RNA binding and chaperone activity of the E. coli cold-shock protein CspA

Abstract

Ensuring the correct folding of RNA molecules in the cell is of major importance for a large variety of biological functions. Therefore, chaperone proteins that assist RNA in adopting their functionally active states are abundant in all living organisms. An important feature of RNA chaperone proteins is that they do not require an external energy source to perform their activity, and that they interact transiently and non-specifically with their RNA targets. So far, little is known about the mechanistic details of the RNA chaperone activity of these proteins. Prominent examples of RNA chaperones are bacterial cold shock proteins (Csp) that have been reported to bind single-stranded RNA and DNA. Here, we have used advanced NMR spectroscopy techniques to investigate at atomic resolution the RNA-melting activity of CspA, the major cold shock protein of Escherichia coli, upon binding to different RNA hairpins. Real-time NMR provides detailed information on the folding kinetics and folding pathways. Finally, comparison of wild-type CspA with single-point mutants and small peptides yields insights into the complementary roles of aromatic and positively charged amino-acid side chains for the RNA chaperone activity of the protein.

Figure 1.

Ribbon representation of the E. coli CspA structure (PDB entry: 1MJC). The CspA orientation has been chosen so that the RNA interaction surface formed by strands β1–β3 is pointing towards the reader. Aromatic and positively charged Lysine side chains at the front side of the protein are also shown.

Figure 2.

Experimental setup for probing RNA refolding and chaperone activity by time-resolved NMR spectroscopy. (A) RNA annealing assay consisting in two complementary RNA hairpins that spontaneously form a hetero-duplex conformation after mixing. (B) Imino-¹H HADSOFAST experiment for sample mixtures of unlabeled and ¹⁵N-labeled RNA. The experiment is based on a SOFAST-HMQC (34,35) sequence with the first ¹H pulse applied with a PC9 shape (41) and a flip angle α = 120°. All ¹H and ¹⁵N 180° pulses are applied with REBURP shape (42). ¹H pulses are typically centered at 12.5 ppm covering a bandwidth of 5 ppm, while the ¹⁵N pulse shape is changed between experiments to achieve HADAMARD sign encoding: (I) no ¹⁵N pulse, ¹⁵N pulse covering a band width of (II) 155 ± 15 ppm (U and G), (III) 161 ± 5 ppm (U only) and (IV) 146 ± 5 ppm (G only). The Imino-¹H HADSOFAST experiment is repeated n-times during the refolding process in an indirect dimension of a 2D NMR experiment. (C) Schematic drawing of the spectral deconvolution process required for obtaining NMR imino ¹H spectra of the three different species: ¹⁵N-U, ¹⁵N-G and ¹⁴N-U/G.

Figure 3.

(A) 1D projections of an imino-¹H HADSOFAST experiment recorded at 950 MHz ¹H frequency during the RNA refolding assay depicted in Figure 2A, with ¹⁵N labeled CB and unlabeled ACB. The RNA concentrations after the mixing are ∼40 μM for CB and ∼60 μM for ACB. Resonances from ACB are annotated in red color, while resonances from CB are annotated in black. Bold letters indicate an NMR signal from the hetero-duplex, while italic letters are used for the annotation of resonances from the RNA hairpins. (B) 2D real-time spectrum of ¹⁴N-bound imino ¹H showing the ACB hairpin decay and duplex buildup (in presence of an equimolar amount of CspA) along the kinetic spectral dimension. Note that the ACB hairpin signals do not decay to zero because of an excess of ACB versus CB in the sample. (C) Normalized kinetics measured for RNA duplex formation in the absence of protein chaperone, in the presence of equimolar amounts of CspA, and a 4-fold excess of CspA. The displayed curves have been obtained by averaging over all individual traces from well-resolved imino ¹H of the RNA hetero-duplex.

Figure 4.

RNA hairpin unfolding and hetero-duplex folding rates, measured for the slow kinetic phase for individual base pair hydrogen bonds, are given and color-coded on the RNA structures in the absence (A) and presence (B) of CspA.

Figure 5.

Imino ¹H NMR spectra of ACB (A) and CB (B) RNA hairpins in the absence (black) and presence (red) of an equimolar amount of CspA. The base regions of 2D ¹H–¹³C correlation spectra are plotted in (C) for ACB and in (D) for CB. The observed spectral changes are summarized in (E) and (F) on the secondary structure of the two RNA hairpins. Closed circles indicate RNA bases with ¹H–¹³C cross peaks that show significant line broadening in the presence of CspA, while dashed circles are used for RNA bases experiencing NMR chemical shift changes. Also the weakening of particular hydrogen bonds as inferred from the imino ¹H spectra and the measured hydrogen exchange rates is highlighted by red color.

Figure 6.

NMR characterization of the CspA–ACB interaction at 20°C. ¹H–¹⁵N BEST-TROSY (28) spectra recorded for CspA in the presence of different amounts of RNA are superposed and color-coded in (A). A large number of amide sites show extensive line broadening in the presence of CspA, while others undergo peak shifts, indicating a rather fast exchange process between the free and the bound form. (B) Fitting the measured chemical shift data for CspA and ACB together, assuming a CspA:ACB stoichiometry of 1:1, yields an apparent K_d of 12 ± 2 μM. (C) Residues with extensive line broadening upon CspA binding (11–14, 16–20, 31–39, 42–44, 63, 69) are highlighted in blue on the ribbon structure of CspA (left panel), while residues undergoing conformational exchange (11–13, 18–21, 30–31, 35, 42–43, 53–56, 66–68) as probed by relaxation-dispersion NMR are shown in red (central panel). In addition, the electrostatic surface potential of CspA is plotted (right panel) using the same orientation of the molecule.

Figure 7.

NMR characterization of the CspA–ACB interaction at 45°C. (A) Superposition of NMR ¹H–¹⁵N correlation spectra recorded for different CspA–ACB mixtures. (B) Chemical shift differences observed between free and bound CspA. The plotted values have been calculated as . The peptide regions with the largest chemical shift changes (Δδ >1) are highlighted in light blue. As expected, this RNA binding site is identical to the results obtained at 20°C (Figure 6C). (C) Intensity ratios measured for individual ¹H–¹⁵N correlation peaks from two CPMG-RD spectra using CPMG frequencies of 33 and 1000 s⁻¹, respectively. CPMG intensity ratios of free CspA are plotted in black, while those of CspA in complex with ACB RNA are shown in red. (D) {¹H}–¹⁵N heteronuclear NOE (HETNOE) data measured for free CspA (black) and CspA in complex with ACB RNA (red).

Figure 8.

Normalized real-time NMR kinetics measured for RNA duplex formation in the presence of (A) CspA-WT (black), CspA-F12I (green), CspA-F18I (red), F-F-F peptide (purple), and (B) PV core (black), triple-K (green) and poly-K (red). In addition, the folding kinetics in the absence of any protein or peptide is shown in gray in both graphs. The displayed data points have been obtained by averaging over all individual traces from well-resolved imino ¹H of the RNA hetero-duplex, and the lines represent the results of a data fit to a bi-exponential model.

Figure 9.

Possible (parallel) duplex formation models. (A) Initially, the two strands can form structurally different encounter complexes (EC-1 and EC-2) with only weak and transient interactions. Then, respective transition states are formed with distinct fast (TS-1) and slow (TS-2) kinetic signatures. Finally, the folding reaction proceeds via disruption of the hairpin base pair interactions, followed by the formation of new intermolecular base pairs leading to the final hetero-dimer RNA. (B) Gibbs-free energy profile corresponding to the folding model presented in (A). Charge compensation reduces the activation energy for encounter state formation, while chaperone-induced base-pair opening lowers the free energy of the transition states.