Mechanism of NanR gene repression and allosteric induction of bacterial sialic acid metabolism

Abstract

Bacteria respond to environmental changes by inducing transcription of some genes and repressing others. Sialic acids, which coat human cell surfaces, are a nutrient source for pathogenic and commensal bacteria. The Escherichia coli GntR-type transcriptional repressor, NanR, regulates sialic acid metabolism, but the mechanism is unclear. Here, we demonstrate that three NanR dimers bind a (GGTATA)₃-repeat operator cooperatively and with high affinity. Single-particle cryo-electron microscopy structures reveal the DNA-binding domain is reorganized to engage DNA, while three dimers assemble in close proximity across the (GGTATA)₃-repeat operator. Such an interaction allows cooperative protein-protein interactions between NanR dimers via their N-terminal extensions. The effector, N-acetylneuraminate, binds NanR and attenuates the NanR-DNA interaction. The crystal structure of NanR in complex with N-acetylneuraminate reveals a domain rearrangement upon N-acetylneuraminate binding to lock NanR in a conformation that weakens DNA binding. Our data provide a molecular basis for the regulation of bacterial sialic acid metabolism.

Fig. 1. NanR binds a conserved operator site in three distinct operons.

a The sialoregulon consists of: nanATEK-yhcH (yellow), which is the core sialic acid catabolic pathway^,; yjhBC (orange), which encodes proteins of unknown function that are hypothesized to process less common variants of sialic acid^,; and nanCMS (blue), which is responsible for outer membrane transport and periplasmic processing^–. b Sequence alignment of the operator sites present in each operon. Conserved nucleotides are marked with an asterisk. c Sequence logo highlights the conservation of the DNA bases within these operator sites (generated using WebLogo). The repeat sequence is shown in gray boxes. d The most common sialic acid, Neu5Ac, is shown in the chair conformation. The α-anomer and thermodynamically favorable β-anomer differ by the stereochemistry at the C2 position, highlighted in red.

Fig. 2. NanR binds DNA with positive cooperativity and nanomolar affinity.

a EMSA of NanR titrated against FAM-5ʹ-labeled (GGTATA)₃-repeat DNA (10 nM, sequence shown below EMSA). Three concentration-dependent complexes are observed (complexes 1–3). Non-specific binding is observed at concentrations of NanR >200 nM. Data presented are representative of at least three independent experiments. b Binding isotherm data from the EMSA (black) and fluorescence-detected analytical ultracentrifugation (AUC) (blue) data best fitted the Hill model (AIC value of 99%), as opposed to a non-cooperative binding model (AIC value of 1%). The mean ± SD of the K_D and Hill coefficient is shown. The independent experiment methods provide complementary data. c Continuous sedimentation coefficient [c(s)] distributions of NanR (gray dash), FAM-5ʹ-labeled (GGTATA)₃-repeat DNA (black dash), and NanR (0.78–794 nM; rainbow) titrated against the FAM-5ʹ-labeled (GGTATA)₃-repeat (80 nM). The binding isotherm data and fit are shown in b. The fitted parameters are shown in Supplementary Table 2. d Continuous sedimentation coefficient [c(s)] distributions of NanR and NanR^33–263 against the FAM-5ʹ-labeled (GGTATA)₃-repeat (3 µM) (blue and orange traces, respectively) and NanR^33–263 alone (gray dash). Sedimentation was monitored via the absorbance of the FAM label at 495 nm. All experimental parameters are summarized in Supplementary Table 3. e, f Deconvoluted sedimentation coefficient distributions resulting from the titration of NanR^33–263 (orange) into (GGTATA)₃-repeat DNA (0.6 µM; black): 1.8 µM NanR^33–263 (e) and 6.0 µM NanR^33–263 (f). A shift in the sedimentation coefficient is observed with increasing NanR^33–263 concentration, consistent with hetero-complex formation. The molar ratio of the integrated peaks (shaded in gray) is that of a dimer. The presence of excess protein free of any co-migrating DNA in (f) indicates that hetero-complex formation has reached saturation. All plots are presented as g(s) distributions with the molar concentration for each interacting partner (protein and DNA) plotted on the y-axis. Hydrodynamic parameters are in Supplementary Table 4.

Fig. 3. NanR dimers assemble on the (GGTATA)₃-repeat operator.

a Sedimentation coefficient distributions of the NanR (blue) and the (GGTATA)₃-repeat DNA operator (black) controls, measured individually at 280 and 260 nm, respectively. b–e Deconvoluted sedimentation coefficient distributions resulting from the titration of NanR into (GGTATA)₃-repeat DNA (0.5 µM): 0.5 µM NanR (b), 1.5 µM NanR (c), 3.0 µM NanR (d), and 5.0 µM NanR (e). A shift in the sedimentation coefficient is observed with increasing NanR concentration, consistent with hetero-complex formation. The molar ratio of the integrated peaks (shaded in gray) and the oligomeric state of each hetero-complex is shown. The presence of excess protein free of any co-migrating DNA in e indicates that hetero-complex formation has reached saturation. All plots are presented as g(s) distributions with the molar concentration for each interacting partner (protein and DNA) plotted on the y-axis. Hydrodynamic parameters are in Supplementary Table 5.

Fig. 4. Crystal structure of NanR in complex with Neu5Ac and Zn²⁺.

a The E. coli NanR monomer has two domains—an N-terminal DNA-binding domain (green) and C-terminal effector-binding domain (beige). The DNA-binding domain contains a highly conserved winged helix–turn–helix motif (left panel) where the wing is defined by an antiparallel two-stranded β-sheet (blue). The C-terminal domain is arranged into an antiparallel, all-α-helical bundle (right inset, rainbow). Helix α4 (purple) is a flexible linker connecting the two domains. b Cartoon/surface representation of the asymmetric domain-swapped dimeric structure formed by an exchange between monomers via the α4-helix (pink). The Neu5Ac-bound and Neu5Ac-free monomers are shown in beige and blue, respectively. c The effector binding site is located within a large polar cavity of the C-terminal domain. The direct or water-mediated hydrogen-bonding residues (gray sticks) that coordinate Neu5Ac in its β-anomeric form and hold Zn²⁺ in an octahedral geometry are indicated, while water molecules are depicted as yellow spheres. d An overlay of the Neu5Ac-bound C-terminal domain (beige) and Neu5Ac-free C-terminal domain (blue) illustrates the effector-induced conformational changes. e An overlay of the Neu5Ac-bound monomer (beige) and Neu5Ac-free monomer (blue) further illustrates effector-induced conformational changes. f Surface depiction of Neu5Ac-bound (beige) and Neu5Ac-free (blue) monomers shows that the binding of Neu5Ac compresses the monomer around the α4-helix (purple) relative to the Neu5Ac-free monomer by 28.3 Å. g Cartoon representation of the interface between the Neu5Ac-bound monomer (beige) and the Neu5Ac-free monomer (blue), facilitated by salt-bridge interactions.

Fig. 5. Cryo-EM structure of the NanR-dimer₁/(GGTATA)₂-repeat DNA complex.

a A 3.9 Å resolution cryo-EM reconstruction of the E. coli NanR-dimer₁/DNA hetero-complex. Reconstruction is shown as transparent isosurfaces fitted with the cartoon representation of the hetero-complex structure. b The (GGTATA)₂-repeat DNA oligonucleotide used to solve this dataset, where each repeat is highlighted red. c Overlay of the C-terminal effector-binding domain from the crystal structure (beige and blue) and the cryo-EM structure (green and black) closely match each other (RMSD = 2.1 Å). d Density in the cryo-EM reconstruction that is hypothesized to be Zn²⁺ (red circle). Coordinating histidine residues are shown as sticks. e Overlay of the N-terminal DNA-binding domain and the flexible α4-linking-helices from the crystal structure (blue) and DNA-bound cryo-EM model (green and black). Note the change in direction of the α4-helix between structures. f Overlay of the interaction between the N-terminal DNA-binding domain and DNA for NanR (green, black, and beige) and FadR (orange and blue) [PDB ID: 1HW2]. A sequence alignment of the N-terminal domain for each structure shows that many of the DNA-binding residues in FadR are conserved in NanR. An asterisk indicates fully conserved residues, a colon indicates conservation between residues of strongly similar properties, and a period indicates conservation between residues of weakly similar properties. Residues proposed to interact with the phosphate backbone are highlighted blue, while residues proposed to interact with the DNA base pairs are highlighted in pink. g–i Close up of the N-terminal domain to highlight the difference in DNA binding between monomer A (green) and B (black). Sidechains of putative DNA-binding residues that could be resolved are shown (e.g., Arg73 and Asn89). A red circle highlights where a sidechain could not be resolved in the opposing monomer.

Fig. 6. NanR undergoes large conformational changes on DNA binding.

a The three structural states of E. coli NanR (Neu5Ac-free (cyan), Neu5Ac-bound (beige), and DNA-bound (green)) are superimposed to highlight the conformational changes that occur as part of the allosteric mechanism. Notably, the N-terminal DNA-binding domain is reoriented via the flexible α4-helices (dash arrows). b The electron density between models shows that the α4-helices (light and dark pink) mediate a domain-swapped interface in the crystal structure (upper panel); however, these are no longer domain swapped when bound to DNA (lower panel). c Schematic to illustrate the structural changes the domains undergo between each state.

Fig. 7. Cryo-EM structure of the NanR-dimer₃/DNA hetero-complex.

a 2D class averages showing three possible NanR dimers bound to DNA in projection (white arrow). b Density map for the NanR-dimer₃/DNA hetero-complex. c Three NanR dimers and a DNA duplex were fitted using a rigid body method. Two NanR dimers are clearly visible within the density (gray isosurface) at each end of the DNA, while the middle dimer is less resolved as a result of orientation bias. The DNA flanking α2-helices provide fit confidence for the middle NanR dimer (green, inset). d Each NanR dimer is offset by approximately one half turn of the DNA helix. e Each GGTATA repeat of the DNA operator (in red) is separated by less than half a turn of the DNA helix. f Zoomed view of the NanR–DNA interface highlighting how each NanR bound to the (GGTATA)₃-repeat operator maintains an analogous binding mode to the NanR-dimer₁/DNA hetero-complex (Fig. 5), where the α3-helix binds in the major groove and the wing motif accommodates the minor groove of DNA. The N-terminal domains are hypothesized to interact with each other, given their proximity, through protein–protein interactions in the higher-order hetero-complex.

Fig. 8. E. coli NanR regulation of gene expression in the sialoregulon.

Based on the data from this study, a schematic of the proposed mechanism for the regulation of gene expression by NanR is provided. a To repress gene expression, dimers of NanR cooperatively and with nanomolar affinity bind to each of the three GGTATA repeats to form a NanR-dimer₃/DNA hetero-complex through rearrangement of their N-terminal DNA-binding domains. This cooperative assembly is believed to be mediated by an N-terminal extension, unique to NanR among closely related GntR-type regulators. b The binding of the allosteric modulator Neu5Ac (beige hexagon) to the C-terminal effector-binding domain of NanR triggers a conformational change that attenuates the protein–DNA interaction. This facilitates a conformational change that results in NanR disengaging from the (GGTATA)₃-repeat operator, relieving repression of gene expression.