Bacterial cell volume regulation and the importance of cyclic di-AMP

SUMMARY

Nucleotide-derived second messengers are present in all domains of life. In prokaryotes, most of their functionality is associated with general lifestyle and metabolic adaptations, often in response to environmental fluctuations of physical parameters. In the last two decades, cyclic di-AMP has emerged as an important signaling nucleotide in many prokaryotic lineages, including Firmicutes, Actinobacteria, and Cyanobacteria. Its importance is highlighted by the fact that both the lack and overproduction of cyclic di-AMP affect viability of prokaryotes that utilize cyclic di-AMP, and that it generates a strong innate immune response in eukaryotes. In bacteria that produce the second messenger, most molecular targets of cyclic di-AMP are associated with cell volume control. Besides, other evidence links the second messenger to cell wall remodeling, DNA damage repair, sporulation, central metabolism, and the regulation of glycogen turnover. In this review, we take a biochemical, quantitative approach to address the main cellular processes that are directly regulated by cyclic di-AMP and show that these processes are very connected and require regulation of a similar set of proteins to which cyclic di-AMP binds. Altogether, we argue that cyclic di-AMP is a master regulator of cell volume and that other cellular processes can be connected with cyclic di-AMP through this core function. We further highlight important directions in which the cyclic di-AMP field has to develop to gain a full understanding of the cyclic di-AMP signaling network and why some processes are regulated, while others are not.

INTRODUCTION

The plasma membrane of bacteria is a semi-permeable barrier that separates the cytoplasm from the external environment (Gram-positive bacteria) or the periplasm (Gram-negative bacteria). The plasma membrane establishes a compartment, the cytoplasm, within which most of the chemistry of the cell takes place. The extracellular environment is dynamic, and the osmolality and other conditions fluctuate. Free water molecules will move across a semi-permeable membrane from a solution with a lower solute concentration (hypotonic) to a solution with a higher solute concentration (hypertonic). Bacterial cells adapt the composition of their cytoplasm in response to these changes. As an example, when a bacterial cell is placed in a hypotonic solution, water flows rapidly into the cell, increasing the cytoplasmic volume (Fig. 1A). Positive turgor pressure increases and in extreme conditions, the cell will lyse. By contrast, when a cell is placed in a hypertonic solution, water flows rapidly out, the cytoplasmic volume decreases and the cell can encounter harmful crowding conditions and ultimately plasmolyse (Fig. 1A) (1, 2). Considering how important cell volume regulation is, it is no surprise that bacterial cells have evolved sophisticated mechanisms to control it (3).

Fig 1.

Cell volume regulation in response to osmotic fluctuations. (A) Schematic representation of the bacterial response to osmotic stress. (B) Overview of the bacterial osmolyte influx and efflux systems. Gray circles = cell wall; blue circles = cell membrane; red squares = cyclic di-AMP; open circles = osmolytes; gray arrows = turgor pressure; dashed arrows = movement of water; yellow star = phosphorylated site. Protein domains: green = RCK_C; light blue = CBS; Purple = USP.

Changes in the extracellular osmolality affect cytoplasmic volume on the sub-second timescale (4). An existing regulator system (e.g., a transport protein) must sense and respond on an equivalent timescale to prevent loss of cell vitality or viability. On a longer timescale (minutes to hours), the expression level of such a system can be tuned. Cells rapidly initiate responses to hypertonic stress by accumulating ions or small, often organic, molecules known as compatible solutes or osmoprotectants. Examples of compatible solutes include sugars like sucrose and trehalose, quaternary ammonium compounds like glycine betaine and carnitine, and amino acids like proline. High concentrations of compatible solutes exert stabilizing effects on proteins and macromolecular complexes, and increase the intracellular osmolality to facilitate reabsorption of water (2). The size of these molecules allows them to diffuse through solutions very quickly, even when they are highly crowded with macromolecules (5–8). If we assume that the length of a bacterial cell is 3 µm and the width is 1 µm, a small molecule will diffuse from one cell pole to another on timescales of less than 100 ms. The diffusion time (t_D) for 3D diffusion is given by <d²>/6D, 2 < d > is the mean distance traveled from the starting point by Brownian motion and D is the lateral diffusion coefficient (D ~ 50 µm²/s for small molecules in the cytoplasm) (9). For an effective response to hypertonic stress, the (internal) diffusion of compatible solutes will generally not be limiting, but the cell needs to rapidly accumulate the molecules to minimize the loss of water (10). Conversely, hypotonic stress requires the rapid export of osmolytes to prevent cells from lysing. Next to import and export of osmolytes, prokaryotic adaptation to osmotic fluctuations requires rearrangement of metabolic activity and adjustment of the bacterial cell wall.

To achieve fast and multi-level adaptation, which is required to counter osmotic fluctuations, bacteria utilize second messengers inside the cell that relay extracellular signals. In many cases, these second messengers are derived from the purine nucleotides ATP and GTP. Examples are cyclic-AMP or -GMP, cyclic di-AMP or -GMP, cyclic GMP-AMP, (p)ppGpp, and Ap4A. Recently, second messengers derived from CTP and UTP have also been described (11). Because of their size and hydrophilicity, they are expected to reach in vivo diffusion coefficients of ~50 µm²/s and hence can travel through the cell and encounter molecular targets on millisecond time scales (6, 12, 13).

Cyclic dimeric adenosine 3′,5′-monophosphate (cyclic di-AMP) is a cyclic di-nucleotide recognized for its pivotal role in bacterial cell volume regulation, for example, Staphylococcus aureus and Streptococcus gallolyticus with elevated cyclic di-AMP levels have a reduced cell size, and many osmolyte transporters involved in cell volume regulation are controlled by cyclic di-AMP (14–19). It also plays a (in)direct role in several other important cell processes, such as cell wall homeostasis, sporulation, DNA repair, virulence, biofilm formation, response to acid and heat stress, nitrogen starvation, central carbon metabolism, and regulation of the day-night cycle in cyanobacteria (20, 21). If the intracellular concentration of cyclic di-AMP is not tightly controlled, it can result in severe adverse effects and cell death. Under standard laboratory growth conditions, cyclic di-AMP is essential (16, 22–29). However, cyclic di-AMP is more accurately described as “conditionally essential,” because under specific growth conditions it is no longer required (18, 30, 31).

A few years ago, it was argued that cell wall metabolism and osmoregulation are tightly connected by cyclic di-AMP which regulates osmolyte transporters to balance the turgor (Fig. 1B) (14). The connection of cell wall metabolism and osmoregulation with other processes regulated by cyclic di-AMP is less clear, and it is an open question, how organisms regulate multiple processes independently using a single fast-diffusing signaling molecule like cyclic di-AMP?

There are several ways in which signaling molecules can control separate processes, namely (i) temporal separation via altered gene expression of cyclic di-AMP metabolic enzymes and receptor proteins; (ii) spatial separation via creation of local cyclic di-AMP pools; and (iii) differences in cyclic di-AMP receptor protein sensitivity via a range of binding affinities for cyclic di-AMP. These different ways are not mutually exclusive and can occur concomitantly in a single species. We will discuss each of them and identify issues with each.

• An example of temporal separation via gene expression is that of cyclic di-AMP synthase CdaS from B. subtilis, of which the gene is expressed exclusively during sporulation (28). If sporulation-specific cyclic di-AMP-binding effectors are expressed concomitantly, only then can the process of sporulation be controlled independently from other cell processes. Note that cyclic di-AMP synthesizing enzyme CdaA and cyclic di-AMP receptors such as pyruvate carboxylase and ion transporters/channels have a basal level of expression and will therefore always be affected.

• Alternatively, intracellular cyclic di-AMP can be spatially separated by the formation of local pools/areas in the cytoplasm which contain high/low concentrations of the nucleotide. This phenomenon is well described for cAMP in eukaryotic cells (13). Here, spatial separation often depends on the colocalization of the metabolic enzymes with the receptor proteins of the second messenger. In this case, a high local concentration of receptors affects the free pool of cyclic di-AMP and hence the signaling process. Thus, the problem of fast diffusion of cyclic di-AMP is overcome by the transient binding of the second messenger to a specific location in the cell. For cyclic di-GMP in bacteria, it has been shown with FRET-based cyclic di-GMP biosensors that there can be a difference in the dinucleotide concentration between a mother and a daughter cell (32, 33). For cyclic di-AMP, it has been suggested that cyclic di-AMP synthesizing enzymes colocalize with a cyclic di-AMP-binding K⁺ transporter for efficient delivery of cyclic di-AMP (34), but given the rapid diffusion of small molecules it is not obvious that co-localization is beneficial. Most cyclic di-AMP cyclases and phosphodiesterases are membrane-embedded, whereas some are DNA bound or localized in the cytoplasm. It is tempting to speculate that this enables spatial separation of the second messenger and the formation of concentration gradients. However, the possibility of spatial separation has not been experimentally established. These experiments would be at the resolution boundary of the in vivo fluorescence techniques that are currently available. This, coupled with a cell volume of only a few fL and a diffusion constant in the order of 50 µm²/s, makes spatial separation difficult to test. Another possibility would be the separation of cyclic di-AMP between the inside and outside of the cell. Later in this review, we will discuss the evidence for cyclic di-AMP efflux. However, no extracellular cyclic di-AMP receptors have been identified yet.

• Cyclic di-AMP could independently regulate multiple processes through the different binding affinities of cyclic di-AMP receptors, similar to (p)ppGpp (35). Under laboratory growth conditions, most cytoplasmic cyclic di-AMP molecules likely exist in a protein/riboswitch-bound state (Box 2; Table 1). Proteins that bind cyclic di-AMP do so with dissociation constants (K_D) spanning more than 3 orders of magnitude (0.020–10 µM), but with no clear correlation between affinity and type of protein (20). Riboswitch ydaO and its homologs show K_D values in the picomolar to nanomolar range, indicating that expression is tightly regulated and more strongly than the activity of proteins that bind cyclic di-AMP with micromolar affinity.

TABLE 1.

Estimated number of cyclic di-AMP molecules per cell

Species	Cyclic di-AMP (µM)	Cell volume (fL)	Cyclic di-AMP molecules per cell
Mycoplasma pneumoniae	0.5–1 (26)	0.05 (36)	15–30
Bacillus subtilis	2–5 (37)	1.5 (36)	1,800–4,500
Staphylococcus aureus	2–8 (38)	0.3 (36)	350–1,450
Synechococcus elongatus	19 (39)	2–3 (40)	22,500–34,500

Despite these three ways to control different cellular processes at the same time in a single organism, many of the processes in which cyclic di-AMP is involved seem to be connected and do not have to be separated in time and place. In the last decades, several reviews (and primary research papers) on the role of cyclic di-AMP have been published, upon which we build (20, 21, 41–46). We describe the relation of cyclic di-AMP with cell volume regulation, cell wall homeostasis, sporulation, DNA repair, nitrogen starvation, central carbon metabolism, and glycogen turnover. We then show that all of these processes are intimately connected to cyclic di-AMP’s core function: cell volume regulation, via a set of shared cyclic di-AMP-regulated proteins. We support this view with a concise summary of quantitative data (e.g., Box 2) and an illustration of the fundamental components of cyclic di-AMP’s signaling network and their functions.

We do not cover the literature that relates cyclic di-AMP to acid stress responses, biofilm formation, virulence, and genetic competence, as there are essentially no cyclic di-AMP receptors or metabolizing enzymes identified that directly connect cyclic di-AMP to these processes. What is known has been reviewed in reference (46). We do recognize that organisms are complex adaptive systems in which cell metabolism is an interconnected, interdependent process; when one component of cell metabolism is altered, other parts can be affected. By analogy, when force is applied to a single thread of the spider’s web, the entire web is able to flex to compensate (to a certain degree). Finally, we highlight outstanding questions in the field which could help locate cyclic di-AMP in the wider context of an organisms‘ array of signaling nucleotides.

CELL VOLUME REGULATION

Mutant strains with altered intracellular levels of cyclic di-AMP have been used to determine the di-nucleotide’s physiological functions. The genes of cyclic di-AMP synthesizing (cyclase) or degrading (phosphodiesterase) enzymes have been inactivated or the proteins have been depleted (transcriptionally) to manipulate the intracellular cyclic di-AMP levels (16, 29, 31, 47). A pioneering study on the functional role of cyclic di-AMP showed that S. aureus cells with high cyclic di-AMP levels have a reduced cell size (16). In most cases, “standard” laboratory growth conditions (rich media) do not allow growth of strains with a high or low cyclic di-AMP level (48, 49). Under specific laboratory growth conditions, namely chemically defined media with increased osmolality, some strains with low levels of cyclic di-AMP remain viable (31). Alternatively, strains can acquire suppressor mutations to cope with the altered cyclic di-AMP levels. Many mutations have been found in genes coding for proteins that are involved in cell volume regulation, such as ion and other osmolyte importers (18, 50, 51). Likewise, three of the four first discovered cyclic di-AMP receptors were cation transporters (17). The correlation between growth conditions and the acquisition of suppressor mutations has provided many insights into how cyclic di-AMP and its effector protein network operate. The experimental data derived from these strains in vivo and the related in vitro biochemical data will be discussed in the following sections. The focus of the following section is on the pivotal and direct role of cyclic di-AMP on cell volume regulation.

The primary response to osmotic upshift

The intracellular osmolyte concentration is involved in sustaining positive turgor pressure. Cations like K⁺ also assist in neutralizing the bulk negative charge of the phosphate backbone of DNA and RNA and the overall anionic proteome (2, 52–54). Under hypertonic conditions, water leaves the cell, and the cytoplasmic volume decreases several percent up to 50% (4, 55, 56) (Fig. 1A). The movement of water occurs on the millisecond timescale, and as a result the cell experiences a rapid decrease in turgor pressure and a significant increase in macromolecular crowding, intracellular solute concentrations, and ionic strength (Box 1) (57). The bacterial response to an osmotic challenge is well documented in the context of gene expression changes (58–60). However, osmoregulatory mechanisms must sense and respond faster, within more relevant timescales to counter the rapid loss of cell volume.

BOX 1. Key terms.

Osmolality—the number of osmoles of solute particles in 1 kg of solvent (Osm/kg).

Osmotic pressure—the pressure that must be applied to a solution to prevent the inward flow of pure solvent across the semi-permeable membrane.

Ionic strength—the effective (and not total) ion concentration of the cytoplasm, expressed in molar unit (M) (61). In equation:

$${\displaystyle {\displaystyle I=1/2\sum _{i=1}^n{c}_i{z}_i^2}}$$ (Eq. 1)

where i is the ion identification number, z is charge of the ion, and c is the concentration (mol/L) of free ion.

Turgor pressure—the force that the cytoplasmic solution applies to the cell envelope structures (plasma membrane and cell wall). Positive turgor refers to a higher pressure inside than outside of the cell. This increase in outward pressure stretches the membrane.

Macromolecular crowding—the high concentration of macromolecules in solutions like the cytoplasm of the cell. In growing Escherichia coli, the concentration of macromolecules in the cytoplasm is in the range of 200–250 mg/mL, which corresponds to an excluded volume of about 20%. The high crowding affects reaction equilibria, the diffusion of proteins, the compaction of the nucleoid, liquid-liquid phase separation, complex formation, and protein folding (62).

Kosmotrope and Chaotrope—Kosmotropes stabilize and chaotropes destabilize macromolecules (e.g., proteins and membranes) by altering the preferential interactions of water with biomolecules in the cytoplasm (63). Kosmotropes are ionic (e.g., Mg²⁺, SO₄^2-) or non-ionic (e.g., glycine betaine, ectoine, proline, and trehalose). Chaotropes are ionic (e.g., NH₄⁺, SCN⁻, H₂PO₄⁻) or non-ionic [e.g., guanidinium, urea (high concentrations), and imidazole].

Proton motive force—The electrochemical proton gradient across the plasma membrane composed of the electrical potential (Δψ) and the pH gradient (ΔpH) across the membrane. Secondary transporters utilize the proton motive force for transport of compounds such as compatible solutes and ions against their concentration gradient. The F₀F₁-ATP synthase utilizes it for the synthesis of ATP. The proton motive force (Δp) of bacteria is typically less than −180 mV (64), and if one proton is co-transported with a neutral solute than at equilibrium, the solute would be accumulated 1,000-fold. In equation:

$${\displaystyle n\mathrm{\Delta }p+{\displaystyle \frac{2.3RT}{F}}\log {\displaystyle \frac{[S{]}_{\text{in }}}{[S{]}_{\text{out }}}}=0}$$ (Eq. 2)

where n is number H⁺ co-transported with the solute S, [S]_in/[S]_out is the maximal accumulation ratio (at equilibrium), R is the gas constant, T is the absolute temperature, and F is the Faraday constant (65).

Potassium is the main cation in the cytoplasm of prokaryotes. The estimated cytoplasmic concentration of potassium under normal laboratory conditions is around 250 mM for E. coli (66), 300 mM for B. subtilis (67), and 500 mM for Corynebacterium glutamicum (68) and Lactococcus lactis (69), whereas extracellular concentrations easily vary from µM to hundreds of mM for these organisms. Bacterial cells commonly import a lot of potassium during an osmotic upshift (70, 71). Counterions, most commonly glutamate, are swiftly imported and/or synthesized to maintain electroneutrality (72). To reduce the cytotoxic effects of potassium accumulation and the increased ionic strength, E. coli and B. subtilis begin accumulating and/or synthesizing neutral compatible solutes such as glycine betaine, trehalose, and other osmolytes to replace K⁺ ions (see below). Hence, K⁺ transport needs to be strictly regulated, which is illustrated by the large number of ATP- and H⁺-driven transporters and potassium channels that play a role in potassium homeostasis (73). This allows the cells to adjust to a broad range of external osmolalities.

Cyclic di-AMP-dependent potassium transport

Many of the links between cyclic di-AMP and potassium transport come from studying mutant strains with altered intracellular levels of cyclic di-AMP. When continuously cultured in adverse conditions, these mutant strains develop gain/loss of function suppressor mutations to survive. Suppressor mutations may occur in either the cdaA or gdpP genes or their regulatory sequences, to adjust the intracellular cyclic di-AMP levels (47, 74). However, many suppressor mutations also arise in genes coding for potassium transporters, potassium channels and other proteins.

To understand why this is the case, it pays to be familiar with the phenotypic changes that occur when the level of cyclic di-AMP is increased or decreased abnormally. Mutant strains with high levels of intracellular cyclic di-AMP have a reduced ability to import potassium. When potassium import is prevented, they decrease in size (31), become sensitive to hypertonic conditions (47, 75, 76), are heat resistant (50, 76), and undergo plasmolysis more easily (18, 29, 47, 74). To survive, they accumulate gain of function mutations in potassium influx systems (50, 51).

By contrast, mutants with low intracellular cyclic di-AMP accumulate potassium to toxic levels and the cell size increases (25, 31, 77, 78), and as a result they grow more slowly and lyse more easily under hypotonic conditions (29). To survive, they accumulate loss of function mutations in potassium influx systems (74) and gain of function mutations in potassium exporters (18). The results of these studies are summarized in Table 2.

TABLE 2.

Summary of the phenotypes and suppressor mutations associated with high and low levels of cyclic di-AMP in various microorganisms

	Cyclic di-AMP level	Phenotype	Gain of function	Loss of function
L. lactis	High	Sensitive to high salt (47, 76)	kupB (51) (salt dependent)	-
	High	Heat resistant (47, 76)	- a	-
	Low	Lysis (47, 74)	-	kupB (74)
L. monocytogenes	High	Sensitive to high salt (75)	-	-
	Low (via cdaA depletion)	Lysis and slower growth (29)	-	-
S. pneumoniae	High		trkH (50) (temperature dependent)	-
S. pyogenes	High (via gdpP inactivation)	Decreased virulence (79)	-	-
B. subtilis	Low	Lysis (18)	nhaK (18) (salt dependent)	-
S. aureus	High (via gdpP inactivation)	Decreased cell size (31)	-	-
	Low (cdaA depletion)	Increased cell size (31, 78)	-	-
M. tuberculosis	Low	Increased cell size (77)	-	-
B. burgdorferi	High	Increased cell length (25)	-	-
S. gallolyticus	High	Increased cell size (15)	-	-

^a -, not applicable.

KtrAB/TrkAH

The TrkAH, KtrAB, and KtrCD proteins are low-affinity potassium import systems (apparent K_M for K⁺ of ~1 mM) (80, 81). Although identified as K⁺/H⁺-symporters, these systems are ATP-gated channels, which depend on the electrical potential for potassium influx (82–84). The proteins are composed of two subunits. TrkA, KtrA, and KtrC are the soluble subunits that contain the RCK_C domains required for gating (Box 3). TrkH, KtrB, and KtrD are the transmembrane proteins required for flux of ions across the membrane (71, 84, 85). These systems primarily operate in conditions where external potassium concentrations are high (86). The gating subunits of each system in B. subtilis (19), Streptococcus mutans (87), Streptococcus pneumoniae (50, 88), Streptococcus agalactiae (24), M. pneumoniae (26), S. aureus (17, 89), Synechocystis sp. PCC 6803 (90), and L. monocytogenes (91) all bind cyclic di-AMP (Fig. 1B). This interaction occurs with high affinity (K_D of 40 nM – 8 µM) and destabilizes the interaction between the gating subunit and the transmembrane protein (20, 50, 88). Here, cyclic di-AMP is bound and potassium influx is inhibited.

KimA

KimA, KupA, and KupB are high-affinity potassium transporters of the KUP (K⁺ uptake) family (92). KimA is a K⁺/H⁺ symporter present in S. aureus (91), B. subtilis (18, 19, 34), and L. monocytogenes (91), and its activity is inhibited by cyclic di-AMP (Fig. 1B). Transcription of kimA is stimulated by low external potassium concentrations (19) (apparent K_M for K⁺ of 140–350 µM (91)) and inhibited by cyclic di-AMP via a riboswitch (18). Similar cyclic di-AMP responsive riboswitch sequences have been found upstream of trk, ktr, kdp, and kup genes (93), suggesting that a similar mechanism may be present in these organisms as well. KupA and KupB of L. lactis (putative K⁺/H⁺ symporters) also bind cyclic di-AMP, resulting in an inhibition of potassium import (Fig. 1B) (94).

KdpFABC and KdpDE

The Kdp system is a P-type ATPase, a high-affinity potassium import system (K_M for K⁺ of ~2 µM) responsible for uptake in low external potassium environments (80, 81, 86). Expression of the kdp genes is induced in response to changes in cell turgor (95). The Kdp system (KdpFABC) is a four-subunit membrane protein complex. KdpA is responsible for transport of ions through the membrane, and is dependent on ATP hydrolysis by KdpB (96, 97). The two-component sensor kinase KdpD (along with KdpE) is responsible for sensing of the ion availability and expression of the kdpFABC genes; KdpD senses Na⁺ and NH₄⁺, and cells accumulate these monovalent ions when K⁺ is limiting (95). Expression of the kdpFABC-operon is upregulated in response to increasing osmolality (86, 98). Below millimolar potassium concentrations (in the micromolar range), expression of the kdpFABC genes is induced to enhance the capacity for potassium import, that is, when the activity of low-affinity transporters becomes too low. Among others, KdpD has been identified as a cyclic di-AMP-binding protein, using a differential radial capillary action of ligand assay (DRaCALA), which allows detection of protein-nucleotide interactions based on the differences in mobility of free and protein-bound nucleotides migrating through a nitrocellulose membrane (17, 99). Proteins form hydrophobic interactions with nitrocellulose membranes but free nucleotides do not. Therefore, when a nucleotide-protein mixture is blotted onto a dry nitrocellulose membrane, the movement of the bound nucleotide will be significantly restricted when compared with a free nucleotide. KdpD binds cyclic di-AMP via its USP domain (K_D for cyclic di-AMP of ~2 µM), which leads to inhibition of kdpFABC-operon transcription (Box 3) (Fig. 1B) (17, 75, 98). Cyclic di-AMP is also able to bind a riboswitch upstream of the kdpFABC, to negatively regulate transcription of the operon (100). Binding of cyclic di-AMP results in reduced potassium import. However, KdpFABC activity itself is not affected by cyclic di-AMP (91).

Magnesium influx

In B. subtilis (19, 101) and synechocystis sp. PCC 6803 (90), magnesium import is inhibited by binding of cyclic di-AMP to the CBS domains of MgtE (Box 3) (Fig. 1B). Magnesium is the only divalent cation that reaches tens of millimolar intracellular levels (101). Although magnesium is not directly involved in osmoadaptation, regulation of magnesium transport could be essential for cell volume regulation because divalent cations contribute four times more strongly to the ionic strength of the cell than monovalent ions do (Box 1).

Potassium export

In B. subtilis, KhtT and the potassium exporter CpaA (cation/H⁺ antiporter) both bind cyclic di-AMP (19). KhtT is a soluble protein that forms a complex with membrane protein KhtU (K⁺/H⁺ antiporter). Complex formation is required for transport activity. Binding of cyclic di-AMP by the RCK_C domain of KhtT enhances potassium export (Box 3) (Fig. 1B) (102). In S. aureus (17, 103) and B. subtilis (19), CpaA also binds cyclic di-AMP through its RCK_C domain with a dissociation constant of 9 µM (Fig. 1B) (103). Given an intracellular cyclic di-AMP concentration of 2–5 μM (Box 2), the majority of cyclic di-AMP is bound to other targets, and with a K_D of CpaA for cyclic di-AMP of 9 µM, the (KhtA and) CpaA exporters will have a low activity under normal growth conditions.

BOX 2. Cyclic di-AMP levels and the number of binding sites.

The intracellular concentration of cyclic di-AMP has been estimated in a range of bacterial species (Table 1). In all of these cases, cyclic di-AMP was isolated via a solvent extraction step, followed by LC-MS/MS (liquid chromatography-coupled tandem mass spectrometry) analysis. This method yields the total cyclic di-AMP pool in vivo and does not distinguish between the proportion of bound and free cyclic di-AMP.

The Subtiwiki database (104) provides the protein abundance for B. subtilis, from which we estimate the maximal number cyclic di-AMP-binding sites in the cell. The numbers are as follows: OpuAA (599; condition CH), OpuCA (248; condition CH), DarB (905; condition CH), MgtE (259; condition CH), KtrC (288; condition MM1), and DarA (1030; condition MM1); the sum of these is 3329. The abundance of other cyclic di-AMP-binding proteins such as KimA, KtrA, CpaA, and KhtT, and mRNA abundance of the cyclic di-AMP responsive riboswitch is missing. If we assume a total number of cyclic di-AMP binding sites of 5,000 and a cytoplasmic volume of 1.5 fL, then the total concentration of cyclic di-AMP binding molecules in B. subtilis would be 5.5 μM. This is a little higher than the measured concentration (Table 1), and thus the majority of the intracellular cyclic di-AMP pool may exist in a protein-bound state. Specific stress conditions [change in internal ionic strength (Box 1), pH, etc.] may affect the dissociation constant for cyclic di-AMP binding and change the distribution of the second messenger over target proteins to provide specific regulation.

The secondary response to osmotic upshift

During osmotic upshift, cells will replace part of the accumulated potassium ions for compatible solutes. The term “compatible solute” was first proposed in 1972 for polyols in yeast which rendered them sugar tolerant: “A compatible solute may be loosely defined as one which, at high concentration, allows an enzyme to function effectively” (105). Examples of compatible solutes are glycine betaine, proline, choline, carnitine, ectoine, trehalose, and other neutral and zwitterionic solutes. These compatible solutes do not perturb the balance of intracellular ionic strength, and they provide an efficient long-term “adaptation” mechanism for sustained growth when the external osmolality remains high for prolonged periods of time. Compatible solutes are low molecular mass molecules that are highly soluble (easily hydrated), non-charged, or zwitterionic kosmotropes that can be accumulated to high (molar) concentrations with little to no negative impact on the physiology of the cell, and they have been shown to stabilize native protein structures via preferential exclusion effects (Box 1) (2, 67, 106, 107). These molecules can either be synthesized de novo in the cell or harvested from the extracellular environment, which is energetically less costly and a more rapid method of acquisition. The best studied example of osmotic stress-induced compatible solute synthesis is the synthesis of proline. Under non-stressed conditions, proline levels are around 20 mM in B. subtilis and increase to 500 mM in osmotically stressed cells due to increased expression of three enzymes which synthesize proline from glutamate (67, 108). Similarly, the expression of all of the main compatible solute importers is regulated by osmotic stress (extensively reviewed in reference (109)).

Compatible solute transporters already present in the membrane of bacterial cells are activated at the moment of osmotic upshift as shown for the glycine betaine transporters OpuA from L. lactis and BetP from C. glutamicum (110–113). Systems that enable biosynthesis or transport of compatible solutes are often considered a secondary response to that of K⁺ accumulation due to the temporal delay required for maximal expression of the corresponding genes (114, 115). Glycine betaine is the most ubiquitously utilized compatible solute in bacteria studied to date, and it is accumulated in response to osmotic upshift. Import systems for glycine betaine are ubiquitous among Firmicutes (116–120) and other microorganisms (121). Glycine betaine accumulation increases the maximal specific growth of wild type L. lactis in conditions of osmotic upshift (0.4 M NaCl) and of increased temperature (40°C) (122). In high osmolality (dehydrating) environments, the presence and import of glycine betaine enables effective cell volume restoration and improves growth (2, 54, 123).

Cyclic di-AMP and compatible solute transporters

Since effective potassium and compatible solute transport are vital for bacteria to survive changes in external osmolality, it is of no surprise that mutant strains with altered intracellular cyclic di-AMP levels develop suppressor mutations, not only in potassium transporters but also in compatible solute and amino acid transporters. The connection between intracellular cyclic di-AMP levels and regulation of compatible solute uptake has been investigated both in vivo and in vitro, at a physiological and molecular level.

In low intracellular cyclic di-AMP mutants, loss of function suppressor mutations have been found in either the glycine betaine transporter gene itself or its transcriptional/translational regulatory sequences, including L. lactis (opuA) (74, 76) and (busR/opuR) (51), S. aureus (opuD) (31, 78) and L. monocytogenes (gbuABC) (49). These transporters are responsible for import of glycine betaine, and therefore, loss of function refers to a reduced ability to import glycine betaine; most of these proteins have cyclic di-AMP-binding domains. These strains also acquire loss of function suppressor mutations in other transporters, such as the oligopeptide transporter in L. monocytogenes (oppABCDF) (30, 49) and the amino acid transporter for glutamine/glutamate in S. aureus (alsT) (31, 78). These transporters are responsible for import of peptides and amino acids, and therefore, loss of function refers to a diminished import of these molecules. By contrast, mutants with high intracellular levels cyclic di-AMP acquire loss of function mutations in eep and pptAB, the systems for peptidase activity and oligopeptide export, respectively, in L. lactis (51). This likely leads to intracellular retention of amino acids and peptides, but this hypothesis has not been verified experimentally.

Cyclic di-AMP binds to the CBS domains of compatible solute importers in L. lactis (112) (OpuA), L. monocytogenes (OpuC) (75), S. aureus (OpuC) (124), and B. subtilis (OpuC) (19) and negatively regulates their transport activity (Box 3) (Fig. 1B). The repressor of opuA (named BusR) is regulated by ionic strength in vitro (125) and functions by inhibiting expression of the opuA operon in low osmolality environments (126). Cyclic di-AMP is bound by the C-terminal RCK_C domain of BusR, with a K_D ~10 µM, resulting in inhibition of opuA transcription and subsequently a decreased glycine betaine uptake (Box 3) (Fig. 1B) (24, 51).

BOX 3. Cyclic di-AMP interacting domains (CBS, RCK_C, USP).

Cyclic di-nucleotide second messenger networks are generally comprised of two parts: (i) the biosynthetic and breakdown enzymes that regulate cyclic di-nucleotide synthesis and degradation and (ii) groups of cyclic di-nucleotide-binding proteins (and riboswitches) and their targets. Cyclic di-AMP is bound by a small number of protein domains, most prominently: the cystathionine β-synthase (CBS) domain (InterPro IPR000644) and the regulator of conductance of K⁺ (RCK_C) domain (InterPro IPR006037) (20, 43, 127).

CBS domains are functionally diverse and low in sequence conservation, but the 3D structures are well conserved; they are present in all kingdoms of life (128). CBS domains can exist as subdomains within soluble proteins or membrane proteins or even as stand-alone polypeptides. CBS domains often appear in tandem. These paired domains are able to form dimeric structures that bind cyclic di-AMP at the dimer interface. The CBS domains are binding a range of adenine-containing purines; CBSs have been found that bind ATP, ADP, AMP, nicotinamide adenine di-nucleotide (NAD), S-adenosyl-l-methionine (SAM), cyclic di-AMP, bis(adenosine)-5′-tetraphosphate (AP4A), 5′-deoxy-5′-methylthioadenosine (MTA), mycophenolic adenine di-nucleotide (MYD), or xanthosine-5′-monophosphate (XMP) (129). In addition, divalent metal ions Mg²⁺ and Mn²⁺, inorganic phosphate, and single-/double-stranded DNA have been found as ligands of CBS domains (128). Ligand binding results in conformational changes that can occur within the CBS domain itself and additionally within more distal parts of the protein. In other cases, it has been proposed that functional changes are conferred by alteration of the protein’s electrostatic properties, such as when a divalent metal ion is bound.

The RCK_C domain is found within many bacterial potassium transport proteins, where it is involved in the ion gating mechanism (84, 130–132). Like the CBS domain, multiple RCK_C domains can be found within a single polypeptide sequence where, upon effector binding, they transmit conformational changes (20). In general, if the protein is involved in potassium import, the binding of cyclic di-AMP downregulates the activity (91). By contrast, if the protein is involved in potassium export, cyclic di-AMP upregulates the activity (102).

In addition to the aforementioned domains, the universal stress protein (USP) domain of KdpD from S. aureus also binds cyclic di-AMP (98). The USP domain (IPR006016) is found in bacterial, archaeal, fungal, insect, and plant stress-responsive proteins (133). The exact function of USP domains is currently not known (134), although activation of the USP domain-containing stress proteins is facilitated by a large number of different stimuli.

High intracellular cyclic di-AMP levels also inhibit the uptake of glutamate via GlnPQ, but the mechanism of inhibition is not clear because the transporter does not have a cyclic di-AMP-binding domain. Hence, the effect of cyclic di-AMP is likely indirect and a consequence of, for example, altered cyclic di-AMP-dependent transport of potassium. A reduced intracellular ionic strength would result in a reduction of glutamate and glutamine import (74).

Cyclic di-AMP independent-compatible solute transport

Four different types of osmotic stress-related compatible solute transporters exist in bacteria: (i) Na⁺-solute-symporters (SSS), like OpuE from B. subtilis (135); (ii) Na⁺-coupled symporters from the betaine-choline-carnitine transporter (BCCT) family, like OpuD from B. subtilis, BetP from C. glutamicum or BetT from E. coli (113, 136, 137); (iii) H⁺-coupled symporters of the major facilitator superfamily (MFS), like ProP from E. coli (138); and (iv) type I ATP-binding cassette (ABC) transporters, like OpuA from B. subtilis and L. lactis, OpuB/C from B. subtilis and S. aureus, ProU from E. coli or GbuA from L. monocytogenes.

Cyclic di-AMP has only been demonstrated to bind members of the OpuA-like ABC-importer group (112). A functional ABC-importer is dimeric and consists of a minimum of three different domains: (i) an intracellular nucleotide-binding domain (NBD) involved in ATP hydrolysis to fuel transport; (ii) a multi-span transmembrane domain (TMD) that translocates the substrate; and (iii) an extracellular substrate-binding domain (SBD) that captures the substrate and delivers it to the TMD (139). Additional accessory domains can be linked to the NBD or TMD, such as the cyclic di-AMP-binding CBS domain (Box 3). Compatible solute ABC-importers can be subdivided into three classes based on sequence conservation and genetic structure. The first class (referred to as OpuC) possesses a heterodimeric TMD. By contrast, the second class (referred to as OpuA) is fully homodimeric. The third class (referred to as YehZYXW) is also heterodimeric but differs from OpuC in sequence conservation. Furthermore, the only identified substrate for this class is the antioxidant ergothioneine (140). For all three classes, it holds that some studied members do not have their primary function in osmoregulation but rather import substrates like choline, proline, and glycine betaine for metabolic purposes. Interestingly, only proteins whose activity or transcription is osmotic stress-responsive have two CBS domains fused to the N-terminus of the NBD (Table S1). Consequently, the CBS domains have been suggested to have a regulatory role in osmotic stress-related transport. Mutational studies in proteins from the OpuA and OpuC classes suggest that the CBS domains are involved in sensing ionic strength, thereby regulating transport activity (141–143). However, recent structures of OpuA, together with biochemical experiments, indicate that the osmosensing takes place in the NBD itself and not in the connected CBS domains (112). Note that E. coli does not contain any cyclic di-AMP metabolizing enzymes but does have CBS domains fused to the OpuA homolog ProU. It is unknown if another nucleotide-based second messenger fulfills the function of cyclic di-AMP in this and related organisms.

Currently, the activity of the Na⁺- and H⁺-coupled compatible solute symporters is not known to be regulated by cyclic di-AMP (Fig. 1B). Although proteins like BetP from C. glutamicum are responsive to cytoplasmic ionic strength or to K⁺ levels, which are themselves directly regulated by cyclic di-AMP (113, 144, 145). Secondary transporters rely on the electrochemical potential of the co-solute (the proton motive force for H⁺-coupled transporters, and the sodium motive force for Na⁺-coupled transporters) to transport their substrate, which may lead to lower levels of solute accumulation than via ATP-coupled systems. As described in Box 1, the proton motive force (Δp) of bacteria is typically less than −180 mV (64), and if one proton is cotransported with a neutral solute (e.g., ProP in E. coli) then at thermodynamic equilibrium the solute would be accumulated 1,000-fold. In comparison, ATP-dependent transporters rely on the phosphorylation potential of ATP (ΔG_P′) to accumulate compounds against their concentration gradient, which in a growing cell can reach −500 mV. This means that compounds can be more than 10⁸-fold when using ATP as the energy source for transport (assuming a stoichiometry of 1 ATP/solute and thermodynamic equilibrium; note that OpuA uses 2 ATP per glycine betaine and the accumulation power is accordingly higher (146, 147)). Transporters hardly ever reach equilibrium in the cell, but these examples illustrate that over-accumulation and thus cell lysis is more likely to occur with ATP-driven uptake systems than with proton or sodium motive force-driven transporters. Hence, an additional level of regulation of ATP-dependent transporters via cyclic di-AMP or other types of second messengers may be needed, whereas the accumulation via secondary transporters may not readily reach the lytic threshold of the cell.

Section summary

We demonstrate that cyclic di-AMP is a master regulator of bacterial cell volume (Fig. 1). It achieves this level of regulation by controlling potassium, compatible solute, and amino acid accumulation along multiple axes: post-translationally via direct binding to transporters and affecting their activity, and transcriptionally via their regulatory components to alter expression. These cyclic di-AMP-dependent effects should ultimately determine the cell’s turgor and provide a mechanism by which acceptable pressures and physicochemical homeostasis (ionic strength, pH, macromolecular crowding) can be maintained in an external environment that fluctuates in osmolality.

Cyclic di-AMP is essential under standard laboratory growth conditions (16, 22–29). However, if the growth media are osmotically stabilized (adjustment of the ionic and non-ionic osmolyte content to reduce the difference in the intracellular and extracellular osmotic potential), cyclic di-AMP is no longer essential (18, 30, 31, 148).

When cyclic di-AMP levels cannot be regulated effectively, cells are no longer viable and suppressor mutations are acquired. These mutations may directly affect the enzymes responsible for the synthesis and degradation of cyclic di-AMP, or alter potassium, compatible solute, and amino acid transport activity. Ultimately, these systems all contribute to the more universal and essential process of cell volume regulation. A process in which cyclic di-AMP plays a critical role.

Transporters that are mainly involved in nutrient uptake and secondary compatible solute transport are not directly regulated in their activity by cyclic di-AMP (Fig. 1B). Probably, there is no direct need to regulate secondary solute transporters, because they do not accumulate solutes to levels that are potentially lethal for the cell.

CELL WALL HOMEOSTASIS

Many microorganisms that have been studied in relation to cyclic di-AMP homeostasis are Gram-positive bacteria. The cell envelope of Gram-positive bacteria consists of the plasma membrane, surrounded by an outer wall. The 20–80 nm thick cell wall provides a robust, physical structure that allows Gram-positive bacteria to maintain high turgor pressures up to ~20 atm (55, 149). Their cell wall is composed of a cross-linked peptidoglycan layer penetrated by lipoteichoic acids and wall teichoic acids. Lipoteichoic acids are anchored to the plasma membrane, and extend through the peptidoglycan layers, whereas wall teichoic acids are attached directly to the peptidoglycan. For more information about the bacterial cell envelope, the following papers are recommended (150, 151).

It is clear that cyclic di-AMP controls cell volume, and, as a result, intracellular osmotic pressure and thus turgor. Hence, it makes sense if there is communication between the systems that regulate cell volume and cell wall synthesis (14). This system would ensure that the synthesis of the cell wall and the internal osmotic pressure are adjusted in concert to avoid cell lysis and to drive other processes like cell division. As an example, if the osmolality of the media is decreased, then water flows into the cell and the turgor pressure increases. A signal to strengthen the cell wall structure to withstand the elevated pressure would prevent cell lysis. It should be noted here that cell wall remodeling is too slow to absorb large osmotic downshifts, but smaller shifts can be accommodated because a lipid membrane can stretch up to ~5% area before the lysis tension is reached (152), and the peptidoglycan structure also absorbs hydrostatic pressure changes (153). Furthermore, non-selective mechanosensitive channels allow the rapid exit of osmolytes, followed by the release of water, to relieve the pressure increases on the sub-second timescale (Fig. 1B). Cyclic di-AMP-dependent inhibition of potassium, amino acid, and compatible solute uptake is a perfect means to control osmolyte fluxes on the (sub)minute timescale and to remodel the cell wall. Below we summarize the links between cyclic di-AMP and cell wall metabolism, but we also refer to another review (14).

Cell wall metabolism, antibiotic resistance, and altered cyclic di-AMP levels

The molecular mechanism(s) directly linking cell wall synthesis and cyclic di-AMP are mostly indirect to date. Evidence for a connection between the two has been obtained from studies that utilize (i) mutant strains with high/low cyclic di-AMP levels, (ii) strains treated with cell wall damaging (β-lactam) antibiotics, (iii) mutant strains with altered cell wall composition, or (iv) a combination of the aforementioned factors.

β-lactam antibiotics are a class of antibiotics characterized by their core ring structure. They function by hindering the maturation of the peptidoglycan layer in Gram-positive bacteria. The structure of these antibiotics closely resembles that of the D-alanine-D-alanine terminus of unconnected peptidoglycan crosslinks. Penicillin-binding proteins catalyze the transpeptidation reaction that connects these crosslinks. When these proteins instead bind the antibiotics, their activity is inhibited and peptidoglycan maturation is not completed. b-lactam antibiotics have a broad range of specificities and affinities for penicillin-binding proteins and their presence ultimately weakens the cell wall; thus, the physical and the mechanical defenses of the cell are compromised.

Mutant strains of L. lactis, B. subtilis, and L. monocytogenes with a reduced level of cyclic di-AMP have an increased uptake of potassium and compatible solutes. As a result, the internal osmotic pressure is expected to be higher. Exposure of these cells to cefuroxime (β-lactam antibiotic) causes the mutant cells to lyse more easily (Fig. 2A) (28, 29, 49, 74, 154), which is known as the cefuroxime-sensitive phenotype. Cells remain viable in the presence of cefuroxime when they develop suppressor mutations (74). Loss of function suppressor mutations have been found in genes for K⁺ uptake (kupB), glutamine and glutamate uptake (glnPQ), and the promoter sequence of the glycine betaine transporter (opuA). These mutations likely reduce the uptake of potassium and compatible solutes, in turn, reducing the turgor pressure. In agreement, strains that harbor these suppressor mutations become less sensitive to cefuroxime and acquire a salt-hypersensitive phenotype (74).

Fig 2.

The effects of osmotic downshift and cell wall damaging antibiotics on cyclic di-AMP mutant strains. (A) Schematic representation of the bacterial response to the presence of cell wall damaging β-lactam antibiotics. Gray circles = cell wall; dashed gray circles = damaged cell wall; blue circles = cell membrane; red squares = cyclic di-AMP; open circles = osmolytes; gray arrows = turgor pressure; dashed arrows = movement of water. The increase/decrease in turgor pressure is inferred from lysis phenotype experiments and the fact that osmolyte transporters are regulated by cyclic di-AMP. (B) Overview of the bacterial components that relate cyclic di-AMP to cell wall metabolism. This includes cyclic di-AMP synthesis, efflux and extracellular breakdown systems, and cell wall synthesis machinery. The term yda0 refers to the cyclic di-amp responsive riboswitch, which in some cases is located before cell wall remodeling enzymes.

Mutants with high levels of intracellular cyclic di-AMP have been constructed in a range of Firmicutes (via deletion/inactivation of gdpP), that is, in B. subtilis (22), L. lactis (74), L. monocytogenes (29), S. aureus (16, 38, 155), S. pyogenes (79), and S. pneumoniae (156). Several of these mutant strains, for example, B. subtilis (28), L. monocytogenes (154), and S. venezuelae (157), display different levels of growth inhibition and other morphological defects (16, 22, 28). In contrast to the low cyclic di-AMP mutants, all mutants with elevated cyclic di-AMP levels have an increased resistance to β-lactam antibiotics (Fig. 2A). Inactivation of gdpP also occurs in β-lactam-resistant clinical isolates of S. pneumoniae (156), presumably to increase the intracellular cyclic di-AMP level and reduce the internal solute concentration, and therefore, reduce turgor. In B. subtilis, Tn7 transposon mutagenesis generated suppressor mutations in a ΔsigM strain overcomes the cefuroxime-sensitive phenotype of the parental strain (22). SigM is one of the main sigma factors involved in the response to cell envelope stress. Suppressor mutations arose in multiple genes, including gdpP (cyclic di-AMP phosphodiesterase), tagA (wall teichoic acid biosynthesis), spo0A (sporulation initiation regulator), and lytE (autolysin). Here, a defect in the cell wall stress response system was compensated for by modulation of intracellular cyclic di-AMP levels (in this case, by inactivating gdpP). This reduced the cefuroxime-sensitive phenotype. When a functional copy of the gdpP gene was reintroduced into this gdpP::Tn7 strain, the cefuroxime-sensitive phenotype returned. These results show that elevated intracellular levels of cyclic di-AMP correlate with increased resistance to a range of β-lactam antibiotics (cefuroxime, aztreonam, cefixime, and moenomycin have been tested). In keeping with this, deletion of the cdaA gene (one of three cyclases in B. subtilis) led to a small increase in sensitivity to cefuroxime treatment (22). While it has not been confirmed experimentally, spo0A may be indirectly controlled by cyclic di-AMP, a suggestion that is supported by the fact that the ΔsigM strain is rescued by the mutation of spo0A (158). More hints supporting this idea are provided in the section on DNA damage repair and sporulation.

In L. lactis, inactivation/deletion of gdpP elevates intracellular levels of cyclic di-AMP and suppressor mutations arise in glmM. One such point mutation (GlmM_I154F) was a gain of function mutation that increased the strength of interaction of GlmM for CdaA (bacterial two-hybrid screen) and led to increased inhibition of CdaA (47). The same strain had threefold higher levels of UDP-NAG, was more susceptible to autolysis (47) and was more resistant to the β-lactam antibiotic penicillin G (76).

In S. aureus, a lipoteichoic acid synthesis deficient strain is not viable when cultured in rich media (16). Lipoteichoic acid is a major component of the cell wall that regulates the autolytic cell wall enzyme (muramidase) activity. When the S. aureus strain was screened for suppressor mutations that allowed it to overcome the growth deficiency, loss of function suppressor mutations occurred in several enzymes, including gdpP (phosphodiesterase) and ktrA (potassium uptake). The gdpP mutant strain had a reduced cell size (~13%) and a 15-fold increase in cyclic di-AMP level. This again suggests the elevation of cyclic di-AMP levels when the cell wall is compromised, which reduces internal volume and likely turgor pressure by decreasing the uptake of K⁺ and compatible solutes. Indeed, increase in cyclic di-AMP level correlates with an increased resistance to β-lactam antibiotics (8-fold for oxacillin and lysostaphin, 32-fold for penicillin G) and an increase in peptidoglycan crosslinking (Fig. 2A) (16). Plating of the gdpP or ktrA loss of function strains on chemically defined media containing 0.75 M NaCl resulted in a 2–3 log reduction in the number of colonies formed, representing a salt hypersensitive phenotype (17). The phenotype could be reversed by complementation with a functional copy of gdpP.

Cyclic di-AMP synthesis and cell wall metabolism

CdaA is the most ubiquitous and widespread class of cyclase enzymes (Fig. 2B) (159). It is found in a vast number of bacterial phyla and some archaeal species (42, 160). The cdaA gene is located in a highly conserved operon (42). The operon most commonly contains the genes for the YbbR domain-containing protein CdaR and the essential cell wall precursor synthesis enzyme GlmM (42). In some species, cdaA and cdaR are fused to form a single gene (cdaAR or cdaC). In general, the operon starts with a single promoter sequence and ends with a single terminator sequence. However, while cdaA and cdaR share a single regulatory ribosome-binding sequence (RBS), the third gene in the operon glmM often has its own RBS. Transcription of the cdaA-operon is controlled by the housekeeping sigma factor A in B. subtilis (22, 28).

CdaR

CdaR consists of an N-terminal signal peptide and, depending on the organism, that can be followed by a single, double, triple (L. lactis and S. aureus), or quadruple (L. monocytogenes and B. subtilis) set of tandem YbbR domains (Fig. 2B). YbbR domains are particularly prevalent in the Firmicute phylum. The function of the domain is currently not known (161). Each individual YbbR domain forms a flat structure of eight β-sheets (161). The precise function of the YbbR domains remains elusive but the fact that the domain is repeated several times in the same protein could indicate a role as a scaffold for protein association (161). Indeed, YbbR monomers of CdaR interact with each other to form multimeric complexes. It is tempting to think that the repeated YbbR domains are used as low-affinity binders of the peptidoglycan layer and thereby sense cell wall integrity or indirectly cell turgor or cell volume changes. It has been suggested that CdaR may play a role as a sensory input for CdaA (28, 42, 154, 162). However, experimental evidence for this hypothesis is not available.

In B. subtilis and L. monocytogenes, CdaA and CdaR have been shown to interact in vivo both by protein co-purification and by bacterial two-hybrid screen (28, 154, 162). It has been suggested that the interaction in L. monocytogenes involves the transmembrane helices of CdaA and a putative membrane-embedded segment of CdaR (154). However, caution should be taken at this point because the transmembrane domain is only 15 hydrophobic residues long and the SignalP-6 program predicts a signal peptide instead of a transmembrane segment in CdaR, which could be cleaved off at position 28–30 in L. lactis, S. aureus, and L. monocytogenes; a signal peptide is not predicted for CdaR in B. subtilis and S. suis (163). Since the hydrophobic region in CdaR is only 15 amino acids in length and is preceded by at least two cationic residues, it is more likely a signal sequence than a transmembrane anchor. If this is true, it remains interesting to see how CdaR stays at the cell surface. Possibly through strong interaction with membrane proteins, the cell wall, or cell wall-anchored proteins.

Despite the apparent physical interaction between the CdaA and CdaR, a cdaR-null strain of L. monocytogenes showed no change in the expression of CdaA and was unaffected. In addition, centrifugal separation of crude membrane extracts showed that the presence/absence of CdaR had no effect on the expression or membrane localization of CdaA (154). Two mutant strains have been constructed from the cdaR-null L. monocytogenes strain (154). In one strain, the cdaA gene was placed under the control of an IPTG inducible promoter. In the other strain, the same was done for the cdaR gene. In the respective strains, either CdaA or CdaR was depleted by removing the IPTG inducer molecule from the culture media. When compared to the wild type, the cdaA-depleted strain contained ~60% less and the cdaA IPTG-induced strain contained ~120% more cyclic di-AMP. In comparison, the parental cdaR-null and cdaR-depleted strains both contained elevated cyclic di-AMP levels, both had ~135% of the wild type. These results suggest that the presence of CdaR negatively regulates CdaA activity in L. monocytogenes (154). Further to this point, when CdaA and CdaR from S. aureus were recombinantly co-expressed in E. coli, CdaR decreased the cyclase activity of CdaA (47). Importantly, and in contrast to the findings in L. monocytogenes and S. aureus, when CdaA and CdaR from B. subtilis were recombinantly co-expressed in E. coli, CdaR was observed to enhance cyclase activity of CdaA; a >20-fold increase in intracellular cyclic di-AMP was measured by LC-MS/MS (28).

As reported in reference (47), cdaR in L. lactis laboratory strains MG1363 and IL1403 contains a premature stop codon in or just after the first YbbR domain, likely resulting in the expression of a truncated polypeptide. The data indicate that a careful investigation of the genome sequence of each strain should be carried out to ensure that data are interpreted accurately.

At present, the role of CdaR is elusive. It is also not clear why opposite effects of CdaR on CdaA have been found. The differences may arise simply from differences in strains and or test conditions. It is possible that CdaR functions under specific growth conditions or in response to specific stimuli/stresses to “fine-tune” CdaA activity.

GlmM

GlmM is a cytoplasmic, soluble protein required for the synthesis of the essential cell wall peptidoglycan intermediate glucosamine-6-phosphate (G6P) from glucosamine-1-phosphate (G1P) (Fig. 2B). Initial indications of a connection between GlmM and cyclic di-AMP were made in a L. lactis strain with an inactive gdpP gene (47). The strain contained an elevated intracellular level of cyclic di-AMP because of the loss of the GdpP phosphodiesterase activity. When cultured in high salt conditions, the strain acquired a number of suppressor mutations, one of which occurred in the open reading frame of glmM and was associated with a reduction in intracellular cyclic di-AMP. Subsequently, a bacterial two-hybrid β-galactosidase screen revealed that GlmM binds to and inhibits CdaA activity (47). The binding and inhibition data have been confirmed for homologs from L. monocytogenes, B. subtilis, and S. aureus. Recombinant expression of the soluble cyclic di-AMP cyclase domain of CdaA (CdaA-DAC) from either L. lactis or S. aureus in combination with the corresponding glmM gene, resulted in a significant reduction in intracellular cyclic di-AMP level (47, 164).

In vitro activity assays with CdaA-DAC of S. aureus either in the presence or absence of GlmM confirmed that GlmM is a potent inhibitor of cyclase activity (Fig. 2B) (164). Interestingly, when the same assay was performed with GlmM from E. coli, no inhibition of cyclase activity was observed. A lack of inhibition by E. coli GlmM was also found with CdaA-DAC from L. monocytogenes (162). Formation of the DAC domain of S. aureus CdaA with S. aureus GlmM has been observed when purified proteins were mixed and assessed via SEC-MALLS (size-exclusion chromatography coupled to multi-angle laser light scattering) and pull-down assays (164). Small-angle X-ray scattering (SAXS) analysis of the structure showed that the CdaA-DAC-GlmM complex is formed from the association of a GlmM homodimer and a CdaA-DAC homodimer, which prevents the formation of higher-order CdaA-DAC oligomers that are required for cyclic di-AMP synthesis (164). The CdaA-DAC-GlmM interaction has also been seen by ion-mobility mass spectrometry analysis.

An interaction between CdaA-DAC and GlmM from B. subtilis was first detected using a bacterial two-hybrid screen (165). Purified CdaA-DAC and GlmM from B. subtilis form a complex with a K_D ~14 µM of CdaA for GlmM (166). SAXS analysis showed that the head-to-head conformation between two diadenylate cyclase active sites, required for cyclase activity, is prevented by GlmM through steric hindrance.

The CdaA-DAC-GlmM interaction was confirmed in L. monocytogenes via bacterial two-hybrid screen and co-purification (162). SEC-MALLS analysis corroborated the previous findings and isothermal titration calorimetry was used to determine a binding affinity of ~1 µM of GlmM for CdaA-DAC. In vitro activity assays performed in the presence of ATP and Mn²⁺ showed that GlmM inhibits DAC domain activity in a concentration-dependent manner. Furthermore, GlmM-dependent inhibition was enhanced in vivo under high salt conditions (0.5 M NaCl). The high affinity of GlmM for CdaA raises the question, how is the strength of the inhibition regulated? Is the binding affinity ionic strength-dependent or are other signals linking cyclic di-AMP synthesis to cell wall remodeling activity? GlmM can be phosphorylated at a conserved serine residue outside the CdaA-GlmM binding interface, which is necessary for its enzymatic activity (167). Interestingly, removal of the Ser phosphorylation site by an alanine substitution in GlmM from Streptococcus suis leads to a reduction of virulence, biofilm formation, osmotic stress tolerance, and an increase in cell volume (168). Although these phenotypes could be a result of a reduced GlmM activity or expression, they also match with those of increased cyclic di-AMP levels. It could well be that the phosphorylation of GlmM is the missing link that is needed to explain how GlmM can conditionally regulate cyclic di-AMP synthesis.

Formation of the CdaA-GlmM complex does not affect the activity of GlmM (164). Furthermore, GlmM activity is not sensitive to cyclic di-AMP (165). However, when cells are exposed to osmotically challenging environments, they experience dramatic changes in their internal pressure. To prevent cell lysis, both the cell volume and cell wall synthesis must be carefully adjusted. It appears that the control mechanisms of these two processes operate in a unidirectional manner, meaning that cell wall-related enzyme GlmM controls cyclic di-AMP synthesis enzyme CdaA, and not vice versa.

However, there is almost no evidence of a feedback mechanism connecting cyclic di-AMP levels to the cell wall synthesis machinery except for the cyclic di-AMP responsive riboswitch that inhibits the expression of a range of enzymes that are involved in cell wall homeostasis (Fig. 2B) (21, 169–172). In fact, in Actinobacteria, almost the complete regulon of the cyclic di-AMP responsive riboswitch is connected to cell wall metabolism (21, 171). Maybe genetic regulation of cell wall synthesis machinery by cyclic di-AMP is sufficient to regulate cell wall maintenance on a minute to hour timescale. However, cell wall metabolism may also be regulated on the protein level, which could be a direction for future research.

Efflux of cyclic di-AMP

Many of the wall remodeling enzymes are located on the outer surface of the cell. Thus, if cyclic di-AMP regulates cell wall metabolism on the protein level, it could be necessary to export cyclic di-AMP. There is some evidence that intracellular cyclic di-AMP can be secreted from the bacterial cytoplasm (Fig. 2B). This idea has been put forward for L. monocytogenes where efflux of cyclic di-AMP would be carried out by multi-drug resistance transporters (MDRs): MdrA (LMO0519), MdrC (LMO2818), MdrL (LMO1409), MdrT (LMO2588), and/or MdrM (LMO1617) (29, 173–175). Overexpression of MdrT increases the level of extracellular cyclic di-AMP, which, in turn, stimulates a stronger immune response from infected host cells and vice versa (173, 174, 176). In B. subtilis, two permeases (YcnB and YhcA) share >50% protein identity with MdrT (158). The ability of cells to attach to plants and to form normal biofilms is affected when the ycnB and yhcA genes are both deleted, indicating genetic redundancy (158). There is significantly less cyclic di-AMP in the medium in the double gene knockout strain but the internal concentration of the nucleotide remains stable. In L. lactis, an EmrB family MDR transporter (llmg1210), and the homolog of MdrT, has been identified as a putative cyclic di-AMP efflux system. Co-expression of llmg1210 and llmg1211 eliminates the negative effects of knocking out all cyclic di-AMP-related phosphodiesterase activity in L. lactis (51). Furthermore, mutant strains of L. lactis that have a high intracellular cyclic di-AMP level develop gain of function suppressor mutations in llmg1210 and llmg1211 (51), presumably to enhance export and reduce the intracellular cyclic di-AMP concentration. Cyclic di-AMP efflux has also been proposed for Chlamydia trachomatis (177) and M. tuberculosis (77). None of the putative efflux systems have been studied and the transport mechanism(s) are unknown.

Cyclic di-AMP also gets degraded by cell wall-anchored phosphodiesterases like CdnP in S. agalactiae (178) and SntA in S. suis (179) (Fig. 2B). Both proteins consist of disordered domains, a calcineurin-like phosphoesterase domain (Pfam PF00149), and a 5′-nucleotidase domain (Pfam PF02872). The presence of both domains together gives the enzymes the ability to hydrolyze AMP further after cyclic di-AMP cleavage, which yields adenosine plus phosphate (178, 179). However, most AMPs are likely further hydrolyzed by the extracellular ectonucleotidase NudP (178, 180). NudP is also involved in the stimulation of CdnP activity (178). However, the molecular details of the stimulation remain unclear. Having an extracellular PDE helps bacteria to reduce the cyclic di-AMP-induced immune response, increasing bacterial virulence, but also supports the notion that cyclic di-AMP is present extracellularly.

Section summary

High cyclic di-AMP mutant strains are resistant to β-lactam antibiotics and low cyclic di-AMP mutant strains are sensitive to β-lactam antibiotics. When the internal osmotic pressure is high due to excessive accumulation of K⁺ and compatible solutes such as glycine betaine (as a result of low cyclic di-AMP) and cell wall targeting antibiotics are present, the cell’s ability to resist lysis is reduced. When the turgor pressure is low due to reduced intracellular levels of K⁺ and glycine betaine uptake (as a result of high cyclic di-AMP), the cells show increased resistance to cell wall-damaging antibiotics (Fig. 2A).

These phenotypes are not evidence of a direct connection between the processes of cell volume regulation and cell wall homeostasis. Instead, they demonstrate that ineffective cell volume regulation indirectly increases the cell’s dependence on the cell wall to resist changes in internal osmotic pressure.

Cell wall metabolism is connected with cyclic di-AMP synthesis through the GlmM-dependent inhibition of CdaA. CdaA may also be regulated by CdaR, which may monitor cell wall integrity via its YbbR domains but the evidence is weak and some of the data are contradictory. Furthermore, cyclic di-AMP inhibits the expression of cell wall remodeling enzymes. There is no evidence that cyclic di-AMP directly binds to proteins that are involved in cell wall remodeling. Cyclic di-AMP is degraded extracellularly by CdnP and SntA (Fig. 2B), which is consistent with the regulation of enzymes involved in cell wall homeostasis.

DNA DAMAGE REPAIR AND SPORULATION

Prokaryotic cells entering sporulation or cell division need to ensure the integrity of the DNA that is transferred to their progeny. Unrepaired DNA may prevent the successful completion of spore formation and reduce the viability of daughter cells or outgrowing spores. Depending on the type of damage, bacteria employ a variety of pathways to slow down the process of cell division and repair the damaged site (181, 182). In several bacterial species, cyclic di-AMP synthesis and a cyclic di-AMP receptor are connected to the processes of DNA-damage repair and sporulation, next to cell volume regulation and cell wall metabolism. It is not obvious why these processes are connected with each other through cyclic di-AMP. In this section, we discuss the available data that connect cyclic di-AMP to DNA damage repair and sporulation, and we propose that these processes are linked to cell volume regulation and cell wall metabolism via cyclic di-AMP.

DisA-mediated DNA damage repair

A player in slowing down spore development or cell division is the octameric DNA integrity scanning protein A (DisA) (Fig. 3A) (183–185). One monomer consists of a DNA-binding domain and a diadenylate cyclase (DAC) domain, which are connected by a triple helix linker domain (183). DisA is expressed both at the onset of sporulation and during the outgrowth of spores (186, 187). Upon entry into sporulation, DisA forms ATP-dependent mobile foci of at least several octamers that together scan the DNA while producing cyclic di-AMP (183, 188). The cyclic di-AMP levels double during sporulation due to the activity of DisA (37). DNA-recombination intermediates such as heteroduplex Holiday junctions are bound by DisA and result in inhibition of the diadenylate cyclase activity, causing a reduction of the cytoplasmic cyclic di-AMP levels (183), which in itself is enough to slow down the sporulation process. Entry into sporulation in B. subtilis is largely dependent on transcriptional regulation via the master sporulation factor Spo0A, which is not directly regulated by cyclic di-AMP but is reduced in activity when cyclic di-AMP levels are lowered (37, 188, 189). Cyclic di-AMP seemingly acts through an unknown receptor upstream of Spo0A.

Fig 3.

The roles of cyclic di-AMP during cell division and sporulation. (A) Overview of the bacterial cyclic di-AMP synthesis and effector proteins that are associated with cell division and/or sporulation. CdaA and DisA = cyclic di-AMP synthase; RadA = branch migration transferase; RecA = multifunctional protein involved in homologous recombination. (B) Schematic representation of the effect of inhibited cyclic di-AMP synthesis on cell division, that is,. when DisA encounters damaged DNA. FtsZ = Cell division protein involved in septum formation; Gray circles = cell wall; blue circles = cell membrane; red squares = cyclic di-AMP; open circles = osmolytes; gray arrows = turgor pressure; brown arrows = DNA. See the legend of Fig. 2 for turgor pressure predictions.

The disA gene is often present in an operon together with genes clpC and radA (188). The protease ClpC is involved in the early induction of spore development by cleaving an anti-sporulating transcription factor (190). However, the activation of the disA gene itself is likely not dependent on sporulation regulators, but more on general stress response factors (188). This could also explain why disA is also present in non-sporulating bacteria and why DisA is not always essential in the sporulation process (20, 42, 157, 188). The branch migration transferase RadA is directly involved in the promotion of heteroduplex formation (Fig. 3A) (191). It binds to RecA-loaded single-stranded DNA that has already formed the heteroduplex. Subsequently, it can unwind the double-stranded DNA to promote further heteroduplex formation (192, 193). Moreover, it can directly inhibit DisA-dependent cyclic di-AMP synthesis, especially in the presence of ssDNA but not in the presence of dsDNA. This explains why RadA does not colocalize with and inhibit DisA under normal in vivo conditions (191, 194).

There is also a direct link between DisA and RecA (Fig. 3A). During replication, RecA binds single-stranded DNA at the site of a double-stranded DNA break and facilitates the migration of the DNA to the homologous strand. Both in vivo and in vitro, DisA directly interacts with RecA and inhibits RecA-mediated strand exchange in B. subtilis, probably allowing different mechanisms to repair the damaged site (195). In B. subtilis, cyclic di-AMP does not appear to have a direct function in this process. However, in M. smegmatis, cyclic di-AMP directly binds to RecA and inhibits its assembly on ssDNA (196). This binding occurs in the non-conserved C-terminus of RecA, which likely explains why the binding is not observed in other species like B. subtilis. The binding of cyclic di-AMP to RecA plays an important role in the evolution of multidrug resistance in M. smegmatis. Multidrug-resistant strains evolve faster from mutant strains of M. smegmatis with high levels of cyclic di-AMP. High levels of cyclic di-AMP inhibit RecA and make cells dependent on error-prone non-homologous end joining, which induces higher mutation rates, and consequently, promotes the generation of multidrug-resistant strains (197). Interestingly, the same region in E. coli RecA binds cyclic di-GMP (196). E. coli lacks cyclic di-AMP synthesizing and degrading enzymes, but some of the functions of cyclic di-AMP may be taken over by cyclic didi-GMP. Besides the direct binding of cyclic di-AMP to RecA in M. smegmatis, DisA stimulates the expression of the recA gene which may occur prior to the formation of DNA-recombination intermediates (196). When a recombination intermediate is bound by DisA, cyclic di-AMP production stops, which increases a RecA-mediated repair response. This is in contrast with the findings in B. subtilis, which argue for species-specific roles of cyclic di-AMP in DNA damage processing. How both RecA and RadA interact with DisA and work together to regulate the cyclic di-AMP pool and the processing of DNA damage remains largely elusive.

Cyclic di-AMP controls cell wall metabolism and osmolyte transport during cell division

All currently known links between DNA damage repair and cyclic di-AMP are via DisA. DisA proteins are encoded in only some species of the phyla Actinobacteria, Dictyoglomi, Firmicutes, Fusobacteria, and Thermotogae (198). Note that, in disA-containing species, such as B. subtilis and S. venezuela, most cyclic di-AMP targets are involved in ion or compatible solute transport (18, 157). It is not obvious why it is advantageous for species to couple DNA integrity control to osmolyte transport and cell wall metabolism through cyclic di-AMP. To get an idea of why a DNA integrity scanning protein produces cyclic di-AMP to regulate osmolyte transport and cell wall metabolism, it is necessary to give some insight in the process of bacterial cell division.

Bacterial cell division relies on a complex interplay of various molecular machines to properly coordinate DNA segregation, cell division, and cell wall remodeling. The details of those processes fall outside the scope of this review, and we refer to references (199–202). In short, at the onset of cell division or sporulation, a dynamic ring of FtsZ is formed via the membrane-anchoring protein FtsA to coordinate and regulate the cell wall remodeling for constriction (203). During division, it is important that cell wall remodeling at other sites of the cell wall is reduced (204, 205). Cell wall remodeling enzymes at the division site will induce invagination of the cell wall, while constriction of the cell membrane may in part depend on the contraction of the FtsZ ring (206). In vitro, FtsZ rings alone can deform liposomes but full constriction has not been shown (207–209). Gram-positive bacteria have a high turgor pressure that would hinder constriction, but it has been postulated that the turgor is on the cell wall and not on the membrane (206, 207). This hypothesis is based on cryo-electron microscopy images where a small “periplasmic” space between the cell wall and membrane is observed. This space would only exist if it is iso-osmotic with the cytoplasm and requires the cell wall to be a proper barrier to prevent leakage of osmolytes, which is a matter under debate. But if the existence of a periplasmic space in Gram-positive bacteria is true, FtsZ may trigger the onset of ring constriction and deform the membrane, even in cells with a high turgor (207–209). In addition, excess phospholipid biosynthesis during division aids the membrane deformation by pushing the membrane inward at the site of contraction and may help overcome a potential turgor pressure (207).

Altogether, we argue that some bacterial species may use DisA-mediated cyclic di-AMP synthesis to progress into cell division by inhibiting osmolyte import and activating osmolyte export, which would reduce the turgor pressure and facilitate FtsZ-dependent constriction (Fig. 3B). In support of this hypothesis, a hypertonic shift supports FtsZ-dependent cell division in E. coli (210). When encountering DNA damage, DisA delays cell division by modulating specific repair mechanisms such as RecA-mediated strand exchange. Furthermore, cyclic di-AMP levels decline, which prevents the progression of cell division (through an unknown mechanism) via Spo0A and increases osmolyte uptake. This induces an outward force on the membrane that would prevent the onset of FtsZ-dependent constriction (Fig. 3B). Additionally, fluctuating levels of cyclic di-AMP will affect cytoplasmic Mg²⁺ levels through MgtE. Because, in vitro, FtsZ only forms functionally contracting rings in a narrow concentration range of Mg²⁺, the cytoplasmic Mg²⁺ concentration would regulate FtsZ activity (200). However, we are not aware of any in vivo studies that link FtsZ regulation to MgtE-dependent Mg²⁺ efflux.

Cyclic di-AMP synthesis by the sporulation-associated enzyme CdaS

Of the three main diadenylate cyclase protein families, CdaS is exclusively involved in the process of sporulation (Fig. 3A). The timing of cdaS expression in the sporulation process differs per species (211, 212). In contrast to DisA and CdaA, interaction partners have not been found for CdaS. Instead, CdaS has two N-terminal α-helices connected to its DAC domain. The helices interact to form lowly active hexamers, whereas amino acid substitutions in the helices can lead to highly active dimers (Fig. 3A) (28, 211). It is unknown what shifts the equilibrium from hexamers to active dimers in the cell. The precise function of CdaS in sporulation is also unknown. The absence of CdaS leads to the inability to germinate in a fraction of the spores, whereas the portion of spores that germinate behave like wild-type spores (211). The function of cyclic di-AMP in these germinating spores remains unresolved.

As for DNA damage repair, we think that the function of cyclic di-AMP during sporulation may also be linked to the regulation of osmolyte transport and cell wall metabolism. First of all, during spore formation, inhibition of osmolyte transport would reduce the turgor and allow for the formation of small and highly dehydrated spores. Second, high cyclic di-AMP levels are known to result in a thick and strong cell wall (16, 175).

Nonetheless, this does not explain why spores with lower cyclic di-AMP levels have a lower germinating potency but behave like wild-type cells as soon as they germinate. Possibly higher potassium levels in the spores explain this phenotype because spores of B. subtilis base their decision to germinate on the cytoplasmic K⁺ content (213). Kikuchi et al. visualize potassium levels with a specific dye, and they have shown that B. subtilis spores integrate extracellular germinant signals over time through the release of K⁺. Deletion of the K⁺ importer ktrC gene reduced the K⁺ levels in spores and increased the potency to germinate in the first pulses, whereas deletion of the K⁺ channel yugO gene leads to a loss of integration capacity over multiple germinant pulses due to a reduced potassium efflux. Likewise, inhibition of potassium efflux by the chemical compound quinine reduced germination probability. Lower cyclic di-AMP levels are predicted to increase the initial potassium concentration in spores and as a result it will take longer before the germination threshold is reached. It remains for further studies to unravel the molecular details of this process and to see if cyclic di-AMP is directly involved in other parts of the sporulation process. For example, several cyclic di-AMP riboswitches are located in front of spore cortex-lytic enzymes, which suggests that cyclic di-AMP could be even more directly involved in spore germination than is appreciated right now (171).

Section summary

A selection of bacteria uses cyclic di-AMP to regulate their DNA damage response, sporulation, and germination. Here, proteins like RecA bind cyclic di-AMP, which inhibits its assembly on ssDNA. Also, the expression of spore cortex-lytic enzymes is directly regulated by the second messenger. Most importantly, cyclic di-AMP regulates osmolyte transport and indirectly cell wall metabolism, both of which are connected to cell division and spore germination. Increased cyclic di-AMP levels are advantageous for cell division, as they could lower the turgor pressure by inhibition of osmolyte import and activation of potassium export, and thus lower the force that needs to be overcome for a cell to constrict (Fig. 3B). In the process of germination, the intracellular potassium concentration is used to determine the onset of germination, which is dependent on the initial K⁺ pool, K⁺ conductance, and membrane potential. Altered cyclic di-AMP levels will affect the intracellular potassium pool and thus the propensity of a spore to germinate.

CYCLIC DI-AMP AND (p)ppGpp CO-ORDINATE NUCLEOTIDE AND AMINO ACID BIOSYNTHESIS

To further appreciate the role of cyclic di-AMP in cell metabolism, two essential systems must be introduced: (i) the tricarboxylic acid cycle (TCA) and (ii) the stringent response. The two systems are interwoven at multiple junctions by two nucleotide signaling networks, cyclic di-AMP and (p)ppGpp, which allows the cell to mitigate the consequences of extracellular stresses.

The TCA cycle regenerates redox cofactors NADH and FADH₂, synthesizes metabolic energy in the form of ATP or GTP, and produces a number of important precursors for biosynthesis in addition to CO₂. Notably, α-ketoglutarate and oxaloacetate are synthesized by isocitrate dehydrogenase and malate dehydrogenase, but oxaloacetate can also be obtained from pyruvate plus bicarbonate via pyruvate carboxylase in a reaction that requires ATP. α-ketoglutarate is required for the biosynthesis of amino acids such as glutamate, and purine-containing nucleotides. Oxaloacetate is required for the synthesis of aspartate (and other amino acids) and pyrimidine-containing nucleotides. Furthermore, oxaloacetate production by pyruvate carboxylase is necessary for replenishing the citric acid cycle in case intermediates are tapped for biosynthetic purposes or when species have an incomplete TCA cycle. The complete tricarboxylic acid cycle is required for oxidative phosphorylation in respiratory organisms (bacteria, eukaryotic cells) (Fig. 4A). Fermentative organisms have additional ways to generate metabolic energy, including amino acid deamination, and amino acid and dicarboxylic acid decarboxylation and they can regenerate redox cofactors in fermentative metabolism. As an example, a comparison of the TCA cycle enzymes that are present/absent in L. lactis, L. monocytogenes, and B. subtilis is shown in Fig. 4B. These differences are important to understand the species-specific effects of cyclic di-AMP. Bacteria that lack a complete TCA cycle are unable to synthesize a number of important metabolites, including precursors for abundant osmolytes such as glutamate, via this metabolic pathway. For instance, L. lactis and L. monocytogenes are only able to synthesize α-ketoglutarate, via the TCA cycle, when pyruvate carboxylase is active.

Fig 4.

Comparison of the tricarboxylic acid cycle in different Firmicutes. (A) Schematic representation of the complete TCA cycle. (B) Comparison of the TCA cycle of B. subtilis (complete), L. monocytogenes (lacks the enzyme α-ketoglutarate), and L. lactis (lacks the enzyme isocitrate dehydrogenase). Red crosses = gene not present in the genome.

The stringent response is a stress-related mechanism that helps bacteria to adjust their metabolism to changes in the availability of nutrients. The stringent response is activated by amino acid starvation, nutrient starvation (carbon, nitrogen, phosphorous), and exposure to antibiotics. (p)ppGpp is the signaling molecule (alarmone), which orchestrates the stringent response. (p)ppGpp levels affect large-scale changes in protein content by regulating, for example, transcription, translation, DNA replication, and nucleotide synthesis among others. The alarmone (p)ppGpp is synthesized and degraded by the RSH superfamily of enzymes. It is synthesized by the enzyme Rel, which transfers phosphate groups from ATP to either GMP, GDP, or GTP to form (p)ppGpp. Genes that code for (p)ppGpp forming enzymes are found in the genomes of both Gram-positive and Gram-negative bacteria, of which B. subtilis and E. coli are best studied. In the following section, we discuss (p)ppGpp signaling in the context of cyclic di-AMP regulation, which is found in Gram-positive bacteria. For a more complete overview of (p)ppGpp signaling, we recommend the following reviews (214, 215).

Cyclic di-AMP inhibits pyruvate carboxylase and (p)ppGpp synthesis

In L. lactis (216) and L. monocytogenes (49, 217), pyruvate carboxylase can directly bind cyclic di-AMP at an allosteric site, resulting in reduction of carboxylase activity (Fig. 4A). In L. lactis, 100 µM cyclic di-AMP gives ~60% inhibition of pyruvate carboxylase activity (216) with two molecules of cyclic di-AMP bound per pyruvate carboxylase tetramer. The pyruvate carboxylase complex displays a K_D for cyclic di-AMP of ~8 µM and ~10 µM in L. lactis and L. monocytogenes, respectively (217).

In B. subtilis, pyruvate carboxylase does not bind cyclic di-AMP directly and regulation by cyclic di-AMP occurs via the intermediary protein DarB (Fig. 4A) (218). Formation of the DarB-pyruvate carboxylase complex is prevented when DarB binds cyclic di-AMP (K_D of 20–50 nM of DarB for cyclic di-AMP) (19, 218, 219). B. subtilis contains a complete TCA cycle and is therefore not reliant on pyruvate carboxylase activity for the biosynthesis of oxaloacetate. Rel-dependent synthesis of (p)ppGpp is stimulated when Rel and DarB form a complex (K_D of ~650 nM of DarB for Rel (218)). However, when DarB binds cyclic di-AMP, the protein is not available for complex formation with Rel and (p)ppGpp synthesis is not stimulated (Fig. 5) (220, 221). Activation of the stringent response enhances the expression of the pyruvate carboxylase gene pycA (49). This means that cyclic di-AMP controls pyruvate carboxylase at multiple levels, via preventing activation of pyruvate carboxylase activity and by inhibition of pycA transcription. In B. subtilis, cyclic di-AMP regulates Rel activity indirectly via DarB. In the same way, in L. monocytogenes (222) and S. agalactiae (223), cyclic di-AMP regulates Rel activity indirectly via a DarB homolog, CbpB (Fig. 5). The CBS domains of CbpB bind cyclic di-AMP (K_D of <10 µM for cyclic di-AMP) (217), preventing formation of the activating CbpB-Rel complex (222). Transcription of rel is itself not cyclic di-AMP-dependent and the protein does not bind cyclic di-AMP directly (38). Interestingly, cyclic di-AMP-deficient strains of L. monocytogenes develop suppressor mutations in both cbpB and rel genes to remain viable (30, 49), and cyclic di-AMP-deficient strains can be constructed if cbpB and cdaA are deleted in parallel (222). Thus, on some occasions the essentiality of cyclic di-AMP is caused by elevated (p)ppGpp synthesis, which leads to large-scale cellular reprogramming. The exact cause of this essentiality is hard to figure out since (p)ppGpp functions over a broad concentration range, affecting different systems at different concentrations.

Fig 5.

Connections between cyclic di-AMP and (p)ppGpp. Gray circle = cell wall; blue circle = cell membrane; GdpP, PgpH, NrnA = cyclic di-AMP phosphodiesterase; Rel = (p)ppGpp synthase/hydrolase; DarB/CnpB = cyclic di-AMP-binding protein and regulator of Rel.

Incomplete TCA cycles and the effect of cyclic di-AMP

The physiological effect of cyclic di-AMP binding to pyruvate carboxylase varies between species and depends on whether the bacteria have a complete tricarboxylic acid cycle or not (224). Two examples have been studied in detail and are discussed here.

L. monocytogenes lacks α-ketoglutarate dehydrogenase, the enzyme that catalyzes the conversion of a-ketoglutarate to succinyl-CoA. This prevents the synthesis of oxaloacetate through the TCA cycle via the metabolic intermediates succinate, fumarate, and malate. Therefore, in L. monocytogenes, pyruvate carboxylase is required to synthesize oxaloacetate (from precursors pyruvate and acetyl-CoA). Oxaloacetate can then be used directly for the synthesis of aspartate, or via citrate and isocitrate in the TCA cycle for the synthesis of α-ketoglutarate and subsequently glutamate (Fig. 4). Hence, a knock-out of pyruvate carboxylase in L. monocytogenes grown on ¹³C-glucose can incorporate the ¹³C in alanine (e.g., from pyruvate), but not in aspartate or glutamate (225). The 60% inhibition of pyruvate carboxylase activity by cyclic di-AMP has a tremendous impact on the biosynthesis of glutamate and glutamine, whereas de novo alanine biosynthesis is not significantly affected (217). In addition, aspartate biosynthesis remains unaffected despite its dependence on pyruvate carboxylase, indicating that pyruvate carboxylase is not the limiting factor in the pathway toward aspartate (217). Whether through direct binding of cyclic di-AMP to pyruvate carboxylase, or binding of cyclic di-AMP to DarB, pyruvate carboxylase activity is never completely inhibited. This may be essential for intermediates of the tricarboxylic acid cycle to be continually replenished, despite activation of stress-related responses, allowing metabolism and energy flow to continue.

L. lactis lacks isocitrate dehydrogenase (Fig. 4B) (226, 227). While the amino acid auxotrophy varies among L. lactis subspecies (228), none of them has the gene for isocitrate dehydrogenase and they are therefore unable to synthesize glutamate from α-ketoglutarate (229) and depend on glutamate (or glutamine) uptake for their growth (Fig. 4). Therefore, in contrast to the pyruvate carboxylase of L. monocytogenes, cyclic di-AMP regulation of pyruvate carboxylase in L. lactis reduces the levels of aspartate, which is accumulated to high levels in L. lactis.

The inability of L. lactis to biosynthesize glutamate leads to the requirement for glutamate/glutamine supplementation of the growth media (Fig. 4) (230). Acquisition of glutamate/glutamine/asparagine occurs through the activity of ABC transporter GlnPQ (231–233). The activity of GlnPQ determines the growth rate of L. lactis when glutamate is present in the medium, especially at neutral or alkaline pH when the fraction of glutamic acid (the species that is transported) is low (233). Elevation of intracellular cyclic di-AMP levels leads to a reduction of glutamate uptake (74), but the molecular basis for the effect is not understood. The effect of cyclic di-AMP may be indirect and relate to a change in charge compensation in the cytoplasm, for instance as a result of a change in the K⁺ pool. There are hints that an analogous metabolic circuit may be present in L. monocytogenes, where RNA levels of genes for glutamate metabolism and transport, mediated by gadD2 (LMO2362) and gadT2 (LMO2363), respectively, are upregulated in a mutant with high levels of cyclic di-AMP (29). Altogether, despite species-specific differences in the TCA cycle, the inhibition of pyruvate carboxylase generally results in lower levels of prevalent cytoplasmic anions, such as glutamate and aspartate. This puts the inhibition of pyruvate carboxylase and the rerouting of central metabolism also in the context of cell volume regulation: cyclic di-AMP inhibits the import of potassium and compatible solutes, activates the export of potassium, and in concert with this, reduces the levels of the main cytoplasmic anions via pyruvate carboxylase.

The reciprocal interaction between the stringent response and cyclic di-AMP turnover

(p)ppGpp and cyclic di-AMP metabolism are connected through the ability of cyclic di-AMP-specific phosphodiesterases (e.g., GdpP (38, 234, 235), PgpH (236), and NrnA (237)] to bind (p)ppGpp (Fig. 5). Whereupon binding their phosphodiesterase activity is inhibited in vitro (38, 234, 236, 237). Importantly, a reciprocal feedback loop exists between cyclic di-AMP and (p)ppGpp metabolism: cyclic di-AMP hydrolysis is inhibited by (p)ppGpp, and (p)ppGpp synthesis is inhibited by cyclic di-AMP (Fig. 5). Surprisingly, a knock out of either cyclic di-AMP synthesizing and hydrolyzing enzymes increase (p)ppGpp levels (30, 38, 238).

The reciprocal interaction between (p)ppGpp and cyclic di-AMP facilitates bacterial adaptation to conditions of potassium deprivation. The structure and performance of ribosomes depend on the presence of potassium (239). Consequently, it proves advantageous for bacteria to synchronize ribosomal synthesis with potassium availability to prevent the formation of defective ribosomes. In line with this, cyclic di-AMP levels in B. subtilis are low in low potassium (0.1 mM) growth media and high in high potassium (5 mM) growth media. Reduced levels of cyclic di-AMP increase (p)ppGpp concentrations through the interaction of apo-DarB with RelA, thus impeding the generation of fresh ribosomes (221). Concurrently, (p)ppGpp upregulates pycA expression, inducing enhanced glutamate synthesis and as a consequence, production of the positively charged amino acids ornithine and arginine derived from glutamate to mitigate the effects of potassium deficiency (218, 240).

Cyclic di-AMP levels have a significant impact on de novo aspartate biosynthesis and uptake of glutamine/glutamate/aspartate, which are amino acids required for nucleotide synthesis in L. lactis (216). De novo purine biosynthesis is a multi-step process involving a number of enzymes which culminates in the synthesis of inosine 5′-monophosphate, the precursor of either AMP or GMP. The primary step in the pathway to synthesis of inosine 5′-monophosphate is carried out by purF. Loss of PurF activity results in a reduction of purine biosynthesis (ATP and cyclic di-AMP) (241). PurF activity is inhibited by binding (p)ppGpp (K_D of ~1.6 µM of PurF for (p)ppGpp) (242). In S. aureus, the deletion of purF and the associated reduction of cyclic di-AMP results in reduction of cell wall teichoic acid content, biofilm formation, cell lysis, and extracellular DNA (241). These associated phenotypes are reversible if exogenous cyclic di-AMP is added to the medium, suggesting that a mechanism for the import of cyclic di-AMP is present in S. aureus or that cyclic di-AMP binds extracellularly and activates a signaling pathway (241). Furthermore, PurR, a repressor of multiple genes involved in purine biosynthesis is able to bind and is activated by (p)ppGpp (243). In E. coli, (p)ppGpp can directly bind RNA polymerase leading to an increased expression of amino acid biosynthesis genes (244). However, in B. subtilis, (p)ppGpp does not bind RNA polymerase. Instead, (p)ppGpp binds and inhibits GTP biosynthetic enzymes downstream of PurR/F in the de novo purine biosynthesis pathway to support metabolic regulation of GTP during both starvation and non-starvation conditions (245).

In B. subtilis and L. monocytogenes, CodY is activated by GTP and branched-chain amino acids (246–249). Activated CodY binds to a conserved DNA sequence found in the promoter region of its target genes, repressing them when nutrients are abundantly available (250, 251). The codY gene is highly conserved among low G + C, Gram-positive phyla, where the protein functions as a switch for large-scale changes in gene transcription in response to nutrient limitation (252). It regulates the expression of peptide import, central carbon metabolism (citB), nucleotide metabolism (guaB and guaC), branched chain amino acid biosynthesis (ilvBHC), and sporulation (rapA) (253, 254). In L. monocytogenes, the absence of cyclic di-AMP leads to a growth defect. The strain accumulates (p)ppGpp and depletes its GTP pool (30). This, in turn, inactivates CodY, allowing unwanted derepression of CodY-controlled genes.

Section summary

In summary, cyclic di-AMP and (p)ppGpp signaling are intimately connected. They engage in direct and indirect reciprocal feedback mechanisms to control their own intracellular levels (Fig. 5). These second messenger levels function to fine-tune the cell’s metabolism via regulation of (i) pyruvate carboxylase and replenishing of the TCA cycle, (ii) de novo amino acid biosynthesis and uptake of aspartate and glutamate, (iii) purine nucleotide biosynthesis enzymes PurF and PurR, and (iv) the synthesis of ribosomes.

Despite organism-specific differences in the TCA cycle, cyclic di-AMP appears to carry out the same function in the Gram-positive organisms studied so far, albeit with minor mechanistic nuances (Fig. 4). In general, cyclic di-AMP inhibits the synthesis of the main cytoplasmic anions glutamate and aspartate, and enhances repression by CodY of the genes involved in peptide and amino acid import via a reduction of (p)ppGpp levels.

GLYCOGEN TURNOVER

Most knowledge on cyclic di-AMP has been derived from studies in Firmicutes. Only in 2018, the first experimental evidence was published on a protein that interacts with cyclic di-AMP in cyanobacteria (39). Cyanobacteria are photoautotrophic prokaryotes, making them very useful as sustainable cell factories for the production of food supplements, bio-ethanol, and a wide range of other chemicals (255). They notably differ from Firmicutes in that they have an outer membrane, a cell wall lacking teichoic acids, and cytoplasmic thylakoid-like membranes, which are necessary for photosynthesis. As photosynthesis can only happen during the day, cyanobacteria dramatically change their metabolism in response to light changes. During the day, the cells grow and divide, and much energy is used for the synthesis of glycogen, which serves as energy storage during the night and as a sink for reducing the power that is produced during photosynthesis in the form of NADPH (256). At night, many metabolic activities are reduced and cells switch to glycogen consumption to generate the energy necessary to survive and grow.

Cyclic di-AMP levels have a circadian rhythm in Synechococcus elongatus PCC 7942 and Synechocystis sp. PCC 6803. Cyclic di-AMP levels in S. elongatus increase within 15 minutes after the onset of darkness (39), whereas the levels in Synechocystis increase at the beginning of the light phase (90). It could be that the cyclic di-AMP fluctuation is species specific, but it could also be that the apparent fluctuations are caused by differences in experimental setup such as the use of different light intensities during the day. Nonetheless, cyclic di-AMP is essential for surviving dark periods in which cyanobacteria switch to a heterotrophic lifestyle. When the cdaA gene is deleted, both S. elongatus and Synechocystis grow like wild-type cells under constant light exposure, but the mutant grows very poorly and bleach (i.e., chlorosis) during cycles of light and darkness. Furthermore, the viability of the S. elongatus knockout strain declines the longer it is exposed to darkness (39), probably because cells fail to accumulate sufficient amounts of glycogen (90). This is because cyclic di-AMP regulates the glycogen-branching enzyme GlgB through the binding of SbtB (Fig. 6).

Fig 6.

The effect of different nucleotides on the regulatory function of SbtB in cyanobacteria during light and dark periods. Gray circle = cell wall; blue circle = cell membrane; SbtA = bicarbonate transporter; SbtB = PII like signaling, regulator of SbtA and GlgB; GlgB = glycogen branching enzyme.

SbtB is a PII-like signal transduction protein, similar to the cyclic di-AMP-binding protein of unknown function PstA/DarA in L. monocytogenes (217, 257), B. subtilis (258), and S. aureus (17, 259, 260). The protein is located in an operon with the Na⁺/HCO₃^- symporter SbtA and can bind AMP, ADP, ATP, cyclic AMP, and cyclic di-AMP in the same binding site. Furthermore, it contains a disulfide bridge at the C-terminus, which can be selectively reduced by thioredoxin TrxA (261). In the oxidized state, SbtB can hydrolyze ADP and ATP to AMP. The AMP-bound protein binds and inhibits the HCO₃⁻ transporter SbtA (262). Accumulation of high cytoplasmic HCO₃⁻ levels is necessary for efficient CO₂ fixation during photosynthesis. Since the expression of trxA is reduced during darkness, the following model has been proposed (Fig. 6): during the day, the expression of trxA is high, which increases the fraction of reduced SbtB. Reduced SbtB can bind but not hydrolyze ATP and ADP, which prevents binding to SbtA and facilitates HCO₃⁻ transport during the day. At night, TrxA levels go down, more SbtB gets oxidized, and through the hydrolysis of ATP and ADP, the fraction of AMP-bound SbtB increases, which inhibits HCO₃⁻ transport via SbtA. In response to high CO₂ levels, cyclic AMP is synthesized, which prevents SbtA inhibition and indirectly facilitates extra HCO₃⁻ uptake during the day (261, 263). Cyclic di-AMP induces a unique conformation of SbtB, which cannot bind to SbtA but enables binding to the glycogen branching enzyme GlgB (Fig. 6) (90). It is not known how SbtB binding affects the activity of GlgB.

Glycogen can account for 10%–50% of the cell’s dry mass (264, 265). During osmotic stress, cyanobacteria use glycogen precursors to produce the compatible solutes sucrose and glucosylglycerol (266). However, the glycogen metabolism itself also contributes to the osmotic pressure of a cell. Sequestering glucose in a very large and dense polymer is a very efficient way to reduce the osmotic pressure induced by glucose (267). Therefore, it makes sense to regulate glycogen turnover and the degree of branching in the context of cell volume regulation. However, it remains to be uncovered which signals directly regulate cyclic di-AMP synthesis and how this affects glycogen production. It is important to note here that energy storage in glycogen is a widespread phenomenon in bacteria, so it would be worth testing if glycogen metabolism is also regulated by cyclic di-AMP in other bacterial phyla.

Like in the other prokaryotic lineages, cyclic di-AMP binds and regulates the Trk potassium channel, Kdp potassium import system, and MgtE magnesium channel in cyanobacteria (39, 90). Both the Trk protein and MgtE contribute to the osmotic stress response by regulating cell volume in cyanobacteria (70, 268–270). Furthermore, the poor growth of a cdaA deletion in S. elongatus could be alleviated by a mutation in the murQ gene which codes for a peptidoglycan degradation enzyme (39). cdaA is not located in an operon together with cdaR and glmM but with lysA and uppS, which are both involved in peptidoglycan homeostasis (271). Thus, like in other prokaryotic lineages, cyclic di-AMP is also involved in cell volume regulation and cell wall metabolism in cyanobacteria.

Section summary

Cyanobacteria use glycogen as an energy storage and sink for reducing power during the day and as an energy source at the night. Cyclic di-AMP regulates glycogen turnover via SbtB, which interacts with the glycogen branching enzyme GlgB when cyclic di-AMP is bound (Fig. 6). It is unclear if this interaction either stimulates or inhibits GlgB. Glycogen accounts for 10%–50% of the cell’s dry mass and its turnover could affect the internal osmotic pressure a lot. Regulation of its turnover through cyclic di-AMP could be necessary for proper cell volume control.

SUMMARY AND OUTSTANDING QUESTIONS

In a span of only 15 years, cyclic di-AMP’s role in microbial cell physiology has been elucidated in great detail in a wide variety of bacteria and some archaea. Although it seems that cyclic di-AMP regulates multiple independent cellular processes at the same time, cyclic di-AMP is best described as a master regulator of cell volume because it inhibits K⁺, Mg²⁺, and compatible solute influx, and activates K⁺ efflux (Fig. 1B and 7). Notable exceptions are secondary compatible solute importers, which are not regulated by cyclic di-AMP, presumably because they accumulate solutes to levels that rarely become lethal for the cell. Regulation of proteins by cyclic di-AMP occurs both at the level of gene expression and enzyme or transport activity. All cyclic di-AMP receptors are very sensitive to fluctuations of cyclic di-AMP levels because under normal laboratory growth conditions, the number of cyclic di-AMP molecules and binding sites are comparable. Changes in cytoplasmic osmolyte content affect the turgor on the cell wall. Hence, it is of no surprise that there is reciprocal feedback between cyclic di-AMP and cell wall metabolism (Fig. 2). For example, cyclic di-AMP inhibits the expression of cell wall turnover machinery, while GlmM (the enzyme that synthesizes cell wall precursor G6P) inhibits cyclic di-AMP synthesis. Various bacterial species connect cyclic di-AMP synthesis with the DNA damage response during cell division and with sporulation (Fig. 3A). Increased cyclic di-AMP levels during cell division reduce the turgor through reduction of the levels of cytoplasmic osmolytes, which enables division, whereas reduced cyclic di-AMP levels increase the levels of osmolytes, and prevent division by the anticipated increased turgor (Fig. 3B). During sporulation, cyclic di-AMP can be used to fine-tune the cytoplasmic K⁺ pool which is used to decide when to germinate.

Fig 7.

Cyclic di-AMP is a master regulator of cell volume. Arrows indicate the direction of the effect.

Not only osmolyte transport but also osmolyte synthesis has to be regulated for cell volume homeostasis. To achieve this, cyclic di-AMP inhibits the synthesis of glutamate or aspartate by inhibiting pyruvate carboxylase activity while repressing amino acid import via inhibition of (p)ppGpp synthesis (Fig. 4 and 5). Furthermore, cyclic di-AMP regulates glycogen turnover in cyanobacteria in a yet unknown manner, but probably also regulates the cytoplasmic osmotic pressure to achieve effective cell volume homeostasis (Fig. 6).

We conclude that cyclic di-AMP is the master regulator of cell volume, which has consequences for cell wall turnover and turgor pressure tuning, and for cell volume-dependent processes such as cell division and spore germination (Fig. 7). Regulation of these metabolic networks requires a very diverse set of systems that are crucially controlled via cyclic di-AMP.

Outstanding questions

• Which parameters directly regulate the synthesis and degradation of cyclic di-AMP?

• Why are most cyclic di-AMP synthesizing and degrading enzymes membrane-bound?

• What is the free intracellular concentration of cyclic di-AMP and how does the free concentration vary when target proteins are up- or down-regulated?

• To what extent do cyclic di-AMP turnover enzymes form protein complexes with cyclic di-AMP receptors?

• How does the turgor pressure vary with cyclic di-AMP concentration?

• Can cell wall metabolism-associated proteins bind and be regulated by cyclic di-AMP?

• Does extracellular cyclic di-AMP have a signaling role?

• What is the second messenger that regulates cell volume homeostasis and cell wall metabolism in α, β, and γ-proteobacteria?

• Does (p)ppGpp act as a bridge between cyclic di-AMP and cyclic di-GMP signaling in organisms that contain all three signaling molecules?

Biographies

Alexander J. Foster will defend his PhD in biochemistry at the University of Groningen and is a post-doctoral researcher in the Microbiology Department. His thesis work focused on the role of cyclic di-AMP in bacterial cell volume regulation and the mechanisms that affect the synthesis and breakdown of cyclic di-AMP. He received his bachelor’s degree in biomedical science from Brunel University (UK), and his master’s degree in bioengineering from Imperial College London (UK). Alex is interested in the application of synthetic biology and high-throughput automation to design novel workflows in the service of improving large-scale biotechnological processes.

Marco van den Noort will defend his PhD thesis in biochemistry at the University of Groningen in May 2024 and is currently a post-doctoral researcher in the Chemical Biology group in Groningen. His doctoral research has been centered around the osmoregulatory ABC-importer OpuA, which is regulated by cyclic di-AMP. Marco holds a bachelor’s degree in Molecular Life Sciences from Utrecht University (NL) and a master’s degree in Molecular Biology and Biotechnology from the University of Groningen (NL). His academic interests revolve around osmoregulation and single-molecule studies of proteins. In his current post-doctoral position, he employs nanopores to study their mechanistic properties on a single-molecule level.

Bert Poolman is a professor of biochemistry at the University of Groningen, NL. He is an elected member of the Royal Netherlands Academy of Arts and Sciences, and Knighted in the Order of the Dutch Lion. Poolman has a track record in metabolic energy conservation, membrane transport, and cell volume regulation as well as the development of innovative technologies in membrane biology and bottom-up synthetic biology. He has advanced the field of membrane transport by combining functional and structural studies. Central questions in his group are as follows: (i) What tasks must a living cell minimally perform and how this can be accomplished with a minimal set of components? (ii) How do molecules permeate biological membranes? (iii) How can one control the volume and physicochemistry of the cell? Website: www.membraneenzymology.com; Wikipage: https://en.wikipedia.org/wiki/Bert_Poolman

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mmbr.00181-23.

Table S1. mmbr.00181-23-s0001.docx.

Comparison of ABC-type compatible solute importers for the responsiveness to osmo-stress and the presence of CBS domains.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.