Knotty is nice: Metabolite binding and RNA-mediated gene regulation by the preQ1 riboswitch family

Abstract

Riboswitches sense specific cellular metabolites, leading to messenger RNA conformational changes that regulate downstream genes. Here, we review the three known prequeosine₁ (preQ₁) riboswitch classes, which encompass five gene-regulatory motifs derived from distinct consensus models of folded RNA pseudoknots. Structural and functional analyses reveal multiple gene-regulation strategies ranging from partial occlusion of the ribosome-binding Shine-Dalgarno sequence (SDS), SDS sequestration driven by kinetic or thermodynamic folding pathways, direct preQ₁ recognition by the SDS, and complete SDS burial with in the riboswitch architecture. Family members can also induce elemental transcriptional pausing, which depends on ligand-mediated pseudoknot formation. Accordingly, preQ₁ family members provide insight into a wide range of gene-regulatory tactics as well as a diverse repertoire of chemical approaches used to recognize the preQ₁ metabolite. From a broader perspective, future challenges for the field will include the identification of new riboswitches in mRNAs that do not possess an SDS or those that induce ligand-dependent transcriptional pausing. When choosing an antibacterial target, the field must also consider how well a riboswitch accommodates mutations. Investigation of riboswitches in their natural context will also be critical to elucidate how RNA-mediated gene regulation influences organism fitness, thus providing a firm foundation for antibiotic development.

Keywords: Shine-Dalgarno sequence; allosteric binding; cooperativity; molecular recognition; protein translation; pseudoknot; queuosine; quintuple base motif; transcription regulation.

Figure 1

Gene-regulatory models of bacterial riboswitches, biosynthesis of queuosine and preQ₁class I covariation models.A, left panel: transcription control entails ligand-dependent partitioning between mRNA folds that support or disfavor polymerase activity. In low levels of ligand (star), the riboswitch aptamer folds poorly, allowing formation of an antiterminator hairpin that favors transcription. When abundant the ligand supports an alternate fold, resulting in a strong terminator helix followed by a polyuridine tract that disfavors transcription. Right panel: translation control entails ligand-dependent partitioning between folds that support or disfavor ribosome access to the Shine-Dalgarno sequence (SDS). In low levels of ligand (star), the riboswitch aptamer folds poorly, allowing access to the SDS for translation initiation. When ligand is abundant an alternative fold favors ligand binding, resulting in SDS burial that attenuates translation. B, the prokaryotic pathway for de novo queuosine biosynthesis begins with GTP, which is converted to preQ₁ (7-deaza-7-aminomethyl guanine) by the ghc1 gene product (GTP cyclohydrolase I) and enzymes encoded by the queCDEF operon (136). The preQ₁ precursor is preQ₀, which contains a nitrile group that is reduced by QueF in the Q pathway (137). However, preQ₀ also serves as a scaffold for other natural products such as toyocamycin and sangivamycin, which have antibacterial properties (136). Bacteria can also salvage preQ₀ and preQ₁ using transporters encoded by queT, yhhQ, and other genes (119, 120, 138). PreQ₁-sensing riboswitches have been found to regulate the biosynthetic queCDEF operon, as well as queT and yhhQ (69, 70, 74). Regardless of the source, preQ₁ is incorporated by the tgt gene product (tRNA-guanine transglycosylase) into the wobble position of specific tRNAs, where it is modified into Q by queA and queG gene products (136). C, covariation models of the three class I preQ₁ riboswitch subtypes based on (69) and rfam.org/family/RF00522. Y symbolizes purine and R is pyrimidine. PreQ_1, prequeosine₁.

Figure 2

Secondary structures, global folds, and expression platforms of class I preQ₁riboswitches.A, secondary structure of the H-type pseudoknot (PK) from Thermoanaerobacter tengcongensis (Tte) based on cocrystal structures of preQ₁-I_II riboswitches (79, 88). Here and elsewhere, colors correspond to specific PK pairing (P) or loop (L) sequences as defined (85); preQ₁ is labeled “Q_1;” base interactions are indicated by Leontis–Westhof symbols (139). The Shine-Dalgarno sequence (SDS) (expression platform) is highlighted yellow; the anti-(a)SDS is highlighted cyan. B, ribbon diagram of the Tte structure (PDB entry 3q50) (78). Helix P2 contains the aSDS-SDS pair (base and ribose rings filled yellow and cyan) that supports the gene-off conformation; the α-preQ₁ ligand is depicted as a surface model (green). C, close-up view of the Tte riboswitch pocket floor showing the conserved G5-C16 pair in helix P1, whose sugar edge interacts with multiple adenines via A-amino kissing interactions (80). The view is rotated −180° about the y-axis from panel B. D, close-up view of the Tte riboswitch pocket ceiling showing canonical and noncanonical pairs in the aSDS–SDS interaction of P2. A C7⁺•G11-C30 triple forms directly above the α-preQ₁ site. The view is rotated 90° about the y-axis from panel C. E–H, diagrams comparable to A–D for the Carnobacterium antarticum (Can) preQ₁-I_I riboswitch (PDB entry 8fb3) (49). I–L, Diagrams comparable to panels A–D for the Escherichia coli (Eco) preQ₁-I_III riboswitch (PDB entry 8fza) (47).

Figure 3

The recurring quintuple-base motif transitions A-rich sequences interacting with the minor groove to major-groove base triples.A, a QBM from the double-ENE RNA stability fold in complex with a 28-mer poly(A) RNA (PDB entry 7jnh) (100). An upper A-form helix (cyan) uses minor-groove interactions with adenines (purple) to transition to a new helix (pale yellow) stabilized by a major-groove U-A•U triple. Nucleotides of the QBM are colored cyan and purple. B, schematic diagram of QBM interactions in panel A. C, QBM of Campylobacter jejuni Cas9 in complex with single-guide RNA (101) (PDB entry 5x2g). Colors are similar to panel A. D, schematic diagram of QBM interactions in panel C. E, QBM of the Eco preQ₁-I_III riboswitch (PDB entry 8fza). Transition to the underlying base triple (pale yellow) has been altered to bind preQ₁ through a base quadruple, located beneath the QBM; this ligand-recognition feature is observed in all class I preQ₁ riboswitches. F, schematic diagram of QBM interactions in panel E. ENE, element for nuclear expression.

Figure 4

Covariation models, secondary structures, global folds, and expression platforms of class II and class III preQ₁riboswitches.A, covariation model of the preQ₁-II riboswitch based on (69, 70) and rfam.org/family/RF01054. B, secondary structure based on the Lactobacillus rhamnosus (Lrh) cocrystal structure (PDB entry 4jf2) (76). C, ribbon diagram depicting the HL_out PK tertiary fold of the Lrh riboswitch. The riboswitch displays a novel mode of preQ₁ recognition compared to class I, dubbed the “γ” mode if binding. D, the ceiling of the preQ₁ binding pocket. E, the expression platform comprises the SDS paired with the aSDS, forming helix P3; the first base of the SDS, A71, forms the floor of the binding pocket directly below preQ₁ (semitransparent green surface) (76). (cont’don next page) F, covariation diagram of preQ₁-III riboswitches based on (69) and rfam.org/family/RF02680. G, secondary structure based on the Faecalibacterium prausnitzii (Fpr) cocrystal structure (PDB entry 4rzd). The paired aSDS-SDS helix P5 is based on the observed crystal contact; the predicted aSDS (cyan) and SDS (yellow) are based on the consensus model in panel F. H, ribbon diagram of the HL_out PK tertiary fold of the Fpr riboswitch (77). The γ-mode of preQ₁ binding is similar to the preQ₁-II riboswitch. I, the ceiling of the preQ₁ binding pocket is made by an A18•A6 pair in helix P1. J, The floor of the binding pocket comprises three underlying layers of major-groove base triples at the interface between J1-2 and P2. K, Computational model of the P5 helix expression platform based on targeted molecular dynamics starting from the co-crystal structure in panel H (77). The first two bases of the SDS are buried. PK, pseudoknot; SDS, Shine-Dalgarno sequence; preQ₁, prequeosine₁.

Figure 4

Covariation models, secondary structures, global folds, and expression platforms of class II and class III preQ₁riboswitches.A, covariation model of the preQ₁-II riboswitch based on (69, 70) and rfam.org/family/RF01054. B, secondary structure based on the Lactobacillus rhamnosus (Lrh) cocrystal structure (PDB entry 4jf2) (76). C, ribbon diagram depicting the HL_out PK tertiary fold of the Lrh riboswitch. The riboswitch displays a novel mode of preQ₁ recognition compared to class I, dubbed the “γ” mode if binding. D, the ceiling of the preQ₁ binding pocket. E, the expression platform comprises the SDS paired with the aSDS, forming helix P3; the first base of the SDS, A71, forms the floor of the binding pocket directly below preQ₁ (semitransparent green surface) (76). (cont’don next page) F, covariation diagram of preQ₁-III riboswitches based on (69) and rfam.org/family/RF02680. G, secondary structure based on the Faecalibacterium prausnitzii (Fpr) cocrystal structure (PDB entry 4rzd). The paired aSDS-SDS helix P5 is based on the observed crystal contact; the predicted aSDS (cyan) and SDS (yellow) are based on the consensus model in panel F. H, ribbon diagram of the HL_out PK tertiary fold of the Fpr riboswitch (77). The γ-mode of preQ₁ binding is similar to the preQ₁-II riboswitch. I, the ceiling of the preQ₁ binding pocket is made by an A18•A6 pair in helix P1. J, The floor of the binding pocket comprises three underlying layers of major-groove base triples at the interface between J1-2 and P2. K, Computational model of the P5 helix expression platform based on targeted molecular dynamics starting from the co-crystal structure in panel H (77). The first two bases of the SDS are buried. PK, pseudoknot; SDS, Shine-Dalgarno sequence; preQ₁, prequeosine₁.

Figure 5

Overview of the α, β, and γ modes of ligand binding by the three classes of preQ₁riboswitches.A–C, the most common mode of preQ₁ recognition, α, occurs in all class I preQ₁ riboswitches as exemplified by the Can preQ₁-I_I, Tte preQ₁-I_II, and Eco preQ₁-I_III riboswitch binding pockets. The guanine-like face of preQ₁ is recognized by a strictly conserved cytosine specificity base that uses WC pairing (e.g., C17 in the Can preQ₁-I_I riboswitch). Each pocket contains a conserved WC-base–paired floor, equivalent to the Can riboswitch G5-C18 pair. Conserved bases recognize the minor-groove-edge equivalent of preQ₁. In all types, the preQ₁ methylamine group hydrogen bonds with the O6 keto moiety of guanine in the pocket floor. The Eco preQ₁-I_III riboswitch also shows hydrogen bonding to the nonbridging phosphate of position 12. D, the β-mode of recognition has been observed only in preQ₁-I_I riboswitches, as exemplified by the Can riboswitch. Unlike the α-mode of recognition, there is no specificity base at the WC face of the β ligand. The minor-groove-edge equivalent of preQ₁ is recognized by conserved bases. The methylamine group of preQ₁ hydrogen bonds to the O4 keto of U16 and the O6 keto group of the underlying α-preQ₁, which stacks beneath the β-preQ₁ ligand. E and F, the γ-mode of recognition represents a third unique way to bind preQ₁. This recognition pattern is characterized by a conserved trans–WC interaction between a cytosine base of the RNA pocket and the guanine-like face of preQ₁, as exemplified by preQ₁-II and preQ₁-III riboswitches. A water molecule at the O6 keto oxygen of preQ₁ mediates a contact with the 2′-hydroxyl of the specificity base. A conserved uracil base from the pocket reads the minor-groove edge equivalent of preQ₁. Watson–Crick; preQ₁, prequeosine₁.

Figure 6

Summary of gene-regulation strategies used by members of the preQ₁riboswitch family. The nomenclature adopted here is based on a recent review of riboswitch-mediated gene regulation see (128). A, preQ₁-I_I riboswitches bind two preQ₁ molecules that stabilize pseudoknot formation to control genes of the queCDEF operon and queT (74). Although the SDS is not strongly buried, this occlusion is sufficient to attenuate translation initiation. Notably, placement of the SDS outside the aptamer still allows gene regulation, suggesting the pseudoknot itself is more important for gene regulation than SDS sequestration (49). B, some preQ₁-I_II riboswitches regulate queT genes by burial of the first two SDS nucleotides in the presence of preQ₁, which inhibits ribosome binding. C, other preQ₁-I_II riboswitches regulate genes of the queCDEF operon by use of an expression platform that controls transcription. The P2 pseudoknot helix is stabilized by preQ₁ binding, which favors formation of an intrinsic terminator hairpin that stops the RNA polymerase from transcribing the downstream gene. D, the preQ₁-I_III riboswitch and preQ₁-II riboswitch regulate yhhQ and queT genes, respectively, by complete sequestration of the associated SDS upon preQ₁ binding. These riboswitches are the quintessential examples of metabolite-sensing RNA regulators that control translation. E, preQ₁-III riboswitches bind preQ₁ using a separate domain that folds as an HL_out pseudoknot. The expression platform contains an SDS at the 3′-end of the riboswitch that docks with the loop of P4 to create an aSDS-SDS helix called P5. PreQ₁ binding stimulates the population of riboswitches undergoing dynamic docking and undocking of P5, which occludes the SDS from the ribosome. PreQ₁, prequeosine₁; SDS, Shine-Dalgarno sequence.