The ever-expanding world of bacterial cyclic oligonucleotide second messengers

Cyclic di-GMP (c-di-GMP) controls transitions from motility to biofilm formation.

The first described cdN in any living organism is 3’−5’, 3’−5’ c-di-GMP. This second messenger was first identified and characterized by Moshe Benziman’s laboratory studying cellulose biosynthesis in Komagataeibacter xylinus [14–16], although the prevalence of this cdN was not widely appreciated until 2000 – 2005 with the advent of bacterial genome sequencing and a renewed interest in understanding the molecular mechanisms that regulate biofilm formation [15]. Analysis of bacterial genomes revealed a widely conserved domain of unknown function that was homologous to Benziman’s c-di-GMP synthases [17–19]. This domain, which possess diguanylate cyclase (DGC) activity, was named the “GGDEF” domain for the key amino acids in its active site [20]. The corresponding c-di-GMP phosphodiesterase (PDE) that degrades c-di-GMP was soon shown to be another widely conserved enzymatic domain that was named the “EAL” domain [21, 22]. Moreover, a second PDE domain termed the HD-GYP domain was demonstrated to also degrade c-di-GMP [23]. GGDEF, EAL, and HD-GYP enzymes are present in the vast majority of bacteria, and in addition to being widely conserved, bacterial genomes can encode up to dozens of enzymes involved in c-di-GMP synthesis or degradation [24]. Seminal studies of biofilm formation and motility in Escherichia coli, Vibrio cholerae, Salmonella Typhimurium sp., and development in Caulobacter crescentus demonstrated that c-di-GMP regulated by DGCs and PDEs was associated with the transitions between motile and sessile lifestyles [19, 25, 26].

C-di-GMP signaling enzymes typically are multidomain proteins consisting of an N-terminal sensory domain and a C-terminal enzymatic domain [15]. The N-terminal domain is proposed to bind specific environmental cues that regulate the synthesis and degradation of c-di-GMP, such that in specific environments a sessile lifestyle is initiated or inhibited. Although a handful of environmental cues are known (for example oxygen [27], spermine [28], bile and bicarbonate [29], and reducing conditions [30]), the majority of environmental signals regulating DGCs and PDEs remain to be discovered. Changes in the intracellular concentration of c-di-GMP is sensed by a myriad of transcription factors, riboswitches, and protein complexes whose activity is controlled by directly binding to c-di-GMP [15, 31]. Although classically known as an inducer of biofilm formation and repressor of motility, c-di-GMP functions as a global regulator controlling numerous phenotypes including but not limited to bacterial predation [32], virulence [33], development [34], DNA repair [35], and cell shape [36]. C-di-GMP signaling systems are primarily found in bacteria, although stalk cell formation in the slime mold Dictyostelium discoideum is dependent on c-di-GMP [37].

Cyclic di-AMP (c-di-AMP).

The next cdN to be discovered in 2008 was 3’−5’, 3’−5’ c-di-AMP, first observed to be synthesized by the DNA integrity scanning protein DisA [38]. As excellent reviews of c-di-AMP have been recently published, we will only briefly summarize it here [39, 40]. C-di-AMP is found in many bacterial and archaeal species, and it is notable for having been identified in many gram-positive bacteria not known to use c-di-GMP [41, 42]. C-di-AMP is synthesized by diadenylate cyclase enzymes (DAC) and degraded by PDEs that have a DHH-DHHA1 domain [39]. C-di-AMP has notable differences from c-di-GMP including that mostbacteria encode only one or a few DACs [41], and c-di-AMP is essential in many but not all bacterial species including the Firmicutes [39, 43]. This requirement for growth stems from c-di-AMP as a regulator of cellular osmolarity that controls the import and export of potassium and other osmoprotective molecules [44]. Although the mechanism of this sensing is not fully understood, high extracellular concentrations of potassium increase DAC activity and intracellular c-di-AMP, which then functions to limit potassium uptake while promoting potassium export [44–46]. C-di-AMP plays an analogous role in controlling the intracellular concentrations of other water-soluble osmoregulatory molecules such as certain amino acids and sugars [45, 47]. Thus, in the absence of c-di-AMP, cells are not able to appropriately balance their osmotic state. Accordingly, c-di-AMP is not essential in rich media with lower salt concentrations or defined minimal media [44, 48].

Cyclic GMP-AMP, the third cdN in bacteria.

Bacterial 3’−5’, 3’−5’ cGAMP (3’3’cGAMP), the third known cdN was first discovered in 2012 from the El Tor Biotype of Vibrio cholerae, which is the causative agent of the 7^th and current cholera pandemic [49]. 3’,3’ cGAMP is synthesized by DncV, which interestingly is encoded on the Vibrio Seventh Pandemic-1 (VSP-1) genomic island that is unique to El Tor [50]. DncV was reported to influence V. cholerae motility and colonization, but the receptor for 3’,3’cGAMP in V. cholerae was unknown [49]. However, we recently demonstrated that 3’,3’ cGAMP directly binds to and activates CapV, a patatin-like phospholipase encoded directly adjacent to dncV on the VSP-1 island. 3’,3’ cGAMP binding to CapV activates this enzyme, leading to degradation of the cell membrane [51]. 3’,3’ cGAMP synthesized by a DncV homolog also modulates biofilm formation in Escherichia coli ECOR1, an animal commensal, suggesting this signal may regulate broader functions [52]. DncV is structurally analogous to the eukaryotic enzyme cyclic GMP-AMP synthase (cGAS), and both enzymes evolved from a common ancestor [53]. Together, they belong to a novel protein superfamily termed cGAS/DncV-like nucleotidyltransferase (CD-NTases), which is genetically conserved throughout all bacterial phyla [54, 55]. In stark contrast to DGC and DAC protein families, CD-NTases produce an extensive array of nucleotide signals utilizing all four ribonucleotides to form linear oligonucleotides to cdN and ctN molecules, allowing for specificity and diversity in downstream pathways [55]. In bacteria, recent in vitro studies revealed the first examples of pyrimidine containing cdNs including cyclic UMP-AMP and cyclic di-UMP, and ctNs such as cyclic tri-AMP (cAAA) and cyclic AMP-AMP-GMP (cAAG) [55–57].

A bioinformatic analysis by Burroughs et al predicted such nucleotide synthases might participate in biological conflicts [54]. Indeed, Cohen et. al. showed that dncV and capV initiate altruistic suicide by restricting cell growth upon phage infection, thereby aborting phage replication and limiting phage spread within the bacterial population [58]. Furthermore, 3’,3’ cGAMP synthesis by DncV is induced upon phage infection via an unknown mechanism [58]. CD-NTases like DncV are found genetically associated with putative effector proteins that suggest their primary function is phage defense (Figure 2). Such signaling modules were renamed as cyclic oligonucleotide (coN)-based anti-phage signaling systems (CBASS) [58]. Millman et al proposed a classification system that organizes CBASS systems according to their operonic architecture, effector function and dominant signaling nucleotide [59]. However, a major outstanding question is how phage infection modulates nucleotide synthesis by CD-NTases.

CD-NTases are associated with a variety of putative effectors such as phospholipases, endonucleases, and transmembrane proteins [54, 55]. Patatin-like phospholipases are the most common effector found associated with CD-NTases. Many effector enzymes are fused to putative cdN and ctN sensing domains such as STING and SAVED, and binding of the nucleotide signal to these domains modulates effector function [54, 55, 60]. The SAVED domain, which contains two CRISPR-associated Rossman folds (CARF), binds the specific nucleotide signal produced by their cognate CD-NTase to activate the fused effector function [57]. This was recently demonstrated with the widespread Cap4 endonuclease that is activated after specifically binding to cyclic tri-AMP synthesized by its CD-NTase, CdnD, following phage infection [57].

The majority of CBASS exist as a two gene operon, comprised of a CD-NTase and effector, predicted to confer phage defense (Figure 2) [54, 55]. Many CBASS operons also encode eukaryotic-like ubiquitination or HORMA/TRIP13 ancillary systems predicted to modulate coN synthesis in phage infection [54, 55]. These accessory genes enhance bacterial resistance to more phage in addition to the CD-NTase and its effector [56, 58]. Bacterial HORMA1 initiates cAAG synthesis by binding its associated CD-NTase and TRIP13 dissociates the two [56]. The role of eukaryotic ubiquitination-like systems, however, remains undefined in bacteria. In eukaryotes, cGAS sensitivity to cytosolic DNA is modulated by E3 ubiquitin ligase and deubiquitinase [61]. Thus, a hypothetical function of eukaryotic-like ancillary proteins might be to modulate CD-NTase sensitivity to activating nucleic acids as in cGAS, or inhibitory molecules such as folates in DncV.

The enzymes that degrade 3’,3’ cGAMP remain relatively unstudied although a few examples of 3’,3’ cGAMP PDEs have emerged. 3’,3’ cGAMP is degraded via hydrolysis by the V-cGAP1/2/3s in V. cholerae and PmxA in Myxococcus xanthus, both of which are HD-GYPs [63, 64]. VcEAL is the first 3’,3’ cGAMP specific EAL type PDE encoded by V. cholerae [65]. Unlike the V-cGAP1/2/3 PDEs, VcEAL and PmxA hydrolyze both cdi-GMP and 3’,3’ c-GAMP.

3’,3’ cGAMP and Hypr GGDEFs.

Bacterial 3’3’-cGAMP has also been studied in delta-proteobacteria. Although, in these bacteria 3’3’-cGAMP is not synthesized by a CD-NTase, but by hybrid promiscuous GGDEF (Hypr GGDEF), a GGDEF-like enzyme with active site residue variations that coordinate synthesis of 3’,3’ cGAMP rather than c-di-GMP (Figure 3) [66]. 3’,3’ cGAMP binds riboswitches to modulate iron(III) oxide metal reduction in Geobacter sulfurreducens and mediates osmotic stress response in M. xanthus [66, 67]. Interestingly, in Geobacter c-di-GMP activates biofilm formation and energy production on electrode surfaces, suggesting 3’,3’ cGAMP and c-di-GMP are antagonistic signaling pathways that induce alternative lifestyles in this bacterium [68]. Structural analyses revealed Hypr-GGDEFs have a symmetric active site whose nucleotide product is dependent on substrate availability, in contrast to DncV, which has an asymmetric active site that preferentially binds ATP and GTP as substrates [68, 69]. However, 3’,3’ cGAMP is the major nucleotide signal for both enzymes.

2’,3’ cGAMP as an immune modulator.

In human cells, a structural isomer of bacterial 3’3’-cGAMP, 2’−5’, 3’−5’-cGAMP, is synthesized by the CD-NTase cGAS upon binding cytosolic DNA introduced via viral invasion or intracellular damage [70]. 2’3’-cGAMP then activates the STING receptor which upregulates Type I interferon production, leading to an anti-viral or anti-cancer response [71–73]. STING is also capable of sensing extracellular c-di-AMP released by the intracellular pathogen Listeria monocytogenes, leading to increased interferon response [42, 73]. A second eukaryotic cdN receptor named RECON, which specifically recognizes adenine containing cdNs, was also recently discovered, suggesting an intricate recognition and response of eukaryotic cells to cdNs [74].

CD-NTase – the newest enzyme family that synthesizes coNs.

In contrast to c-di-AMP and c-di-GMP, 3’,3’ cGAMP regulation and effector functions are not as well described. Given the diversity of CD-NTases, characterization of these signaling modules is an exciting new frontier in understanding the evolution of coN signaling systems across biological kingdoms and its adaptation to modulate specific response pathways in different environments from seawater to human hosts. Many CD-NTases that have been identified do not exhibit nucleotide synthesis activity and their function is unknown. Evidence of CD-NTases mediating signaling pathways vital for cellular adaptation and virulence present novel therapeutic targets amid growing concerns with antibiotic resistance. In theory, abortive infection driven by CBASS can be manipulated to limit infection by human pathogens.