Identification of core allosteric sites through temperature- and nucleus-invariant chemical shift covariance

Abstract

Allosteric regulation is essential to control biological function. In addition, allosteric sites offer a promising venue for selective drug targeting. However, accurate mapping of allosteric sites remains challenging since allostery relies on often subtle, yet functionally relevant, structural and dynamical changes. A viable approach proposed to overcome such challenge is chemical shift covariance analysis (CHESCA). Although CHESCA offers an exhaustive map of allosteric networks, it is critical to define the core allosteric sites to be prioritized in subsequent functional studies or in the design of allosteric drugs. Here, we propose two new CHESCA-based methodologies, called temperature CHESCA (T-CHESCA) and CLASS-CHESCA, aimed at narrowing down allosteric maps to the core allosteric residues. Both T- and CLASS-CHESCAs rely on the invariance of core inter-residue correlations to changes in the chemical shifts of the active and inactive conformations interconverting in fast exchange. In T-CHESCA the chemical shifts of such states are modulated through temperature changes, while in CLASS-CHESCA through variations in the spin-active nuclei involved in pairwise correlations. T- and CLASS-CHESCAs, as well as complete-linkage CHESCA, were applied to the cAMP-binding domain of the exchange protein directly activated by cAMP (EPAC), which serves as a prototypical allosteric switch. Residues consistently identified by the three CHESCA methods were found in previously identified EPAC allosteric core sites. Hence, T-, CLASS-, and CL-CHESCA provide a toolset to establish allosteric site hierarchy and triage allosteric sites to be further analyzed by mutations and functional assays. Furthermore, the core allosteric networks selectively revealed through T- and CLASS-CHESCA are expected to facilitate the mechanistic understanding of disease-related mutations and the design of selective allosteric modulators.

Keywords: CHESCA; EPAC; NMR; allosteric regulation; allostery; cAMP; dynamics; signaling.

Figure 1

(A) Coupled activation and binding equilibria underlying the allosteric regulation of EPAC1. IL denotes the ionic latch region of EPAC1-CNBD, which is displaced upon activation by cAMP. (B) Overlay of the EPAC-CNBD holo, active state (dark blue, PDB: 3CF6) and the apo, inactive state (light blue, PDB: 1O7F). Dashed boxes indicate the phosphate binding cassette (PBC), the base binding region (BBR), and the hinge region. The PBC and the hinge are in the “out” conformation in the apo, inactive state, whereas in the holo, active state they are in the “in” conformation. The endogenous ligand, cAMP, is shown in stick representation docked between the BBR and the PBC. (C) Molecular structures of cAMP and cAMP analogs. cAMP analogs include covalent modifications marked by circles. Sp-cAMPS and Rp-cAMPS include a sulfur atom in place of the axial and equatorial oxygens, respectively, while 2′-OMe-cAMP exhibits a methoxy group in place of the 2′-hydroxyl group. The extent of relative activation for each cyclic nucleotide is qualitatively described by the k_max scale above the structures. Shades of red also reflect the extent of EPAC1 activation. (D) Scheme illustrating three representative residues denoted as i, j, and k subject to linear versus nonlinear changes in chemical shift changes across different perturbations. (E) ¹H and ¹⁵N combined chemical shift correlation plots between residues undergoing a concerted chemical shift response to the perturbations in (C) result in a high degree of correlation. In (F) and (G), lower degrees of correlation are observed between residues with chemical shift changes that deviate from linearity due to deviations from the two-state fast exchange model. To see this figure in color, go online.

Figure 2

(A) Overlay of unreferenced ¹H-¹⁵N HSQC spectra acquired for the cAMP-bound (holo) EPAC1-CNBD at different temperatures. (B) Example of inter-residue pairwise correlations for a residue pair with high Pearson’s correlation coefficient at low and high temperatures. (C) Example of a residue pair with high correlation coefficient at low temperature but significantly lower correlation coefficient at higher temperature. Labeled circles represent the different EPAC states at 290 K (purple) and 316 K (red). The black dotted lines mark the difference in HSQC positions between the same state at the two temperatures. (D)–(F) Chemical shift correlation (CHESCA) matrices for the EPAC1-CNBD at (D) 298 K (E) 306 K, and (F) 310 K. The CHESCA R value legend in (F) indicates the color code for absolute correlation coefficients above or equal to 0.98 (red, positive; blue, negative). The secondary structure of the apo EPAC1-CNBD is mapped on the matrix, whereby gray boxes represent α helices and green boxes correspond to β sheets. Regions highlighted in gray denote the phosphate binding cassette (PBC), ionic latch (IL), base binding region (BBR), and the hinge helix. Additional CHESCA matrices acquired at different temperature values are available in the supporting material (Fig. S2). To see this figure in color, go online.

Figure 3

Three-dimensional bar plots showing the percentage distribution of residue pair Pearson’s correlation coefficients (R) at different temperatures ranging from 290 K (dark blue) to 316 K (red) for (A) ¹⁵N-only based CHESCA, (B) ¹H-only based CHESCA, and (C) combined chemical shift (¹⁵N-¹H)-based CHESCA. To see this figure in color, go online.

Figure 4

The average R value () for residue pair correlations was calculated across the five temperatures and plotted against the standard deviation of R (σ) in (B). Each circle represents a residue pair: orange (blue) circles correspond to residue pairs with negative (positive) mean R values. The black dashed lines (i.e., σ = <R> if |<R>| < 0.4 and σ = 1 − <R> if |<R>| > 0.6) capture the general approximate trend: residue pairs with high <R> values consistently exhibit low standard deviations, as illustrated in the zoomed images in (A and C), while residue pairs with lower <R> values tend to result in higher and more variable σ values. To see this figure in color, go online.

Figure 5

Schematic diagram explaining the rationale of the CLASS-CHESCA. (A) For each residue pair, four Pearson's correlation coefficients (R_NH–R_NN) are calculated, where R_NH corresponds to the correlation coefficient between the nitrogen chemical shift of the first residue in the pair and the proton chemical shift of the second residue, R_HN is for the proton chemical shift of the first residue and the nitrogen chemical shift of the second residue, R_HH is for the proton chemical shift of the first residue and the proton chemical shift of the second residue, and R_NN is for the nitrogen chemical shift of the first residue and the nitrogen chemical shift of the second residue. (B) Residue pairs are then divided into five different classes (0–4) depending on how many of the correlation coefficients (R_NH–R_NN) are greater than a set threshold (R_th), e.g., 0.98. (C) Examples of residue pairs that fall under different classes. For each class, one residue pair is provided and the four different correlations (corresponding to R_NH–R_NN) are shown in different columns. The perturbation states are indicated by color-coded circles, as in Fig. 1. (D) CLASS-CHESCA matrix, where the different colors relate to the type of class that the residue pair correlation belongs to, as shown by the legend on the right of the panel. The secondary structure of the EPAC1-CNBD is mapped on the matrix whereby gray boxes represent α helices and brown boxes correspond to β sheets. Regions highlighted in gray denote the phosphate binding cassette (PBC), ionic latch (IL), base binding region (BBR), and the hinge. To see this figure in color, go online.

Figure 6

(A) Venn diagram of CHESCA-identified residues, including the proposed T-, CL-, and CLASS-CHESCA ensembles aimed at identifying the core allosteric network of EPAC1. The cyan ensemble refers to classes 3 and 4 of the CLASS-based CHESCA at 306 K, the green ensemble is for the set of residues from T-CHESCA with mean Pearson’s coefficient greater than 0.98 and standard deviation less than 0.02 across temperatures, and the dark (light) gray ensemble describes the residues identified from complete (single)-linkage clustering, i.e., CL (SL)-CHESCA, at 306 K. Core residues that are common to all ensembles are highlighted in orange. (B)–(F) The residues identified from each ensemble are mapped on the structure of EPAC1-CNBD and represented as surfaces colored as in (A). (F) Map of the core residues highlighted in orange on the Venn diagram with respective zoomed in panels. Notations for the structural elements are as in Fig. 1, A and B. To see this figure in color, go online.