Structure of the Eukaryotic Thiamine Pyrophosphate Riboswitch with Its Regulatory Ligand

See allHide authors and affiliations

Science  26 May 2006:
Vol. 312, Issue 5777, pp. 1208-1211
DOI: 10.1126/science.1128451


Riboswitches are untranslated regions of messenger RNA, which adopt alternate structures depending on the binding of specific metabolites. Such conformational switching regulates the expression of proteins involved in the biosynthesis of riboswitch substrates. Here, we present the 2.9 angstrom–resolution crystal structure of the eukaryotic Arabidopsis thaliana thiamine pyrophosphate (TPP)–specific riboswitch in complex with its natural ligand. The riboswitch specifically recognizes the TPP via conserved residues located within two highly distorted parallel “sensor” helices. The structure provides the basis for understanding the reorganization of the riboswitch fold upon TPP binding and explains the mechanism of resistance to the antibiotic pyrithiamine.

Riboswitches are conserved regions of mRNA that bind specific metabolites and regulate gene expression (14). This RNA-based mechanism of genetic control is believed to be of ancient origin (5) and is broadly distributed among bacteria, where it regulates ∼4% of all genes [for reviews, see (68)]. Riboswitches use diverse mechanisms to alter gene expression, including sequestration of the ribosome binding site (1, 9), formation of a transcription-terminating hairpin structure (1012), or direct cleavage of their mRNAs, thus functioning as ribozymes (13, 14).

Thiamine pyrophosphate (TPP) is an essential cofactor in bacteria, archaea, and eukaryotes. Its production is tightly regulated by TPP-binding riboswitches, which have been identified in thiamine-biosynthetic genes in all kingdoms (15). The mechanism of the regulation of gene expression by TPP has been most extensively studied for the Escherichia coli TPP riboswitch. Binding of the ligand to the 5′-untranslated region of the thiM gene, which is involved in the biosynthesis of thiamine, turns the riboswitch structure “off” and reduces translation of the mRNA by sequestering the ribosome binding site (1). Eukaryotic riboswitches conform to the consensus sequence and secondary structure of the TPP riboswitches found in bacteria and archaea. They bind TPP with a similar affinity and undergo the same conformational changes as their bacterial counterparts (15). In fungi and plants, TPP riboswitches are found either in introns or in the 5′- or 3′-untranslated regions of their target genes where they regulate, for example, mRNA splicing (15, 16). TPP-responsive riboswitches display an apparent dissociation constant of ∼50 nM, with the binding of TPP to the riboswitch dependent on the presence of Mg2+ ions (1, 15, 17). Such a high affinity relies on specific recognition of the pyrimidine ring and the pyrophosphate group of TPP (1). TPP riboswitches are attractive targets for antimicrobial drugs such as pyrithiamine, which exerts its toxic effects by binding TPP riboswitches in bacteria and fungi (16, 18).

Here, we describe the crystal structure of the complex between TPP and the eukaryotic Arabidopsis thaliana riboswitch (AtRs), which is located in the 3′ region of the thiC gene. The ThiC protein homolog in E. coli catalyzes the conversion of 5-aminoimidazole ribotide to hydroxymethyl pyrimidine phosphate in the thiamine-biosynthetic pathway. Because of its localization adjacent to the polyadenylate tail, it has been proposed that the TPP binding to the riboswitch regulates mRNA processing and stability of the thiC genes in plants (15). The riboswitch structure reveals a complex overall fold of AtRs and the details of interactions that govern its high specificity for TPP. The architectural features of AtRs provide the basis for understanding the reorganization of the mRNA fold upon TPP binding and explain the resistance to the pyrithiamine antibiotic.

For AtRs structure determination, we chose an RNA segment representing the complete sequence of the A. thaliana TPP-binding riboswitch (Fig. 1A). The in vitro–transcribed RNA was refolded in the presence of TPP and crystallized from 1,6-hexanediol at neutral pH. Phases were calculated using multiple wavelength anomalous dispersion data collected from a crystal derivatized with osmium. The experimentally phased electron density map allowed tracing of the phosphodiester backbone and defining the register of the sequence based on the discrimination between purines and pyrimidines (fig. S1A). The model was completed by iterative rebuilding guided by sigma-a–weighted difference Fourier maps. The TPP molecule was positioned in the last stage of refinement into the simulated annealed omit map (fig. S1B). The final model contains two copies of the full-length RNA in complex with TPP and refines to a final R/Rfree of (18.3/25.0)% (19).

Fig. 1.

Structure of the TPP riboswitch from A. thaliana. (A) Secondary-structure diagram of the TPP-binding domain located in the 3′ region of the thiC gene. For crystallization, the P1 helix was shortened by one and the 5′ sequence replaced by two guanines (boxed, dash lines), and the extension following P3 was replaced by the L3 tetraloop (boxed, solid lines). Color code: Pyrophosphate sensor helix (blue); pyrimidine sensor helix (pink); residues involved in pyrophosphate (green asterisks) or pyrimidine binding (red asterisks); conserved nucleotides (red). (B) Overall structure of the A. thaliana riboswitch. The TPP riboswitch binds TPP with its pyrophosphate and pyrimidine sensor helices. Bulges J2/3 and J4/5, which are contacting TPP (yellow), introduce sharp kinks into the sensor helices. Both sensor helices are linked via the J2/4 junction and switch helix P1, which is formed by base pairing of the 5′ and 3′ ends of the RNA.

The highly conserved secondary structure of the TPP-binding riboswitches consists of five helices, termed P1 to P5 (Fig. 1A). The helix P1 is formed by the 5′ and 3′ end of the riboswitch RNA and is expected to be disrupted in the absence of TPP. In E. coli, alternative structures of the “switch” P1 helix have been shown to expose the ribosome binding site (1). Helices P1, P2, and P4 together with junctions J1/4 and J2/4 form the central three-way junction. Bulges J2/3 and J4/5 connect helices P2 with P3 and P4 with P5, respectively. Terminal loops L3 and L5 close helices P3 and P5. The structure of AtRs has overall dimension of 56 Å by 40 Å by 22 Å and consists of two major helices, which are arranged in a parallel manner (Fig. 1B). One of them is formed by coaxial stacking of the switch helix P1 and helices P2 and P3, and the other consists of stacked helices P4 and P5. Such an organization of a three-way junction has been observed in several other RNA structures, including the guanine riboswitch, despite its less complex secondary structure (2022). Helices P2/P3 and P4/P5 are severely distorted by internal bulges J2/3 and J4/5, which are involved in TPP binding (Fig. 1B). Helices P2/P3 and P4/P5 are joined at their base by junction J2/4, which forms a sharp kink.

According to their interaction with different parts of the TPP, the parallel helices P2/P3 and P4/P5 can be considered as “pyrimidine sensor” and “pyrophosphate sensor” helices, respectively (Figs. 1B and 2). The TPP binding sites on each of the two sensor helices form deep binding pockets, which tightly accommodate the pyrimidine ring on one side and the pyrophosphate on the other.

Fig. 2.

Detailed views of TPP binding by the sensor helices. (A) The pyrophosphate sensor helix contacts the pyrophosphate moiety of TPP via a deep cleft formed by bases G64, C65, G66, and G48 of the J4/5 bulge. (B) G64, C65, G66, and G48 form a network of hydrogen bonds coordinating the putative Mg2+ ion (red sphere) and the pyrophosphate. (C) The pyrimidine ring of TPP is inserted into the pyrimidine sensor helix at the J2/3 bulge region by stacking between bases G30 and A31. (D) G28 adopts a “flipped out” conformation. This allows base pairing with the pyrimidine ring of TPP. The ring is additionally stabilized by a hydrogen contact with the ribose of base G11.

The TPP bound to the AtRs riboswitch has its pyrimidine and thiazole rings in an extended conformation resembling those observed in structures of free TPP (Fig. 1B) (23). This is different from the V-conformation of enzyme-bound TPP, where it functions as a cofactor (24). The pyrophosphate group adopts a bent conformation observed in both the free and the enzyme-complexed form of TPP (Fig. 2B) (23, 24).

The high specificity of AtRs for the TPP is explained by inspection of the contact area (Fig. 2). The contacts between the pyrophosphate of the TPP and the AtRs riboswitch are localized in the internal loop J4/5 in the pyrophosphate sensor helix P4/P5 (Fig. 2A). Residues G64, C65, and G66 from the J4/5 bulge interact with the pyrophosphate moiety directly (Fig. 2, A and B), and residues G66 and G48 coordinate a putative bridging Mg2+ ion (Fig. 2B). The Mg2+ ion has been modeled based on high-resolution Mg2+-bridged pyrophosphate structures as found in the Protein Data Bank and subjected to refinement. An alternative conformation of the pyrophosphate that includes an additional Mg2+ ion would also explain the electron density but did not refine stably at the current resolution. The large J2/3 bulge is responsible for direct recognition of the pyrimidine ring of the TPP. Residues U27 to A33, which form the longer segment of the bulge, fold in a helical stack extruded from the major groove of the pyrimidine sensor helix (Fig. 2C). Such arrangements of vertically stacked nucleotide bases were also observed in the crystal structures of several tRNAs and in the hairpin ribozyme (25, 26). The extrusion starts with a trans-Watson-Crick/Hoogsteen base pair between residues U27 and A31 and ends with a cis-Watson-Crick/Hoogsteen base pair between U12 and A33. Residue A31 adopts a C2′-endo conformation, allowing the stacking of the additional TPP pyrimidine ring between bases G30 and A31 (Fig. 2C). The flipped G28 forms specific hydrogen bonds with the pyrimidine ring. This interaction mimics a Watson-crick/sugar edge base pair (Fig. 2D). The N1′ position of the TPP pyrimidine ring forms a hydrogen bond with the 2′ oxygen of the conserved residue G11 (Fig. 2D). The central thiazole ring of TPP is in proximity of the phosphate group of residue C45. However, this contact is unlikely to be very discriminating because biochemical experiments have shown that TPP-related molecules containing larger rings replacing the thiazole moiety can also be accommodated by the riboswitch (1).

In addition to the TPP-mediated contacts, two other major interaction areas connect the sensor helices. The first region includes contacts between loop L5 and the base of helix P3 that resemble GAAA tetraloop/receptor interactions (Figs. 3A and 4A) (27). These contacts are probably only present when the TPP is bound to the internal loops of the sensor helices and properly orients the tip of loop L5 (Fig. 3A). A similar observation was made in the case of the Hammerhead ribozyme where interactions between terminal loops stabilize the functional conformation (28, 29). The second region of interaction involves residues in the three-way junction (Fig. 3A). This region is located between the TPP-binding part of both sensor helices and the switch helix P1. Residues A72, G42, C38, and G8 build an interaction platform, which stabilizes the kink in the J2/4 junction between the two sensor helices (Fig. 3, B and C). The extensive contacts within the interaction platform include residues from distant parts of the primary structure. This observation underscores the critical role of the platform in stabilizing the TPP-bound state of the riboswitch.

Fig. 3.

Major contact areas between the sensor helices are located outside the TPP binding site. (A) Surface representation of the separated pyrophosphate- and pyrimidine-binding helices viewed from the interface. Loop L5 from the pyrophosphate-binding helix contacts helix P3 from the pyrimidine-binding regions via hydrogen bonds (green). The three-way junction forms a large contact area between the sensor helices including strong base-pair and stacking contacts (orange). TPP (yellow) is clamped between the sensor helices. Both sensor helices contain deep binding pockets (red). (B) Central role of the conserved residue A72 in the formation of the three-way junction. At the J2/4 junction, the RNA backbone undergoes a sharp kink at residue U41 between the pyrophosphate (blue) and pyrimidine (pink) sensor helices. A72 (yellow) serves as an assembly platform by forming hydrogen bonds with G8, C38, and G42 (green) and stacking with A44 and the ribose of U39 on both sides of the kink region. (C) The detailed view of the plane around A72 shows the extended network of hydrogen bonds formed with bases C38, G42, and G8, thereby joining together four different RNA strands.

Fig. 4.

Model of the coupling between TPP binding and structural rearrangements in the TPP riboswitch. (A) Summary of the contacts observed between the pyrophosphate (blue) and pyrimidine (pink) sensor helices, TPP, and the switch region. Watson-Crick base pairs are indicated in black, nonstandard base pairs are shown in green, and hydrogen bonds are indicated with thin black lines. TPP is schematically drawn in yellow: phosphates, triangles; thiazole ring, pentagon; pyrimidine base, hexagon. The Mg2+ ion is shown as a red dot. Contacts between nucleotides and magnesium-bound TPP are highlighted in red. Red double-lines indicate stacking interactions with the ligand. Base A72, which forms the crucial assembly platform, is shown in yellow. (B) Upon TPP binding (1), bulges J2/3 and J4/5 rearrange, thereby closing the sensor helix clamp and forming new hydrogen bonds at the tip of loop L5. The TPP-induced parallel positioning of the sensor helices stabilizes the three-way junction with A72 at the core of the interaction platform (2) and promotes the formation of the switch helix P1 (3), which turns the riboswitch “off.”

Residue A72 is mutated in a pyrithiamine-resistant strain of Aspergillus oryzae (16). The phosphorylated form of the well-known antibiotic pyrithiamine (PTPP) exerts its activity by directly interacting with TPP riboswitches (18). Both A72 and C38 mutations have also been selected in a screen for pyrithiamine resistance in bacteria (18). These nucleotides are key residues within the interaction platform, which stabilizes the three-way junction (Fig. 3, B and C). Biochemical experiments indicate that TPP and PTPP are still able to bind to the riboswitch despite the mutations (18). Our structural data are in agreement with these experiments and show that the TPP binding pocket would remain unaffected by the A72 or C38 mutations. Instead, these mutations likely disturb the correct folding of the three-way junction and prevent the coupling between TPP binding and the formation of the switch helix P1.

The structure of the AtRs riboswitch described here provides structural insights into the TPP-induced mechanism of shifting between its ligand-free “on” and the ligand-bound “off” conformational states. In the thiM gene from E. coli, where the TPP riboswitch is also found, the “on” state promotes translation of the mRNA, whereas the “off” state inhibits translation. Previous in-line probing experiments on the E. coli TPP riboswitch, in which the spontaneous RNA cleavage rate in the presence and absence of the ligand was monitored, revealed reduced cleavage of the “switch” helix P1 residues upon binding of the TPP (1). This indicates that helix P1 is not formed in the “on” state. Based on the structural and biochemical data, we propose a model for the sequence of events leading to the “off” state of the riboswitch (Fig. 4B): (1) TPP binding promotes the parallel disposition of the sensor helices. (2) Consequently, the interaction platform is assembled, forming a strong kink at the J2/4 junction. (3) The folding of the three-way junction reduces the entropic penalty for the formation of the switch helix P1. Therefore, if the formation of the three-way junction is impaired by one of the mutations, the TPP will still be able to bind to the sensor helices but will nevertheless be unable to turn the riboswitch “off.”

The structure of the AtRs riboswitch reveals how thiamine pyrophosphate is recognized with high specificity and high affinity, rationalizes the mechanism of resistance to the well-known antibiotic pyrithiamine, and demonstrates which regions of the riboswitch are critical for the stability of its “off” conformation. The results presented here provide a good starting point for structure-based in vivo and in vitro experiments aimed at studying the mechanism of TPP riboswitch-regulated gene expression in general.

Supporting Online Material

Materials and Methods

Fig. S1

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article