Thermal adaptation of structural dynamics and regulatory function of adenine riboswitch

ABSTRACT

Ligand binding and temperature play important roles in riboswitch RNAs’ structures and functions. However, most studies focused on studying structural dynamics or gene-regulation function of riboswitches from the aspect of ligand, instead of temperature. Here we combined NMR, ITC, stopped-flow and in vivo assays to investigate the ligand-triggered switch of adenine riboswitch from 10 to 45°C. Our results demonstrated that at single-nucleotide resolution, structural regions sensed ligand and temperature diversely. Temperature had opposite effects on ligand-binding and gene-regulation of adenine riboswitch. Compared with higher temperature, the RNA bound with its cognate ligand obviously stronger, while its regulatory capacity was weakened at lower temperature. In addition, application of specific-labelled RNAs to the stopped-flow experiments identified the real-time folding of the specific positions upon ligand addition at different temperatures. The kissing loop and internal loop at the riboswitch responded to ligand and temperature differently. The distinct thermo-dynamics of adenine riboswitch exposed here may contribute to the fields of RNA sensors and drug design.

Introduction

Riboswitches are cis-acting RNAs that usually located at the 5ʹ-UTR regions of mRNA, controlling the gene expression of the downstream genes in response to specific ligands [1–4]. They exert gene regulatory functions primarily at transcriptional or translational level and have not been discovered in mammals, and therefore, has potential to be applied as antibiotics, molecular sensors, and gene regulators [5]. A common riboswitch consists of a ligand-sensing aptamer and an expression platform [6]. Helices, kissing loops, internal loops or pseudoknots are structural building blocks in most riboswitches, which can affect conformational stability and gene regulation efficiency of riboswitches [7].

The add adenine riboswitch from Vibrio vulnificus controls translation of adenine deaminase by binding with adenine [8]. Its aptamer domain has been investigated extensively through biophysical and biochemical techniques including crystallography, NMR spectroscopy and single-molecule methods [9–13]. The crystal structure of adenine riboswitch aptamer domain was solved firstly in 2004, which is characterized by three-way junction with three helices stacked in a coaxial manner and supported by a kissing loop (Fig. 1) [9,14]. The specific ligand, adenine is docked at the internal loop, forming base pairs with four nucleotides in the internal loop (Fig. 1). Real-time crystallographic study acquires two ligand-free, one intermediate and one ligand-bound conformation of the aptamer domain [10,15]. The binding pocket at the internal loop is highly ordered although with different folding modes in the ligand-free and ligand-bound conformations. The crystallographic and NMR studies support that Mg²⁺ is not enough to induce the compact folding of the internal loop. However, the kissing loop is formed at the presence of Mg²⁺ without adenine [10,11,15]. The kissing loop is also affected by helices P2 and P3, and mutations at P2 can impact the kissing loop significantly [11,16]. Besides, kinetic studies indicate that the kissing loop and three-way internal loop are two core structural motifs in RNA folding, and stopped-flow spectroscopy demonstrates a faster rate constant of the binding pocket than that of the kissing loop upon adenine at room temperature [17,18]. Crystal structures of the full-length adenine riboswitch are unavailable, probably due to flexibility of the RNA. And compared with its aptamer, much less studies have been done for the full-length adenine riboswitch [19]. The formation of the kissing loop in adenine riboswitch is dependent on magnesium ion at smFRET, as other riboswitches [20]. Single molecule force spectroscopy (SMFS) technique reveals the folding sequence of the full-length adenine riboswitch, in which helices P2 and P3 are prior to the kissing loop interaction and the helix P1 [21–23].

Figure 1.

Structures of add adenine riboswitch. (A) and (B) Primary secondary structures of FriboA in the presence of adenine (Translation ‘ON’ state) and in the absence of adenine (Translation ‘OFF’ state). Helices P1, P4 and P5 (black), P2 (green), P3 (orange), kissing loop (KL, red) and internal loop (IL, blue) are marked. The aptamer domain is boxed by dash line, ribosome binding site (RBS) in expression platform is highlighted by magenta. (C) Watson-Crick kissing loop of G37-C61 and G38-C60 (coloured) in the crystal structure of the aptamer domain (PDB: 5SWE). (D) Four residues U22, U47, U51 and U74 at the binding pocket form H-bonds with adenine (yellow) in the crystal structure (PDB: 5SWE)

Adenine riboswitch has been reported to be a temperature sensing RNA [24,25]. However, the intrinsic relationship between the structural cores, kissing loop and internal loop, and how they mutually affect each other are still elusive. Besides, the structural dynamics and influence on gene regulation of adenine riboswitch at various temperatures are not clarified neither. Elucidation of these puzzles is essential for illuminating the molecular mechanism of riboswitch RNAs and benefit for broadening the applications of RNAs as biosensors and drugs. In this work, we combined NMR, ITC, stopped-flow spectroscopy and in vivo experiments to study the roles of kissing loop on ligand binding and gene regulation of the full-length adenine riboswitch at various temperatures. NMR spectra provided conformational information of adenine riboswitch at atomic resolution and revealed response of individual nucleotides to ligand and temperature [26]. The stopped-flow fluorescent experiments monitored the real-time change of specific-labelled adenine riboswitch at single-nucleotide resolution, identified different thermal-sensing patterns of the kissing loop and internal loop, and supported that response of the internal loop to adenine was prior to the kissing loop at different temperatures. The factor of temperature was particularly concerned in this work, and the conformational dynamics of kissing loop and internal loop of adenine riboswitch were differentiated with temperature. And our results supported that the mutual connection between the kissing loop and internal loop, in which the formation of the internal loop would facilitate folding of the kissing loop, and the stability of the kissing loop would promote ligand binding and gene regulation. In short, our research identified the key roles of kissing loop and temperature in ligand binding and gene regulation, provided a new perspective to illuminate the regulatory mechanism of riboswitches, and may be helpful to facilitate the studies of RNAs with kissing loops or thermo-sensitive characteristics.

Results

Ligand-induced conformation change of full-length adenine riboswitch at low temperature

The add adenine riboswitch studied in this work is from Vibrio vulnificus, and it contains both the aptamer domain and the expression platform (Fig. 1) [8]. Adenine binds to the aptamer domain specifically, rearranging the expression platform and regulating its downstream genes at translational level consequently [27,28]. As shown in Fig. 1A, upon adenine binding, the RBS site (magenta) is located at the single-stranded region, which is capable of interacting with ribosome and translating the downstream genes. On the contrary, without adenine binding, the RBS site is placed at the double-stranded region, inhibiting translating of the related genes (Fig. 1B). The RNA construct, FriboA used in this work is composed of 113 nucleotides, and its aptamer domain is boxed at Fig. 1A. Based on the crystal structure of the aptamer domain bound with adenine (PDB ID: 5WSE), the aptamer domain forms a compact ‘fork like’ structure with a kissing loop (KL) between L2 and L3 (Fig. 1C). Fig. 1D is the adenine-binding pocket at internal loop (IL).

The imino regions of ¹H¹⁵N-HSQC spectra of FriboA at 2 mM MgCl₂ with and without adenine were collected at 10°C (Fig. 2). Without adenine, the observed imino signals, U17, U77 and G78 were from helix P1 (grey columns, diagonal filled), imino signals U28, U31, U39-41, G42 and G44 (light green columns, diagonal filled), G59 and U68 (orange columns, diagonal filled) were from helix P2 and P3, respectively (Fig. 2C). The weak but defined G37 and G38 peaks (pink columns, diagonal filled) supported that the KL was partially formed although it was possibly flexible or in low population at 2 mM Mg²⁺ and no adenine. The lack of imino peaks from the IL was expected in the absence of ligand. The slight imino peaks of G115 (grey column, diagonal filled at Fig. 2C) indicated that a small fraction of the regions around the RBS site was folded as double strands, and efficient translation of the downstream genes was supposed at the absence of adenine.

Figure 2.

NMR spectra of FriboA with and without adenine at 10°C. Imino regions of the ¹H¹⁵N-HSQC spectra of FriboA in the absence of adenine (A) and presence of adenine (B) at 2 mM Mg²⁺. The imino signals are marked as grey for helix P1 and expression platform, light green for helix P2, light orange for helix P3, pink for KL and cyan for IL. Imino signals from the co-exist conformation in both ligand-free and ligand-bound states are marked with asterisks. (C) The normalized intensities of the imino intensities without adenine (diagonal filled columns) and with 1.2 eq adenine (full-filled columns). The colour scheme is as defined in (A) and (B)

As 0.6 mM adenine titrated to 0.5 mM FriboA at 10°C, obviously more peaks showed up, which indicated that FriboA folded more compactly upon adenine addition (Fig. 2B,Fig. 2C). Decent imino signals from the IL, including U22, G46, U47, U49 and U51 (cyan, filled columns) appeared with adenine addition (Fig. 2B,Fig. 2C), which supported that the ligand-binding pocket at the IL interacted with adenine. In the meanwhile, signals from the KL (G37, G38), helices P2 (U25, 28, 31, 39–41, G42, G44) and P3 (G57, 59, U68, 71, G72) were further enhanced or appeared upon adenine addition. This suggested that these regions were further stabilized with adenine as reported elsewhere [29]. Noteworthy, adenine addition possibly led to slight dissociation of the double-stranded region of expression platform, which was identified from the slim reduction on the peak intensities of G115. That’s to say, based on the NMR spectra at 10°C, adenine was likely to elevate the translation of the genes following the adenine riboswitch although the regulatory capability of adenine on the RNA was not significant at the temperature as low as 10°C. It’s worth to point it out that the peak intensities of U91, U92 and U106 were barely affected by ligand addition, which matched that the base-pairing environment around these positions were not affected upon ligand although U91 and U92 were weak probably due to flexibility (Fig. 1A,Fig. 1B). More than one signal was assigned to a single base, for example, G44, G46, U70, G72, U91, U92 and G110 (Fig. 2), which indicated that multiple conformations co-existed as reported [24]. And the ligand-bound conformation became predominant with adenine addition at FriboA as reported elsewhere [24].

In a short summary of the NMR spectra at 10°C, the helices P1, P2 and P3 were pre-organized at the absence of adenine, P1 except U75 (adjacent to the binding pocket) was barely affected by adenine while P2 and P3 were further stabled by adenine addition. The presence of G37 and G38 at the absence of adenine supported that 2 mM Mg²⁺ was capable of folding the KL as reported [10,15], however, the KL got further stabilized or intensive by adenine. The compact folding of the IL was observed solely with adenine. Besides, the peaks nearby the KL and IL, including U75, U31, U40, U41, G42 and G57 at helices P1, P2 and P3 became stronger with adenine, too. Surprisingly, only slight change was observed at the expression platform after adding adenine, which indicated that the RBS site at the expression platform may not be efficiently regulated by adenine at the temperature as low as 10°C.

Ligand-induced conformation change of full-length adenine riboswitch at high temperature

Compared with 10°C, without adenine, fewer imino peaks were detected at FriboA at 30°C, and the peaks from the KL were too weak to be detected (Fig. 3A,Fig. 3C). In addition, the peaks close to the KL, including U31, U39-41 at P2, G59 and U68 at P3 were unobservable at 30°C, too (Fig. 3A,Fig. 3C). Therefore, at 30°C and the absence of adenine, the KL was not formed or too dynamic to be detected by NMR, and the stability of helices P2 and P3 were also impacted adversely. Another obvious difference between 10 and 30°C was G115 from the expression platform. At the absence of adenine, G115 at helix P4 was significantly stronger at 30°C than 10°C (Fig. 4A). This indicated that higher population of double-stranded region around the RBS site was present at 30°C. Hence, lack of adenine, the translation of the genes located at downstream of FriboA worked better at the lower temperature.

Figure 3.

NMR spectra of FriboA with and without adenine at 30°C. Imino regions of the ¹H¹⁵N-HSQC spectrum of FriboA in the absence of adenine (A) and presence of adenine (B) at 2 mM Mg²⁺. The imino signals are marked as black for helix P1 and expression platform, green for helix P2, orange for helix P3, red for KL and blue for IL. (C) The normalized intensities of the imino intensities without adenine (diagonal filled columns) and with 1.2 eq adenine (full-filled columns). The colour scheme is as defined in (A) and (B)

Figure 4.

The imino intensities of FriboA with and without adenine at 10°C and 30°C. (A) Normalized imino intensities of FriboA in the absence of adenine at 10 and 30°C. (B) Normalized imino intensities of FriboA in the presence of adenine at 10 and 30°C. The colour scheme is as defined in Figures 2 and Figures 3

Upon the addition of adenine, FriboA went through a tremendous change and the observable imino signals were significantly increased at 30°C (Fig. 3B,Fig. 3C). The KL (G37 and 38) and IL (U22, 47, 49 and 51) strengthened obviously with adenine at 30°C. Interestingly, the change of the KL was higher at 30°C than 10°C (Fig. 4A,Fig. 4B), while the IL, especially U49 and 51 was opposite, with a more drastic change upon adenine addition at 10°C. What’s more, distinct from slight change at 10°C, G115 from the expression platform reduced greatly with the addition of the ligand at high temperature. This agreed with the regulation mechanism of FriboA, in which the region around the RBS site at the expression platform of FriboA sensed its specific ligand and dissociated from helical region to turn on the translation in vivo.

Generally speaking, based on the NMR spectra, most imino peaks at the aptamer domain strengthened or appeared with adenine at both 10 and 30°C (Fig. 4), while the imino peaks at the expression platform weakened with adenine. The intensities of the imino peaks at the IL and KL increased with adenine adding at 10 and 30°C indicating they were further stabilized by adenine. However, comparing the change caused by adenine, adenine binding affected folding of the KL more significantly at 30°C. While the imino peaks of the IL were observed at the presence of adenine only, and they were more stable at 10°C. This indicated that FriboA probably bound with adenine with higher affinity and led to higher stability of the IL at low temperatures. Another important investigation was that the conformational switch of expression platform was affected by both ligand addition and temperature. The imino intensity of G115 reduced more significantly after adding adenine at 30°C than 10°C, which indicated that both ligand and high temperature favoured the conversion from double-strand to single-strand of G115 along with the RBS site. Hence, the gene regulation of FriboA should be performed more effectively at the high temperature in the case of 10 vs. 30°C.

Stability of the KL affects the ligand binding and gene regulation of FriboA

In Figs. 2 and Figures 3, the peaks of KL and IL at FriboA were both strengthened upon adenine addition at both 10 and 30°C, and we proposed that these two regions were possibly interrelated. To further identify if the KL and the IL were mutually affected, we measured the binding affinity of FriboA and its mutant, M1 (C60G, C61G) by ITC (Isothermal Titration Calorimetry). High binding between adenine and FriboA was measured with K_d = 354.6 nM at ITC (Fig. 5A). Destruction of the KL at M1 led to drastic reduction on binding affinity, and interaction between adenine and M1 could barely be detected (Fig. 5B).

Figure 5.

The influence of kissing loop on ligand binding and gene regulation of FriboA. (A) ITC curves of adenine titration to FriboA in the presence of 2 mM Mg²⁺ at 25°C. The calculated K_d was 354.6 nM with N = 0.97. (B) ITC curves of adenine titration to M1 in the presence of 2 mM Mg²⁺ at 25°C. No binding was detected for M1. (C) In vivo production of RFP downstream FriboA or its mutants with and without adenine. The plasmid constructs contain adenine riboswitch or its mutants and reporter RFP are overexpressed without or in the presence of 1 mM adenine at 37°C with 2 mM Mg²⁺. FriboA exhibits regulatory function with nearly two-time increase of the RFP intensity (grey column). Disturbing the kissing pairs greatly impaired the gene regulation capability of the riboswitch and reduced the production of RFP. Cyan, pink, and light green columns represented for M1, M2 and M3 mutants, respectively. (D) The mutations of M1, M2 and M3 were diagrammed

To clarify if the KL also interacted with the gene regulation of the adenine riboswitch, we also ran in vivo assays of FriboA and several mutants with mutations at the KL. The plasmid constructs contained the adenine riboswitch or its mutants and the downstream reporter, RFP (Red Fluorescence Protein) were overexpressed in E. coli, and the translated RFP was quantified at different adenine concentrations (Fig. 5C,Fig. 5D). For FriboA, the expression level of its downstream RFP was improved over two times when 1 mM adenine was added to the cell culture at 37°C (grey column, Fig. 5C). The regulatory function of the mutant M1 was hindered based on the fact that no apparent RFP fluorescence change was observed after adding 1 mM adenine to the culture (cyan columns, Fig. 5C). And the impaired gene regulation was also observed at the mutant M2 (C60A, C61A) (pink columns, Fig. 5C). Surprisingly, the mutations at M3 with switching G and C at the KL were unable to recover the regulation function of adenine riboswitch (light green columns, Fig. 5C), which indicated that the formation of the KL was also affected by interactions with the neighbouring nucleotides. ITC and in vivo results demonstrated that disruption of the KL may impede folding of the IL and even the aptamer domain, resulting in the loss of adenine-binding and gene regulation capacity of adenine riboswitch.

Thermo-sensitivity of the KL, IL and expression platform of FriboA

Our NMR data revealed that the adenine riboswitch was not only sensitive to its cognate ligand, adenine, but also to temperature (Figs. 2, Figs. 3 and Figs. 4). To further verify the thermo-sensitivity of FriboA, we performed ITC experiments, stopped-flow and in vivo assays at different temperatures. As shown in ITC experiments, the binding affinity of adenine to FriboA decreased with the K_d value increased from 354.6 to 1897.5 nM as temperature rose from 25 to 37°C at 2 mM Mg²⁺ (Fig. 6A and Table S2). We also collected the stopped-flow traces of FriboA mixing with adenine analogue, 2-aminopurine (2AP) from 20 to 45°C (Fig. 6B and Table S3). In the stopped-flow experiments, the change of fluorescence became weaker as temperature rose. The ITC and stopped-flow results matched the NMR data, in which the imino peaks at the IL changed less intensively at the high temperature (Fig. 4). Except for in vitro tests, we also ran in vivo assays at different temperatures to determine whether temperature would affect the gene regulation. As a result, the expression level of RFP increased with adenine addition at both 20 and 37°C, however, opposite from the binding affinity, RFP expression was up-regulated with greater ratios as 1 mM adenine was added into the culture at 37 than 20°C (Fig. 6C). The in vivo results agreed with the NMR data that G115 at the expression platform shifted greater at high temperature (Fig. 4).

Figure 6.

Temperature affects binding affinity, structural dynamics and gene regulation of FriboA. (A) ITC analysis monitoring the adenine binding of FriboA at 25°C (blue) and 37°C (orange) at 2 mM Mg²⁺. The calculated K_d values were 354.6 and 1897.5 nM at 25°C and 37°C, respectively. (B) The stopped-flow traces of FriboA upon 2AP binding. As temperature rose from 20°C to 45°C, less fluorescence change range was observed. (C) In vivo gene regulation is affected by temperature. The plasmid construct with adenine riboswitch and reporter RFP was overexpressed at 20°C or 37°C. The expression level of RFP increased about 1.7 times (black) at 1 mM adenine at 20°C, less than the increase of 2.4 times for RFP expression at 37°C (orange). (D) Stopped-flow fluorescence trajectories of 0.5 μM 35AP-FriboA upon rapidly mixing with 0.5 mM adenine in the presence of 2 mM Mg²⁺ at 20°C (purple) and 37°C (red). The fluorescence of 35AP-FriboA decreased with k_obs = 0.67 s⁻¹ and 2.52 s⁻¹ at 20°C and 37°C, respectively. (E) Stopped-flow fluorescence trajectories of 0.5 μM 52AP-FriboA upon rapidly mixing with 0.5 mM adenine in the presence of 2 mM Mg²⁺ at 20°C (olive) and 37°C (blue). The fluorescence of 52AP-FriboA decreased with k_obs = 6.32 s⁻¹ and 21.84 s⁻¹ at 20°C and 37°C, respectively. (F) Stopped-flow curves of 0.5 μM 35AP-FriboA upon rapidly mixing with 2 mM Mg²⁺ without adenine at 20°C and 37°C. The fluorescence decreased with k_obs = 1.13 s⁻¹ and 9.37 s⁻¹ at 20°C (light purple) and 37°C (light red), respectively. The curves at (B), (D), (E) and (F) were merged at 0 s for better comparison

The ITC and in vivo assays demonstrated that the KL formation was necessary for ligand binding of FriboA. To further investigate the intrinsic relation between the KL and IL, we introduced 2AP to site 35 nearby the KL (35AP-FriboA) and site 52 at the IL (52AP-FriboA) by PLOR to monitor the real-time response of the KL and IL to adenine at 20 and 37°C in the stopped-flow experiments (Fig. 6D,Fig. 6E). 0.5 μM 35AP- and 52AP-FriboA were pre-incubated with 2 mM Mg²⁺ beforehand. When mixed with adenine, the fluorescence of 35AP-FriboA decreased with k_obs = 0.67 s⁻¹ and 2.52 s⁻¹ at 20 and 37°C, respectively (Fig. 6D and Table S4). For 35AP-FriboA, mixing with adenine at higher temperature brought more remarkable change. This matched the NMR observations on the KL (Fig. 4). Obviously, the IL responded to adenine with higher sensitivity at lower temperature since stronger fluorescence was measured at 20°C. And the kinetic rate constants for 52AP-FriboA were k_obs = 6.32 s⁻¹ and 21.84 s⁻¹ at 20 and 37°C, respectively (Fig. 6E). By using specific-labelled samples, the real-time response of the KL and IL were tracked individually. Although they showed unique thermo-sensitivity, the observed rate constants of the IL were faster than the KL at the tested temperatures. This indicated that adenine triggered structural switch of the IL prior to the KL, and folding of the IL upon adenine addition may further stabilize the KL. This is similar as the folding orders of guanine riboswitch observed by single-molecule analysis [30].

Discussion

Like other riboswitches, adenine riboswitch has a conserved aptamer domain and a flexible expression platform. Only the atomic-resolution structure of the aptamer domain is available, and it contains a kissing-loop, a three-way internal loop, and three helices. The aptamer domain of the adenine riboswitch is responsible for ligand binding. The structural rearrangement of the expression platform can promote or forbid gene expression of its downstream genes. The adenine riboswitch, especially its aptamer domain has been extensively studied, however, the interconnection and individually unique thermo-sensitivity of the crucial structural motifs, including the kissing loop, the internal loop and expression platform are still elusive.

The kissing loop unit is a well-known constituent part for adenine riboswitch structurally. Our ITC and in vivo results demonstrated the necessity of the kissing loop functionally, and destroying of the kissing loop led to the loss of binding and gene regulatory functions. Furthermore, the stability of kissing loop, internal loop and expression platform were greatly affected by ligand and temperature. NMR data supported that the aptamer domain showed less stability at the absence of adenine, while the expression platform formed a higher portion of helices without adenine. And adenine was a trigger for the dissociation of expression platform, but further stabilizing the aptamer domain of FriboA. From ITC and stopped-flow experiments, the pre-folded kissing loop favoured the ligand binding of the internal loop, and the stronger binding of adenine to the internal loop at lower temperatures may be partially due to the higher portion of the kissing loop without adenine (Fig. 4A). In return, compactness of the internal loop contributed to further stabilization of kissing loop. With consideration of temperature, adenine affected the kissing loop and the internal loop differently. Both NMR and stopped-flow results supported that adenine addition brought greater change on the kissing loop at higher temperatures. On the contrary, based on ITC, NMR and stopped-flow data, conformational rearrangement of the internal loop was more significant at lower temperature.

The opposite response of the kissing loop and the internal loop seemed irrational since that it is the folding of internal loop upon adenine binding affected the base pairing of the kissing loop based on the stopped-flow results. With consideration of the NMR results that the kissing loop was observed at 10°C but barely detected at 30°C in the absence of adenine and with 2 mM Mg²⁺, we supposed that the stability of the kissing loop probably also impacted by Mg²⁺, and very likely, Mg²⁺-triggering stability of the kissing loop have different responses on temperatures from adenine. And indeed, lower temperature fitted better for kissing-loop folding at the presence of Mg²⁺ (Fig. 6F). While summarized the effect of Mg²⁺ and adenine, the kissing loop of FriboA was still more stable at lower temperature, similarly as the internal loop. The expression platform is related directly with the regulatory function of FriboA, and based on NMR, its structure switched more obviously at the high temperature. And the adenine-dependent conformation change was also reflected by the in vivo gene regulation experiment. In vivo data supported that under the control of FriboA, RFP production increased more greatly at higher temperature. It is proposed that the stability of the expression platform was adversely related with that of the aptamer domain, which ensured adenine binding can stabilize the aptamer domain but at the same time, destabilized the expression platform to perform gene regulation function at physiological temperature of cells.

Temperature plays critical and non-uniform roles in structures and functions of the adenine riboswitch. On one hand, from structural view, temperature affected different core regions of the adenine riboswitch diversely. Based on NMR and stopped-flow experiments, the internal loop responded to ligand more sensitively at low temperature, while the population of the kissing loop was fluctuated more intensively with adenine at high temperature. Different from the aptamer domain, which tended to be stabilized at low temperature, the expression platform contained more double-stranded helices at high temperature in our NMR and in vivo assays. On the other hand, from functional view, temperature brought contrary impacts on ligand-binding and regulatory function of the adenine riboswitch. From ITC and stopped-flow studies, low temperature was preferred for RNA-adenine binding. And from our in vivo data, the translation level was relatively high in the absence of adenine at low temperature, however, the translation was up-regulated less significantly upon adenine addition at low temperature. The diverse behaviours of temperature may be beneficial for balance the stability and dynamic folding of the full-length RNA molecule at different cellular environments. Understanding the intrinsic relationships among basic structural motifs, ligand interactions and functions are crucial to elucidate the dynamics and molecular mechanism of RNAs. The distinct thermo-sensitivity of adenine riboswitch exposed in this work may contribute to the studies of thermometer RNAs, and has significant implications for RNA fields, such as RNA sensors.

Materials and methods

RNA preparation

All RNAs in this work were in-house prepared. The unlabelled and ¹³C¹⁵N-labelled adenine riboswitch samples were prepared by in vitro transcription with T7 polymerase as described previously [31]. The sequences of the DNA templates used to produce the RNAs were listed in Table S1. The crude products were purified by denaturing 12% PAGE (polyacrylamide gel electrophoresis) as reported [31]. The fluorophore RNA samples (35AP- and 52AP-FriboA) were generated by PLOR and purified by HPLC as previously described [31,32]. The purified samples were then exchanged to suitable buffer and stored at −80°C. Except noted, all RNAs were incubated at 85°C for 5 min, and then cooled down to room temperature before use.

NMR spectroscopy

All ¹H¹⁵N-HSQC spectra were collected at Bruker Avance 900 MHz spectrometer equipped with a triple-resonance cryo-probe (Bruker, Germany). 0.5 mM ¹H¹⁵N-FriboA were exchanged to the NMR buffer (10 mM K₂HPO₄, 30 mM KCl, 2 mM MgCl₂, pH 6.8) containing 10% D₂O. NMR data were processed and analysed by using the software programmes Topspin 3.6 (Bruker, Germany) and Sparky (University of California, US).

ITC measurements

ITC (Isothermal Titration Calorimetry) experiments of FriboA and its mutants were performed in the buffer containing 10 mM HEPES, 100 mM KCl, 2 mM MgCl₂, pH 7.5 on a MicroCal ITC200 calorimeter (General Electric, USA) at 25°C and 37°C. 2 μL, 300 μM adenine was injected to the cuvette containing 280 μL, 30 μM RNA for 20 times with 90 s interval between injections. The titration between the buffer and RNA was referred as the background for ITC. The ITC data withdrew the background were fitted by Origin ITC software (OriginLab, USA).

Stopped-flow fluorescence experiments

The stopped-flow fluorescence experiments were run at the buffer (10 mM HEPES, 100 mM KCl, 2 mM MgCl₂, pH 7.5) equally mixing RNA with 2.5 μM 2AP or 0.5 mM adenine at an SX 20 stopped-flow spectrometer (Applied Photophysics Ltd., UK). 1 μM non-labelled or 0.5 μM 2AP-labelled RNA were used in the stopped-flow fluorescence experiments. Data were collected with 300 nm excitation and a 360 nm long-pass filter. The average data of over 3 replicates and were fitted by using Origin (OriginLab, USA).

In vivo assay of adenine riboswitch-controlled gene expression

The sequence of adenine riboswitch was inserted between Enhanced Green Fluorescent Protein (EGFP) and Red Fluorescent Protein (RFP) genes at the vector pE1K, generating the vector pE1K_add. E. coli DH5α strain carrying the recombinant plasmid was grown in minimal media and cultured at 20°C and 37°C. 0 ~ 1 mM adenine was added to the culture. The fluorescence of EGFP (the internal control) and RFP was measured at 510 nm (λ_ex = 480 nm) and 610 nm (λ_ex = 580 nm), respectively, by a Cytation^TM 5 multi-mode reader (BioTek, USA). All experiments were performed in triplicate.

Funding Statement

This work was supported by National Natural Science Foundation of China [Grant no. 31872628 and 32071300 to Y. Liu] and the Zhangjiang Laboratory Foundation [Grant no. Y93Z011D01 to ZJ. Liu].

Disclosure statement

The authors have no conflict of interest.

Supplementary material

Supplemental data for this article can be accessed here.