SEARCH

SEARCH BY CITATION

Keywords:

  • alternative splicing;
  • aptamer;
  • metabolism;
  • plant;
  • riboswitch;
  • thiamin

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Riboswitches are RNA elements that bind small molecules and in turn regulate gene expression. This mechanism allows the cell to sense the intracellular concentration of these small molecules. A particular riboswitch typically regulates its adjacent gene by altering the transcription, the translation or the splicing of this gene. Recently, a riboswitch that binds thiamin pyrophosphate (TPP) was characterized and found to regulate thiamin biosynthesis in plants and algae. Furthermore, it appears that this element is an essential regulator of primary metabolism in plants. Manipulation of endogenous riboswitch activity resulted in metabolic phenotypes that underlined the role of these elements and their ligands in preserving metabolic homeostasis. This situation supports the hypothesis that riboswitches could be remnants of the most ancient metabolic regulators. Here, we review the mode of action of the plant and algal TPP riboswitch and its relevance to the metabolic network. We also discuss the potential engineering of riboswitches as metabolite sensors in plants and platforms for gene control. Whether additional such RNA-based mechanisms exist in plants and in algae is still an open question, yet, the importance of these elements to metabolic regulation is beyond doubt.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Riboswitches are natural RNA sensors that affect post-transcriptional processes via their capacity to bind small molecules. These are structured and conserved elements located in the untranslated regions (UTR) and introns of mRNAs in all three kingdoms of life. They were first described over a decade ago and have the amazing capability to convert a change in the intracellular level of a small molecule into RNA structural rearrangement and thereby into a change in gene expression (Mironov et al., 2002; Nahvi et al., 2002; Winkler et al., 2002a,b). Each riboswitch class recognizes a metabolite with high selectivity and regulates adjacent gene expression in cis, with only few known exceptions (André et al., 2008; Loh et al., 2009). Typically, the genes regulated by a given riboswitch are involved in the biosynthesis, catabolism, signaling or transport of the riboswitch ligand, potentially creating a feedback to maintain adequate levels of this molecule. The structure of a riboswitch consists of a sensory domain and a regulatory domain. The former is commonly called an ‘aptamer’ (although this term typically refers to a synthetic RNA molecule designed to bind a metabolite), and is responsible for small molecule (later referred as ligand) binding. The aptamer exhibits the majority of the sequence conservation in the riboswitch. The regulatory domain, typically resides adjacent to the aptamer, and is termed an ‘expression platform’, and transduces small molecule binding into a genetic regulatory signal. Ligand binding to the aptamer induces a structural rearrangement of the riboswitch in a nascent RNA molecule, and this conformational change immediately alters adjacent gene expression. To date, riboswitches were demonstrated to modulate gene expression by affecting transcription, translation, splicing, or RNA stability (Roth and Breaker, 2009; Lee et al., 2010).

Riboswitches react instantly to variations in intracellular metabolite concentrations. Bacterial riboswitches, located in the 5′ UTRs of mRNAs, are intrinsically synthesized early during transcription. RNA folding is 2–3 orders of magnitude faster than the rate of transcription, and base–base recognition takes place as soon as the emerging strand of RNA reaches sufficient length to permit folding. Therefore, a riboswitch can respond to metabolite concentration before the full-length mRNA is produced and authorize transcription to stop or continue (Cruz and Westhof, 2009). This mechanism relies merely on RNA-metabolite interaction and allows bacteria to avoid wasting energy in the production of unnecessary full-length mRNAs (Breaker, 2011). There are, to date, over 20 experimentally validated riboswitch classes, which recognize structurally diverse effector molecules including nucleobases, amino acids, cofactors, metal ions, and second messengers (Breaker, 2011). However, the only forms of eukaryotic riboswitch found to date are those that recognize thiamin pyrophosphate (TPP) as a ligand (Sudarsan et al., 2003). Interestingly, it was recently demonstrated that riboswitches can also recognize several related chemical compounds, rather than just one metabolite. This situation allows the integration of information from intermediates and thus several locations in a metabolic pathway and facilitates its regulation (Watson and Fedor, 2011; Moulin et al., 2013). Furthermore, recent advances in sequencing technologies and metagenomic studies uncovered the presence of thousands of representatives of some of these classes in the current DNA sequence databases (Breaker, 2011). The known riboswitch classes probably represent only a tiny proportion of the total that exists in the biosphere, and many new classes are likely to be discovered in the future (Breaker, 2011).

The regulatory activities of riboswitches are executed in various ways and include ligand-dependent formation of mutually exclusive RNA conformations. The most simplistic regulatory activity of these elements consists of the metabolite-induced shift between terminator or anti-terminator hairpins (Mironov et al., 2002; Rodionov et al., 2002; Winkler et al., 2002b). In another straightforward regulatory mechanism, metabolite binding to the riboswitch structure triggers the sequestration or release of the ribosome-binding sites (Rodionov et al., 2002; Winkler et al., 2002a). Conversely, the glucosamine-6-phosphate (GlcN6P) riboswitch exhibits ribozyme activity and couples metabolite sensing with mRNA cleavage. Ligand binding induces mRNA cleavage within the riboswitch sequence to repress gene expression (Winkler et al., 2004). Riboswitches can also regulate gene expression by a complex cascade of events that requires association of several RNA components. In one example, riboswitches for the second messenger, cyclic di-guanosyl-5′-monophosphate (c-di-GMP), are located adjacent to group I self-splicing introns. Upon ligand binding, RNA conformational changes allow GTP to attack the intron's 5′ splice site. This action causes the self-excision of the intron, which in turn brings together two distantly located halves of the ribosome-binding site to generate a translatable mRNA (Lee et al., 2010; Chen et al., 2011). In another example, the structural rearrangement adopted by the lysine riboswitch upon ligand binding can simultaneously inhibit translation initiation and expose RNase E cleavage sites located in the riboswitch expression platform (Caron et al., 2012). Interestingly, some riboswitches are found in tandem where the two aptamer units can recognize either the same or different ligands. Following this pattern, many glycine riboswitches consist of two glycine sensors separated by a short linker (Mandal et al., 2004) and are capable of intricate tertiary interactions (Huang et al., 2010; Butler et al., 2011). Alternatively, tandem riboswitches that bind different ligands modulate gene expression in response to both compounds, thus operating as a two-input logic gate (Sudarsan et al., 2006). To date, only one example exists of a trans-acting riboswitch. In Listeria monocytogenes, SAM-binding to its aptamer induces transcription termination of the SreA and SreB genes. The resulting short RNA then pairs with the 5′ UTR of the mRNA-encoding virulence regulator PrfA and down-regulates its expression at the translational level (Loh et al., 2009). Following this example, it was speculated that more trans-acting metabolite-dependent RNA regulators will be discovered (Serganov and Nudler, 2013).

The TPP riboswitch is the most widespread riboswitch and it is found in both pro- and eukaryotes. In fungi, algae and plants, TPP riboswitches regulate alternative splicing to induce the formation of mRNA molecules that are likely to be unstable. This result is achieved either by the introduction of upstream start codons that results in the translation of a non-functional open reading frame (ORF) in fungi and algae (Cheah et al., 2007; Croft et al., 2007), premature translation termination in algae (Croft et al., 2007), or a modified 3′ UTR in plants (Bocobza et al., 2007; Wachter et al., 2007). The recent reports on this eukaryotic riboswitch provided a first glance at such mechanisms in which plants utilize RNA–small molecule interaction to control metabolic pathways. At this stage it is not clear, whether additional, similar riboswitch-based mechanisms are present in plants or other eukaryotes.

In this review, we describe in detail the structure and the mode of action of the single plant riboswitch characterized to date. We further discuss the significance of this element in the maintenance of adequate thiamin levels and in the upkeep of homeostasis of plant central metabolism. The last section of this review describes the potential of engineering riboswitches as in vivo metabolite sensors and molecular platforms for gene control.

The Plant TPP Riboswitch

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Structure of the 3′ untranslated region of the plant THIC gene

In all plant species investigated to date, the TPP riboswitch resides in the 3′ untranslated region (UTR) of the THIC gene. Surprisingly, not only the presence of this element, but also the entire organization of the 3′ UTR region of the THIC genes is conserved in plants (Figure 1); (Bocobza et al., 2007; Wachter et al., 2007). The conservation of the various components of the THIC 3′ region (introns, exons, TPP riboswitch, splice sites, poly-adenylation signal, polypyrimidine tract etc.) and the conservation of the distances between them, indicates the importance of the structure of this region for TPP-mediated gene regulation (Wachter et al., 2007). In plants, the THIC stop codon is immediately followed by an intron (i.e. int1) of variable length. This intron is followed by a short exon (about 100 bp) and the following intron (i.e. int2; approx. 150 bp in length), is either retained or spliced out depending on the TPP levels in the nucleus. The TPP riboswitch is typically located on the 3′ region of int2. In many plant species, but not in all cases, the 3′ splice site of int2 also resides on the P2 helix of the riboswitch (Wachter et al., 2007). Furthermore, a poly-adenylation signal resides in this second intron, and this signal remains in the THIC mRNA only when int2 is retained, at low TPP levels. In such a case, poly-adenylation occurs at a fixed position, 25 bp after the signal. However, when TPP levels are high, int2 is spliced out, no poly-adenylation signal remains on the THIC mRNA, a longer 3′ UTR is produced and poly-adenylation occurs at various positions (Bocobza et al., 2007; Wachter et al., 2007). Remarkably, the TPP aptamer, which resides over the 3′ splice site of int2, is not present on the mRNA that originates from intron retention because poly-adenylation occurs upstream from it and remains only partially on the mRNA that result from intron splicing. This partial aptamer was found to be dysfunctional (Wachter et al., 2007), indicating that the TPP aptamer is present in its active form only on the pre-mRNA. This finding suggested that only the TPP aptamer present on the pre-mRNA is able to bind its ligand and direct splicing of int2 inside the nucleus. Thus, once splicing has occurred, the dysfunctional aptamer no longer affects the fate of the resulting mRNAs.

image

Figure 1. Structural conservation in the plant THIC 3′ UTR containing the thiamin pyrophosphate (TPP) riboswitch. The 3′ UTR is depicted downstream of the THIC gene stop codon. Grey boxes and thick black lines represent exons and introns, respectively. The alternatively spliced intron (int2) is represented by dashed lines (the 5′ and 3′ splice site are depicted as 5′ss and 3′ss, respectively). The conserved elements are represented by logos generated from an alignment of the THIC genomic sequences of Arabidopsis thaliana, Solanum lycopersicum, Ricinus communis, Setaria italica, and Sorghum bicolor. The aptamer of the TPP-binding riboswitch consists of five helices termed P1 to P5 with their complementary helices marked with an apostrophe, and the junction J2/3 that connects helices P2 with P3. poly-Ad, poly-adenylation.

Download figure to PowerPoint

Mode of action of the plant TPP riboswitch

How the structural rearrangement of the TPP aptamer affects splicing is not fully understood to date, but the strong conservation of the entire THIC 3′ region suggests a common mechanism in all plants. Sequences on the 5′ side of the P4-P5 riboswitch stems are complementary to the sequences of the 5′ splice site of int2, indicating that the two could interact directly. This hypothesis was confirmed using an in-line probing experiment that showed the stabilization of the 5′ splice site at low TPP concentration (Wachter et al., 2007). This result further suggested that the TPP aptamer, in its unbound state, may prevent splicing by interacting directly with the 5′ splice site of int2. Similarly, in Neurospora crassa the TPP aptamer interacts with a conserved sequence element located near a 5′ splice site to form a base-paired structure and regulate alternative splicing (Cheah et al., 2007; Li and Breaker, 2013). Conversely, determination of the bound state of the TPP aptamer structure of Arabidopsis (Thore et al., 2006) revealed that the 3′ splice site of int2 is exposed in this state (Bocobza et al., 2007). In addition, TPP causes an increase in the structural flexibility of the nucleotides immediately upstream of the 5′ splice site and also at the splice site junction (Wachter et al., 2007). Taken together, it is likely that the TPP aptamer directs the splicing of int2 by the mutually exclusive formation of structures that either mask or expose the splice sites under low or high TPP levels, respectively. The binding of TPP to its aptamer affects the accessibility of the splice sites to the spliceosome and governs the removal of this intron. Interestingly, removal of the 3′ splice site of int2 did not prevent intron splicing but leads to the usage of a different 3′ splice site (Wachter et al., 2007). This situation indicates that the structural rearrangement of this RNA region upon TPP-binding predisposes the RNA molecule to splicing. Furthermore, additional structural changes in the 5′ region flanking the riboswitch also reveal that ligand-dependent modulation outside the TPP aptamer might be important for the control of the 5′ splice site structure (Wachter et al., 2007).

How alternative splicing of the second intron in the 3′ region of THIC (int2) affects gene expression is a different, but yet not completely answered, question. Experiments using a reporter gene fused to the 3′ regions of THIC corresponding either to the retained (short) or to the spliced (long) variant, suggested that the extended length of the THIC 3′ UTR affects the steady-state level of this transcript. Specifically, reporter transcripts containing the long THIC 3′ UTR (in which int2 is spliced out) accumulate at lower steady state levels than the respective transcripts that contain the short THIC 3′ UTR (where int2 is retained; Wachter et al., 2007). The reduced stability of the intron-spliced variant was further confirmed by measuring the decay rate of the THIC alternative variants using the transcription inhibitor actinomycin D. The results showed that the relative stability of the long THIC 3′ UTR variant (where int2 is spliced out) is lower than that of the short THIC 3′ UTR variant (in which int2 is retained; Bocobza et al., 2007). Thus, intron splicing, which yields the long 3′ UTRs, might increase the THIC transcript turnover.

Recent works described the role of alternative splicing in entering transcripts into the nonsense mediated decay (NMD) process (Kalyna et al., 2012; Drechsel et al., 2013). NMD therefore might take part in the mechanism controlling the turnover of the long THIC transcripts. Recent findings in our group indicated that NMD pathway mutants accumulate the intron-spliced transcripts even when intracellular TPP concentrations are low (unpublished data). Interestingly, in all plants observed, a polypyrimidine tract was present ~20 bp upstream the TPP aptamer and an additional one was located immediately downstream the TPP aptamer (Figure 1). Polypyrimidine tract-binding proteins were reported to be involved in the regulation of pre-mRNA splicing in plants (Stauffer et al., 2010; Rühl et al., 2012), RNA poly-adenylation, and in mRNA stability (Sawicka et al., 2008). To date, the involvement of this polypyrimidine element in riboswitch function, in RNA turnover, or in the regulation of intron splicing remains to be determined.

Another unresolved question is how TPP enters the aptamer binding pocket or makes initial contacts with the aptamer. Comparison of the free and TPP-bound aptamer models revealed that the helical elements involved in metabolite recognition are not pre-arranged in a side-by-side orientation prior to ligand binding (Baird et al., 2010). Furthermore, this aptamer was found to undergo large-scale folding and considerable compaction in response to TPP binding (Lang et al., 2007; Kulshina et al., 2010) in an ‘outside-in’ folding trend (Micura et al., 2013). In its bound state, the TPP aptamer adopts a ‘tuning fork’ shape in which two long parallel helices recognize the ligand (Thore et al., 2006). The central region of one helix directly connects to the pyrimidine moiety of TPP, while the other contacts the pyrophosphate of TPP, which is bound as an Mg2+ chelate. Thus, TPP is completely encapsulated by the RNA aptamer. These two long helices also connect a short third helix, which contains the 5′ and 3′ termini of the aptamer. Notably, this three-helix junction consists of only seven nucleotides, and does not interact with TPP (Baird et al., 2010), as reported in other riboswitches (e.g. c-di-GMP). The metabolite-induced structural rearrangement of the TPP aptamer is variably tuned to function more as dimmers or rheostats rather than binary on/off power switches. Hence, gene expression varies according to the ligand concentration and to the affinity of the aptamer for its ligand. This situation allows the TPP riboswitch to constantly convert changes in TPP concentration into appropriate gene expression to answer the evolutionary need for precise genetic regulation (Baird et al., 2010).

The Role of Riboswitches in the Metabolism of their Corresponding Ligand

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Structured RNAs are readily adaptable because compensatory mutations and base isostericity will directly affect their conformation (Cruz and Westhof, 2009). Likewise, riboswitches are very amenable, and this has facilitated, through the selection of appropriate mutations, the efficient fine-tuning of these elements to reach the most physiologically relevant affinity for their ligand. This phenomenon is well illustrated in RNA thermometers, where variation in the length and strength of base-pairing, tunes helix melting and regulation to an optimal temperature response (Grigg and Ke, 2013). Similarly, the mode of riboswitch action is based on the direct structural response of the RNA molecule to the variations of ligand concentration and does not involve protein interaction; thus riboswitches can modulate gene expression quickly at minimum energy cost. They are consequently excellent candidates to regulate the expression of genes involved in the metabolism of primary metabolites whose levels must be most optimally controlled. A natural riboswitch thus maintains adequate expression level of its regulated genes. Since these genes are usually involved in the biosynthesis, catabolism, signaling, or transport of the riboswitch ligand, this mechanism allows a cell to preserve homeostasis of a given primary metabolite (i.e. a riboswitch ligand) with minimal energy cost. Hence, riboswitches, particularly in the case of eukaryotes add an additional layer of control on essential, super-regulated metabolic pathways.

In some organisms, certain classes of riboswitches can control a number of different genes or operons with the same aptamer domain (e.g. the purine- and S-adenosylmethionine (SAM)-binding riboswitches). In some cases, riboswitches have optimized gene expression according to the function of their adjacent gene(s) (Mulhbacher and Lafontaine, 2007; Tomsic et al., 2008). This is well illustrated in Bacillus subtilis, in which at least 11 SAM-binding riboswitches control their own transcriptional unit with different ligand-binding affinities despite nearly invariant secondary structure and nucleotide compositions in the effector-binding site (Tomsic et al., 2008). Specifically, SAM-binding riboswitches controlling SAM biosynthetic genes exhibit a higher affinity than those controlling genes involved in SAM transport. As a consequence, B. subtilis preferentially imports exogenous SAM supplies rather than performs de novo SAM biosynthesis (Tomsic et al., 2008).

Since the characterization of the first riboswitch a decade ago, extensive research efforts have been devoted to the determination of the molecular mechanisms that underlie the capacity of riboswitches to regulate gene expression. However, little information was known with respect to the participation of these RNA regulatory elements in the metabolic function of the corresponding ligands. Two recent reports have described the consequences of TPP riboswitch malfunction on thiamin metabolism in plants and algae. In plants, the TPP riboswitch directs THIC expression in response to TPP levels (Bocobza et al., 2007; Wachter et al., 2007). Unlike in algae (see below), there is no known mutant that harbors a deficient TPP riboswitch. Thus, to study the biological role of this element, transgenic Arabidopsis lines specifically altered in function of their TPP riboswitch were generated (Bocobza et al., 2013). This result was obtained by generating expression cassettes containing the Arabidopsis THIC (AtTHIC) gene promoter, coding and 3′ UTR regions (the latter containing the TPP riboswitch element) with or without an A to G point mutation in the TPP riboswitch (i.e. A515G, relative to the stop codon) that reduces its activity (Sudarsan et al., 2005). These cassettes were introduced independently into the background of an Arabidopsis mutant in which a T-DNA insertion abolishes AtTHIC expression (Kong et al., 2008). The results showed that, in Arabidopsis, TPP riboswitch deficiency elicits intron retention and thereby triggers THIC overexpression by elevating the level of the more stable retention variant. Moreover, higher overall levels of AtTHIC expression were observed in the riboswitch-deficient plants throughout the whole day period, with no change in amplitude and frequency of AtTHIC circadian oscillations. Notably, riboswitch-deficient plants also exhibit elevated levels of thiamin monophosphate (TMP), thiamin and TPP, synthesized in this order (Bocobza et al., 2013). While TMP levels increase almost three-fold, only a moderate augmentation of thiamin and TPP levels is observed. Thiamin pyrophosphate serves as an obligatory ligand for the key enzymes involved in both the TCA cycle and the pentose phosphate pathway. Interestingly, thiamin requiring enzymes extracted from plants deficient in riboswitch activity display higher in vitro enzymatic activities in the presence of increasing TPP concentrations as compared with the control plants. It has been suggested that TPP could not accumulate as much as its precursor (TMP) because TPP usage was also increased in these plants (Bocobza et al., 2013). Thus, riboswitch-dependent regulation of thiamin biosynthesis in the nucleus (where splicing occurs) seems to be highly efficient to maintain appropriate TPP levels in mitochondria and chloroplasts, compartments in which this co-factor molecule is required.

In Chlamydomonas, the TPP riboswitch regulates the THIC and the THI4 genes, which encode proteins producing the pyrimidine and thiazol moiety of the thiamin molecule, respectively. Increased TPP levels cause alternative splicing events that decrease the level of the THIC and THI4 transcripts. The Chlamydomonas pyr1 mutant harbors a single point mutation in the P2 stem loop of the THI4 TPP aptamer that prevents TPP binding (Croft et al., 2007). This mutation causes the accumulation of THI4 transcripts in pyr1. Unexpectedly, THIC transcripts also accumulate in pyr1 despite their functional riboswitch, likely to balance thiamin metabolism and avoid accumulation of pathway intermediates. As a result, pyr1 displays elevated levels (5- to 10-fold) of thiamin monophosphate (TMP), thiamin and TPP. Thus, it was concluded that, in Chlamydomonas, the TPP riboswitch regulates the expression of thiamin biosynthetic genes to ensure optimal metabolism of thiamin esters. Additionally, the Chlamydomonas THIC riboswitch can use as ligands both an intermediate (hydroxymethylpyrimidine pyrophosphate, HMP-PP) and the final product (TPP) of the thiamin biosynthetic pathway. Thus, if pyrimidine production (i.e. HMP-PP) exceeds that of thiazol, then HMP-PP binds to the aptamer and down-regulates THIC expression, which in turn reduces HMP production. This facilitates the balance between the two branches of thiamin biosynthesis (the pyrimidine and thiazol branches) and allows optimum thiamin biosynthesis (Moulin et al., 2013).

The study of TPP riboswitches deficient organisms point to how such elements may serve as a buffer mechanism that prevents thiamin deficiency or surplus (Bocobza et al., 2013). The de-regulation of thiamin metabolism caused by a single point mutation in the TPP riboswitch demonstrates that riboswitches control the expression level of essential metabolic genes to govern the metabolic pathways of their ligands. It also suggests that the TPP riboswitch has evolved to bind its ligand with optimal affinity so that in nuclei regulation of thiamin biosynthesis controls thiamin supply for the mitochondria and chloroplast.

Endogenous Riboswitch Activity and Homeostasis in Core Metabolism

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Remarkably, all riboswitch ligands known to date (nucleobases, amino acids, cofactors, metal ions, and second messengers) represent core/primary or most vital cell metabolites and metabolic pathways. This situation is exemplified in mycoplasmas, which are considered to have arisen by degenerative evolution, leading to the loss of ancestral genes, to the reduction of its genome, and therefore to the retention of a limited metabolic network. Notably, the remaining metabolic pathways in mycoplasmas are those involved in the metabolism of riboswitch ligands known to date in prokaryotes (Figure 2). For such an organism that harbors a relatively limited metabolic network, riboswitches provide an excellent solution to sense metabolism and regulate gene expression accordingly because riboswitch binding does not require the formation of protein complexes or full-length RNA transcripts. This situation raises the hypothesis that riboswitches may have been used to maintain metabolic homeostasis in ancient organisms and to a lesser extent in extant bacteria. During evolution, these RNA-based elements may have evolved to regulate a smaller number of genes but the mechanism involved in gene regulation may have become increasingly more complex. It is therefore likely that riboswitches have been replaced by their protein counterparts that now regulate metabolism. While riboswitch mode of action has been studied extensively, only recent reports described the global, network-level metabolic consequences of riboswitch malfunction. Both in algae and in plants riboswitch deficiency altered thiamin metabolism (see above). Although no negative effect on the growth of the Chlamydomonas mutant (i.e. pyr1) was observed (Moulin et al., 2013), Arabidopsis plants harboring the deficient TPP riboswitch displayed a strong physiological and metabolic phenotype (Bocobza et al., 2013). In these plants, a general increase in the plant respiration rate was observed, together with the accumulation of six amino acids (alanine, β-alanine, aspartate, threonine, proline, and tryptophan) and the reduction of seven other amino acids [γ-aminobutyric acid (GABA), glycine, methionine, histidine, glutamine, tyrosine, and phenylalanine]. Moreover, riboswitch-deficient plants accumulated significantly less isoprenoids including chlorophyll a and b, β-, δ- and γ-tocopherol, β-cryptoxanthin, iolaxanthin and neoxanthin as compared with control plants, which likely to have reduced their photosynthetic rate. Consequently, riboswitch-deficient Arabidopsis plants displayed leaf chlorosis, growth retardation and delayed flowering (Bocobza et al., 2013). It was further concluded that de-regulation of thiamin metabolism in Arabidopsis leads to a stronger flux through central metabolism (Bocobza et al., 2013). The same study suggested that riboswitches are essential regulators of core/primary metabolism, reinforcing the hypothesis that they could be ancient metabolic control systems that maintain homeostasis.

image

Figure 2. Riboswitches mediate gene control in essential pathways of core/primary metabolism. The scheme represents an overview of core/primary metabolism (reference pathway; KEGG, http://www.genome.jp/kegg/), in which the pathways present in Mycoplasma mycoides are in color. Each riboswitch class is highlighted by the name of its ligand (white text over a black background). Riboswitches control the levels of enzymes cofactors, nucleotides, carbohydrates, and amino acids. Metal ions binding riboswitches were omitted. 2′-dG, 2′-deoxyguanosine; AdoCbl, Adenosylcobalamin; ci-di-GMP, cyclic di-guanosyl-5′-monophosphate; FMN, flavin mononucleotide; GlcN6P, glucosamine-6-phosphate; MoCo, molybdenum co-factor; PreQ1, pre-queuosine; SAH, S-adenosylhomocysteine; SAM, S-adenosylmethionine; THF, tetrahydrofolate; TPP, thiamin pyrophosphate.

Download figure to PowerPoint

Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

There is a growing interest in the utilization of RNA-based strategies to engineer synthetic gene controllers that can be employed to program cellular networks with novel biological functions. Because of its chemical properties, RNA is an attractive candidate to design synthetic systems for gene regulation. The four RNA building blocks can interact via hydrogen bonding, base-stacking, and electrostatic interactions. This has allowed the development of models, based on the optimization of energies contributed by the Watson-Crick AU and GC base pairs as well as the GU wobble pair, to predict RNA secondary structure (Liang et al., 2011). Furthermore, because RNA folding is primarily dictated by its secondary structure, a primary sequence can be used in silico to predict RNA tertiary structure and generate native-like RNA conformations (Das and Baker, 2007; Parisien and Major, 2008). Computational tools were developed to assist the rational design of genetic controllers and facilitated the engineering of functional RNA molecules to regulate gene expression (Liang et al., 2011). However, given the difficulties predicting non-canonical base pair contributions, the molecular basis to adjust the activity of an RNA-based control element (i.e. a riboswitch) is poorly understood (Grigg and Ke, 2013).

Strategy for the development of artificial riboswitches

To develop riboswitch-based platforms for gene regulation, a natural or a synthetic aptamer should be linked to a gene-regulatory component in a suitable genetic context. Importantly, the conformational change associated with ligand binding to the aptamer should be designed to affect the activity of a gene-regulatory component, in a way that gene expression will be regulated post-transcriptionally in response to the changes in intracellular ligand level. To this end, functional RNA components can be taken directly from natural systems (as in using an existing aptamer and its endogenous ligand or a known terminator hairpin) or designed and assembled artificially. A synthetic aptamer can be isolated using systematic evolution of ligands by exponential enrichment (SELEX) technology. In SELEX, successive rounds of RNA isolation and amplification, using standard reverse transcription and polymerase chain reaction processes are used to select rare functional sequences out of large libraries. To construct a riboswitch, the aptamer (natural or synthetic) should then be bridged with a linker to a post-transcriptional regulatory element (e.g. terminator hairpin). The linker sequence participates in the structural changes of the RNA-based system to transfer the conformational changes of the aptamer to the regulatory elements. It should therefore be designed in silico, and improved in vivo using selection and screening strategies (Buskirk et al., 2004). Optimization of the linker using the finest library design is crucial to achieve desired riboswitch functions. For example, an in vivo genetic screen of a linker library (using bacteria or yeast) was used to develop new riboswitches (see below), but also it can provide new functions to natural RNA components (i.e. turning an ‘on’ into an ‘off’ riboswitch (Nomura and Yokobayashi, 2007; Muranaka et al., 2009).

Moreover, native riboswitch-based mechanisms can guide us in engineering artificial platforms. The SAM and guanine riboswitches from different loci in the Bacillus subtilis genome exhibit a wide range of transcript levels of adjacent genes, of transcriptional read-through efficiencies, and of binding affinities (Mulhbacher and Lafontaine, 2007; Tomsic et al., 2008). Because the aptamer is invariant in this organism, the differential response of these riboswitches is tuned by sequence changes in the nucleotides adjacent to the aptamer (Stoddard et al., 2013). Thus, the activity of an RNA-based gene regulator (natural or synthetic) can be adjusted by optimizing the nucleotides adjacent to the ligand-binding pocket.

Engineered riboswitches in unicellular organisms

In bacteria, ribosome loading onto mRNA is directly affected by the structure of the ribosome-binding site. The relative ease in modulating this structure has encouraged researchers to design RNA-based metabolite-dependent gene regulators relying on the control of translation initiation. In E. coli, such devices were produced by coupling a theophylline aptamer to a ribosome-binding site through a linker sequence capable of structural rearrangements. The accessibility of the ribosome-binding site can be altered either by ligand-dependent changes in base-pairing interactions between the aptamer and linker (Desai and Gallivan, 2004) or by local nucleotide shifts within the linker (Suess et al., 2004). In another example, the TPP riboswitch was linked to a hammerhead ribozyme to generate a synthetic RNA-based TPP biosensor that represses gene expression in response to TPP in E. coli (Wieland et al., 2009). In an additional study, a linker was optimized to modulate the function of this artificial riboswitch. Specifically, a sequence within the riboswitch was randomized and the resulting library was screened for new synthetic devices that exhibit TPP-dependent gene activation (Nomura and Yokobayashi, 2007). This emphasizes the flexibility of riboswitch mechanisms to implement various modes of gene regulation.

In yeast, metabolite-responsive natural RNA-based gene controlling elements are less prevalent than in prokaryotes and gene-regulatory mechanisms are more complex. Thus, different design approaches have been taken to engineer artificial riboswitches in this organism. In one example, the RNA aptamer for tetramethylrosamine (TMR) was integrated directly into the stem of a transcription activator. Part of the stem was also randomized to facilitate the screening for TMR-responsive gene-regulatory activity (Buskirk et al., 2004). In a second designed approach, an aptamer connected to a hammerhead ribozyme through a linker, was placed in the 3′ UTR of a target gene. The linker was optimized to induce a structural rearrangement that activates ribozyme self-cleavage upon binding of the ligand to its aptamer and this, in turn, lowers gene expression. This system used theophylline and tetracycline as ligands and could also function as logic gates (AND, NOR, NAND, or OR gates) as well as exhibiting cooperativity (Win and Smolke, 2008). Interestingly, as ribozyme activity is determined merely by RNA properties, and is independent of the cell-specific gene-regulatory machinery, this device can be used across different organisms including mammals (Chen et al., 2010).

Engineered riboswitches in higher organisms

In plants, the design and engineering of a riboswitch-based gene control platform is highly challenging due to the lack of information on natural metabolite-responsive RNA elements and the complexity of the post-transcriptional gene regulation machinery. Yet, chloroplasts have a separate gene expression mechanism in which translational control is much more prevalent and is largely compatible with the one found in prokaryotes. Based on the above, the theophylline-responsive control device engineered in E. coli (Suess et al., 2004) was successfully optimized in silico to meet the requirements of translational regulation in chloroplasts, and the resulting synthetic riboswitch was stably transformed into the tobacco chloroplast genome (Verhounig et al., 2010). Remarkably, this synthetic RNA-based metabolite-dependent gene regulator responded to exogenously applied theophylline and repressed gene expression in chloroplasts through translational regulation. Furthermore, a repressible gene expression system was engineered in Chlamydomonas reinhardtii chloroplast to inhibit protein synthesis in response to TPP levels. In this system the 5′ UTR of the CrTHI4 gene, which contains a TPP riboswitch (Croft et al., 2007), was fused to the NAC2 gene, so that TPP supplementation stopped the expression of the RIBOSOMAL PROTEIN S12 gene (RPS12), which encodes a plastid ribosomal protein, and RPOA gene, which encodes the α-subunit of the chloroplast bacterial-like RNA polymerase. Using this system the authors studied these two essential chloroplast genes (Ramundo et al., 2013). In addition, a theophylline riboswitch was designed to mediate translation in a eukaryotic cell-free translation system (wheat germ extract; Ogawa, 2011). In this system, ligand binding to the aptamer modulated the internal ribosome entry site to promote translation.

Our laboratory has recently performed experiments towards engineering plants with a riboswitch-based TPP sensor. To this end, the YFP reporter (YELLOW FLUORESCENT PROTEIN) gene fused to the AtTHIC 3′ UTR was expressed under the CaMV-35S promoter in Arabidopsis. This construct was previously shown to respond to exogenous TPP levels, however, due to the endogenous TPP levels in wild-type plants, the response was mild (Bocobza et al., 2007). Thus, we introduced the same construct into a known thiamin deficient mutant (Atthi1; Li and Rédei, 1969) background. These plants were then exposed in vitro to increasing TPP concentrations and levels of thiamin esters (thiamin, TMP, TPP) as well as YFP fluorescence were measured. The results showed that augmentation of TPP levels caused a strong reduction in YFP reporter levels (Figure 3(a)). When Atthi1 plants were exposed to low exogenous TPP concentrations (from 0 to 100 nM) a linear correlation was observed between the exogenous and the intracellular TPP levels (Figure 3(b)). This suggested that under limited thiamin resources, all thiamin esters are converted into the active form of the molecule (i.e. TPP). It is not known to date whether TPP enters the cells in this form. Since thiamin deficient mutants can be rescued by all three thiamin esters, TPP might be hydrolyzed into thiamin prior to being absorbed. Furthermore, when Atthi1 plants were exposed to higher TPP concentrations (from 100 to 500 nM), TPP levels reached a plateau and thiamin started to accumulate in the plant tissues. This suggested that TPP cannot accumulate in the plant tissue beyond a certain level, but instead excess thiamin esters are stored in the form of thiamin molecules (Figure 3(b)).

image

Figure 3. Riboswitch-based detection of thiamin deficiency in a thiamin mutant. The results presented were obtained with 2-week-old Atthi1 mutant Arabidopsis seedlings that expressed the YELLOW FLUORESCENT PROTEIN (YFP) reporter gene fused to the native AtTHIC 3′ UTR (driven by the CaMV-35S promoter). (a) Phenotype (top) and YFP fluorescence (bottom) observed in the transgenic plants exposed to increasing thiamin pyrophosphate (TPP) concentrations. Similar results were obtained in two additional independently transformed lines. (b) Thiamin and TPP levels were measured by high pressure liquid chromatography (HPLC) analysis in transgenic plants that were exposed to increasing TPP concentrations. Values are presented as means ± standard error (SE) using five independent biological replicates (n = 5). (c) Thiamin and TPP levels were measured by HPLC analysis in wild-type plants grown in the absence of TPP in the medium. Values are presented as means ± SE using five independent biological replicates (n = 5). (d) Relative YFP fluorescence (measured using total protein extracts obtained from the transgenic plants) as a function of the TPP concentrations present in the media on which the plants were grown. Values are presented as means ± SE using five independent biological replicates (n = 5). (e) Relative YFP fluorescence, measured in Figure 1(d), as a function of the intracellular TPP concentrations measured in Figure 3(b).

Download figure to PowerPoint

Notably wild-type Arabidopsis plants grown in the absence of exogenous TPP accumulate thiamin and TPP (Figure 3(c)). These wild-type plants accumulate thiamin esters to the same extend as Atthi1 mutant plants fed with about 150 nm TPP. Levels of TMP remained below detection in this assay. Interestingly, while the fluorescent signal correlated with exogenous TPP concentration following the trend of exponential decay (Figure 3(d)), this signal correlated with the intracellular TPP levels in a linear manner (Figure 3(e)). Given this linear correlation, we suggest that the TPP riboswitch (in the Atthi1 background) can be used as a ‘biosensor’ to monitor endogenous TPP levels in plants. Moreover, as intracellular TPP levels could not exceed a certain limit (0.7 ng/mg FW), it can be concluded that this riboswitch has evolved to respond at least to all physiological TPP levels that can occur in the plant cell (i.e. the largest range of possible TPP levels observed in this mutant). It remains to be determined whether other synthetic RNA-based metabolite-dependent gene regulators retain this capacity in other organisms. This result also indicated that the dynamics of ligand uptake should be taken into account when using a RNA-based biosensor to estimate ligand concentration in a medium.

Thus, despite the higher complexity of eukaryotic systems it is possible to engineer such regulatory mechanisms including in plants. The resulting genetically-encoded metabolite-dependent gene regulators can be used to report on metabolic pathway activity when for example linked to a reporter (Michener et al., 2012). More generally, it might be employed to control cellular function of choice depending on the downstream gene they control. This has potentially great biotechnological and medical applications. Furthermore, altering the affinity of a riboswitch to its natural ligand could deregulate its biosynthesis and hence be used to increase ligand levels. This strategy can be exploited for the large-scale production of, for example, nutritionally important ligands such as vitamin B1 (i.e. thiamin) through manipulating the TPP riboswitch.

Conclusions

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

Since the discovery of the first riboswitch over a decade ago, the research in this field has exposed fascinating and novel mechanisms of metabolite-dependent gene regulation. In contrast, only limited research was dedicated to the investigation of their physiological role at the organism level. Recently, the metabolic alterations that occur in plants harboring a deficient TPP riboswitch were described and indicate the importance of this element in maintaining homeostasis of primary metabolism. Notably, these elements are prevalent in bacteria, and a class of riboswitch exists for most if not all of their metabolic pathways. Thus, we wish to support the premise that riboswitches are ancient regulators of metabolic pathways and essential to maintain metabolic homeostasis. It is also reasonable to suggest that these elements were selected against in the course of evolution and consequently replaced by their protein counterparts (i.e. receptors and transcription factors). Saying this, possibly no additional bona fide riboswitch will be discovered in higher organisms whereas many other, yet undescribed, RNA and small molecule interaction-based mechanisms do exist. Above and beyond, the elucidation of riboswitch mechanisms has greatly assisted in the development of synthetic RNA-based genetic regulators and more artificial riboswitches will likely be reported in the future with implication in biotechnology and medicine.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References

We wish to thank Tom and Sondra Rykoff Family Foundation, Roberto and Renata Ruhman, the Adelis Foundation, Leona M. and Harry B. Helmsley Charitable Trust, Minna James Heineman Stiftung and the Raymond Burton Plant Genome Research Fund for supporting the work in the A.A. lab. A.A. is an incumbent of the Peter J. Cohn Professorial Chair. S.E.B. received the Weizmann Institute, faculty of life sciences Dean Fellowship for part of his work on riboswitches in plants.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. The Plant TPP Riboswitch
  5. The Role of Riboswitches in the Metabolism of their Corresponding Ligand
  6. Endogenous Riboswitch Activity and Homeostasis in Core Metabolism
  7. Engineering Riboswitch-based Platforms for in Vivo Gene Control and Metabolite Sensing
  8. Conclusions
  9. Acknowledgements
  10. Conflict of Interest
  11. References