilibrium skewed toward I. After phosphorylation, conformational exchange virtually disappears (Fig. 3C), suggesting that the equilibrium is far shifted toward the active conformation (>99%). The population of active species in wild-type, unphosphorylated NtrCrcan be estimated to be between 2 and 10% because we were able to solve the structure by NMR. The 1H–1H NOE intensities in the region of conformational change were comparable to those observed for P-NtrCr, which populates the active form almost exclusively. In contrast, NOE spectra on the double mutant resulted in a loss of NOEs in the region of conformational exchange. The inactive conformation must therefore be the dominant species in wild-type NtrCr, with the active conformation populated to at least 2% to yield the observed R ex values. Consequently, the position of the average signals in Fig. 4B reflects the approximate equilibrium constants between I andA for all forms. An equilibrium constant of about 1 can be determined for NtrCr (D86N/A89T) (21). Phosphorylation of this mutant form increases its activity only to that of the phosphorylated wild type and results in chemical shifts identical to P-NtrCr, supporting the population shift model. In summary, NtrCr interconverts between the two functionally important conformers and not among many random substates. Phosphorylation does not induce a new structure, but rather shifts a preexisting equilibrium.

If there is already a significant fraction of active molecules before phosphorylation, why is there no basal transcriptional activity and how can a sharp switch be accomplished? Oligomerization of full-length NtrC is essential for biological output (22). The active conformation of NtrC's receiver domain triggers oligomerization through interaction with the central domain. Therefore, the activity for full-length NtrC is highly sigmoidal in concentration (23), ensuring a sharp signal response. Our data suggest that the population of active conformers in NtrCr is below the response threshold. In contrast, the single-domain response regulator CheY, which does not oligomerize upon phosphorylation, shows a low level of activity in the unphosphorylated state (24), directly reflecting the population of active conformers. In fact, microsecond time-scale motions have been reported for unphosphorylated CheY (25) and Spo0F (20), suggesting that the two-state exchange model might be true for response regulators in general. Our study of NtrCr dynamics in three defined functional states, combined with knowledge of the inactive and active protein conformations, enables the direct observation of the functional equilibrium process. A movie presentation of the activation process can be found at (26).

The observation of persistent motion in the β4–α4 loop and part of β5 in P-NtrCr (Fig. 3C) might seem to contradict the basic idea of a two-state equilibrium. R ex in the β4–α4 loop arises from unwinding the NH2-terminal portion of helix 4 upon phosphorylation (8). A likely source of the persistent R ex in β-strand 5 is motion of the side chain of Tyr101(Y101). The dynamic behavior around Y101 in all three forms of NtrCr indicates that the conformational exchange of this conserved aromatic side chain is not simply coupled to the equilibrium between I and A described above (27). The combination of the relaxation and activity data allowed the discrimination between two modes of motion, one that is correlated to the activation process and a second, uncoupled motion.

Our NMR relaxation experiments demonstrate that activation of the NtrC regulatory domain occurs in the micro- to millisecond range, and not in pico- or nanoseconds, where the dynamics do not change between unphosphorylated and phosphorylated NtrCr. In fact, functionally important protein motions are likely to be in this time regime because many biological processes occur in micro- to millisecond time scales.

We propose that stabilization of preexisting conformations may be a fundamental paradigm for ligand binding. For calmodulin, a similar conformational exchange process was detected using NMR relaxation experiments (28, 29). Overall, the concept of “allosteric” regulation of single domains has not been widely considered. NMR provides a good tool to study conformational equilibria; however, to identify functionally important motions it is necessary to combine structural, dynamic, and functional information.

  • * Present address: Department of Biochemistry, Medical College of Wisconsin, Milwaukee, WI 53226, USA.

  • To whom correspondence should be addressed. E-mail: dkern{at}brandeis.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article