Molecular impact of covalent modifications on nonribosomal peptide synthetase carrier protein communication

Abstract

Nonribosomal peptide synthesis involves the interplay between covalent protein modifications, conformational fluctuations, catalysis, and transient protein-protein interactions. Delineating the mechanisms involved in orchestrating these various processes will deepen our understanding of domain-domain communication in nonribosomal peptide synthetases (NRPSs) and lay the groundwork for the rational reengineering of NRPSs by swapping domains handling different substrates to generate novel natural products. Although many structural and biochemical studies of NRPSs exist, few studies have focused on the energetics and dynamics governing the interactions in these systems. Here, we present detailed binding studies of an adenylation domain and its partner carrier protein in apo-, holo-, and substrate-loaded forms. Results from fluorescence anisotropy, isothermal titration calorimetry, and NMR titrations indicated that covalent modifications to a carrier protein modulate domain communication, suggesting that chemical modifications to carrier proteins during NRPS synthesis may impart directionality to sequential NRPS domain interactions. Comparison of the structure and dynamics of an apo-aryl carrier protein with those of its modified forms revealed structural fluctuations induced by post-translational modifications and mediated by modulations of protein dynamics. The results provide a comprehensive molecular description of a carrier protein throughout its life cycle and demonstrate how a network of dynamic residues can propagate the molecular impact of chemical modifications throughout a protein and influence its affinity toward partner domains.

Keywords: adenylation domain; allostery; antibiotics; aryl carrier protein; nonribosomal peptide synthetase; nuclear magnetic resonance (NMR); protein dynamic; protein-protein interaction; structural biology.

Figure 1.

Covalent modifications of the yersiniabactin synthetase ArCP during NRPS synthesis. A PP transferase (PPT) converts apo-ArCP (A) to holo-ArCP (B) upon attachment of PP to Ser⁵². An adenylation domain loads salicylate (hexagon) on holo-ArCP (C). This A-domain first adopts an adenylation conformation (E) to activate the substrate through ATP (star) into a high-energy adenylate, SalAMP, where AMP is represented as a triangle. Next, a thioester conformation (F) allows for tethering the substrate to holo-CP with release of AMP. A third conformation may exist for the free domain (D). The proximal and distal amide groups in PP, discussed in the legend to Fig. 5, are labeled with p and d, respectively.

Figure 2.

Structure and dynamics of apo-ArCP. A, the mean from the 20 lowest-energy conformers (see supplemental data) is shown. The N-terminal end of loop 1 is colored yellow, and its C-terminal end is orange. The phosphopantetheinylation site is highlighted by an asterisk. B, alternative view obtained by rotation of 60° along a vertical axis. Shown are the order parameter (C) and exchange rates (D) obtained by model-free analysis of ¹⁵N relaxation for the apo-ArCP core, excluding disordered N and C termini. Full profiles can be seen in supplemental Figs. S2 and S3. E, 3D representation of ArCP dynamics. A thicker sausage indicates a lower-order parameter and picosecond-to-nanosecond dynamics, whereas microsecond-to-millisecond fluctuations are color-coded according to the magnitude of R_ex, from yellow to red. F, correlation of ArCP dynamics with structural variations in crystal structures of carrier proteins in NRPS multidomains. Thirteen carrier proteins were aligned according to their α-helical core (transparent) and compared with the 3D representation of ArCP dynamics. The ribbon in αI has been removed to highlight the fluctuation of its orientation.

Figure 3.

Binding of YbtE and ArCP in apo-form (A, D, and G), holo-form (B, E, and H), and SalNH-loaded form (C, F, and I). A–C, fluorescence anisotropy titrations show that YbtE binds to holo- and SalNH-ArCP with dissociation constants 10-fold lower than that of apo-ArCP. Error bars, S.D. of six (apo-ArCP) or five (holo- and SalNH-ArCP) titrations. The concentration of ArCP was 100 nm. D–F, isothermal titration calorimetry reveals that the interaction with apo-ArCP is driven entirely by entropy and slightly endothermic (D), whereas those of holo- and SalNH-ArCP are exothermic and supplemented with favorable entropy. ArCP was in the cell and at a concentration of 40 μm. The top panels show baseline-corrected raw data, and the bottom panels show the integrated heats. G–I, different forms of ArCP show different spectroscopic responses to binding by YbtE. Select regions of HN-HSQCs of 0.1 mm ¹⁵N-ArCP in the presence of YbtE at stoichiometries of 1:0 (red), 1:0.125 (purple), and 1:0.25 (cyan) are shown for apo-ArCP (G), holo-ArCP (H), and SalNH-ArCP (I). All spectra were scaled to identical contour levels.

Figure 4.

Spectroscopic perturbations induced by YbtE binding to apo-ArCP (A, D, G, and J), holo-ArCP (B, E, H, and K), and SalNH-ArCP (C, F, I, and L). A–C, CSPs and R_int plotted on the structures of ArCP. Global maxima (accounting for all forms) are used. CSPs are represented by a color gradient from blue (no CSP) to red (global maximum CSP in D–F), and R_int is represented by the thickness of the sausage, with the thickest line representing the maximum value in G–I. *, position of Ser⁵². D–F, CSPs representing shifts in HN-HSQCs of 1:0.125 ArCP/YbtE. For all forms, perturbations occur on α2 and α3, but they localize on the surface for apo-ArCP. G–I, relative change in intensity upon the addition of 0.125 eq of YbtE, R_int. Only specific residues are affected in apo-ArCP, but most residues in α2, α3, and loop 2 show dramatic changes in intensity for holo-ArCP and loaded ArCP. J–L, comparison between spectroscopic perturbations due to YbtE binding and a complex trapped in crystallographic studies for apo-SalNH (J), holo-SalNH (K), and SalNH-ArCP (L). Each form of ArCP (Protein Data Bank entries 5TTB (apo; this work), 2N6Y (holo), and 2N6Z (loaded)) was aligned onto EntB-ArCP in the EntE-EntB complex, Protein Data Bank entry 3RG2. For holo-ArCP (K), the phosphopantetheine group is shown in green when docked with ArCP and in black when extended between the N-terminal A(N) and C-terminal A(C) subdomains of EntE. The adenylate mimic used in the crystallographic study is shown in red spheres.

Figure 5.

Phosphopantetheinyl amide NMR signals from holo-ArCP show a complex response to binding by YbtE. A–E, proximal amide group (p in Fig. 1). A, two signals can be observed in free holo-ArCP (100 μm). As the concentration of YbtE is increased to 12.5 μm (B), 25 μm (C) 37.5 μm (D), and 50 μm (E), the signal marked with a blue plus shifts significantly and decreases in intensity, whereas the minor signal, marked with a black cross, shows little change in chemical shift. New signals also begin to appear, although it is not possible to assign them to either the major or minor PP signal. F–J, the distal group (d in Fig. 1) is also subject to an unusual perturbation, although spectral crowding hampers the interpretation. The question marks emphasize perturbations of ambiguous origin. The cross and plus symbols are not meant to be related with those used in A–E.

Figure 6.

Modulation of protein dynamics imparted by covalent modifications of ArCP. A–C, 3D representation of significant variations resulting from phosphopantetheinylation (A), salicylate (Sal) attachment (B), and combined effect of S-salicylyl-phosphopantetheine (C). The thickness of the sausage reflects the magnitude of differences in order parameters (H–J, corresponding to A–C, respectively), with a thicker line emphasizing residues varying by more than one S.D. value from the mean, accounting for errors (shown with black dots in H–J). The color, from pale to dark, reports on changes in R_ex. D–F, 3D representations of the dynamics of apo-ArCP (D), holo-ArCP (E), and loaded ArCP (F). G, apparent allosteric effect resulting from the modulation of dynamics shown in A–C. H–J, differences in order parameters, S², used to generate the 3D representations in A–C. Dashed horizontal lines, limits for one S.D., used to emphasize residues with significant changes.