Structure of the Escherichia coli ProQ RNA-binding protein

Abstract

The protein ProQ has recently been identified as a global small noncoding RNA-binding protein in Salmonella, and a similar role is anticipated for its numerous homologs in divergent bacterial species. We report the solution structure of Escherichia coli ProQ, revealing an N-terminal FinO-like domain, a C-terminal domain that unexpectedly has a Tudor domain fold commonly found in eukaryotes, and an elongated bridging intradomain linker that is flexible but nonetheless incompressible. Structure-based sequence analysis suggests that the Tudor domain was acquired through horizontal gene transfer and gene fusion to the ancestral FinO-like domain. Through a combination of biochemical and biophysical approaches, we have mapped putative RNA-binding surfaces on all three domains of ProQ and modeled the protein's conformation in the apo and RNA-bound forms. Taken together, these data suggest how the FinO, Tudor, and linker domains of ProQ cooperate to recognize complex RNA structures and serve to promote RNA-mediated regulation.

Keywords: FinO; ProQ; RNA chaperone; protein–RNA interactions; regulatory RNA; riboregulation.

FIGURE 1.

Domain organization and NMR spectroscopy of ProQ. (A) Linear domain representation of ProQ, with the ProQ/FinO N-terminal domain (NTD, blue), the disordered linker region (yellow), and the C-terminal domain (CTD, green). (B) Sequence alignment of ProQ proteins from Salmonella enterica, E. coli, Legionella pneumophila, Neisseria meningitdis, and FinO from E. coli. Sequence regions corresponding to the NTD, linker, and CTD are highlighted by boxes colored as in A. Electro-positive residues predicted to be involved in RNA binding (from Fig. 2) are indicated with asterisks. (C) ¹⁵N relaxation data for full length apo ProQ. Per residue plots for T₁, T₂, and the {¹H}-¹⁵N heteronuclear NOE ratio are shown in the top, middle, and bottom panels, respectively. Individual data points are colored blue, yellow, or green to signify whether they correspond to residues from the NTD, linker, or CTD, respectively.

FIGURE 2.

The NMR structure of the N-terminal domain of ProQ. (A) Ensemble of the top five solutions shown as cartoon representation in three orientations, and colored as rainbow from blue to red. (B) Electrostatic surface representation of the ProQ NTD, three views as in Figure 2A (generated in Coot) (Emsley et al. 2010). Electro-negative patches are colored red and electro-positive patches are colored blue. Residues forming putative RNA-binding surfaces are indicated with arrows.

FIGURE 3.

Structural comparison of the N-terminal domain of ProQ with other proteins. The lowest energy model of the E. coli ProQ NTD is shown in two orientations in cartoon representation colored as rainbow (blue–red, N-C termini). Chains A and F of N. meningitidis NMB1681, and E. coli FinO are shown in the same orientations, also colored as rainbow from blue to red as a comparison.

FIGURE 4.

The NMR structure of the C-terminal domain of ProQ. (A) Ensemble of the top five solutions shown as cartoon representation and colored as rainbow from blue to red. (B) Electrostatic surface representation of the ProQ CTD (two orientations as in Fig. 4A). Charged surface exposed amino acids are labeled.

FIGURE 5.

Structural comparison of the C-terminal domain of ProQ with other proteins. The lowest energy model of the ProQ CTD is shown in two orientations in cartoon representation colored in blue to red rainbow. Tudor domain 2 of human PHD finger protein 20 (PDB 3QII) and E. coli Hfq (PDB 4PNO) are shown in the same orientations, also colored as rainbow from blue to red. A UMP molecule bound in the Hfq model is shown as stick representation.

FIGURE 6.

Hfq and ProQ do not physically interact directly and compete for interaction with the malM 3′ UTR. ProQ and Hfq were mixed at a 1:6 molar ratio and purified by gel filtration chromatography. The chromatogram is shown in A, and SDS–PAGE gel of the eluted fractions is shown in B. MalM 3′ UTR, ProQ, and Hfq were mixed in equimolar ratios, run on a native polyacrylamide gel, and stained for RNA (C) and protein (D). N.B. Full length apo-ProQ does not enter the native gel.

FIGURE 7.

HDX mapping of RNA-binding surfaces on ProQ. Two views of the NMR models of ProQ N- and C-terminal domains are shown as cartoon representation with semitransparent surfaces. Regions protected upon binding to either SraB or cspE 3′ UTR in HDX experiments are colored with a heat-map (gray = less/no protection; blue = more protection). The boundaries between the heat-map divisions are shown on the bar at the bottom.

FIGURE 8.

Solution shape of ProQ by small-angle X-ray scattering. (Top) The molecular envelope for ProQ is shown as semitransparent green spheres, with manually docked NMR models of the N- and C-terminal domains shown as solid blue surfaces. (Bottom) An extended 63 amino acid peptide is shown in scale to the upper panels to illustrate the potential distance the unmodeled linker region can span. The annotated sizes are direct measurements of the shown DAMMIF model. The average values from fitting the SAXS data for the D_max and R_g are 173.0 and 43.3 Å, respectively.

FIGURE 9.

Solution shape of ProQ in complex with the sRNA SraB. (A) Three views of the molecular envelopes of ProQ:SraB and apo ProQ, shown as green spheres and to the same scale. The NMR models of the N- and C-terminal domains of ProQ are shown as blue surfaces and are manually docked into the SAXS envelope of apo-ProQ. (B) Secondary structural model of SraB, generated by RNAfold. The annotated sizes are direct measurements of the shown DAMMIF models. The average values for the D_max and R_g from fitting the ProQ:SraB SAXS data are 254.0 and 54.36 Å, respectively.

FIGURE 10.

Docking of ProQ and SraB into the ProQ:SraB solution envelope. Two views of the solution envelope of ProQ:SraB (as in Fig. 7) shown as gray spheres, with manually docked NMR models of the ProQ N- and C-terminal domains shown as solid blue and green surfaces, respectively. A 63 amino acid extended peptide is shown as yellow ribbon to represent the ProQ linker region. A model of SraB generated by SimRNAweb with some helix positions rotated is manually docked into the SAXS envelope and is shown as cartoon representation.