Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press Structure and ligand binding of the ADP-binding domain of the NAD+ riboswitch Lin Huang1, Jia Wang and David M. J. Lilley* Cancer Research UK Nucleic Acid Structure Research Group, MSI/WTB Complex, The University of Dundee, Dow Street, Dundee DD1 5EH, U.K. * To whom correspondence should be addressed 1. Current address : Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation, Medical Research Center, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou 510120 , P. R. China and RNA Biomedical Institute, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou 510120, P. R. China Keywords : Riboregulation; RNA structure; X-ray crystallography; metal ions; nucleotide binding Running title: The structure of the NAD+ riboswitch Submitted to the RNA journal 20 January 2020 Editorial enquiries to Professor Lilley: Tel: (+44)-1382-384243 Email: [email protected] Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press ABSTRACT The nadA motif is the first known NAD+-dependent riboswitch, comprising two similar tandem bulged stem-loop structures. We have determined the structure of the 5’ domain 1 of the riboswitch. It has three coaxial helical segments, separated by an ACANCCCC bulge and by an internal loop, with a tertiary contact between them that includes two C:G base pairs. We have determined the structure with a number of ligands related to NADH, but in each case only the ADP moiety is observed. The adenosine adopts an anti conformation, and forms multiple hydrogen bonds across the width of the sugar edge of the penultimate C:G base pair of the helix preceding the bulge, and the observed contacts have been confirmed by mutagenesis and calorimetry. Two divalent metal ions play a key structural role at the narrow neck of the bulge. One makes direct bonding contacts to the di-phosphate moiety, locking it into position. Thus the nucleobase, ribose and phosphate groups of the ADP moiety are all specifically recognised by the RNA. The NAD+ riboswitch is modular. Domain 1 is an ADP binding domain that may be ancient, and could potentially be used in combination with other ligand binding motifs such as CoA. INTRODUCTION Riboswitches are cis-acting elements that occur predominantly in the 5’-untranslated regions of bacterial mRNA to control gene expression (1-3). The RNA folds to create a specific binding site for a small molecule such as a metabolite or ion so as to change the structure and thereby alter the rate of either transcription or translation of the downstream gene. In general the gene product will be functionally related to the small molecule that binds the riboswitch, such as being an enzyme in its biosynthetic pathway. Approximately 40 classes of riboswitches have been identified to date. While the biosynthesis of many co-enzymes in bacteria is controlled by riboswitches, until recently none had been identified for the coenzyme nicotinamide-adenine dinucleotide (NAD). This is an important coenzyme that is involved in oxidation-reduction reactions in cellular metabolism (Figure 1A), involved in transfer of electrons by switching between two oxidation states NAD+ and NADH. However, in the course of a searching intergenic RNA sequences, Weinberg et al (4) identified a putative folded motif in the 5’-UTR of the nadA genes involved in NAD+ biosynthesis within the Acidobacteria phylum (5), that might act as a riboswitch. Very recently the Breaker laboratory have presented evidence that _______________________________________________________________________________ Huang, Wang & Lilley The structure of the NAD+ riboswitch 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press the nadA motif serves as an NAD+-dependent riboswitch to control expression of NAD+ biosynthetic genes (6). They identified over 100 unique sequences of this motif with an average length of 160 nucleotides, and deduced a consensus sequence and putative secondary structure. Examination of the consensus sequence of the nadA motif reveals that it has two domains of similar sequence and secondary structure, each comprising a stem-loop with a conserved bulge of sequence ACANCCCC (Figure 1B). Domain 1 consists of three helices P1, P1a and P1b, where P1 and P1a are interrupted by the bulge, and P1a and P1b by a small conserved internal loop. Domain 2 is very similar, but lacks the internal loop, and has a putative ribosome binding site on the 3’ side of the P1 helix. At first sight the tandem architecture is reminiscent of the glycine riboswitch that binds two molecules of glycine cooperatively to narrow the concentration range over which it switches (7-10). However, using in-line probing Malkowski et al (6) measured binding isotherms that provided no evidence for cooperative binding. Their probing experiments also indicated that only the ADP moiety of NADH (and ADP itself as well as a number of derivatives including AMP and ATP) bound to domain 1. In addition, they could find no evidence for binding of nicotinamide or other NAD+-related compounds to domain 2 despite its similarity in sequence to domain 1. We therefore crystallized the NAD+ riboswitch domain I bound to NADH, ADP and other NAD+- related compounds, and solved their structures by X-ray crystallography. We find that helices P1, P1a and P1b are coaxial, with a base paired interaction between the ACANCCCC bulge and the internal loop generating a pseudo-three-way junction that creates a binding site for ADP and related compounds. However no interaction with the nicotinamide moiety of NAD+ or NADH can be observed in any complex. The adenine nucleoside of the ligand is hydrogen bonded to the riboswitch RNA, and the phosphate groups interact directly with one of two magnesium ions bound in the narrow neck of the ACANCCCC bulge. RESULTS Crystallization of the ADP-binding domain of the NAD+ riboswitch A 52 nt RNA with the sequence of the Candidatus koribacter versatilis NAD+ riboswitch domain 1 was made by chemical synthesis and co-crystallized with a number of ligands related to NAD+ (Table S1). 5-Bromocytosine was incorporated at position 3 or positions 42 and 45 in the P1 helix in order to provide phase information. These species crystallized in space group I222, with good resolution, and some of the structures were solved using SAD from the bromine atoms (Table S2). _______________________________________________________________________________ Huang, Wang & Lilley The structure of the NAD+ riboswitch 3 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press The overall structure of the ADP-binding domain of the NAD+ riboswitch The structure of domain 1 with bound NADH ligand is shown in Figure 2. Ligand binding is discussed in detail below, but we note that electron density is only visible for the ADP moiety of the NADH. The three helices P1, P1a and P1b are coaxially stacked in the structure, effectively forming a single long helix. The conserved 8-nucleotide ACANCCCC bulge (beginning at A8) at the internal end of P1 extends away from the helix. A8 and C9 are stacked on each other, A10 lies inside the bulge, while A11 is extended away from the body of the RNA. This is then followed by a sharp turn whereby the loop passes back towards the helix. The nucleobases of C12-C15 are stacked (the C4 stack), and C15 is also stacked on A10. Thereafter G16 is paired with C45 to form the first base pair of the P1a helix. Between P1a and P1b helices is a small internal loop containing a conserved GYG (beginning at G36) that lies on the opposite strand to the 8-nucleotide bulge. C37 is base paired with G23, but G36 and G38 are base paired with C14 and C13 respectively in the C4 stack of the 8-nucleotide bulge. This explains why the length of the P1a helix is strongly conserved at 7 bp, as this is required to place the GYG and C4 sequences on the same side of the helix in order to form the two cross-strand base pairs. Furthermore, the great majority of the conserved nucleotides are contained within the bulged region and the interacting P1a-P1b loop (Figure S1). The 8-nucleotide bulge loop has an extremely narrow neck, with the phosphates at nucleotides 7, 8 and 9 approaching those at 14 and 15 with an average distance of 5 Å. The proR non-bridging oxygen atoms of A8 and C15 are just 3.9 Å apart, and the proS oxygen atoms of A7 and C15 are separated by 4.4 Å. However, electrostatic repulsion between the strands will be reduced by the binding of a number of metal ions to the RNA in this region. The backbones of the 5’ and 3’ ends of the loop are bridged by hydrated metal ions making both inner and outer sphere interactions with the RNA and the phosphate groups of the ligand (Figure 3A). The experimental phasing map shows unambiguous electron density corresponding to the metal ions and inner sphere coordinated water molecules (Figure 3B, Figure S2). This is supported by electron density calculated from anomalous scatter from Mn2+ ions diffused into the crystals that is observed in the same locations (Figure S3). Metal ion M1 bonds directly to the proR oxygen atoms of A8 and C9 plus N7 of A10. The inner sphere water molecule apical with the C9 proR oxygen is hydrogen bonded to C14 proR, so bridging the two strands in the neck of the loop. Metal ion M2 is bonded directly to the conserved A8 proS, and to non-bridging oxygen atoms of each phosphate group of the ADP moiety of NAD+.