c-di-AMP modulates Listeria monocytogenes central metabolism to regulate growth, antibiotic resistance, and osmoregulation

Summary

Cyclic di-adenosine monophosphate (c-di-AMP) is a conserved nucleotide second messenger critical for bacterial growth and resistance to cell wall-active antibiotics. In Listeria monocytogenes, the sole diadenylate cyclase, DacA, is essential in rich, but not synthetic media and ΔdacA mutants are highly sensitive to the β-lactam antibiotic cefuroxime. In this study, loss of function mutations in the oligopeptide importer (oppABCDF) and glycine betaine importer (gbuABC) allowed ΔdacA mutants to grow in rich medium. Oligopeptides were sufficient to inhibit growth of the ΔdacA mutant and we hypothesized that oligopeptides act as osmolytes, similar to glycine betaine, to disrupt intracellular osmotic pressure. Supplementation with salt stabilized the ΔdacA mutant in rich medium and restored cefuroxime resistance. Additional suppressor mutations in the acetyl-CoA binding site of pyruvate carboxylase (PycA) rescued cefuroxime resistance and resulted in a 100-fold increase in virulence of the ΔdacA mutant. PycA is inhibited by c-di-AMP and these mutations prompted us to examine the role of TCA cycle enzymes. Inactivation of citrate synthase, but not down-stream enzymes suppressed ΔdacA phenotypes. These data suggested that c-di-AMP modulates central metabolism at the pyruvate node to moderate citrate production and indeed, the ΔdacA mutant accumulated 6-times the concentration of citrate present in wild-type bacteria.

Abbreviated Summary

c-di-AMP is an essential nucleotide second messenger in many bacteria. In this report, Whiteley et al. demonstrate that suppressor mutations in the acetyl-CoA binding site of Listeria monocytogenes pyruvate carboxylase restore growth, resistance to cell-wall antibiotics, and virulence of a mutant unable to produce c-di-AMP. These data suggest that c-di-AMP modulates central metabolism at the pyruvate node, which is necessary for optional growth, osmoregulation, and virulence.

Introduction

Bacterial nucleotide second messengers are signaling molecules that act as allosteric regulators of proteins and riboswitches. The best characterized include: cAMP, which is synthesized during glucose starvation, (p)ppGpp which is synthesized in response to amino acid starvation as part of the stringent response, and cyclic di-GMP, which modulates the transition from sessile to planktonic growth (Kalia et al., 2013). More recently, cyclic diadenosine monophosphate (c-di-AMP) and cyclic AMP-GMP have been added to the growing list of cyclic dinucleotide second messengers (Romling, 2008; Davies et al., 2012). However, unlike cAMP and (p)ppGpp, which are synthesized upon stress, c-di-AMP is required for growth in conventional media and the cues that regulate its synthesis and degradation are less well-defined (Corrigan and Gründling, 2013). Production of c-di-AMP from two ATP molecules is catalyzed by diadenylate cyclases (DAC) that share a common protein domain and are distributed in the genomes of virtually all Gram-positive bacteria, some Gram-negative bacteria, and archaea (Witte et al., 2008).

In Firmicutes (e.g. bacilli, streptococci, staphylococci, and listeria), c-di-AMP is often essential (Commichau et al., 2015; Cheng et al., 2015). Decreased c-di-AMP levels result in growth defects, susceptibility to cell wall-targeting antibiotics, dysregulation of potassium import, and alterations in salt tolerance (Corrigan and Gründling, 2013; Zhu et al., 2015; Moscoso et al., 2015). Mutations in c-di-AMP specific phosphodiesterases increase c-di-AMP levels, suppress growth defects of lipotechoic acid biosynthesis mutants, and increase resistance to β-lactam antibiotics such as the β-lactam cephalosporin cefuroxime (Corrigan et al., 2011; Corrigan and Gründling, 2013). These data support a correlation between c-di-AMP levels and resistance to cell wall-targeting antibiotics that is conserved among Firmicutes. In many of these organisms the diadenylate cyclase DacA (also known as CdaA) is encoded within the same operon and interacts with GlmM (Gundlach et al., 2015b; Zhu et al., 2015; Rismondo et al., 2016), an enzyme responsible for the conversion of glucosamine-6-phosphate to glucosamine-1-phosphate, an early step of peptidoglycan biosynthesis (Barreteau et al., 2008). It is unclear if this interaction is important for c-di-AMP associated cell wall defects.

Listeria monocytogenes is a Gram-positive bacterium that has a biphasic lifestyle as a ubiquitous saprophyte and as a facultative intracellular pathogen of humans and animals (Tilney and Portnoy, 1989; Freitag et al., 2009). L. monocytogenes has a single diadenylate cyclase (DacA) and two c-di-AMP-specific phosphodiesterases (PdeA and PgpH) (Woodward et al., 2010; Witte et al., 2013; Huynh et al., 2015). L. monocytogenes mutants with altered expression of either DacA or the phosphodiesterases revealed that both high or low c-di-AMP levels result in slower growth (Witte et al., 2013; Huynh and Woodward, 2016). To dissect the role(s) played by c-di-AMP in L. monocytogenes, Sureka et. al. identified c-di-AMP binding proteins by affinity chromatography (Sureka et al., 2014). Most strikingly, they identified that c-di-AMP is an inhibitor of pyruvate carboxylase (PycA) and dacA depletion showed corresponding metabolic imbalances that resulted in bacteriolysis. In L. monocytogenes, PycA is the primary anabolic source of oxaloacetate, an intermediate of the TCA cycle and amino acid biosynthesis (Schar et al., 2010). However, it is unclear why increased oxaloacetate was toxic. Sureka et al. also identified a homolog of PstA, a protein identified in a systematic search for c-di-AMP binding proteins in Staphylococcus aureus (Corrigan et al., 2013). PstA is a small (11.8 kDa), widely distributed, PII-like protein (Forchhammer, 2008), that has been identified and structurally characterized in Bacillus subtilis, S. aureus, and L. monocytogenes (Müller et al., 2015; Campeotto et al., 2015; Gundlach et al., 2015a; Choi et al., 2015). Other PII-like proteins modulate nitrogen metabolism via protein-protein interactions and are modulated by a cognate small molecule (Ninfa and Jiang, 2005). PstA protein-protein interactions are likely inhibited by c-di-AMP (Choi et al., 2015), however, these interactions have not been identified and PstA function remains poorly characterized.

c-di-AMP also interacts with the cystathionine-β-synthase (CBS) domain-containing proteins of unknown function CbpA and CbpB, and a carnitine importer (OpuCA)(Sureka et al., 2014; Huynh et al., 2016; Schuster et al., 2016). Carnitine is a “compatible solute” similar to glycine betaine, which act as osmoprotectants and are accumulated by cells during adaptation to osmotic stress (Wood, 2015). Osmotic stressors alter the extracellular concentration of solutes, which affect the hydration of the cytosol and cellular turgor pressure (Wood, 2011). Adaptation to osmotic upshift (extracellular increases in osmotic pressure) starts with import/synthesis of K+ and glutamate followed by the accumulation of compatible solutes (Wood, 1999). c-di-AMP inhibits OpuCA and in related organisms inhibits potassium import (Corrigan et al., 2013; Moscoso et al., 2015; Huynh et al., 2016), indicating a role for c-di-AMP in osmoregulation.

Our lab has taken a genetic approach to dissect the role(s) of c-di-AMP in the physiology of L. monocytogenes. Mutants lacking c-di-AMP (ΔdacA) cannot form colonies on conventional rich medium unless they accrue one or more suppressor mutations (Whiteley et al., 2015). For example, mutants unable to synthesize (p)ppGpp grew in rich medium suggesting that elevated (p)ppGpp levels were partially responsible for the growth defect of ΔdacA and that these organisms were experiencing an unknown stress that leads to increased production of (p)ppGpp. c-di-AMP and (p)ppGpp levels appeared to be inversely correlated. Whereas (p)ppGpp is essential for growth in synthetic but not rich media, c-di-AMP was identified as essential for growth in rich but not synthetic media (Whiteley et al., 2015). ΔdacA mutants can be constructed and maintained without the appearance of suppressor mutations if bacteria are cultivated in a synthetic medium. Although ΔdacA mutants grow on synthetic medium they are highly susceptible to cefuroxime. In this study, we characterized additional suppressor mutations that allowed for growth in rich medium and identified mutations in the oligopeptide transporter (oppA-F), the glycine-betaine transporter (gbuA-C), cbpB, pstA, and pycA. Only null mutations in pstA, point mutations in PycA that blocked acetyl-CoA mediated allosteric activation, or addition of salt to media suppressed sensitivity to cefuroxime. These data suggest that ΔdacA mutants are unable to regulate intracellular osmotic pressure, which becomes growth inhibitory when oligopeptides or cefuroxime are present. Further, c-di-AMP-dependent modulation of the TCA cycle altered osmotic pressure and was important for virulence of L. monocytogenes.

Results

c-di-AMP is essential for growth in rich medium and resistance to cefuroxime

L. monocytogenes encodes one di-adenylate cyclase (DacA), and previous research demonstrated that dacA was conditionally essential for growth in rich, but not defined synthetic media. We have extended these observations by constructing ΔdacA mutants in Listeria synthetic medium (LSM, see Table S1 and Experimental Procedures for details) that promoted enhanced growth of both wild-type and ΔdacA bacteria and was easily adapted to solid-agar and liquid culture. To analyze growth in rich medium, ΔdacA mutants were constructed in LSM, grown in LSM-culture overnight, and 10-fold serial dilutions were plated on LSM-agar and the L. monocytogenes conventional rich medium brain heart infusion (BHI)-agar (Fig. 1A). This assay compared the ability of L. monocytogenes mutants to grow in rich vs. defined solid media. Whereas wild-type bacteria formed an equivalent number of colonies on LSM and rich media, the ΔdacA mutant formed over 10,000-fold fewer colonies on rich medium (Fig. 1A and 1B). Growth in rich medium was restored by expressing the native diadenylate cyclase dacA or the distantly related diadenylate cyclase disA from B. subtilis (Fig. 1A and 1B)(Witte et al., 2008). DacA and DisA share a diadenylate cyclase protein domain, but DisA, unlike DacA, is not membrane-localized.

Figure 1. Growth of c-di-AMP-deficient mutants in rich medium and resistance to cefuroxime.

(A and B) Mutants constructed and grown overnight in LSM were serially diluted 10-fold in a 96-well plate with PBS then 5 μL of each dilution was spotted onto either LSM or BHI agar. Images were taken and CFU were enumerated after 48 hrs of incubation at 37 °C. (C) Immunoblot of DacA and P60 (loading control) proteins for the strains indicated, grown to mid-log in LSM at 37°C. Data are representative of three independent experiments. (D) Antibiotic sensitivity measured by disk diffusion of 125 μg of cefuroxime on LSM-agar for the indicated L. monocytogenes strains measured at 48 hrs. (B and D) Data are mean ± standard error of the mean (s.e.m) of at least three independent experiments.

DacA interacts with the peptidoglycan biosynthesis enzyme GlmM (Gundlach et al., 2015b; Zhu et al., 2015; Rismondo et al., 2016), however, the consequences of this interaction are unclear. When grown in LSM, the ΔdacA mutant was extremely sensitive to the cell wall-targeting β-lactam antibiotic cefuroxime (Fig. 1D) and it is unclear if this phenotype might be due to DacA regulation of GlmM. The impact of this interaction was interrogated by constructing a strain of L. monocytogenes that expressed a catalytically inactive allele of dacA, (D171A). The D171A mutation disrupts the catalytic capability of DacA (Rosenberg et al., 2015), but did not affect DacA synthesis (Fig. 1C). The dacA^D171A mutant was equivalent to the ΔdacA mutant for growth in rich medium and resistance to cefuroxime (Fig. 1B and 1D). These data, along with ability of disA to complement ΔdacA, suggest that targets of c-di-AMP, not DacA physical interactions, are responsible for cell wall homeostasis and growth in rich medium.

Suppressor mutations allow ΔdacA to grow in rich medium

All ΔdacA mutants recovered after growth in rich medium harbored suppressor mutations (Whiteley et al., 2015). When overnight cultures were serially diluted onto solid media, 10⁹ cfu mL^-1 were recovered on LSM while 10⁵ cfu ml^-1 were recovered on rich media, suggesting that suppressor mutations arose at a rate of approximately 10^-4 in rich medium (Fig. 1B). Suppressor mutations in relA such as R295S were previously investigated, and are sufficient to ameliorate dacA essentiality, demonstrating that increased (p)ppGpp is growth inhibitory in the absence of c-di-AMP (Whiteley et al., 2015). The remaining loss-of-function suppressor mutations found in multiple strains were identified at 4 loci: the oligo-peptide permease (opp)(Borezee et al., 2000), the glycine betaine importer (gbu)(Ko and Smith, 1999), and the genes of unknown function CbpB and PstA (Fig. 2A-F). These mutations will be investigated here. Loss-of-function mutations in genes encoding the Opp were found in over 94% of ΔdacA suppressor mutants (Whiteley et al., 2015). The Opp is a five-subunit active importer of oligopeptides, consisting of an extracellular solute binding protein (OppA), transmembrane permeases (OppBC), and two ATPases (OppDF)(Fig. 2A and 2B)(Borezee et al., 2000). Mutations in any of the Opp components, with the exception of the ATPases, are sufficient to inactivate Opp import of peptides. The gbuABC operon encodes a glycine betaine importer homologous to opuAABC in B. subtilis, that is important for survival during osmotic shock (Ko and Smith, 1999; Sleator et al., 2003). CbpB and PstA were identified as c-di-AMP-binding proteins and their functions remain poorly characterized (Sureka et al., 2014; Choi et al., 2015).

Figure 2. ΔdacA suppressor mutants.

(A and B) Illustration of oligopeptide permease (Opp) operon and protein subunits. (C and D) Illustration of glycine-betaine importer (Gbu) operon and protein subunits. (E) Operon of c-di-AMP binding protein B (cbpB) which encodes a c-di-AMP binding, CBS domain containing, protein of unknown function. ccpC, catabolite control protein and LysR family transcriptional regulator (Kim et al., 2005); dapH, 2,3,4,5-tetrahydropyridine-2,6-dicarboxylate N-acetyltransferase; dapL, N-acetyl-diaminopimelate deacetylase (similar to dapL in B. subtilis). (F) Operon of PII-like signal transduction protein A (pstA) a c-di-AMP binding, PII-like, protein of unknown function. lmo2694, hypothetical ornithine/lysine/arginine decarboxylase (similar to yaaO in B. subtilis); tmk, thymidylate kinase. (A, C, E, and F) Mutations previously identified (Whiteley et al., 2015) and not drawn to scale. (G) Enumeration of CFU on indicated media for L. monocytogenes strains constructed in LSM. “+” vs “Δ” indicate a mutation in a wild-type or dacA-deficient background. Data are mean ± s.e.m of at least three independent experiments.

ΔdacA suppressor mutants harbored mutations in either a single locus (relA or opp) or multiple loci (gbu, cbpB, and pstA). Deletion mutations for oppB, gbuABC, cbpB, and pstA, along with the previously investigated relA^R295S allele, were constructed in wild-type and ΔdacA backgrounds and characterized for growth in rich medium. Strikingly, each of these null mutations independently suppressed dacA essentiality in rich medium (Fig. 2G). These phenotypes could be fully complemented in ΔgbuABC and ΔpstA, and partially complemented in relA^R295S due to the dominant nature of the mutation (Fig. S1)(Whiteley et al., 2015). We hypothesize that ΔcpbB was only partially complemented due to low expression of the complementation construct. The ΔoppB mutant could not be complemented due to toxicity to E. coli (see Experimental Procedures for details)(Fig. S1).

ΔdacA mutants are unable to grow in the presence of peptides or resist cefuroxime due to defects in osmoregulation

We hypothesized that substrates of the Opp and Gbu transporters were toxic to ΔdacA mutants and first analyzed the role of imported oligopeptides. In other organisms, Opp-imported oligopeptides have been described as either peptide pheromones, peptide fragments of peptidoglycan, or nutritive oligopeptides (Maqbool et al., 2011). We reasoned that the former two types of peptides are derived from L. monocytogenes and would likely still be synthesized in LSM but that ΔdacA is unable to grow in rich medium due to the presence of nutritive oligopeptides. The impact of nutritive oligopeptides was tested by supplementing LSM with a tryptic digest of casein, which prevented the growth of ΔdacA but not ΔdacAΔoppB or ΔdacA complemented with disA (Fig. 3A). Similarly, the ΔdacA mutant was unable to grow in the presence of the synthetic oligopeptides “KLLLLK”, “KAAAAK”, and “AQ”, but grew well when the molar equivalent of alanine and glutamine were provided as free amino acids (Fig. 3B). The growth-restrictive phenotype of synthetic peptides was not observed for wild-type or ΔdacAΔoppB bacteria (Fig. 3B). These data suggested that dipeptides were efficiently imported by the Opp and that increased transporter activity or synthesis may underlie phenotypes of the ΔdacA mutant. The activity of the Opp transporter was measured by disk diffusion of the toxic peptide antibiotic bialaphos (Perego et al., 1991). In this assay, no zone of clearance was observed for the ΔoppB mutant and the ΔdacA mutant was modestly more sensitive (Fig. S2A), revealing that Opp activity may be slightly increased in the ΔdacA mutant, but that the growth-inhibitory effect of peptides in the absence of c-di-AMP is also the product of changes in bacterial physiology. In line with this hypothesis, all identified suppressor mutations enabled ΔdacA mutants to grow in LSM supplemented with oligopeptides (Fig. S2B), but did not considerably alter sensitivity to bialaphos (data not shown).

Figure 3. Growth of ΔdacA in media containing oligopeptides and salt.

(A-D) Enumeration of CFU on indicated media for L. monocytogenes strains constructed in LSM. Dashed line indicates limit of detection (Lod). (B) “KLLLLK”, “KAAAAK”, and “AQ” are sequences of synthetic oligopeptides added, and “A&Q” is the molar equivalent of free amino acids (E) Cefuroxime disk diffusion on LSM-agar supplemented with the indicated concentration of NaCl. All data are mean ± s.e.m of at least three independent experiments.

The predicted substrate of the Gbu transporter is glycine betaine, a compatible osmolyte that allows bacteria to cope with osmotic stress and modulate the water content/turgor pressure of the cell. Mutations in gbuABC are predicted to decrease the internal osmotic pressure of ΔdacA mutants; thus, we hypothesized that dacA is conditionally essential due to large differences in internal and external osmotic pressure that are exacerbated by import of glycine betaine. In support of this hypothesis, addition of NaCl and KCl to rich medium, which increases extracellular osmotic pressure, restored growth of ΔdacA mutants in rich medium (Fig. 3C and 3D). From these data we speculated that a general role for c-di-AMP in coping with osmotic pressure might underlie other ΔdacA phenotypes, such as sensitivity to cefuroxime. This hypothesis is supported by literature demonstrating that peptides and amino acids act as osmoprotectants (Maria-Rosario et al., 1995), and that c-di-AMP directly inhibits carnitine import via OpuCA (Huynh et al., 2016), Intriguingly, supplementation of LSM with NaCl rescued the sensitivity of the ΔdacA mutant to cefuroxime but had no effect on wild-type L. monocytogenes (Fig. 3E). These data suggest that c-di-AMP modulates bacterial physiology to decrease internal osmotic pressure, which is important for growth on peptides and resistance to cell wall-acting antibiotics.

PstA mutations suppress ΔdacA sensitivity to cefuroxime

c-di-AMP appeared to alter internal osmotic pressure through a yet unidentified mechanism. Our analysis next aimed to identify mutations that suppressed ΔdacA cefuroxime sensitivity. Mutations in relA, oppB, gbuABC, cbpB, and pstA suppressed growth of ΔdacA in rich medium (Fig. 2G) but only the ΔpstA mutation suppressed the sensitivity of the ΔdacA mutant to cefuroxime (Fig. 4). This phenotype was complemented by over-expressing pstA containing a C-terminal strep(II)-tag (SII) from a neutral locus, confirming the affinity-tagged fusion protein was biologically active (Fig. 4). PstA appeared to have a central role in ΔdacA phenotypes; however, despite an abundance of structural information for PstA and its homologs, its function is poorly understood (Müller et al., 2015; Campeotto et al., 2015; Gundlach et al., 2015a; Choi et al., 2015). We hypothesized that in the absence of c-di-AMP, PstA-protein interactions were stabilized, which led to an inability of ΔdacA to grow in rich medium and to resist cefuroxime. Accordingly, we performed SPINE affinity purification of PstA from L. monocytogenes and a yeast 2-hybrid to identify PstA-interacting proteins (Herzberg et al., 2007). Unfortunately, we were unable to confirm any protein-protein interactions with high confidence.

Figure 4. ΔdacA suppressor mutant resistance to cefuroxime.

Cefuroxime disk diffusion on LSM-agar of L. monocytogenes strains. Data are mean ± s.e.m of at least three independent experiments.

Additional suppressor mutations that confer resistance to cefuroxime map to the acetyl-CoA binding site of pyruvate carboxylase

To gain further insight into the function of c-di-AMP, we screened for suppressor mutations that phenocopied ΔpstA and suppressed toxicity of rich medium and β-lactam antibiotics. We constructed a ΔdacA mutant merodiploid for pstA (ΔdacA p-pstA, to avoid isolating further pstA mutations), selected cefuroxime-resistant colonies, and screened for growth in rich medium (Fig. 5A). Genome sequencing of suppressor mutants revealed that 15 of 16 strains harbored mutations in pyruvate carboxylase (pycA)(Table 1). PycA converts pyruvate and bicarbonate to oxaloacetate using a biotin cofactor with hydrolysis of ATP. In L. monocytogenes PycA is the primary source of oxaloacetate due to an incomplete TCA cycle and other metabolic insufficiencies (Fig. 5B) (Eylert et al., 2008; Schar et al., 2010).

Figure 5. Suppressor mutations of ΔdacA cefuroxime sensitivity.

(A) Illustration of suppressor analysis. (B) Schematic of central metabolism in L. monocytogenes. Grey arrows with red X indicate enzymes not encoded in the L. monocytogenes genome, bold labels indicate enzyme names, and non-bold labels indicate metabolites. Underlined labels indicate metabolic pathways providing or using precursors/products of the enzymes shown. OAA, oxaloacetate; CIT, citrate; ICI, isocitrate; αKG, 2-oxoketoglutarate; SUC, succinate; CoA, coenzyme A; FUM, fumarate; MAL, malate. (C) PycA suppressor mutations from Table 1 on color-coded protein domains showing the biotin carboxylase (BC), pyruvate carboxylase tetramerization (PT), carboxyltransferase (CT), and biotin carboxyl carrier protein (BCCP) domains. Figure not drawn to scale. (D) Crystal structure of PycA (PDB: 4QSH, (Sureka et al., 2014)) from L. monocytogenes with modeled suppressor mutations on all four monomers. Monomer 1 is colored as in (C) and only mutations on this monomer are labeled. The resolved c-di-AMP (cdA) molecules are shown in red. (E) Crystal structure of PycA homolog from S. aureus (PDB: 3HO8, (Yu et al., 2009)) with modeled suppressor mutations at homologous residues on all four monomers. Monomer 1 is colored as in (C) and only mutations near the acetyl-CoA binding site are labeled on this monomer. The resolved coenzyme A ligands are also shown. (F) Detailed view of (E) modeling interactions between the S. aureus arginine residues homologous to R367 and R1051 with coenzyme A. Dashed red lines hydrogen bonds. (D, E, and F) Visualizations made with PyMol software. (G) Enzymatic activity of recombinant PycA in the absence or presence of the allosteric activator acetyl-CoA. Data are the mean ± s.e.m. of three independent experiments and p values were calculated using a heteroscedastic Student's t-test; ** p < 0.001, ns p > 0.05.

Table 1. Suppressor mutations identified in ΔdacA p-pstA mutants capable of growth on rich media and resistance to cefuroxime.

Strain	Genome Coordinates	Reference	Allelea	Gene Name	10403S Locus	EGD-e locus	Protein name	Uniprot Entry	Locus amino acid change
DP-L6544	1081790	T	A	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Val241Glu
DP-L6545	1082155	G	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Val363Leu
DP-L6546	1082155	G	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Val363Leu
DP-L6547	1082168	G	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Arg367Leu
DP-L6548	1082168	G	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Arg367Leu
DP-L6548	1272594	C	T
DP-L6549	1082408	C	A	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Thr447Lys
DP-L6550	1082408	C	A	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Thr447Lys
DP-L6551	1082939	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Pro624Leu
DP-L6552	1082939	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Pro624Leu
DP-L6553	1082939	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Pro624Leu
DP-L6554	1083536	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Pro823Leu
DP-L6555	910001	G	A	rsbU	LMRG_02316	lmo0892	Sigma-B regulation protein, phosphatase	A0A0H3GEW1_LISM4	Gly79Ser
DP-L6555	1083536	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Pro823Leu
DP-L6556	1083536	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Pro823Leu
DP-L6557	1084219	C	T	pycA	LMRG_00534	lmo1072	pyruvate carboxylase	A0A0H3GJD4_LISM4	Arg1051Cys
DP-L6558	1852607	G	T	stp	LMRG_00968	lmo1821	serine/threonine phosphatase	A0A0H3GDQ3_LISM4	His41Asn

Genome sequencing results of suppressor mutants. Some strains harbored identical mutations and it is unclear if these strains were derived from the same parent suppressor mutant or arose independently of one another.

^aNucleotide variations as compared to the parent strain

PycA is allosterically regulated by a diverse set of metabolites, including inhibition by c-di-AMP (Sureka et al., 2014) and activation by acetyl-CoA. All of the identified PycA suppressor mutations encoded point mutations, which were modeled onto the crystal structure of tetrameric PycA from L. monocytogenes (Fig. 5C and D) (Sureka et al., 2014). None of the mutations were located near the c-di-AMP-binding site or on the biotin carboxyl carrier protein (BCCP) to which the biotin cofactor is attached (Fig. 5D). Suppressor mutations in PycA were next modeled onto the homologous residues in the structure of S. aureus pyruvate carboxylase (Fig. 5E) (Xiang and Tong, 2008). Many of the mutations clustered near the acetyl-CoA binding site, which was previously revealed in this structure (Fig. 5E) (Yu et al., 2009) and in the structure of Rhizobium etli pyruvate carboxylase (Maurice et al., 2007), but such a structure for L. monocytogenes PycA is currently not available. Acetyl-CoA allosterically activates pyruvate carboxylase by interacting with four conserved arginine residues to induce a conformational change proposed to increase catalytic activity (Yu et al., 2009). Two suppressor mutations, R1051C and R367L, alter those arginine residues and are predicted to disrupt interactions with the phosphates of acetyl-CoA (Fig. 5F, modeling the homologous residues of S. aureus pyruvate carboxylase). In a previous report, mutating another arginine residue in the binding pocket greatly reduced the sensitivity of pyruvate carboxylase to acetyl-CoA (Xiang and Tong, 2008). Enzyme activity of recombinant L. monocytogenes PycA was measured in an in vitro assay and more than doubled in the presence of 100 μM acetyl-CoA (Fig. 5G). The R1051C and R367L mutations abrogated acetyl-CoA-mediated activation (Fig. 5G). The R1051C mutation decreased enzyme activity relative to wild-type enzyme without ligand. Interestingly, the R367L mutation resulted in increased activity compared to the wild-type enzyme without ligand but remained less than wild-type enzyme with acetyl-CoA (Fig. 5G).

The R1051C and R367L mutations were selected from the screen and reconstructed in L. monocytogenes by complementing a ΔpycA strain. PycA protein levels were unaffected by these mutations (Fig. 6A) and when dacA was deleted, the PycA mutations recapitulated phenotypes from the genetic screen (Fig. 6B and 6C). PycA is an integral member of the “pyruvate node” encompassing the pyruvate dehydrogenase complex and PycA (Fig. 5B), which is central to growth of L. monocytogenes in both nutrient poor conditions and within a mammalian host (O'Riordan, 2003; Schar et al., 2010). c-di-AMP is a negative regulator of the pyruvate node and decreases flux from pyruvate into the TCA cycle (Sureka et al., 2014). Mutants lacking pycA, dacA, or mutants expressing the catalytically dead DacA were avirulent (Fig. 6D). We hypothesized that c-di-AMP levels modulate PycA to balance allosteric activation by acetyl-CoA during infection. Accordingly, mutations that disrupted PycA activation by acetyl-CoA suppressed the virulence defect of the ΔdacA mutant by approximately 100-fold (Fig. 6E). These data suggest a model where PycA is over-active in the absence of c-di-AMP due to activation of the protein by acetyl-CoA, which underlies the ΔdacA mutant's sensitivity to rich medium, susceptibility to β-lactam antibiotics, and virulence defects.

Figure 6. ΔdacA suppressor mutations in PycA.

(A) Immunoblot of PycA (biotin cofactor using streptavidin) and P60 (loading control). Data are representative of three independent experiments. (B) Enumeration of CFU on indicated media for L. monocytogenes strains constructed in LSM. “+” vs “Δ” indicate a mutation in a wild-type or dacA-deficient background. (C) Cefuroxime disk diffusion on LSM-agar of L. monocytogenes strains. (B and C) Data are the mean ± s.e.m. of at least three independent experiments. (D and E) CFU recovered from spleens 48 hours post-infection of CD-1 mice infected with 10⁵ CFU of the indicated strains. Data are pooled results from two independent experiments of n = 5 mice, black bars indicate median, p values were calculated using a heteroscedastic Student's t-test; ** p < 0.001, ns p > 0.05.

TCA cycle intermediates are toxic to ΔdacA mutants

c-di-AMP inhibits PycA and in a dacA depletion mutant the increased production of oxaloacetate leads to an accumulation of glutamate/glutamine (Fig. 5B) (Sureka et al., 2014). Disruption of citrate synthase, the first step of the L. monocytogenes TCA cycle, is sufficient to abolish the enhanced production of glutamate/glutamine (Sureka et al., 2014). Glutamate has a well-documented role in osmoregulation (Wood, 2015), and we hypothesized that changes in glutamate levels might underlie the defects of the ΔdacA mutant for growth in rich medium and resistance to cefuroxime. Indeed, citZ mutations allowed ΔdacA to grow in rich medium and conferred resistance to cefuroxime (Fig. 7A and B). However, mutations in citB and citC, the next two steps of the TCA cycle (Fig. 5B), did not phenocopy the citZ mutation (Fig. 7A and B). These results suggest that accumulation of citrate, not downstream metabolites such as glutamate, are responsible for the observed ΔdacA phenotypes. In support of this hypothesis, the ΔdacA mutant accumulated increased intracellular citrate compared to wild-type L. monocytogenes, which was similar to ΔdacA citZ∷himar1 and ΔdacAΔpycA p-pycA^R1051C (Fig. 7C).

Figure 7. TCA cycle intermediates restrict growth in the absence of c-di-AMP.

(A) Enumeration of CFU on indicated media for L. monocytogenes strains constructed in LSM. (B) Cefuroxime disk diffusion on LSM-agar of L. monocytogenes strains. dacA vs. ΔdacA and “+” vs “Δ” indicate a mutation in a wild-type or dacA-deficient background. (C) Intracellular citrate levels normalized to wild-type L. monocytogenes, p values were calculated using a heteroscedastic Student's t-test; * p < 0.05. All data are mean ± s.e.m of at least three independent experiments.

Discussion

Cyclic-di-AMP has emerged as a vital second messenger in many Gram-positive and some Gram-negative bacteria (Commichau et al., 2015). In L. monocytogenes the sole c-di-AMP synthase, DacA, is essential on rich medium, but not on a defined synthetic medium (Whiteley et al., 2015). To gain insight into the essential role of c-di-AMP, we extensively characterized suppressor mutations that bypassed the growth defect of the L. monocytogenes ΔdacA mutant in rich medium. Mutations in the oligopeptide transporter, oppA-F, allowed ΔdacA mutants to grow in rich medium, and peptides represented a growth restrictive medium component. Mutations in the osmoprotectant glycine-betaine transporter, gbuABC, also rescued the growth of the ΔdacA mutant, suggesting that defects in osmoregulation prevent growth in rich medium. Consistent with this hypothesis, the addition of salt rescued the growth of the ΔdacA mutant in rich medium. Deletion of genes encoding the c-di-AMP binding proteins PstA and CbpB also rescued growth; these highly conserved proteins have unknown function and remain poorly characterized. However, unlike the other suppressor mutations, ΔpstA also suppressed the extreme sensitivity of the ΔdacA mutant to the cell wall-acting antibiotic cefuroxime. Another genetic screen for suppressor mutations that allowed for growth in rich medium and in the presence of cefuroxime revealed point mutations in PycA that disrupted its activation by acetyl-CoA. Since c-di-AMP is an allosteric inhibitor of PycA, we reason that these suppressor mutations dial-down PycA activity and thereby restore metabolic homeostasis in the ΔdacA mutant. Taken together, these data suggest that c-di-AMP is necessary to couple metabolism with osmoregulation, which is important for resistance to cell-wall active antibiotics and for growth in the presence of osmolytes such as oligopeptides.

There is considerable evidence that c-di-AMP plays a major role in osmoregulation in multiple organisms. In Lactococcus lactis and L. monocytogenes, c-di-AMP-specific phosphodiesterase mutants are unable to grow in the presence of high salt concentrations (Smith et al., 2012; Zhu et al., 2015; Huynh et al., 2016). In B. subtilis, a c-di-AMP depletion mutant required the addition of sucrose for growth (Luo and Helmann, 2012). In L. monocytogenes, a dacA depletion strain lyses in rich medium, but is rescued by the addition of salt (Witte et al., 2013). Not surprisingly, c-di-AMP often controls the activity or synthesis of proteins involved in osmoregulation. For example, in S. aureus, c-di-AMP binds to and controls KtrA, a potassium importer (Corrigan et al., 2013; Kim et al., 2015), while in both L. monocytogenes and S. aureus, c-di-AMP binds to and blocks the activity of OpuCA, a carnitine importer that contributes to osmoregulation (Huynh et al., 2016; Schuster et al., 2016). In other organisms, c-di-AMP controls homologous genes (opuC, gbu, opp, and ktrA) by binding to the yuaA/ydaO riboswitch to block transcription (Block et al., 2010; Nelson et al., 2013). Our study demonstrated that the growth defects of the ΔdacA mutant were a consequence of defects in osmoregulation caused by oligopeptides in the medium. These results support the assertion of Amezaga et al. and others, that oligopeptides play a major role in L. monocytogenes osmoregulation (Maria-Rosario et al., 1995; Sleator et al., 2003). However, in the absence of c-di-AMP, oligopeptides appear to contribute to osmotic stress and inhibit growth.

c-di-AMP also plays an important role in resistance to cell wall-acting antibiotics such as cefuroxime. For example, in S. aureus, B. subtilis, L. monocytogenes, E. faecalis, and S. pyogenes, mutants in c-di-AMP-dependent phosphodiesterase are more resistant to cell wall-acting antibiotics (Griffiths and O'Neill, 2011; Luo and Helmann, 2012; Witte et al., 2013; Cho and Kang, 2013; Miller et al., 2013). As shown in this study, ΔdacA mutants are extremely sensitive to cefuroxime, yet no direct interactions between c-di-AMP and cell wall-associated enzymes have been identified. Interestingly, dacA is in an operon with glmM, which encodes phosphoglucosamine mutase, an enzyme important for amino sugar incorporation into peptidoglycan (Mengin-Lecreulx and van Heijenoort, 1996; Barreteau et al., 2008). DacA interacts with GlmM in multiple organisms (Gundlach et al., 2015b; Rismondo et al., 2016), and GlmM inhibited DacA in L. lactis (Zhu et al., 2015). Therefore, it is reasonable to postulate that DacA may also inhibit GlmM to alter the abundance of peptidoglycan precursors. However, in this study a point mutation in the DacA active site was sufficient to render L. monocytogenes sensitive to cefuroxime, and while we cannot rule out a reciprocal DacA inhibition of GlmM, these data failed to demonstrate a role for DacA protein in cefuroxime resistance independent of c-di-AMP. We speculate that the ΔdacA mutant is extremely sensitive to cefuroxime as a consequence of defects in osmoregulation, although direct effects of c-di-AMP on cell wall biosynthesis and peptidoglycan crosslinking, as has been suggested in S. aureus (Corrigan et al., 2011), cannot be excluded.

The most striking result in this study was that single amino acid mutations in PycA that block allosteric activation by acetyl-CoA restored growth, osmoregulation, cefuroxime resistance, and virulence. Pyruvate carboxylase is an evolutionarily ancient enzyme conserved in many different clades of bacteria, Archaea, and mammals (where it is located in the mitochondria) (Jitrapakdee et al., 2008). In humans, mutations affecting pyruvate carboxylase expression or activity disrupt gluconeogenesis and result in mitochondrial diseases such as lactic acidosis, hypoglycemia, and mental retardation (Robinson, 2006). One such mutation found in humans (R451C) is similar to the two pycA mutations investigated here (R367L/R1051C) that suppress ΔdacA phenotypes (Wexler et al., 1998). Interactions between these arginine residues and acetyl-CoA are proposed to cause a conformational change in the PycA biotin carboxylase domain dimer, activating biotin carboxylase activity through a yet unknown mechanism (Yu et al., 2009). The R367L mutation may correspond to an intermediate form of the activated confirmation because despite being unaffected by acetyl-CoA, R367L is more active than wild-type enzyme without ligand. Since PstA mutations have the same phenotype as the PycA suppressor mutations, it is conceivable that PstA interacts with PycA. However, neither co-IP or addition of recombinant PstA into PycA enzyme assays demonstrated any interaction (data not shown). Alternatively, PstA may interact with the pyruvate dehydrogenase complex (PDHC) that catalyzes acetyl-CoA production from pyruvate and was identified in both yeast 2-hybrid and PstA pull-downs (Data not shown). Perhaps PstA activates PDHC and the increased acetyl-CoA activates PycA. However, this has not been experimentally confirmed and biochemistry with the massive PDHC complex (estimated at >4 mega daltons) is challenging (Mattevi et al., 1992).

c-di-AMP appears to regulate pyruvate carboxylase in L. monocytogenes to balance allosteric activation by acetyl-CoA. Hyper-activity of PycA that occurs in the absence of c-di-AMP leads to increased flux through the oxidative branch of the TCA cycle, resulting in the increased production of glutamate/glutamine (Sureka et al., 2014). Mutations in citZ rescue the growth of ΔdacA mutants and decrease levels of glutamate/glutamine, suggesting that their accumulation is toxic. Unexpectedly, in this study, mutations of genes downstream of CitZ did not restore growth of the ΔdacA mutants. We suggest that accumulation of citrate (or a downstream metabolite other than isocitrate or glutamate/glutamine) causes osmotic stress and inhibits growth in the absence of c-di-AMP. Citrate can modulate osmosis through an intrinsic affinity for water molecules, which has been utilized for osmotic water purification by coupling citrate to beads and for osmotically-induced diarrhea where magnesium citrate is used clinically (Schiller, 1999; Na et al., 2014). In addition, citrate has direct roles in bacterial metabolism as an inhibitor of phosphofructokinase (Sonenshein et al., 2015), a chelator of iron and other divalent cations (Pechter et al., 2013), and as an allosteric regulator of CcpC, a transcriptional activator/inhibitor of citCZ and citB in B. subtilis and L. monocytogenes (Sonenshein, 2007; Mittal et al., 2012). Of note, ccpC is encoded in an operon with cbpB, although the relevance of this organization has not been investigated.

The stimuli that alter c-di-AMP levels are poorly defined. However, cross regulation between c-di-AMP and (p)ppGpp has demonstrated that (p)ppGpp inhibits both PdeA and PgpH (Huynh and Woodward, 2016), which predicts that c-di-AMP levels increase during starvation. This nucleotide crosstalk may explain why L. monocytogenes and E. faecalis mutants unable to regulate (p)ppGpp are unable to survive osmotic stress and are more sensitive to β-lactam antibiotics (Okada et al., 2002; Abranches et al., 2009; Yan et al., 2009). Conversely, depletion of c-di-AMP in L. monocytogenes and high levels of c-di-AMP in L. monocytogenes and S. aureus lead to an increase in (p)ppGpp (Liu et al., 2006; Corrigan et al., 2015; Whiteley et al., 2015). The stress that induces the stringent response when c-di-AMP levels are altered has not been characterized and while changes in metabolism are one possibility there is also evidence in the literature that alterations in osmotic pressure might affect (p)ppGpp levels (Yan et al., 2009). Progress in understanding how c-di-AMP levels are regulated will be aided in the future by continued development of nucleotide biosensors (Kellenberger et al., 2015) and by studying regulation of c-di-AMP synthases and hydrolases.

Although c-di-AMP is clearly an allosteric regulator of metabolic function, we discovered it as a secreted L. monocytogenes activator of host innate immunity and there is evidence that it is also secreted by both Mycobacterium tuberculosis and Chlamydia trachomatis during infection (Woodward et al., 2010; Barker et al., 2013; Yang et al., 2014; Dey et al., 2015). It remains a mystery why bacteria secrete c-di-AMP, although it has been suggested that secretion may represent an additional level of regulation (Huynh and Woodward, 2016). Since c-di-AMP is secreted through multidrug efflux pumps that are activated by toxic compounds such as bile (Quillin et al., 2011), perhaps secretion provides a mechanism to couple the response to noxious compounds with metabolic regulation. Alternatively, c-di-AMP secretion may occur in response to osmotic stress during infection. In multiple organisms, c-di-AMP clearly regulates potassium transport (Nelson et al., 2013; Peng et al., 2015; Moscoso et al., 2015). Since both bacterial and host cells maintain high extracellular sodium levels and high intracellular potassium concentrations, the transition of L. monocytogenes from the environment to the cytosol of a macrophage involves a large increase in potassium, which might lead to c-di-AMP secretion that inadvertently triggers a host innate immune response.

Experimental Procedures

Ethics statement

This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All protocols were reviewed and approved by the Animal Care and Use Committee at the University of California, Berkeley (AUP-2016-05-8811).

Bacterial culture conditions

The L. monocytogenes strains used in this study (Table 2) were derived from wild-type 10403S (Becavin et al., 2014), were maintained in brain heart infusion (BHI, Difco) for dacA+ strains, and were maintained on Listeria synthetic media (LSM) for dacA- strains, in 1-3 mL of media in 14 mL (17 × 100 mm) culture tubes, at 37°C while shaking at 220 rpm, unless specified otherwise. “Overnight” bacterial cultures were defined as 16-20 hrs post-inoculation with a single colony. The tryptic digest of casein (Tryptone, US Biotech Sources, BTS) was catalog #T01PD-500. The E. coli strains used in this study (Table 3) were used for propagating plasmids and conjugation and maintained in Lysogeny Broth (LB, Miller) at 37°C while shaking. Antibiotics were used at the following concentrations: carbenicillin (100 μg/mL, Gold Biotechnology), streptomycin (200 μg/mL, Sigma), chloramphenicol (7.5 μg/mL for L. monocytogenes and 10 μg/mL for E. coli, Sigma), erythromycin (1 μg/mL, Sigma), kanamycin (15 μg/mL in LSM, 30 μg/mL in BHI, Gold Biotechnology), and tetracycline (2 μg/mL, Sigma). All bacterial strains were frozen at -80°C in either BHI or LSM, with 40% glycerol.

Table 2. L. monocytogenes strains used in this study.

DP number	Straina	Description	Reference
	10403S	Wild-type L. monocytogenes	(Becavin et al., 2014)
DP-L6325	ΔdacA-kanR.MM1	Marked deletion of dacA (lmo2120), constructed on previous minimal medium	(Whiteley et al., 2015)
DP-L6378	ΔdacA-kanR	Marked deletion of dacA, constructed on LSM, made by transducing with DP-L6325	This study
DP-L6412	ΔdacA-kanR tRNAArg∷pLIV2.tcR-dacA	DP-L6378 integrated with plasmid from DP-E6543	This study
DP-L6413	ΔdacA-kanR tRNAArg∷pPL2t-pH2-disA	DP-L6378 integrated with plasmid from DP-E6397	This study
DP-L6390	dacAD171A	Chromosomal, catalytically dead, dacA point mutation made using DP-E6389	This study
DP-L6414	dacAD171A tRNAArg∷pLIV2.tcR-dacA	DP-L6390 integrated with plasmid from DP-E6543	This study
DP-L6530	dacAD171A tRNAArg∷pPL2t-pH2-disA	DP-L6390 integrated with plasmid from DP-E6397	This study
DP-L6291	relAR295S	relA suppressor allele reconstructed in wild-type	(Whiteley et al., 2015)
DP-L6383	ΔdacA-kanR relAR295S	DP-L6291 transduced with lysate from DP-L6378	This study
DP-L6539	ΔdacA relAR295S tRNAArg∷pPL2t-relA	DP-L6383 integrated with plasmid from DP-E6334	This study
DP-L6272	ΔoppB	In-frame deletion of oppB (lmo2195)	(Whiteley et al., 2015)
DP-L6542	ΔdacA-kanR ΔoppB	DP-L6272 transduced with lysate from DP-L6378	This study
DP-L6346	ΔcbpB	In frame deletion of cbpB (lmo1009), made using DP-E6343	This study
DP-L6380	ΔdacA-kanR ΔcbpB	DP-L6346 transduced with lysate from DP-L6378	This study
DP-L6534	ΔdacA-kanR ΔcbpB tRNAArg∷pPL2t-pH-cbpB	DP-L6380 integrated with plasmid from DP-E6408	This study
DP-L6347	ΔpstA	In frame deletion of pstA (lmo2692), made using DP-E6344	This study
DP-L6381	ΔdacA-kanR ΔpstA	DP-L6347 transduced with lysate from DP-L6378	This study
DP-L6386	ΔdacA-kanR ΔpstA tRNAArg∷pPL2t-pH-pstA-SII	DP-L6381 with integrated plasmid from DP-E6376	This study
DP-L6384	ΔdacA-kanR tRNAArg∷pPL2t-pH-pstA	DP-L6378 with integrated plasmid from DP-E6375	This study
DP-L6394	ΔgbuABC	In frame deletion of gbu operon (lmo1014-1016), made using DP-E6387	This study
DP-L6395	ΔgbuABC ΔdacA-kanR	DP-L6394 transduced with lysate from DP-L6378	This study
DP-L6531	ΔdacA-kanR ΔgbuABC tRNAArg∷pPL2t-gbuABC	DP-L6396 integrated with plasmid from DP-E6406	This study
DP-L6348	ΔpycA	In frame deletion of pycA (lmo1072), made with DP-E6345	This study
DP-L6399	ΔpycA tRNAArg∷pPL2t-pycA	DP-L6348 integrated with plasmid from DP-E6398	This study
DP-L6400	ΔdacA-kanR ΔpycA tRNAArg∷pPL2t-pycA	DP-L6399 transduced with lysate from DP-L6378	This study
DP-L6535	ΔpycA tRNAArg∷pPL2t-pycAR1051C	DP-L6348 integrated with plasmid from DP-E6410	This study
DP-L6536	ΔdacA-kanR ΔpycA tRNAArg∷pPL2t-pycAR1051C	DP-L6535 transduced with lysate from DP-L6378	This study
DP-L6537	ΔpycA tRNAArg∷pPL2t-pycAR367L	DP-L6348 integrated with plasmid from DP-E6411	This study
DP-L6538	ΔdacA-kanR ΔpycA tRNAArg∷pPL2t-pycAR367L	DP-L6537 transduced with lysate from DP-L6378	This study
DP-L6403	citC∷himar	Transposon mutation in citC (LMRG_01401∷471, lmo1566)	(Reniere et al., 2016)
DP-L6404	ΔdacA-kanR citC∷himar1	DP-L6403 transduced with lysate from DP-L6378	This study
DP-L6540	ΔcitB	In frame deletion of citB (lmo1641), made with DP-E6409	This study
DP-L6541	ΔdacA-kanR ΔcitB	DP-L6540 transduced with lysate from DP-L6378	This study
DP-L6401	citZ∷himar1	Transposon mutation in citZ (LMRG_01400∷143, lmo1567)	(Sureka et al., 2014)
DP-L6402	ΔdacA-kanR citZ∷himar1	DP-L6401 transduced with lysate from DP-L6378	This study
DP-L6533	ΔdacA-kanR citZ∷himar1 tRNAArg∷pPL2t-citZC	DP-L6402 integrated with plasmid from DP-E6407	This study

^aThe ΔdacA-kanR marked deletion nomenclature has been simplified in the body of paper as “ΔdacA”

Table 3. E. coli strains and plasmids used in this study.

DP number	Description	Reference
XL1-Blue	Cloning; recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′ proAB lacIqZΔM15 Tn10 (Tetr)]	Stratagene
SM10	Conjugation; thi-1 thr-1 leuB6 tonA21 lacY1 supE44 recA λ- integrated [RP4-2-Tcr∷Mu] aphA+ (Kmr) Tra+	(Simon et al., 1983)
DP-E6333	pPL2t, a derivative of pPL2 tetracycline resistant in L. monocytogenes	(Whiteley et al., 2015)
DP-E6340	pKSV7-oriT	(Camilli et al., 1993)
DP-E6543	pLIV2t-dacA (lmo2120)	(Witte et al., 2013)
DP-E6397	pPL2t-pH1-disA (BSU00880)	This study
DP-E6389	pKSV7-dacAD171A (lmo2120)	This study
DP-E6334	pPL2t-relA	(Whiteley et al., 2015)
DP-E6343	pKSV7-ΔcbpB (lmo1009)	This study
DP-E6408	pPL2t-cbpB	This study
DP-E6344	pKSV7-ΔpstA (lmo2692)	This study
DP-E6376	pPL2t-pH3-pstA-SII	This study
DP-E6375	pPL2t-pH3-pstA	This study
DP-E6387	pKSV7-ΔgbuABC (lmo1014-1016)	This study
DP-E6406	pPL2t-gbuABC	This study
JW363	pET28a-pycA	(Sureka et al., 2014)
DP-E6345	pKSV7-ΔpycA (lmo1072)	This study
DP-E6398	pPL2t-pycA	This study
DP-E6410	pPL2t-pycAR1051C	This study
DP-E6411	pPL2t-pycAR367L	This study
DP-E6409	pKSV7-ΔcitB (lmo1641)	This study
DP-E6407	pPL2t-citZC (lmo1567-1566)	This study

Listeria synthetic medium

The LSM recipe was devised by starting with previously published defined media (Welshimer, 1963; Premaratne et al., 1991; Phan-Thanh and Gormon, 1997). First, the excess riboflavin was removed, and the phosphate buffer was reduced and replaced with MOPS, as noted for a more minimal defined media (Tsai and Hodgson, 2003). By using MOPS the medium no longer precipitated at 2× concentration, presumably due to decreased formation of insoluble phosphate salts, making the medium more suitable for media-agar. Iron citrate was replaced with iron chloride and decreased to 50 μM without effects on growth. By comparison to B. subtilis defined (SD) media (Vasantha and Freese, 1980), salts of manganese, zinc, and calcium were included as trace metals. In addition, other trace metals, citrate (to keep metals in solution), and additional salt were added based on other published defined media. Finally, multiple stock solutions were combined for simplicity with no observed problems.

LSM was prepared by combining 8 individual concentrated stock solutions (prepared ahead of time and stored at 4 °C) with fresh glutamine and cysteine, see table S1. Stock solutions were filter sterilized and stored at 4 °C, with the exception of MOPS, Glucose, and Phosphate which were stored at room temperature. Each stock was added at the appropriate dilution factor and in the order listed in table S1, dissolving glutamine and cysteine immediately prior to filter sterilization. The final LSM is stable at 2× concentration and for approximately 6-8 weeks. LSM may be prepared from an amino acid stock solution with either the “minimum” 8 amino acids that are important for L. monocytogenes growth or the remaining 18 amino acids (other than glutamine and cysteine, which are added fresh). These stocks are referred to as “minimum” amino acid stock (used to make iLSM) or “complete” amino acid stock (used to make cLSM). Including all amino acids resulted in much faster, robust growth but may be inappropriate for some metabolic studies. In this work only cLSM was used and referred to as “LSM”.

LSM-agar plates are prepared by combining stock solutions with the fresh ingredients to a final 2× concentration, filter-sterilized, warmed to 37°C, combined with an equal volume of molten autoclave-sterilized 2× agarose (10 g/L at 1×), and poured at 15 mL/plate. Agarose must be used, not agar, which is not sufficiently pure. It is recommended that the LSM-agarose be kept warm (>55°C) while preparing the plates.

Plasmid construction and cloning

DNA manipulations and cloning were performed as previously described (Whiteley et al., 2015). Briefly, vectors were constructed by either Gibson assembly using the Gibson Assembly Ultra (Synthetic Genomics) according to manufacturer's instructions or by restriction digest and DNA ligation, see associate oligonucleotide description (Table S2). Unless otherwise stated, all reagents and kits were from New England Biolabs. PCR was performed using Kapa HiFi polymerase (Kapa Biosystems) and all vectors were transformed into XL1-Blue for construction.

All knock-in vectors were constructed by ligating recombinant DNA into the multiple cloning site of pPL2t 5′ of the B. subtilis rrnB transcriptional terminator. pPL2t is a derivative of the integration vector pPL2 that provides chloramphenicol resistance in E. coli and tetracycline resistance in L. monocytogenes (Lauer et al., 2002; Whiteley et al., 2015). The vectors pPL2t-cbpB, pPL2t-gbuABC, pPL2t-pycA, pPL2t-pycA^R1051C, pPL2t-pycA^R367L, and pPL2t-citZC were constructed by amplifying each complemented gene with its endogenous promoter(s) (see Table S2), as annotated (Wurtzel et al., 2012). Wild-type gDNA was used as a template with the exception of the pycA mutants, where suppressor mutant gDNA was used instead. The vector pPL2t-oppAB (lmo2196-5) was constructed such that the promoter of oppA could read-through the oppA coding sequence to oppB, which is important for sufficient oppB expression (Borezee et al., 2000). This vector was difficult to construct and Sanger sequencing revealed multiple mutations and IS elements disrupting oppA that lead to insufficient oppB expression. The oppA gene appears toxic to E. coli and thus a complement vector for oppB could not be constructed. The vectors pPL2t-pH1-disA, pPL2t-pH2-pstA-SII, and pPL2t-pH2-pstA were constructed by fusing the gene of interest to a constitutive promoter. The P_spac-hy (P_hyper) promoter was selected as the basic promoter and lacO sites were added on either side of the core promoter sequence to decrease expression in E. coli. Twenty nucleotides of the actA UTR were used for the ribosomal binding site. Variations were made to these sequences to optimize expression. pH1 corresponds to “aattgtgagcgctcacaattttgcaaaaagttgttgactttatctacaaggtgtggcataatgtgtgtaatTATGAGCGCTCACAATTCTCGAGAGCGTATCACGAGGAGggagtcgac” prior to the ATG of disA, and pH2 corresponds to “aattttgcaaaaagttgttgactttatctacaaggtgtggcataatgtgtgtagcgtatcacgaggagggagtataa” prior to the TTG of pstA; both promoters were synthesized as gBLOCKS (IDT) and pre-inserted into pPL2t. The amino acid sequence for the strep(II) tag was “AAASWSHPQFEK” and the nucleotide sequence was “gcggccgcaagctggAGCCATCCACAATTCGAAAAATAA”.

All knock-out vectors were constructed by ligating recombinant DNA in the multiple cloning site of the suicide vector pKSV7-OriT (Smith and Youngman, 1992; Camilli et al., 1993; Kline et al., 2015). The DNA fragment constructed for in-frame deletions consisted of 1-0.75 kb of amplified genomic DNA 5′ (using primers A&B) and 3′ (using primers C&D) of the deleted gene (see Table S2). Sequence overlap extension (SOE) PCR was used to join the two upstream and downstream fragments, maintaining the first and last six codons of the deleted gene. For allelic exchange of the dacA^D171A mutation, the aspartate codon (GAT) was mutated to alanine (GCG) using SOE PCR.

Although all L. monocytogenes strains used were a derivative of 10403S, by convention, genes are referred to by the EGD-e ordered loci names. For reference: dacA, lmo2120, LMRG_01274; oppB, lmo2195, LMRG_01637; relA, lmo1523, LMRG_01447; gbuA, lmo1014, LMRG_02114; pstA, lmo2692, LMRG_02239; cbpB, lmo1009, LMRG_02109; pycA, lmo1072, LMRG_00534; citZ, lmo1567, LMRG_01400; citC, lmo1566, LMRG_01401; citB, lmo1641, LMRG_01325.

L. monocytogenes strain construction

Strains were constructed by standard methods (Reniere et al., 2016). Plasmids were introduced to L. monocytogenes by transconjugation using donor E. coli SM10 and the appropriate L. monocytogenes recipient strain by standard methods (Simon et al., 1983; Burke et al., 2014). Chemically competent E. coli SM10 were first transformed with either pPL2t-based or pKSV7-OriT-based plasmids and mixed 1:1 with the appropriate L. monocytogenes recipient strain on non-selective solid media for 4-24 hours. Successful trans-conjugates were selected by streaking donor/recipient mixture on selective solid media containing streptomycin (to select against E. coli) and either tetracycline (for pPL2t) at 37°C or chloramphenicol (for pKSV7-OriT) at 30°C. Single colonies appeared in 24-48 hours, were selected, re-streaked on the appropriate selective media/temperature, and single colonies were again isolated. The resulting knock-in strains were stable and not maintained on antibiotics after initial construction. The resulting pKSV7-OriT trans-conjugate was then re-streaked onto solid media containing streptomycin and chloramphenicol at the 42°C, a temperature that is non-permissive for plasmid DNA replication. Single colonies appeared in 24-48 hours and were purified by re-streaking twice, sequentially, in the same selective conditions. The resulting L. monocytogenes strain harbored an integrated pKSV7-OriT and was inoculated into non-selective media at 30 °C (the temperature permissive for pKSV7-OriT replication) to enrich for strains where the plasmid had excised from the chromosome and been cured from the strain. Under these conditions bacteria were cultured sequentially 4-8 times, inoculating with approximately 10⁷ colony forming units (cfu) into 2 mL of non-selective media at 30°C. Successful excision of pKSV7 from the chromosome was screened for by patch-plating for chloramphenicol sensitive strains, then screened by PCR for successful allelic exchange. Strains were confirmed by a second round of PCR using A and D primers for the KO of interest and, when necessary, Sanger sequenced.

See table 2, L. monocytogenes strains used in this study, for details of strain construction. dacA was deleted by transducing (see section on Generalized Transduction) a ΔdacA deletion marked with a kanamycin resistance gene (kanR) as previously described (Whiteley et al., 2015) in LSM. The resulting strains were re-streaked for single colonies in LSM + kanamycin. The ΔdacA-kanR marked deletion nomenclature was simplified in the body of this paper to “ΔdacA”. When dacA mutant strains were complemented, conjugation and integration of pPL2t was performed on LSM + tetracycline. When allelic exchange was performed for the dacA^D171A mutation, the passages at 30°C in non-selective media were performed in LSM and plated on LSM-agar.

Immunoblots

Immunoblots were performed as previously described (Reniere et al., 2015). Briefly, bacteria overnight cultures were diluted 1:10 into media, incubated for two hours (BHI) or four hours (LSM) at 37°C while shaking (for isolation of log-phase cultures), and 1 OD of bacteria was harvested by centrifuge after adjusting the volume for OD normalization. The cells were washed once in PBS, resuspended in 0.1% NP-40 with 1 mM PMSF, and lysed by bead beating for 20 min with approximately 100 μL of 0.1 mm Zirconia/Silica beads (BioSpec). The lysate was centrifuged for >30 min and the supernatant was treated with TCA to 10% for one hour at -20 °C. The protein pellet was collected by centrifugation and washed twice with ice-cold acetone, followed by air-drying the protein pellet. The pellet was resuspended in 50 μl LDS (Invitrogen) with 5% BME, boiled for 20 min, and analyzed by SDS-PAGE.

Primary antibody against P60 (mouse, monoclonal, Adipogen) was used at a 1:5,000 dilution and DacA (rabbit anti-sera, a generous gift of Fabian Commichau) was used as previously reported (Rismondo et al., 2016). P60 is a loading control and is constitutively expressed. For secondary antibodies, fluorophores conjugated to goat anti-mouse, goat anti-rabbit, or streptavidin (Licor) were used according to manufacturer's instructions. Immunoblots were visualized and quantified using an Odyssey Imager.

Disk diffusions

Disk diffusions were performed as previously described (Durack et al., 2015), with the following changes. Approximately 10⁷ CFU from an overnight culture were diluted in 3 mL of molten top-agar (0.8% agarose, 0.8% NaCl) and evenly spread over LSM-agar plates. After the agar had solidified, Whatman paper disks soaked in 125 μg of cefuroxime or 50 μg of bialaphos (Gold Biotechnology), unless specified otherwise, were placed on top of the bacteria-agarose layer. The zone of inhibition was measured as area by imaging plates after 48 hours of incubation at 37 °C and quantifying using ImageJ software (Schneider et al., 2012).

Quantification of mutant growth in various media and synthetic peptides

Overnight cultures in LSM were serially diluted in a 96-well plate 1:10 with PBS, ranging from 10^-1 to 10^-8. 5 μL of each culture-dilution were spotted on solid media plates using a multichannel pipette. Plates were imaged and CFU per dilution were quantified for LSM, BHI, and LSM + 1% w/v tryptone after 48 hours at 37 °C.

For synthetic peptides, 0.0075 g hexapeptide (0.25% w/v final concentration, 4.48 mM for KAAAAK and 3.44 mM for KLLLLK) or 0.03 g dipeptide (1% w/v final concentration, 46 mM for AQ) were dissolved in 1.5 mL 2× LSM and the LSM-peptide product was filter-sterilized (0.2 μm). Specific peptides were: KAAAAK (MW=558.68 g mol^-1 >61% pure, Mimotopes), KLLLLK (MW=727.00 g mol^-1 >72% pure, Mimotopes), and AQ (MW=217.22 g mol^-1 >98% pure, Sigma). Lysines on N and C termini of hexapeptides were added to increase solubility. The LSM-peptide mixture was then warmed to 37 °C, combined with 1.5 mL molten autoclave-sterilized 2× agarose, and distributed into 6-well (9.5 cm² well^-1) tissue culture dishes (Corning) at 2.5 ml well^-1. Overnight cultures in LSM were serially diluted in a 96-well plate 1:10 with PBS, ranging from 10^-1 to 10^-7. For strains able to grow in the presence of tryptone, 5 μL of culture-dilution at 10^-5, 10^-6, and 10^-7 were spotted on the media-agar in wells using a multichannel pipette. For strains unable to grow in the presence of tryptone, 5 μL of culture-dilutions 10^-1, 10^-2, and 10^-3 were spotted likewise. All strains, including wild-type L. monocytogenes, grew more slowly when synthetic peptides were added to the media. Thus in experiments comparing growth in media containing synthetic peptides, dishes were imaged and CFU were enumerated after 72 hours at 37 °C, including LSM and BHI controls.

Generalized transduction

Generalized transductions were performed as previously described (Zemansky et al., 2009). Briefly, phage was cultivated from donor L. monocytogenes strains that were cultured overnight at 30°C, without shaking, and tilted slightly. 1 mL of overnight culture was pelleted and resuspended in 1 mL fresh LMC (LB broth, 10 mM CaCl₂, 10 mM MgSO₄), diluted 1:100 into 4 ml of molten LB Top-Agar with phage (0.7% agar, LB broth, 10 mM CaCl₂, 10 mM MgSO₄, approximately 10⁶ PFU of ΦU153 (Hodgson, 2000)) and spread over LB-agar plates. After 24 hours at 30°C, 5 mL of TM buffer (10 mM Tris, 10 mM MgSO₄, pH 7.5) was added to the plates and incubated at room temperature overnight (>6 hours) to “soak out” phage. The phage lysate was collected filter sterilized. The following day, a recipient strain overnight culture was harvested and approximately 5×10⁸ recipient bugs were resuspended in 100 μL LMC and 10 μL of the phage lysate was added (approximately 10⁷ PFU of phage). After incubating for 30 min at room temperature, the mixture was spread over selective solid media. himar1 transposons (erythromycin resistant) were recovered after 48 hours at 37°C, ΔdacA-kanR transductants were recovered 48 hours at 37°C on LSM plus kanamycin, single colonies were re-streaked on selective solid media and, when necessary, confirmed by PCR/Sanger sequencing. When phage lysate for ΔdacA-kanR was made, all LB and selective media were replaced with LSM and LSM was used in place of LMC. ΦU153 was propagated on the hyper-susceptible strain SLCC-5764 a.k.a. “MACK”.

ΔdacA p-pstA suppressor generation and identification

Overnight cultures of ΔdacA p-pstA were plated on LSM containing 150 μg/ml cefuroxime (approximately 10× the MIC of ΔdacA, determined as previously described (McKay and Portnoy, 2015)). Cefuroxime-resistant colonies were isolated, cultured in LSM, and dilutions were spot-plated on BHI plates. Resistance to cefuroxime was verified by antibiotic disk diffusion on LSM and compared to ΔdacA mutants. Mutants identified in this screen were genome sequenced as previously described (Burke and Portnoy, 2016). Briefly, mutants were cultured overnight in 5 mL of LSM and genomic DNA was extracted (MasterPure Gram Positive DNA Purification Kit, Epicentre) according to manufacturer's instructions. gDNA was then submitted for library preparation and Illumina sequencing (single read 50) at the UC Berkeley QB3 Genomics Sequencing Laboratory using. Data was assembled and aligned to the 10403S reference genome (GenBank: GCA_000168695.2) demonstrating >50× coverage. SNP/InDel/structural variations from the wild-type strain were determined (CLC Genomics Workbench, CLC bio).

Pyruvate carboxylase purification and enzyme assay

The wild-type PycA protein was expressed and purified as previously described (Sureka et al., 2014). pycA mutations were made using the QuikChange kit (Stratagene) and confirmed by Sanger DNA sequencing. The mutant alleles were expressed and purified using the same protocol as the wild-type protein. The catalytic activities of wild-type and mutant proteins were assessed based on a published protocol (Modak and Kelly, 1995), which couples oxaloacetate production to the oxidation of NADH by malate dehydrogenase (Sigma), followed spectrophotometrically by the decrease in absorbance at 340 nm. The activity was measured at room temperature in a reaction mixture containing 100 mM Tris (pH 7.5), 200 mM NaCl, 5 mM MgCl₂, 50 mM NaHCO₃, 50 mM (NH₄)₂SO₄, 5 units of malate dehydrogenase, 4 mM NADH, 5 mM pyruvate, 2 mM ATP, and 1 μM PycA (based on the monomer). The reaction was initiated by addition of ATP.

In vivo infections

Mouse models of infection were performed as previously described (Rae et al., 2011; Archer et al., 2014), with the following alterations. CD-1 mice were purchased from Charles River and infected at 8-12 weeks of age with 10⁵ CFU of L. monocytogenes intravenously (i.v.) via the tail-vein. Mice were euthanized 48 hours post-infection, organs were homogenized in 0.1% IGEPAL CA-630 (Sigma), and dilutions were plated in LSM for all strains except ΔpycA, which does not grow on LSM and instead was plated on BHI.

Citrate quantification

Bacterial strains were diluted from overnight culture into fresh LSM at an initial OD of approximately 0.1 and cultured to mid-log (OD₆₀₀=1.0) in LSM (250 mL flask without baffles, 25 mL of media, 220 rpm shaking, 37 °C), harvested bacteria were washed in PBS, resuspended in an equal volume of citrate assay buffer (see below), and immediately frozen in liquid nitrogen. Bacteria were then thawed and lysed by bead-beating using 0.1 mm Zirconia/Silica beads (BioSpec) for 15 minutes. Citrate concentrations were determined using Citrate Assay Kit (Sigma) according to the manufacturer's instructions.

Statistical Analysis

Statistics were performed as previously described (Mitchell et al., 2015), with the following alterations. Where appropriate data were compared using a 2-tailed heteroscedastic Students t-test.