Small RNA regulators in bacteria: powerful tools for metabolic engineering and synthetic biology
Small RNAs, a large class of ancient posttranscriptional regulators, have recently attracted considerable attention. A plethora of small RNAs has been identified and characterized, many of which belong to the major small noncoding RNA (sRNA) or riboswitch families. It has become increasingly clear that most small RNAs play critical regulatory roles in many processes and are, therefore, considered to be powerful tools for metabolic engineering and synthetic biology. In this review, we describe recent achievements in the identification, characterization, and application of small RNAs. We give particular attention to advances in the design and synthesis of novel sRNAs and riboswitches for metabolic engineering. In addition, a novel strategy for hierarchical control of global metabolic pathways is proposed.
KeywordsSmall RNARiboswitchGene expressionPosttranscriptional regulationMetabolic engineeringSynthetic biology
In recent years, it has become evident that posttranscriptional regulation mediated by RNA regulators is critical to many cellular processes in both prokaryotic and eukaryotic kingdoms (Grosshans and Filipowicz 2008). Hundreds of candidate regulatory RNA genes in many bacteria have been predicted and characterized using a variety of experimental tools, such as direct labeling and RNA sequencing, Rnomics (shotgun cloning), DNA microarray, and Genomic SELEX technologies and bioinformatic approaches (Gottesman 2005; Li et al. 2013b; Livny and Waldor 2007; Pichon and Felden 2008). RNA regulators have been exploited for use in molecular engineering and compared with other regulatory systems; posttranscriptional regulation using small RNAs is cost-effective (Altuvia and Wagner 2000). Furthermore, cells are able to use these small molecules to rapidly respond to various environmental signals and stresses. According to their mechanisms of action, these RNA regulators can be divided into two main classes: small noncoding RNAs (sRNAs) and riboswitches. In this review, we summarize recent advances in the discovery, design, synthesis, and application of sRNAs and riboswitches. We also predict future directions for the application of small RNA regulators in metabolic engineering and synthetic biology.
sRNAs in bacteria and their mechanisms of action
Over the past decades, sRNAs in bacteria (especially in Escherichia coli) have been predicted and experimentally investigated. In 1967, the first bacterial sRNA 6S RNA, encoded by the E. coli gene ssrS, was discovered and characterized (Hindley 1967; Wassarman and Storz 2000). Many studies have shown that 6S RNA is highly abundant during transition from the exponential to the stationary growth phase (Decker and Hinton 2009; Trotochaud and Wassarman 2004; Wassarman 2007). 6S RNA can directly interact with the σ70-RNA polymerase and downregulate the transcription of many σ70-dependent promoters, leading to long-term cell survival during nutrient deprivation (Geissen et al. 2010; Klocko and Wassarman 2009; Steuten et al. 2013). Based on a genome-wide transcriptional analysis, Neusser et al. (2010) demonstrated that the effect of 6S RNA on transcription is not strictly confined to σ70-dependent promoters but is also involved in stationary phase adaptation. Interestingly, 6S RNA can also dissociate from σ70-RNA polymerase during nutritional upshift, enabling cells to resume rapid growth after starvation. This indicates the crucial role of 6S RNA in universal transcriptional regulation (Steuten et al. 2013; Wurm et al. 2010).
In 1984, the first chromosomally encoded sRNA, MicF (174 nucleotides), which inhibits translation of the OmpF mRNA, was identified in E. coli (Mizuno et al. 1984). By 2000, about 13 sRNAs had been serendipitously discovered by traditional methods (for instance separation by polyacrylamide gel electrophoresis) (Masse et al. 2003). Since 2001, many bioinformatic methods designed to analyze intergenic genomic regions (Livny and Waldor 2007; Tjaden et al. 2006; Vejnar et al. 2013; Wassarman et al. 2001) and experimental strategies, such as comparative genomics and microarray technology, Rnomics, and deep sequencing (Altuvia 2007; Huttenhofer and Vogel 2006; Sharma and Vogel 2009), have been developed and used to identify new sRNAs at the genome level in many model organisms, including E. coli (Livny and Waldor 2007), Bacillus subtilis (Irnov et al. 2010), Pseudomonas aeruginosa (Sonnleitner et al. 2012), Staphylococcus aureus (Geissmann et al. 2009), and Listeria monocytogenes (Christiansen et al. 2006). To date, more than 80 sRNAs have been experimentally identified in E. coli. With the ongoing development of genomics, more and more sRNAs will be predicted and identified. This will not only enrich the sRNA database (Li et al. 2013b) but will also provide new insights into their mechanisms of function and their roles in cellular pathways.
Applications of sRNAs
Genetic circuit engineering with sRNAs
Since the discovery of sRNAs, researchers have sought to exploit them to mimic bacterial gene regulation to control genes of interest at the posttranscriptional level (Table 2). Coleman et al. (1984) were the first to develop and introduce this approach for the inhibition of target genes. By inserting artificial sequences (which complement the ribosome regions of Ipp, OmpC, and OmpA mRNAs) between two stem loops at both 5′ and 3′ termini of the naturally occurring MicF sRNA, they successfully repressed the expression of Ipp, OmpC, and OmpA. In recent years, great advances have been achieved in the emerging field of synthetic biology (Chen et al. 2010; Rodrigo et al. 2013), particularly in the field of synthetic RNA biology (Benenson 2012; Chappell et al. 2013). By analyzing the common structural characteristics of natural E. coli sRNAs, Man et al. (2011) developed a semi-rational strategy to design artificial sRNAs for specific gene silencing in Gram-negative bacteria. Many synthesized sRNAs effectively suppressed the expression of an exogenous EGFP gene and the endogenous uidA gene in E. coli. However, because of the high diversity of sRNA sequences and non-stringent target base pairing, and the many factors involved in this complicated process, deficiencies in the efficacy of rational design soon became apparent. Accordingly, a high-throughput screening strategy combined with rational design was shown to be more efficient. Using a fused fluorescent protein as reporter, Sharma et al. (2012) successfully isolated and characterized two artificial sRNAs that can posttranscriptionally repress the endogenous genes, ompF and fliC, respectively. Meanwhile, Rodrigo et al. (2012) reported a fully automated methodology for the design and experimental validation of synthetic sRNAs in a cellular environment. By combining Watson–Crick interactions (Ellington 2007) with physicochemical principles and structural constraints, the authors engineered sRNA sequences that can implement predefined interactions. The authors also demonstrated that energy of formation and activation are sufficient criteria to engineer RNA interaction and regulation in bacteria (Rodrigo et al. 2012).
sRNAs as powerful tools for metabolic pathway engineering
Metabolic engineering using an antisense RNA strategy has been performed in Clostridium acetobutylicum (Desai and Papoutsakis 1999). By over-expressing long antisense RNAs, the activities of the enzymes butyrate kinase, phosphotransbutyrylase, and phosphotransacetylase were significantly decreased, which resulted in increased production of acetone and butanol. Similarly, several antisense RNAs have been designed and overexpressed in E. coli to silence acetate kinase and phosphotransacetylase. This reduced carbon flux to acetate and increased heterologous gene expression (Kim and Cha 2003; Nakashima and Tamura 2009; Nakashima et al. 2006). Impressively, Srivastava et al. (2000) indirectly improved the yield of active organophosphorus hydrolase by downregulating sigma 32 (RpoH) with antisense RNA, indicating that modulation of transcriptional sigma factors with novel sRNAs might be a potential strategy for metabolic engineering. Although this antisense strategy has been applied, such antisense RNAs have not been thoroughly investigated with respect to unknown specificities and variable efficacies (Sharma et al. 2012). Consequently, sRNAs of short length are more attractive for gene regulation and pathway engineering.
To precisely investigate the functions and roles of these sRNAs, a stringent overexpression vector was designed and constructed that uses L-arabinose as an inducer (Majdalani et al. 1998). Accordingly, Kang et al. (2012) introduced and overexpressed the sRNA, RyhB, in an aerobic platform E. coli strain (Kang et al. 2009). As predicted, high levels of the target product, succinate, were accumulated, suggesting that sRNAs with specific functions are powerful tools for metabolic engineering in E. coli. More recently, Li et al. (2013a) reported that by modulating the heme biosynthesis pathway by constitutive expression of RyhB, the titer of 5-aminolevulinic acid was increased by 16 % compared with the parental strain, which contained the genes, hemAM, hemL, and rhtA (Kang et al. 2011). By comparing the levels of the sRNA, SgrS, in E. coli K-12 (JM109 and MG1655) and E. coli B (BL21), Negrete et al. (2013) reduced acetate secretion from E. coli K-12 by over-expressing SgrS. To rationally engineer an optimized pathway for the production of target chemicals, Na et al. (2013) selected the natural sRNA, MicC, to act as a scaffold and synthesized 130 sRNAs composed of two parts: a scaffold MicC sequence and a target-binding sequence. By fine-tuning the target chromosomal gene expression, the titers of tyrosine and cadaverine were substantially increased. These results demonstrated that sRNAs are versatile tools for use in metabolic engineering and synthetic biology (Yoo et al. 2013; Chen et al. 2010; Rodrigo et al. 2013), and their application will be greatly enhanced with the development of sRNA databases (Mehta et al. 2013).
Improved stress resistance with sRNAs
In nature, many sRNAs are involved in controlling a variety of stress responses, including envelope stress, temperature stress, and acid stress (Hoe et al. 2013). In the enterobacter, E. coli and Salmonella, at least 12 sRNAs, InvR, MicA, MicC, MicF, OmrAB, RprA, DsrA, RseX, SdsR, CyaR, VrrA, and RybB, have been proposed as regulators of the envelope stress response (Boehm and Vogel 2012; Coornaert et al. 2010; Vogel and Papenfort 2006). For instance, DsrA, an 87-nucleotide sRNA, is recognized as a multifunctional genetic regulator in E. coli and several studies have demonstrated that this sRNA plays crucial roles in response to many environmental stresses (Lease et al. 2004). In the fermentation process, it is critical to maintain a robustness and prolonged productivity under stressful conditions. Thus, improvement of a strain’s tolerance would be beneficial for bioprocessing applications. In E. coli, the global transcription regulator, RpoS, is posttranscriptionally upregulated by the sRNA DsrA, RprA, and ArcZ (Majdalani et al. 2002; Papenfort et al. 2009; Sledjeski et al. 1996), and downregulated by OxyS (Zhang et al. 1998). More recently, Gaida et al. (2013) reported that simultaneous overexpression of DsrA, RprA, and ArcZ substantially increased acid tolerance and provided protection against carboxylic acid and oxidative stress during the exponential phase. Similarly, applying this multi-posttranscriptional regulation strategy, Kang et al. (2008) constructed a stress-induced system for the high-level production of chemicals, including polyhydroxybutyrate and 1,3-propanediol (Liang et al. 2011b). Therefore, improvement of stress resistance with sRNAs is an effective strategy.
Potential applications of sRNAs: global regulation in vivo
Riboswitches are noncoding RNA sequences located in UTRs of mRNAs and can specifically control the expression of a large number of genes in response to changing cellular conditions (Barrick and Breaker 2007; Henkin 2008; Serganov 2009). Unlike the sRNAs discussed above, riboswitches are cis-acting and can directly bind small molecules (ligands) to cause allosteric rearrangement of their structures and a quick controllable response (Vitreschak et al. 2004). Therefore, riboswitches are considered to be one of the oldest universal genetic factors and are widely distributed in archaea, bacteria, fungi, and plants (Barrick and Breaker 2007).
Riboswitches in bacteria and their mechanisms of action
Riboswitches identified and characterized
Thiamine biosynthesis or transport
Bacteria, archaea, fungi, and plants
Miranda-Rios et al. (2001)
Riboflavin biosynthesis and transport
Winkler et al. (2002b)
Cobalamin biosynthesis; transport of cobalamin
Nou and Kadner (2000)
Methionine biosynthesis and transport; SAM metabolism
Grundy and Henkin (1998)
Wang et al. (2008)
Purine metabolism and transport
Mandal et al. (2003)
Adenine metabolism and transport
Mandal et al. (2003)
Mandal et al. (2003)
Roth et al. (2007)
Lysine biosynthesis, transport and catabolism
Mandal et al. (2003)
Lipfert et al. (2007)
Glucosamine-6-phosphate synthesis and transport
Klein and Ferre-D'Amare (2006)
Physiologic adaptation and virulence
Sudarsan et al. (2008)
Regulski et al. (2008)
Barrick et al. (2004)
Ames and Breaker (2011)
In most cases, natural riboswitches negatively regulate gene expression. Direct interaction with a specific metabolite causes changes to the secondary structures of the RBS and AUG sequences, which eventually results in translation repression (Fig. 1d). To date, only three riboswitch activators, pbuE, ydhL, and gcvT, which are involved in adenine (pbuE and ydhL) and glycine (gcvT) synthesis, have been identified (Breaker 2008). With specific aptamer interaction, translation of the corresponding gene was triggered and strengthened (Fig. 1e). Although it was generally accepted that the formation of a Watson–Crick base pair between the ligand and the aptamer is specific, recent studies show that riboswitches can reprogram ligand specificity by folding diverse secondary structures (Serganov and Patel 2012), indicating that this approach to posttranscriptional regulation is much more flexible than expected.
Application of riboswitches
Driven by the identification of new natural riboswitches and increased understanding of their functional mechanisms, many studies have focused on the design and engineering of novel riboswitches for wide-ranging applications.
Construction of biosensors for environmental monitoring
Application examples of sRNAs and riboswitches
Expression of an antisense RNA to silence the butyrate kinase gene
Increase of acetone and butanol
Desai and Papoutsakis (1999)
Downregulation of the sigma factor 32 with antisense RNAs
Increase of active organophosphorus hydrolase
Srivastava et al. (2000)
Expression of antisense RNAs to silence the genes pta and ackA
Decrease of acetate and increase of recombinant proteins
Kim and Cha (2003)
Design and expression of effective PTasRNAs to silence the ackA gene
Repression of the AckA activity
Nakashima et al. (2006)
Construction of a stress-induced system based on regulation of the sigma factor 38
Production of chemicals such as polyhydroxybutyrate
Kang et al. (2008)
Overexpression of antisense RNAs to silence the genes mutS, mutD and ndk
Increase of mutation frequencies
Nakashima and Tamura (2009)
Overexpression of the sRNA RyhB
Increase of succinate
Kang et al. (2012)
Overexpression of the sRNA RyhB
Regulation of the heme biosynthesis and increase of 5-aminolevulinic acid
Li et al. (2013a)
Synthesis of sRNAs to repress the target chromosomal genes
Increase of tyrosine and cadaverine
Na et al. (2013)
Overexpression of the three sRNAs DsrA, ArcZ and RprA
Increase of acid stress resistance
Gaida et al. (2013)
Engineering of riboswitches
Monitor of specific small molecules
Desai and Gallivan (2004)
Design of antiswitches (trans-action)
Control of eukaryotic gene expression
Bayer and Smolke (2005)
Synthesis of a theophylline-response riboswitch to regulate the gene cheZ
Navigation of bacteria with small molecules
Topp and Gallivan (2007)
Development of ribozyme switches with a sensor domain (aptamer) and an actuator domain (hammerhead ribozyme)
Regulation of cell growth and in vivo sensing of metabolite production
Win and Smolke (2007)
Cloning the sensor of the coenzyme B12-response riboswitch
Monitor of the synthesis and the import of coenzyme B12
Fowler et al. (2010)
Combination of in vitro and in vivo selection to identify novel riboswitch
Construction of recombinant strains to seek and destroy atrazine
Sinha et al. (2010)
Development of a novel synthetic riboswitch
Positive response to 2,4-dinitrotoluene
Davidson et al. (2013)
De novo synthesis of riboswitch modulated by theophylline
Regulation of gene expression at the transcriptional level
Wachsmuth et al. (2013)
Construction of a riboselector with modules of riboswitch and selection
Enrichment of lysine and tryptophan metabolite-producing microbes
Yang et al. (2013)
Generating novel biosensors for metabolic pathway engineering
Construction of genetic circuits or optimization of metabolic pathways requires precise and dynamic control of gene expression (Dahl et al. 2013). Over the past decade, many novel riboswitches that act as regulatory elements have been developed, screened, and characterized using the SELEX strategy (Kiga et al. 1998), which has recently been expertly reviewed (Blouin et al. 2009; Henkin 2008; Joyce 2007; Serganov and Patel 2007; Suess and Weigand 2008). To date, many riboswitch control systems have been constructed in several model organisms (Bayer and Smolke 2005; Topp et al. 2010; Win and Smolke 2007) and applied in many areas of metabolic engineering (biosynthesis, bioremediation, and health and medicine) (reviewed in Liang et al. 2011a) (Table 2). To simplify the engineering process, Qi et al. (2012) constructed several ligand-sensing sRNA regulators by substituting the riboswitch expression platform with naturally occurring trans-sRNAs to fine-tune the intracellular transcriptional cascades. Compared with previous approaches, this new strategy is much more modular, flexible, and versatile. Similarly, Ceres et al. (2013) generated novel active riboswitches by the modular selection of different readout domains (expression platforms) to regulate translation both in vitro and in vivo. More importantly, this technique enables rapid engineering of novel functional RNAs (Ceres et al. 2013). To expand the toolbox for the rational design and construction of biosynthetic and regulatory networks, riboswitches have the potential to be transcriptional regulators, although several in vivo screening approaches have failed. Excitingly, by applying the well-characterized theophylline aptamer as sensor, Wachsmuth et al. (2013) successfully constructed functional transcriptional riboswitches incorporating in silico designed RNA sequences as expression platforms. This work demonstrated that riboswitches can also be useful tools for regulating gene expression at the transcriptional level.
Although rational metabolic engineering approaches are preferred for the optimization of target pathways and to improve product yields, the desired cellular phenotypes are not always generated because regulation of intracellular metabolic fluxes is usually much more complicated than expected (Chen and Nielsen 2013; Dahl et al. 2013; Pfleger et al. 2006). As a result, many directed evolution strategies (Alper et al. 2006; Isaacs et al. 2011; Santos and Stephanopoulos 2008; Wang et al. 2009) combined with high-throughput screening methods have been employed (Dietrich et al. 2010; Scheel and Lutke-Eversloh 2013; Tannler et al. 2008). Recently, Yang et al. (2013) developed a synthetic RNA device (called a riboselector) comprising a riboswitch (Fig. 4c) (for sensing the intracellular concentration of a specific metabolite and actuating the regulation of gene expression based on metabolite signals) and a selection module (to enable high-throughput screening) (Fowler et al. 2010; Muranaka et al. 2009) to expedite the evolution of metabolite-producing microbes. As expected, by applying this novel riboselector, they successfully enriched for pathway-optimized E. coli strains producing L-lysine and L-tryptophan. Moreover, this riboselector device can provide the foundation for an efficient screening method for strain improvement.
In bacteria, approximately 10–20 % of all genes are predicted by high-throughput computational and experimental approaches to code for RNAs with critical regulatory roles in metabolic, physiological, and pathogenic processes (Romby and Charpentier 2010). Consequently, identification of novel regulatory RNA molecules and documentation of their various functions and modes of action would greatly enrich the toolbox of metabolic engineering. Traditionally, efforts to regulate and optimize genetic circuits and metabolic pathways were mainly concentrated at the transcriptional and translational levels. In comparison, the new posttranscriptional approaches show many advantages, such as rapid response, flexible and precise control, ready restoration, and no metabolic burden. It is, therefore, reasonable to predict that with the discovery of increased numbers of diverse natural sRNAs and riboswitches, and the development of synthetic biology and systems biology (Heinemann and Sauer 2010; Keasling 2012; Kohlstedt et al. 2010; Lee 2011; Nielsen and Pronk 2012), novel sRNAs and riboswitches with specific function can be designed, synthesized, and applied for rational, hierarchical, and dynamic regulation of metabolic pathways (Figs. 3 and 4d). This would enable the rapid development of robust strains for specific purposes. In addition, research in this area will contribute to the understanding of the control mechanisms of natural biological networks.
This work was financially supported by the Major State Basic Research Development Program of China (973 Program, 2012CB720802, 2014CB745103, 2013CB733902, 2013CB733602), the National Natural Science Foundation of China (31200020, 31130043), the National High Technology Research and Development Program of China (863 Program, 2011AA100905), the Program for Changjiang Scholars and Innovative Research Team in University (No. IRT1135), the National Science Foundation for Post-doctoral Scientists of China (2013M540414), the Jiangsu Planned Projects for Postdoctoral Research Funds (1301010B), and the 111 Project.