Shining the Light on Cyclic di-GMP Dark Matter

ABSTRACT

Bacterial cyclic di-GMP signaling networks often consist of dozens of components, and the majority of these components have no observable function. Dahlstrom et al. (J. Bacteriol. 200:e00703-17, 2018, https://doi.org/10.1128/JB.00703-17) explored the function of every component of the Pseudomonas fluorescens cyclic di-GMP network under 188 different growth conditions and identified activities for 80% of the network. They further demonstrated that multiple mechanisms function in tandem to control the activity of the network in different environments.

TEXT

Our universe is full of dark matter. In astrophysics, dark matter is defined by its measurable gravitation impact but cannot itself be observed. But living systems contain their own version of “biological dark matter,” which consists of predicted genes and proteins that can be observed but whose impacts and contributions to the organization of the surrounding environment are not known. Cyclic di-GMP (c-di-GMP) signaling networks typify this type of biological dark matter, as a majority of the individual components involved in these systems have no readily observable effect on bacterial physiology. In this issue of the Journal of Bacteriology, a systematic analysis by Dahlstrom et al. from the laboratory of George O'Toole begins to shed new light on the function of c-di-GMP dark matter (1).

Discovered by members of Moshe Benziman's laboratory (2), c-di-GMP has been well studied over the last 15 years and is now appreciated as the keystone signal controlling a bacterium's transitions between a sessile biofilm state and a free-swimming motile lifestyle. c-di-GMP signaling systems are predicted to be found in ∼80% of all bacteria based on genome analysis (3). c-di-GMP is synthesized by diguanylate cyclase (DGC) enzymes consisting of a GGDEF domain and degraded by phosphodiesterase (PDE) enzymes containing either a conserved EAL domain or HD-GYP domain (4). These enzymes are modular in nature, as the vast majority of them are fused to an N-terminal “signaling” domain that contains different small-molecule binding domains or inner membrane-spanning helices. Hybrid GGDEF-EAL “dual-domain” proteins are also widespread, and these can function as DGCs or as PDEs or as sensors of c-di-GMP in a manner that is dependent upon conditions (5, 6). To add to this complexity, bacteria encode multiple DGCs or PDEs, with some bacterial genera such as Pseudomonas or Vibrio containing dozens of such enzymes (3). Thus, fundamental issues in this field include why there are so many inputs to the system and how the flow of information in the network is specifically controlled.

These issues have begun to be addressed in some bacterial species by attempting to define the DGCs and PDEs that make the greatest contribution to c-di-GMP-related phenotypes. For example, analysis of individual DGC and PDE mutants in Pseudomonas aeruginosa, Vibrio cholerae, or Escherichia coli identified a few DGCs or PDEs that controlled biofilm formation and motility, allowing models of the c-di-GMP signaling network to be constructed (7–9). However, a common feature of these analyses is that the majority of the DGCs and PDEs in any one bacterium do not appear to significantly contribute to c-di-GMP-regulated phenotypes. Consistent with these studies, experiments in Pseudomonas fluorescens performed in the O'Toole laboratory determined that only 9 of the 43 putative DGC/PDE domains significantly impacted biofilm formation (10). The caveat concerning these experiments is that they were done under only one or a few growth conditions that represented typical bacterial growth medium used in laboratory studies. Bacteria live in an ever-changing world, and a fundamental challenge for bacteria is to appropriately respond and adapt to these fluctuating conditions. Thus, Dahlstrom et al. hypothesized that the function of the c-di-GMP dark matter in P. fluorescens might be illuminated by exploring its impact in a variety of different environments.

P. fluorescens is an excellent system to explore this hypothesis, as biofilm formation under all conditions tested is dependent on surface presentation of LapA adhesion. LapA localization is controlled by c-di-GMP through inside-out signaling via the inner membrane c-di-GMP binding dual-domain LapD protein and the periplasmic LapG protease. At low c-di-GMP concentrations, LapG does not interact with LapD and cleaves LapA from the cell surface. When c-di-GMP concentrations increase, LapD sequesters LapG and LapA is presented on the cell surface, leading to surface adherence and biofilm formation (5). Thus, the simplicity of the manner in which P. fluorescens builds a better biofilm allowed the researchers to focus on the nature of the intracellular state of c-di-GMP signaling in different environments. In total, the authors quantified biofilm formation of 50 different P. fluorescens null mutations in the genes encoding each GGDEF domain, EAL domain, and GGDEF-EAL dual-domain protein and putative c-di-GMP effectors in 188 different environments. These environments were created using a 96-well Biolog system, in which each well contains a different organic compound in a common minimal glycerol-based growth medium. The inclusion of glycerol as a common carbon source allowed the researchers to quantify the formation of P. fluorescens biofilms in response to each of the unique Biolog compounds without also requiring the bacterium to utilize them as a nutrient source.

Compared to the wild-type strain, 39 of the 50 mutants exhibited a significant impact on biofilm formation as summed across all 188 environments. The number of environments in which a given gene had an impact varied greatly, ranging from as low as one to over one hundred. Thus, consistent with the model that c-di-GMP functions to sense changing environments, an activity could then be detected for the majority of these enzymes. As the Biolog system changed environments by the addition of organic compounds, it is likely that the remaining enzymes/effectors that lacked activity function in other environments in which P. fluorescens has evolved. Surprisingly, 10 strains with mutations in GGDEF-encoding genes showed significantly increased biofilm formation in some environments, bucking the dogma that these enzymes must positively contribute to biofilm formation through the synthesis of c-di-GMP. Perhaps these GGDEFs are contributing to biofilm dispersal, as has been reported for NicD in P. aeruginosa (11). Furthermore, deletion of dual-domain enzymes primarily led to increased biofilm formation, suggesting that these dual-domain enzymes function predominantly as PDEs in P. fluorescens.

Having thus identified a phenotype for most of the c-di-GMP dark matter in P. fluorescens, the authors sought to understand how a changing environment impacted the activity of the various GGDEF domain enzymes, EAL domain enzymes, dual-domain enzymes, and effector proteins. One can speculate concerning three major, non-mutually exclusive mechanisms underpinning the changes in these observed activities: (i) altered protein production through transcriptional regulation, (ii) nutrient cues directly controlling enzyme activity, or (iii) changes in protein complex formation in different environments (Fig. 1) (12). To explore the first of those mechanisms, the authors selected 46 different compounds that significantly altered wild-type biofilm formation to various degrees from low levels to very high levels and directly measured RNA transcripts of the genes coding for GGDEF domain, EAL domain, and dual-domain proteins and putative effectors. Although there were a few notable exceptions, in general, the level of transcript change for any gene was modest and typically less than 2-fold. However, the authors did find six nutrients in which significant groups of genes were coregulated, suggesting that the response to some environments is minimal transcriptional regulation of a subset of network components rather than robust regulation of a limited number of network components. A comparison of the results from P. fluorescens to transcriptomic data from the related species P. aeruginosa showed similar conclusions; the magnitude of changes in transcription was modest, but similar sets of genes might be coregulated in the two organisms. Furthermore, changes in transcription of given genes rarely correlated with their impact on biofilm formation. Thus, transcriptional regulation was observed but did not appear to be the primary driver of environment-specific responses.

FIG 1.

The contribution of GGDEF domain (circle), EAL domain (triangle), or dual-domain proteins to c-di-GMP signaling in different environments could be modulated through increased production via transcriptional regulation, increased activity by interacting with environment-specific cues, or changes to protein complex formation. The report by Dahlstrom et al. suggests that all of those mechanisms contribute simultaneously to modulating the c-di-GMP network of P. fluorescens.

The evidence therefore suggests that environment-specific signals could directly modulate the activity of c-di-GMP-related proteins. One mechanism by which this may occur is changes in the physical interaction between c-di-GMP network components. Protein-protein interactions represent an emerging concept in understanding c-di-GMP signaling specificity, and such interactions have been demonstrated in E. coli, Xanthomonas axonopodis, and P. fluorescens (13–15). To test all potential interactions in the network, the authors cloned each GGDEF domain, EAL domain, and dual-domain protein and PilZ into both the bait and prey vectors and performed the herculean feat of assessing all possible interactions in nearly 2,000 bacterial two-hybrid assays. These experiments demonstrated extensive interactions between different components of the network. Dual-domain proteins showed the greatest degree of interaction, with some proteins participating in 7 to 18 interaction pairs. Thus, these proteins may coordinate “hubs” of c-di-GMP signaling. Interactions with GGDEF enzymes were highly variable, with some of these proteins exhibiting extensive interactions and others showing very few. Finally, EAL and PilZ proteins rarely interacted with other members of the network. In some cases, these interactions correlated with both proteins impacting biofilm formation in the same direction (either positively or negatively) or experiencing the same direction of transcriptional control, but this was not always the case. Note that the protein-protein interaction network was determined for only one environment, and it is likely that interacting partners can change in response to environmental inputs.

The report by Dahlstrom et al. shows that understanding a given c-di-GMP network must be accompanied by the realization that its nature is highly dependent on the growth environment. Any model of a c-di-GMP signaling network can be applied only to the conditions under which it was obtained. Therefore, statements asserting that a given GGDEF domain, EAL domain, or dual-domain protein is a “dominant” member of the network should be tempered or qualified. To understand the true impact of c-di-GMP signaling on the biology of a given bacterium, one must appreciate, and model, the relevant environments in which the c-di-GMP signaling network evolved.

It is clear that the impact of GGDEF domain, EAL domain, and dual-domain proteins changed in different environments in the absence of transcriptional change or complex formation. These findings further highlight the discovery that environmental signals could be funneled into the network by acting as ligands directly binding to and changing the activity of DGCs or PDEs, and such mechanisms have previously been demonstrated (16–18). Additionally, environmental cues can also impact host-derived signals such as quorum sensing autoinducers or central metabolites that modulate the activity of DGCs and PDEs (19). Uncovering the molecular mechanism by which the environment impacts activity remains a daunting challenge given the breadth and diversity of c-di-GMP signaling components, but, ultimately, it is this information that is necessary to understand the physiological role of these systems.

The authors demonstrated that transcriptional regulation of c-di-GMP network components is not a major factor driving differential activity in P. fluorescens and P. aeruginosa but rather that complex formation is a central theme. A take-home message from the article by Dahlstrom et al. is that no one mechanism to control the network can explain environmental control of protein activity and that in c-di-GMP systems these mechanisms occur in tandem to modulate signaling. The contribution of transcription, modulation of protein activity, and complex formation to other complex bacterial c-di-GMP networks remains to be determined, but the report by Dahlstrom et al. offers a means to shine a light on c-di-GMP dark matter in other bacterial species.

The views expressed in this Commentary do not necessarily reflect the views of the journal or of ASM.