|Home | About | Journals | Submit | Contact Us | Français|
The note by Leduc and Roberts (14) in this issue of the Journal of Bacteriology sheds new light (i) on the mechanisms of action of the bacterial second messenger, cyclic dimeric GMP (c-di-GMP), and (ii) on the functions of the well-known family of transcription factors. The authors describe Clp from the plant pathogen Xanthomonas axonopodis pv. citri as a new c-di-GMP-binding protein. Surprisingly, Clp belongs to the CRP/CAP (cyclic AMP receptor protein or catabolite activator protein) family of transcription factors. To fully appreciate the implications of this finding, we will take a closer look at the veteran signal transduction actors (CRP-like proteins) and at their newly found calling (c-di-GMP binding).
Let's start with c-di-GMP. This cyclic dinucleotide was discovered over 20 years ago by Moshe Benziman and his colleagues at Hebrew University of Jerusalem (21, 22). However, until recently, c-di-GMP remained unknown to the majority of microbiologists. These days, c-di-GMP is at the height of its fame. No wonder—it has been shown to play key roles in a number of important decisions that bacteria make while adjusting their different lifestyles. One of the best-understood lifestyle changes affected by c-di-GMP involves a transition from the motile single-cellular state to the surface-attached multicellular state that can ultimately lead to biofilm formation. This transition has been explored most extensively in the Proteobacteria (reviewed in references 10 and 20). c-di-GMP has also been shown to affect bacterial virulence (reviewed in reference 26), development (18), cell cycle progression (5), cell differentiation (16), long-term survival (13), etc. There is little doubt that the number and diversity of phenomena found to be affected by c-di-GMP will increase as c-di-GMP continues to be “discovered” in new bacterial species. After all, c-di-GMP is just a messenger, and as such, it can be adapted to receive messages from various inputs and to deliver them to diverse outputs.
The field of c-di-GMP-dependent signaling has experienced remarkable progress in recent years. We have learned a great deal about how c-di-GMP is made by diguanylate cyclases containing the GGDEF domain, how it is hydrolyzed by phosphodiesterases containing either the EAL or HD-GYP domain, how these enzymes work at the atomic resolution level, and how some of them are regulated by external factors. We have learned that c-di-GMP acts globally, at the whole-cell level, as well as locally, akin to signaling molecules in eukaryotes. We have learned about several kinds of protein domains and motifs that bind c-di-GMP (10, 20). However, we are just beginning to uncover the mechanisms through which c-di-GMP works, and that is where the contribution of Leduc and Roberts lies.
Prior to the report by Leduc and Roberts, we knew about the following sites and protein domains involved in c-di-GMP binding (Fig. (Fig.1):1): (i) the PilZ (Pfam: PF07238) (6) domain (1, 24); (ii) the I site, an allosteric site for feedback inhibition present in the GGDEF-domain diguanylate cyclases (3), which is also present in the degenerate GGDEF domains that have lost enzymatic activity (15); (iii) the enzymatically inactive EAL domains, which retain ability to bind c-di-GMP but can no longer hydrolyze it (17); and (iv) the enzymatically inactive HD-GYP domains, which can bind c-di-GMP. The proteins belonging to the latter category have yet to be experimentally characterized, but their existence is readily predictable. These protein domains and binding sites are specific for c-di-GMP and can be deduced from sequence analysis, with at least some certainty. There is also a class of c-di-GMP-specific riboswitches, whose sequences are also identifiable, that control gene expression in a c-di-GMP-dependent manner (25).
And then there are “unpredictable” c-di-GMP-binding proteins. The first such protein, FleQ from Pseudomonas aeruginosa, was discovered by Hickman and Harwood (11). FleQ is a transcriptional regulator that works with a σ54-type factor. It represses expression of the flagellar biosynthesis genes and activates expression of the polysaccharide pel biosynthesis genes involved in biofilm formation. c-di-GMP decreases the affinity of FleQ to the pel promoter. The discovery of FleQ broke the orderly view of c-di-GMP-binding proteins that had just started to emerge. The paper by Leduc and Roberts is in some ways a sequel to the FleQ story. It describes a CRP-like transcription factor whose affinity to DNA is regulated by c-di-GMP (Fig. (Fig.11).
The CRP family proteins are expected to bind cAMP. The notion that Clp, which shares 45% sequence identity with the Escherichia coli CRP, binds and responds to a different nucleotide looks at first glance like “molecular treason.” However, a closer look reveals that the matter is not quite as dramatic as it first appears.
CRP consists of two protein domains, the cNMP-binding domain (PF00027), involved in cyclic nucleotide binding, and the so-called Crp domain (PF00325), a DNA-binding domain specific to the CRP family. The cNMP-binding domain can bind cAMP, but it can also bind cGMP. Therefore, with respect to the ligand specificity of the cNMP-binding domain, a cAMP/cGMP dichotomy already exists. Furthermore, the cNMP-binding domain may contain Fe-S clusters (in the proteins of the FNR family ) or bind heme (in CooA-like proteins discovered in the Roberts laboratory earlier ). To our knowledge, the existence of cGMP in bacteria has not been shown. However, the c-di-GMP-dependent signaling systems are ubiquitous (7, 23). In hindsight, the cNMP-binding domain would seem a logical candidate for c-di-GMP-binding, but this idea had been overlooked prior to the study by Leduc and Roberts (14).
These authors had strong phenomenological support for the hypothesis that Clp may work as a c-di-GMP-dependent transcription factor. A clp gene knockout in Xanthomonas campestris pv. campestris results in decreased virulence gene expression (4), and so do mutations in the c-di-GMP phosphodiesterase genes rpfG and ravR (8). These observations imply that Clp might mediate c-di-GMP-dependent transcription of the virulence genes. The authors dismissed cAMP as a possible effector of Clp function, because Xanthomonas lacks genes for cAMP synthesis or hydrolysis and cAMP has never been detected in this genus.
To test the hypothesis that Clp is a c-di-GMP-binding factor, Leduc and Roberts first overexpressed and purified this protein from X. axonopodis, a close relative of X. campestris. Second, they showed that Clp binds a fragment of DNA containing the CRP binding site, which is identical to the binding site of X. campestris Clp. It is noteworthy that the affinity of apo-Clp for DNA, ca. 60 nM, is comparable to that of CRP-cAMP complex. This observation indicates that Clp does not need a ligand for DNA binding, in contrast to other CRP-like proteins. Third, the authors showed that c-di-GMP inhibits DNA binding by Clp, which is consistent with both the phenotypic data and the strong apo-Clp binding to DNA. The inhibitory concentration of c-di-GMP in vitro is ca. 1 μM, which is within the physiologically relevant range (10). Allosteric inhibition, as opposed to activation, is not very common among transcription factors, and Clp is the first representative of the CRP family whose binding to DNA is allosterically inhibited.
To demonstrate the specificity of Clp-c-di-GMP interactions, the researchers performed two sets of tests. First, they tested the effects of different nucleotides, including cAMP and linear diguanylate pGpG, on DNA binding by Clp. None of the tested compounds affected Clp at physiologically relevant concentrations. However, at very high (millimolar) concentrations, cAMP competed with c-di-GMP, which implies that the sites for cAMP and c-di-GMP may overlap. The second set of experiments probed whether known cAMP-dependent transcription activators, E. coli CRP and Pseudomonas aeruginosa Vfr, can bind c-di-GMP. The results were negative. Therefore, both sets of specificity controls support the conclusion that Clp is distinctly different from other CRP-like factors by its preference for c-di-GMP and unusual mode of action. This is the essence of the Leduc and Roberts' study (14)—concise, straightforward, and elegant.
What's next? For one thing, it is important to know how Clp binds c-di-GMP. To what extent does the c-di-GMP-binding site overlap with the regions comprising the cAMP-binding sites in “classic” CRP factors? How does c-di-GMP binding inhibit Clp binding to DNA? It can be anticipated that future structure-function analyses will give us answers to these questions.
What are the implications of these findings for other organisms? Xanthomonas is not unique in lacking cAMP-dependent signaling. The genomes of many other bacteria have no identifiable adenylate cyclases or cAMP hydrolases. However, they often contain multiple CRP-like proteins. Do some of these CRP-like proteins bind c-di-GMP? In our opinion, the answer is almost certainly positive. What processes these new c-di-GMP-binding CRP-like proteins regulate and how they function mechanistically remain to be investigated.
Now that Leduc and Roberts have expanded the realm of the c-di-GMP-binding proteins, one wonders what other protein domains have been overlooked as potential c-di-GMP-binding domains. One reasonable candidate is the ubiquitous GAF domain (PF01590), which is known to bind cAMP and cGMP, among other small molecular ligands (2). Perhaps a distant cousin of GAF, the PAS domain (9, 27), is another. Whether either of these domains will be found to bind c-di-GMP, or whether Mother Nature will demonstrate her imaginative superiority once again, the expansion of the list of c-di-GMP receptors is very welcome. It helps resolve the existing imbalance between large numbers of enzymes involved in c-di-GMP synthesis and hydrolysis in many proteobacteria and small numbers of c-di-GMP receptors. For example, E. coli K-12 has almost 30 GGDEF or EAL domain proteins, the majority of which are involved in c-di-GMP synthesis or hydrolysis. However, it has only two known c-di-GMP-binding proteins (24). The work by Leduc and Roberts assures us that there have to be more than that, and so the quest for new types of c-di-GMP-binding proteins continues.
What does this study teach us about the CRP family transcription factors? First, it reveals the increased importance of this family. The new calling of some of its members, c-di-GMP binding, broadens the known range of regulatory decisions they make in bacteria. Second, it implies that we have not yet uncovered the full potential of this family. The CRP-like proteins can bind cyclic mononucleotides, Fe-S clusters, heme, and now c-di-GMP. Discoveries of new ligands may lie ahead. We should not underappreciate the flexibility and adaptability of veteran signal transduction actors. They can play new roles and play them spectacularly, which is why they have been so evolutionarily successful.
The c-di-GMP work in my laboratory is supported by NSF (MCB 0645876) and USDA Cooperative State Research Education Extension Service (AD-417) via University of Wyoming Agricultural Experimental Station.
I thank Jill Zeilstra-Ryalls for editorial suggestions.
The views expressed in this Commentary do not necessarily reflect the views of the journal or of ASM.
Published ahead of print on 11 September 2009.