PerspectiveBacterial Signal Transduction

Cyclic-di-GMP Reaches Out into the Bacterial RNA World

See allHide authors and affiliations

Science Signaling  23 Nov 2010:
Vol. 3, Issue 149, pp. pe44
DOI: 10.1126/scisignal.3149pe44


The ubiquitous bacterial signaling molecule bis-(3′-5′)-cyclic guanosine monophosphate (c-di-GMP) has brought second messenger signaling back onto the agenda of molecular microbiologists. This is due not only to its general role in promoting biofilm formation, but also to the increasingly diverse array of effector molecules bound by c-di-GMP and of the target processes affected. Effectors include diverse transcription factors and proteins that directly interact with complex cellular machineries, as well as RNA molecules that act as riboswitches to regulate transcriptional elongation or translation. This flexibility in c-di-GMP action enables it to control diverse molecular processes in bacterial cells. New evidence further extends this range to include a c-di-GMP riboswitch linked to a self-splicing intron that has been “domesticated” by its carrier, the pathogenic bacterium Clostridium difficile, to serve in the control of expression of a downstream gene.

After leading a shadow existence in the scientific literature for almost 20 years, bis-(3′-5′)-cyclic guanosine monophosphate (c-di-GMP) has recently moved to center stage. It is now recognized as a “life-style” switch molecule that controls the transition between the motile single-cellular planktonic state of bacteria and their adhesive multicellular existence in a biofilm. Moreover, the enzymes that “make and break” c-di-GMP are ubiquitous and present in high numbers in many bacteria (1). c-di-GMP is generated from two guanosine triphosphate (GTP) molecules by diguanylate cyclases (DGCs) containing GGDEF (Gly-Gly-Asp-Glu-Phe) domains, a highly conserved motif that constitutes the active site or “A-site” of these enzymes (Fig. 1). Cleavage of c-di-GMP is catalyzed by two types of phosphodiesterases (PDEs) that contain either EAL (Glu-Ala-Leu) or HD-GYP (His-Asp, Gly-Tyr-Pro) domains (2). The cellular concentration of c-di-GMP is dynamically controlled by multiple environmental and cellular signals that regulate the abundance and activities of the antagonistically acting DGCs and PDEs, with equilibrium concentrations also determined by product inhibition of most of the DGCs (through a secondary c-di-GMP binding site known as the “I-site”).

Fig. 1

The diversity of c-di-GMP–binding effectors enables the bacterial second messenger c-di-GMP to affect multiple cellular processes. Transcription is regulated by different types of c-di-GMP–binding transcription factors. c-di-GMP–controlled riboswitches regulate transcriptional termination or antitermination or translational initiation. Activity, localization, or proteolysis of target proteins or activities of larger cellular structures are controlled by c-di-GMP–binding effector proteins that act by direct interaction with their targets. Riboswitch structures are based on (16). Proteins are represented by ovoids with colors indicating different types of c-di-GMP–binding domains. Parentheses indicate partial degeneration of GGDEF or EAL domains, which eliminates enzymatic activity but still allows binding of c-di-GMP. R denotes A or G; Y denotes C or U.


Although GGDEF, EAL, and HD-GYP domains are easy to recognize and fairly abundant, the identity of c-di-GMP–binding effectors has remained more elusive. The first-identified and currently best-characterized effector is the PilZ domain, which is allosterically controlled by c-di-GMP (3) and can affect the activities of enzymes such as cellulose synthase (4) or of larger cellular structures such as the flagellar basal body complex (57). c-di-GMP–controlled transcription factors include FleQ (8) and PelD (9), which control exopolysaccharide synthesis in Pseudomonas aeruginosa, and Clp in the plant pathogen Xanthomonas axonopodis, which belongs to the CRP family, classically known to bind to adenosine 3′,5′-monophosphate (cAMP) (10). Intriguingly, slightly degenerate and therefore enzymatically inactive GGDEF and EAL domains can evolve into effectors as long as they retain their ability to bind c-di-GMP (11, 12). Finally, RNA molecules have joined the effector “club” with the description of the first c-di-GMP–controlled riboswitch (c-di-GMP-I), the phylogenetically widespread “GEMM” RNA domain present in the 5′-untranslated regions of various mRNAs (13). The GEMM domain folds into a complex secondary and tertiary structure or “aptamer” that forms a tight and selective pocket, in which c-di-GMP is bound by base pairing as well as by stacking interactions (14, 15).

Lee et al. (16) have now analyzed a second type of c-di-GMP–binding riboswitch (c-di-GMP-II), which was detected in a bioinformatic search for new putative riboswitches encoded in bacterial and archaeal genomes (17). This c-di-GMP-II element is mainly found in anaerobic Gram-positive members of the Clostridiales family. As expected for a riboswitch, it is usually located upstream of an open reading frame (ORF), consistent with a potential role in transcriptional termination-antitermination or translational control (16). In one case, however, the c-di-GMP aptamer region was found to be separated from the downstream ORF (encoding a putative surface protein) by an intervening sequence of about 600 nucleotides that resembled group I self-splicing ribozymes. These self-splicing introns occur in various bacteriophage genes (18, 19) and occasionally even in core genes of bacterial genomes (2022). The initial step in splicing at group I ribozymes is a nucleophilic attack by the 3′-OH group of an exogenous guanosine (such as that provided by GTP) on the 5′-splice site to release the 5′-end of the intron now decorated with an extra guanosine (23).

Lee et al. not only provide evidence that intron excision occurs in vivo, but also show that in vitro binding of c-di-GMP to the c-di-GMP-II region determines where this initial attack by GTP—and therefore self-cleavage of the mRNA—takes place. In the absence of c-di-GMP, GTP-mediated cleavage occurs four nucleotides upstream of the initiation codon of the coding sequence and is not followed by a splicing step. In contrast, c-di-GMP binding to its aptamer in the 5′-region of the mRNA results in a secondary structure rearrangement that allows GTP to attack at a position immediately downstream of the aptamer sequence, thereby attaching guanosine to the now free 5′-end of the intron, which in a second step can attack the 3′-splice site. This releases the intron and splices the exons together.

These alternative reactions should have striking consequences for the translation of the following ORF. In the precursor mRNA, the weak start codon (UUG) is not only preceded by a very weak ribosome binding site (the Shine-Dalgarno or SD sequence) but is also trapped in a double-stranded structure that inhibits translation. GTP-induced cleavage in the absence of c-di-GMP releases the start codon but completely eliminates any SD sequence; thus, this processed mRNA should be translationally “dead.” However, c-di-GMP−GTP-induced splicing generates a strong SD sequence (AGGAGG, with the first five nucleotides derived from the 3′-end of the c-di-GMP aptamer, namely the 5′-exon) at an optimal distance upstream of the initiation codon, which should promote translational initiation.

These findings are remarkable for a number of reasons. Self-splicing introns, which often are located on transposons, are usually regarded as “selfish” DNA elements, but in this case, the intron has been “domesticated” by its bacterial host to play a productive role in gene regulation. That these introns can sometimes be beneficial for their carriers is also suggested by the observation that self-splicing of phage T4 group I introns, which requires ongoing translation, is reduced in stationary-phase bacterial host cells, thereby rendering the phage dormant until nutritional conditions improve again (19). Moreover, with the control of self-splicing of a group I intron, yet another RNA-based molecular mechanism is added to the growing list of target processes controlled by c-di-GMP. Future studies will be important to demonstrate the physiological roles of these processes in vivo. Yet it already seems clear that c-di-GMP delves deeply into the RNA world inside the bacterial cell—which, after all, is not entirely surprising given the fact that c-di-GMP itself is a tiny RNA molecule.

In parallel, various reports have provided evidence that c-di-GMP can act locally in complexes in which DGC, PDE, effector, and direct target components physically interact (12, 24, 25). This sequestration enables the use of the same biochemical principle to control distinct pathways in parallel without “cross talk” (1). So far, direct interactions or even covalent linkage of the relevant domains (4) have been described only for c-di-GMP signaling proteins. However, the involvement of c-di-GMP in various RNA-based mechanisms suggests that GGDEF, EAL, and HD-GYP domain proteins may be found in combination with the cellular RNA-processing machinery as well.

Finally, c-di-GMP signaling raises an interesting evolutionary perspective. As outlined above, there seem to be hardly any constraints with respect to the molecular mechanisms that can be affected by c-di-GMP. However, when it comes to the larger physiological context of c-di-GMP signaling, evolution has been extremely conservative. Wherever we look, c-di-GMP seems to control processes involved in lifestyle switching—in particular by interfering with motility and promoting adhesion and biofilm formation (1, 26, 27). This also holds true for its role in virulence, because chronic infections are often associated with biofilm formation (28), and even for degenerate GGDEF or EAL domain proteins that are enzymatically inactive and act by directly interacting with other macromolecules (29, 30). Although evolution seems to rather freely play around at the molecular level of c-di-GMP action, the physiological ability to form biofilms seems to have been a major and constant selective pressure during eons of bacterial evolution.


Funding: c-di-GMP–related research in the author’s laboratory is funded by the Deutsche Forschungsgemeinschaft, an ERC Advanced Grant, and the Fonds der Chemischen Industrie.

References and Notes

View Abstract

Stay Connected to Science Signaling

Navigate This Article