Chemical proteomics reveals a second family of cyclic-di-AMP hydrolases

Cyclic-di-AMP is a relative newcomer to the large family of nucleotide-based signaling molecules, which notably includes cyclic-AMP, cyclic-di-GMP, and ppGpp(p) (1). Cyclic-di-AMP was first implicated as a possible signaling molecule when it was unexpectedly found bound to the active site of the Bacillus subtilis DisA protein, the prototype for a diverse family of diadenyl cyclases (DAC) (2). Shortly thereafter, cyclic-di-AMP was detected in bacterial cells and shown, in the case of Listeria monocytogenes, to trigger the innate immune response in infected cells (3). To date, all identified cyclic-di-AMP synthases have a DAC domain homologous to that found within DisA (1). Signaling, of course, requires mechanisms for both synthesis and degradation of the signal. The best characterized cyclic-di-AMP phosphodiesterase (PDE) is the B. subtilis GdpP(YybT) protein, which contains an active site structure with DHH-DHHA1 domains (4). All previously characterized cyclic-di-AMP PDEs have a similar catalytic region, although the overall protein architecture varies. In PNAS, Huynh et al. now provide evidence for a new class of cyclic-di-AMP–specific PDE, which functions—together with a GdpP ortholog (PdeA)—to degrade cyclic-di-AMP in L. monocytogenes (5). This enzyme was previously defined as a metal-dependent hydrolase that affected ppGpp levels and was therefore designated PgpH (6). Huynh et al. (5) demonstrate that L. monocytogenes PgpH is the primary cyclic-di-AMP hydrolase during growth in broth, whereas PdeA is the dominant activity during intracellular growth in mammalian cells.

Regulation of Cyclic-di-AMP Degradation

The PgpH protein was initially implicated in cyclic-di-AMP homeostasis when it was found to bind specifically to cyclic-di-AMP–coupled Sepharose and confirmed to interact with cyclic-di-AMP in solution studies (7). This type of chemical proteomics approach has been instrumental in the elucidation of cyclic-di-GMP signaling pathways, and is also proving powerful with cyclic-di-AMP, as shown originally for Staphylococcus aureus (8). PgpH contains seven transmembrane segments linked to a cytosolic His-Asp (HD) -containing active site (a member of the 7TMR-HD family of receptors). Huynh et al. demonstrate that PgpH is a Mn²⁺-dependent hydrolase with high specificity for cyclic-di-AMP and no detectable activity for ppGpp (5). The mechanism by which PgpH might affect ppGpp levels, as suggested by prior work, remains unresolved. However, ppGpp does bind to PgpH and allosterically inhibits cyclic-di-AMP hydrolysis. It

Huynh et al. now provide evidence for a new class of cyclic-di-AMP–specific PDE, which functions—together with a GdpP ortholog (PdeA)—to degrade cyclic-di-AMP in L. monocytogenes.

is thus plausible that elevated cyclic-di-AMP, in a pgpH mutant strain, may also inhibit the hydrolysis of ppGpp. There is also suggestive evidence for cross-talk between the stringent response (mediated by ppGpp) and cyclic-di-AMP signaling in B. subtilis, where ppGpp has been shown to inhibit cyclic-di-AMP hydrolysis by the GdpP hydrolase in vitro (4). The in vivo consequences of such cross-talk, and whether it is general for other cyclic-di-AMP hydrolases, remains to be investigated.

In addition to the suggested regulation by ppGpp, it is likely that other mechanisms serve to regulate cyclic-di-AMP degradation, although the details have yet to be fully appreciated. For example, expression of B. subtilis GdpP is regulated by an antisense RNA, which is itself controlled by an alternative σ subunit of RNA polymerase (σ^D), best known for its role in regulation of cell motility and autolysin expression (9). Biochemical studies suggest that GdpP activity may be allosterically regulated by ligand binding to a heme-containing PAS domain in addition to competitive inhibition by ppGpp, as noted above (10). The in vivo roles of these regulatory domains, and even the identity of the relevant ligands, are not completely clear. Whether or not PgpH is also subject to regulation is not yet known, but certainly the presence of an extracellular domain and homology to the ubiquitous seven transmembrane (7TMR) family of receptors provides suggestive evidence that the hydrolytic activity may be regulated by ligand binding.

Regulation of Cyclic-di-AMP Synthesis

The synthesis of cyclic-di-AMP is mediated by one or more DAC. The B. subtilis DisA protein, a DNA-integrity scanning protein that acts as a checkpoint for entry into sporulation, was the first characterized (2). Subsequent studies have revealed a diversity of DACs, all of which contain a DAC catalytic domain similar to DisA (and hence designated as the DisA_N family in Pfam). In B. subtilis, there are three different synthases designated DisA, DacA, and DacB. These serve as prototypes for the three major DAC subfamilies that together account for nearly 99% of the DisA_N family proteins identified in databases (1). The cytosolic DisA protein, found mostly in spore-formers, also contains a DNA-binding domain and couples cyclic-di-AMP synthesis to the integrity of the DNA and acts as a checkpoint protein to restrict growth in response to DNA damage. The major cyclic-di-AMP synthase during growth in B. subtilis is DacA(CdaA) (11). DacA is a transmembrane protein widely distributed among bacteria and, in many cases, including L. monocytogenes, appears to function as the sole synthase. DacB(CdaS) family proteins appear to be largely limited to the Bacillaceae and function during sporulation. As noted for cyclic-di-AMP hydrolases, the synthases are also subject to complex regulation involving interactions with other proteins, DNA, or other regulatory ligands (2, 12–14). An important direction for future research is to define how environmental and physiological signals are sensed, and how these signals regulate both the synthesis and degradation of cyclic-di-AMP.

Looking Forward: Effects of Cyclic-di-AMP on Cell Physiology and Host Interactions

With the identification of a second family of hydrolases for cyclic-di-AMP (5), and the characterization in several systems of one or more DACs (1), we have now identified the key enzymes that regulate the level of cyclic-di-AMP in many bacterial systems. However, our understanding of how cyclic-di-AMP functions in regulating cell physiology is far less advanced. Studies of several systems have demonstrated that dysregulation of cyclic-di-AMP signaling (either overproduction or underproduction) leads to pleiotropic phenotypes, including altered sensitivity to osmotic stress, cell wall defects, and antibiotic sensitivity (11, 12, 15, 16). Most dramatically, in several systems bacteria are unable to grow in the complete absence of cyclic-di-AMP (1), implying a role in one or more essential functions.

Chemical proteomics has led the way in helping unravel the downstream targets of cyclic-di-AMP. Many of these have been identified by either affinity purification (as used for the detection of PgpH), or by screening of expression libraries for cyclic-di-AMP interacting proteins. Some of the targets identified to date include the Mycobacterium DarR transcription factor (17), the B. subtilis KtrA potassium importer and other proteins containing homologous nucleotide binding domains (8), a widespread PII-like signal-transduction protein (PstA/DarA) (8, 18–20), and L. monocytogenes pyruvate carboxylase (LmPC) (7). During intracellular growth, cyclic-di-AMP is also secreted by L. monocytogenes and subsequently interacts with the host receptor STING to trigger a host response dominated by type I interferon (3).

The second major class of targets for cyclic-di-AMP signaling in bacterial cells is the YdaO family of riboswitches, which directly bind cyclic-di-AMP to regulate gene expression (21). Because riboswitches, in contrast to protein targets, can be readily identified bioinformatically, the appreciation of YdaO family riboswitches as a direct cyclic-di-AMP target greatly extends the number of known genes and processes affected by cyclic-di-AMP. The two most numerically dominant classes of genes associated with YdaO riboswitches are transporters, including potassium uptake systems, and proteins implicated in cell wall homeostasis (21).

Chemical proteomics combined with structural biology have led the way in advancing our understanding or cyclic-di-AMP signaling. The initial appreciation of cyclic-di-AMP as a possible signaling molecule in bacteria emerged, as noted above, during the structural analysis of DisA. We now have an abundance of structures of cyclic-di-AMP bound to interacting macromolecules, including the DisA synthase (2), PgpH hydrolase HD domain (5), LmPC (7), PstA (18–20), and the YdaO riboswitch (22, 23). This wealth of structural detail is welcome, and provides insights into both protein and RNA-based mechanisms of nucleotide specificity. However, the mechanisms that serve to integrate cyclic-di-AMP into the physiology of the cell have largely remained elusive. Two major challenges remain for the future. First, we must define those signals that control the activity of the various DAC synthetic enzymes as well as the GdpP and PgpH class hydrolytic enzymes. Second, we must unravel the effects of cyclic-di-AMP on its targets (both protein and RNA) and downstream effectors to elucidate the mechanisms by which this signaling molecule helps cells adapt to their changing environment. For a signaling molecule discovered little more than 6 y ago, we have come a long way in a short time, but it seems clear that many more exciting advances are yet to come.