OUP user menu

The CodY pleiotropic repressor controls virulence in gram-positive pathogens

Ludwig Stenz, Patrice Francois, Katrine Whiteson, Christiane Wolz, Patrick Linder, Jacques Schrenzel
DOI: http://dx.doi.org/10.1111/j.1574-695X.2011.00812.x 123-139 First published online: 1 July 2011


CodY is involved in the adaptive response to starvation in at least 30 different low G+C gram-positive bacteria. After dimerization and activation by cofactor binding, CodY binds to a consensus palindromic DNA sequence, leading to the repression of approximately 5% of the genome. CodY represses the transcription of target genes when bound to DNA by competition with the RNA polymerase for promoter binding, or by interference with transcriptional elongation as a roadblock. CodY displays enhanced affinity for its DNA target when bound to GTP and/or branched chain amino acids (BCAA). When nutrients become limiting in the postexponential growth phase, a decrease of intracellular levels of GTP and BCAA causes a deactivation of CodY and decreases its affinity for DNA, leading to the induction of its regulon. CodY-regulated genes trigger adaptation of the bacteria to starvation by highly diverse mechanisms, such as secretion of proteases coupled to expression of amino acid transporters, and promotion of survival strategies like sporulation or biofilm formation. Additionally, in pathogenic bacteria, several virulence factors are regulated by CodY. As a function of their access to nutrients, pathogenic gram-positive bacteria express virulence factors in a codY-dependant manner. This is true for the anthrax toxins of Bacillus anthracis and the haemolysins of Staphylococcus aureus. The purpose of this review is to illustrate CodY-regulated mechanisms on virulence in major gram-positive pathogens.

  • Staphylococcus aureus
  • Listeria monocytogenes
  • Streptococcus pneumoniae
  • Sreptococcus pyogenes
  • Bacillus anthracis
  • Clostridium difficile


A major limitation on bacterial growth is the availability of nutrients. Depending on the nutrient supply, the bacterium has to adjust its gene expression. There is no need to express genes implicated in the division process if the environment is not sufficiently rich in nutrients to feed the ‘mother’ bacteria. This adaptation to the nutritional status is triggered in gram-positive bacteria by CodY. The codY gene was identified for the first time in 1993 in Bacillus subtilis (Slack et al., 1993). Homologues of codY were then found exclusively among other low G+C gram-positive bacteria. Putative GTP-binding motifs were detected in the CodY peptide sequence that may be involved in GTP binding. The binding of this nucleotide induces increased DNA-binding properties of CodY. CodY is able to dimerize (Blagova et al., 2003; Levdikov et al., 2006; Levdikov et al., 2009), forming a dimer interface with highly conserved residues found across CodY homologues (Fig. 1) (Levdikov et al., 2006). Each monomer of CodY is able to bind one branched chain amino acid (BCAA) cofactor (Levdikov et al., 2009). Two major domains in CodY are responsible for two distinct physical interactions. One interaction involves the binding of a BCAA to the ‘cGMP-specific and -stimulated phosphodiesterases, Anabaena adenylate cyclases and Escherichia coli FhlA’ (GAF) domain of CodY localized at the N-terminal region of the protein. As a consequence of the binding, a conformational change activates DNA binding, leading to the formation of a hydrophobic pocket surrounding the bound BCAA (Fig. 1) (Levdikov et al., 2009). The second interaction involves a helix–turn–helix (HTH) motif localized at the C-terminal region of CodY. The HTH motif binds to a consensus palindromic DNA sequence and represses the transcription of genes located downstream of the binding site (Guedon et al., 2001; Ratnayake-Lecamwasam et al., 2001; Molle et al., 2003; Shivers & Sonenshein 2004; Bennett et al., 2007; Dineen et al., 2007; Belitsky & Sonenshein 2008; Hendriksen et al., 2008; Hsueh et al., 2008). The B. subtilis CodY HTH sequence –ASKIADRV [Helix 1], GITR [turn], SVIVNALR [Helix 2] — is extremely well conserved (conserved amino acids are depicted in bold) suggesting that CodY homologues recognize and bind target promoters in a similar manner (Joseph et al., 2005; Levdikov et al., 2006).

Figure 1

Structure of Bacillus subtilis CodY and physical interactions. The diagram illustrates the secondary structure of the 259 amino acids B. subtilis CodY with α-helices (▪), β-sheets (▪) and η-turns (□) as determined by crystal structure analysis (Levdikov et al., 2006; Levdikov et al., 2009). The interface region between two CodY homodimers is presented in blue as well as the amino acids important for the dimerization process. In green are depicted the G1, G3 and G4 putative GTP-binding motifs. The lines below the GAF and HTH domains show their location on the peptide sequence of CodY. Black lines illustrate amino acids in the CodY sequence that interact with BCAA according to their position, also shown in black. In gray are depicted the amino acids needed for DNA binding. Dimerization: Arg8, Glu144, Gln15 and Thr148 amino acids are important for CodY dimerization according to crystal structure and sequence alignments (Levdikov et al., 2006). HTH motif: B. subtilis CodY helix–turn–helix (HTH) motif is located in the C-terminus and was characterized as residues 203–226 according to crystallography studies, and residues 200–224 by comparison of HTH regions of characterized transcription factors (Dodd & Egan 1990; Joseph et al., 2005; Levdikov et al., 2006). Ala207, Asp208, Arg214 and Ser215 amino acids are essential for DNA-binding properties of CodY according to site-directed mutagenesis combined to DNA-binding experiments (Joseph et al., 2005). GAF domain: in B. subtilis, the binding of BCAA to CodY triggered extending refolding of the binding site loops linking β3 and β4 sheets with up to 15 Å movement. This refolding appeared responsible for the formation of a hydrophobic binding pocket in the GAF domain of CodY only present when the BCAA cofactor is present. The core and its dimerization interface are unchanged during the refolding process (Levdikov et al., 2009). More precisely in the GAF domain, F40, F71 and F98 amino acids are essential for CodY-BCAA binding, according to site-directed mutagenesis and in vitro CodY–DNA footprinting experiments (Villapakkam et al., 2009). GTP-binding: Comparison of putative GTP-binding motifs in CodY homologues leads to the identification of three conserved motifs with a total of 10 amino acids highly conserved; motif G1 (GXXXXGXT), motif G3 (DXXG) and motif G4 (NKXDTQ), where X is a nonconserved amino acid (Ratnayake-Lecamwasam et al., 2001; Handke et al., 2008). These motifs are typically found within a single domain in small GTPase proteins and are dispersed in CodY, the G1 motif being localized in the N-terminal moiety of CodY whereas G3 and G4 are localized in the C-terminal part of the protein (Levdikov et al., 2006). The implications of these motifs for cofactor binding to CodY remain controversial (Handke et al., 2008).

CodY senses the intracellular nutrient status of bacteria by BCAA binding

Bacteria take up nutrients from the environment to survive and multiply. When nutrients become limiting, bacteria stop dividing and population size remains stationary. A hallmark of this transition is the decrease of intracellular levels of BCAAs (Mitani et al., 1977; Freese et al., 1979; Lopez et al., 1979, 1981; Ochi et al., 1982; Soga et al., 2003; Handke et al., 2008; Tojo et al., 2008). The decreasing concentration of BCAAs in the starved bacteria causes a transcriptional derepression of the direct targets of CodY by detachment of the inactivated dimers from DNA-binding sites. CodY seems to act as a nutritional sensing molecular system leading to bacterial adaptation with newly expressed genes in starved low G+C gram-positive bacteria. Mechanistically speaking, binding of BCAAs occurs in the N-terminal GAF domain of CodY and induces a conformational change that is thought to affect the C-terminal DNA binding domain, thereby increasing the affinity of CodY for its DNA-binding box (den Hengst et al., 2006; Levdikov et al., 2006, 2009). GAF domains are present in numerous signalling and sensory proteins, in transcription factors, and in the c-di-GMP signalling mechanism (Zoraghi et al., 2004; Levdikov et al., 2006, 2009). The combination of HTH and GAF domains within a single polypeptide chain was not previously reported in other proteins among all sequenced organisms; it seems to be specific to CodY and restricted to low G+C gram-positive bacteria (Levdikov et al., 2006). The common characteristics of all GAF domains that are present in proteins belonging to all the three kingdoms of life are a β-sheet and two α-helices forming superimposable three-dimensional structures (Levdikov et al., 2006). The crystal structure of the putative GAF domain of B. subtilis CodY was superimposable with the GAF domain of the YKG9 protein found in Saccharomyces cerevisiae. This superposition validated that a GAF domain is present in CodY (Levdikov et al., 2006).

In order to investigate which BCAA is needed for CodY activity, Lactococcus lactis CodY was tested for ilvD promoter affinity with gel retardation assays performed in the presence of leucine, valine and isoleucine (Guedon et al., 2005). Only the presence of isoleucine significantly enhanced CodY binding, whereas leucine and valine showed lower and insignificant increases in DNA binding (Guedon et al., 2005). Isoleucine also seemed to be the major cofactor for CodY in Staphylococcus aureus (Pohl et al., 2009). Interestingly, excess isoleucine in the growth culture media lead to abnormal activation of CodY, inhibiting growth. This growth inhibition might be due to blockage of metabolic pathways involved in amino acid synthesis (Guedon et al., 2005). That specific isoleucine-induced and CodY-related growth inhibition was observed in L. lactis and S. aureus (Chambellon & Yvon 2003; Guedon et al., 2005; Pohl et al., 2009). But, isoleucine is not the only BCAA able to activate CodY homologues. In B. subtilis, valine was also able to bind CodY leading in this case to a 18-fold higher affinity for DNA (Levdikov et al., 2006). The most precise description of CodY-isoleucine binding comes from X-ray crystallography with B. subtilis CodY showing that the isobutyl side chain of isoleucine is enveloped in the hydrophobic pocket formed by the GAF domain of CodY, whereas the carboxyl groups of isoleucine engage in polar interactions with nearby residues (Levdikov et al., 2006). Note that conformational changes resulting from the binding of BCAAs to CodY can affect residues at great distances from the corepressor-binding site, as demonstrated by the effect of N-terminal GAF-binding BCAAs on the conformational changes in the C-terminal DNA-binding domain (Levdikov et al., 2006). Finally, an additive effect of GTP and BCAA concentration on the binding of B. subtilis CodY to DNA was reported in the promoter region of B. subtilis ylmA (Shivers & Sonenshein 2005; Belitsky & Sonenshein 2008; Handke et al., 2008).

CodY senses the intracellular energetic status of bacteria by binding GTP

As reported for BCAA, the intracellular concentration of GTP also decreases in starved bacteria entering the stationary phase of growth. For example, intracellular GTP concentration was estimated to be between 2 and 3 mM during the exponential phase of growth in B. subtilis and decreases to 300 µM during the early-stationary phase (Soga et al., 2002; Handke et al., 2008). The discrimination of genes repressed by BCAA-CodY or GTP-CodY is not yet fully understood. By binding GTP, CodY directly senses the energetic status of the cell and may control the expression of developmental genes implicated in morphological differentiation. Sporulation or expression of flagella were reported in starved B. subtilis, leading to different bacterial adaptation mechanisms: adopting a tougher survival form or swarming towards a healthier environment, respectively (Bennett et al., 2007). The binding site of GTP in the CodY protein remains unknown, but it is definitely not located in the GAF domain (Levdikov et al., 2006, 2009; Handke et al., 2008). Three motifs encountered in small GTPase proteins, called G1, G3 and G4, are conserved among CodY homologues and are potentially implicated in GTP binding (Fig. 1) (Ratnayake-Lecamwasam et al., 2001; Handke et al., 2008). Interestingly, in the G1 motif of L. lactis CodY, a homologue that lacks the ability to be activated by GTP, the first conserved guanine residue (G) is replaced by a serine residue (S) (Fig. 1) (Petranovic et al., 2004). Moreover, substituting a glycine in the sixth position of the G1 motif of CodY to asparagine greatly decreased CodY affinity for GTP (Slack et al., 1995; Ratnayake-Lecamwasam et al., 2001). Small GTPase proteins hydrolyse GTP whereas CodY probably does not. The separation of binding and catalytic steps has been studied in other proteins that interact with nucleic acids, such as recombinases (Whiteson & Rice 2008). Indeed, nonhydrolysable analogues of GTP [guanosine 5′-(γ-thio)triphosphate, guanosine 5′(βγ-imido)triphosphate] were able to activate CodY at similar concentrations than those at which GTP was able to activate CodY for DNA binding (Handke et al., 2008). An extensive study was performed by Handke et al. (2008) to investigate the ability of molecules resembling GTP to activate CodY for DNA binding in B. subtilis. GTP and dGTP are able to induce CodY binding to the ilvB promoter, suggesting that CodY does not discriminate between ribose and desoxyribose forms of GTP (Handke et al., 2008). But because GTP concentration is much higher than dGTP concentration within cells, as reported for example in Salmonella typhimurium, GTP is probably the physiological substrate of CodY (Bochner & Ames 1982). Intriguingly, ATP — in contrast to CTP and UTP — was able to compete with GTP for binding to CodY but without increasing the affinity of CodY for DNA (Ratnayake-Lecamwasam et al., 2001; Handke et al., 2008). Finally, an additional ppppG synthetic molecule was able to activate CodY binding to DNA even more efficiently than GTP (Handke et al., 2008).

In addition to CodY, another regulatory system is triggered simultaneously whenever the intracellular GTP levels decrease. This adaptation was called the ‘stringent response’ (ST) due to the strength of the observed global adjustment of gene expression (Shivers & Sonenshein 2005; Sonenshein 2005; Handke et al., 2008; Lemos et al., 2008; Tojo et al., 2008; Geiger et al., 2010). Upon nutrient starvation, the intracellular pool of amino acids decreases, leading to increases in accidental binding of uncharged tRNA (a tRNA without amino acid) to the ribosome. Then, the ribosome-associated protein RelA mediates the synthesis of two different alarmones, corresponding to a kind of intracellular signalling molecules produced by bacteria and plants in function of environmental factors, also called magic spots: the guanosine pentaphosphate (pppGpp) and the guanosine tetraphosphate (ppGpp). These alarmone molecules are the products of enzymatic phosphorylation of GTP and GDP, taking phosphate groups from ATP hydrolysis. The reaction is mediated by RelA and SpoT in gram-negative bacteria and RelA/SpoT homologues (RSH) in gram-positive bacteria (Mechold et al., 1996; Wendrich & Marahiel 1997; Mittenhuber 2001; Magnusson et al., 2005). Subsequently, ppGpp binds to the RNA polymerase and redirects transcription from growth-related genes to genes involved in stress resistance and starvation survival (Sonenshein 2005). In other words, genes implicated in macromolecular biosynthesis are downregulated whereas genes involved in amino acid biosynthesis and stress tolerance are upregulated (Magnusson et al., 2005). Therefore, in response to starvation, macromolecular stocks of energy and nutrients are mobilized and metabolism of amino acids is activated, as well as stress tolerance mechanisms (Magnusson et al., 2005). The ST decreases intracellular GTP levels and also decreases GTP synthesis by inhibiting the inosine monophosphate dehydrogenase (Sonenshein 2005). Thus the ST is related to the derepression of the CodY regulon in gram-positive bacteria. However, the alarmones pppGpp and ppGpp are not able to bind and activate CodY directly (Handke et al., 2008). In summary, CodY is also activated by GTP, which probably binds inside motifs resembling those found in small GTPases. The ST can derepress the CodY regulon by decreasing the intracellular GTP levels.

DNA fixation sites and conserved targets of CodY

HTH motifs are founded in transcription factors belonging to the tree kingdoms of live and are implicated into the DNA-binding activity of these HTH carrying proteins, resulting to a regulation of the transcription of genes (Pabo & Sauer 1984; Aravind et al., 2005). Generally, helix 1 has a stabilizing role and sits above the major groove of DNA whereas helix 2, the so-called recognition helix, forms specific interactions with the major groove of DNA; both helices are at a 120° angle (Pabo & Sauer 1984; Joseph et al., 2005). Through its HTH motif, CodY binds to a palindromic DNA sequence (AATTTTCWGAAAATT). This was first established in L. lactis when searching for a de novo motif in front of CodY-regulated genes, and then across a variety of CodY-containing bacteria (Serror & Sonenshein 1996a, b; Guedon et al., 2001; Bergara et al., 2003; den Hengst et al., 2005a, b; Guedon et al., 2005; Majerczyk et al., 2010). Each monomer of CodY likely recognizes half of the symmetrical DNA-binding site, as reported for other DNA-binding proteins such as the well-known oncogenic transcription factors Fos and Jun, many restriction enzymes and telomerases (Risse et al., 1989; Gipson et al., 2007). In addition, investigations based on ‘protein stitchery’, corresponding to artificial combinations of various half dimmers recognizing new DNA-binding sites, revealed the design superiority of palindromic DNA sequences for protein recognition (Park et al., 1993). For CodY, experiments performed with substitutions of DNA base pairs have shown that (1) the two primary GG of the G1 motif (Fig. 1) remain important for DNA binding of GTP-CodY, (2) increasing or decreasing the similarity between CodY-binding sites to the consensus recognized sequence increased (or decreased, respectively) the affinity of CodY for DNA (Guedon et al., 2005; Belitsky & Sonenshein 2008), (3) varying the location of the artificial CodY box by five nucleotides (half of a helical turn of DNA) from the correct position destroys CodY regulation, suggesting that CodY-mediated repression might be helix-face-dependant (den Hengst et al., 2005b), and (4) inserting an artificial CodY-binding site in front of a gene that is not regulated by CodY leads to artificial CodY-mediated regulation of that gene, whereas removing the binding site of a CodY-regulated gene leads to the loss of CodY regulation (den Hengst et al., 2005b). Altogether, these experiments show that the palindromic DNA sequence is ‘an independent functional motif’ responsive to CodY protein in vivo, according to Chris D. den Hengst et al. (2005a, b), and therefore, it is generally accepted that a gene derepressed in the absence of CodY and possessing a CodY box in its promoter should be a direct target of CodY (den Hengst et al., 2005b; Guedon et al., 2005; Belitsky & Sonenshein 2008; Hendriksen et al., 2008). Surprisingly, the AATTTTCWGAAAATT sequence is not present in the genome of B. subtilis, but appears to harbour homologues containing up to five mismatches (Kunst et al., 1997; Belitsky & Sonenshein 2008). A CodY box containing two mismatches compared with the consensus is present in the promoter region of B. subtilis ylm. Base substitution leading to only one mismatch resulted in CodY–DNA binding both in absence and presence of BCAA and GTP. Thus, the regulation of the targeted gene by cofactors sensing was not possible, suggesting that B. subtilis CodY displays too high affinity for the consensus binding motif (Belitsky & Sonenshein 2008). Note that sometimes, CodY–DNA binding can interfere with other regulators. For example, in the promoter region of dpp (dipeptide permease operon), CodY binds to a DNA sequence that overlaps with the binding region of another regulator called AbrB (Belitsky & Sonenshein 2008). The strength of CodY–DNA binding was estimated in three independent studies and results in similar Kd values in the nanomolar range (Shivers & Sonenshein 2004; Joseph et al., 2005; Hendriksen et al., 2008). Nonspecific DNA-binding affinity normally has dissociation constants in the micromolar range, while typical specific DNA binding is low nanomolar (Lodish Biochemstry textbook), and exceedingly tight DNA binding or covalent bonds are in the picomolar range (Nalefski et al., 2006). These ranges of Kd depend on the in vivo concentrations of the protein [P] and its binding site [L] calculated with the formula (Nalefski et al., 2006): Embedded Image CodY–DNA interaction is therefore within the expected range for a specific DNA-binding interaction (Shivers & Sonenshein 2004; Joseph et al., 2005; Hendriksen et al., 2008). Additionally, increasing CodY concentrations in vitro can lead to additional CodY-binding sites displaying various affinities for CodY, as in the promoter region of Listeria monocytogenes argC (Joseph et al., 2005; Dineen et al., 2007).

Among low GC gram-positive bacteria, the CodY sequence, its cofactors, the recognized DNA palindrome and the dimerization process are all relatively well conserved. Therefore, it is likely that some CodY-regulated genes are also conserved across species, illustrating a probable basic and ancestral function triggered by CodY. The CodY regulons, mainly determined by microarray experiments comparing wild type and codY mutants, were determined in S. aureus, B. subtilis, L. lactis, L. monocytogenes and Streptococcus pneumoniae. Similarly, searches for CodY DNA-binding motifs were performed in 12 different genomes belonging to Lactococcus, Streptococcus, Staphylococcus, Bacillus, Listeria, Enterococcus and Clostridium species. This resulted in the discovery of a consensus DNA-binding sequence (containing up to three nucleotide mismatches) upstream of genes encoding BCAA synthesis proteins that were determined to be differentially expressed by microarray, and included secreted proteases and oligopeptide transporters. Thus, the ancestral function of CodY probably aims at increasing the pool of intracellular BCAAs by different mechanisms (Molle et al., 2003; Guedon et al., 2005; Bennett et al., 2007; Hendriksen et al., 2008).

CodY exerts feedback control on the pathways involved in the biosynthesis of its direct effectors or cofactors (Guedon et al., 2005); it affects operons involved in the synthesis of isoleucine, leucine and valin (ilv-leu) as well as the oligopeptide permease (opp), all repeatedly identified as targets of CodY (Slack et al., 1993; Slack et al., 1995; Serror & Sonenshein 1996a, b; Ratnayake-Lecamwasam et al., 2001; Blagova et al., 2003; Molle et al., 2003; Shivers & Sonenshein 2004, 2005; Joseph et al., 2005; Sonenshein 2005; Levdikov et al., 2006; Shivers et al., 2006; Bennett et al., 2007; Dineen et al., 2007; Majerczyk et al., 2007; Belitsky & Sonenshein 2008; Handke et al., 2008; Pohl et al., 2009; van Schaik et al., 2009; Villapakkam et al., 2009; Majerczyk et al., 2010). In general, CodY regulons are composed of hundreds of genes that usually overlap across bacterial species when concerning genes implicated in the amino acid metabolism. However, it is not surprising that CodY-regulated virulence mechanisms remain highly organism-specific (Molle et al., 2003; Guedon et al., 2005; Bennett et al., 2007; Hendriksen et al., 2008). Finally, the major metabolic regulators CodY, the catabolite control protein A (CcpA) and the transcriptional nitrogen regulator A (TnrA) control the transcription of the ilv-leu operon together. This is a major intersection of metabolic pathways, and a way to link nitrogen and carbon metabolism in gram-positive bacteria (Grandoni et al., 1992; Tojo et al., 2004; Shivers & Sonenshein 2005).

Regulation and physiology of CodY

Numerous studies have attempted to reveal how CodY regulates the expression of genes, but the regulation of CodY itself still remains poorly understood. The ATPase chaperone ClpC was reported to repress the expression of CodY in S. aureus (Luong et al., 2011). However, from a mechanistic point of view, there is no requirement for CodY regulation because the presence of cofactors should be sufficient to influence the activity of CodY. Nevertheless, CodY boxes were detected in the promoter region of codY genes in bacteria belonging to Lactococcus, Streptococcus, Enterococcus, Staphylococcus and Clostridium genera (Guedon et al., 2005), allowing for negative feedback regulation. The most convincing evidence for autoregulation of codY expression comes from experiments performed in L. lactis, where CodY is able to bind to a box located upstream of its ORF, in the presence of isoleucine (den Hengst et al., 2005b). Replacement of that box with an unrelated sequence prevents CodY binding to its promoter. Moreover, lacZ fused in frame to disrupted codY resulted in an increase of β-galactosidase activity (den Hengst et al., 2005b). These results support the hypothesis that CodY itself provides negative feedback regulation of codY (den Hengst et al., 2005b). The autoregulation of codY was also validated experimentally in Streptococcus pyogenes and S. pneumoniae (Hynes 2004; den Hengst et al., 2005b; Hendriksen et al., 2008). During exponential growth, activated CodY should be expressed at low levels due to autorepression. However, when nutrients become limiting and CodY is inactivated, its expression should increase.

Interestingly, CodY is sometimes expressed independently, sometimes cotranscribed with other genes (pncA, aat, xerC, clpQ and clpY), and sometimes both modes of expression are detected; but the biological significance of the cotranscription is not known. First, codY is probably cotranscribed with downstream pncA, encoding a pyrazinamidase/nicotinamidase in Streptococcus mutans (Lemos et al., 2008). Second, codY is expressed both monocistronically and with downstream aat genes encoding an aspartate aminotransferase in a dicistronic form in S. pyogenes, but only the monocistronic mRNA of codY appears to be negatively autoregulated. Third, in S. aureus, codY is cotranscribed with a tyrosine recombinase and two ATP-dependent proteases in a long transcript but also in a monocistronic form (Malke et al., 2006; Pohl et al., 2009). In cases of cotranscription, it is necessary to control for the possibility that the observed phenotype is not due to a polar effect, as described in the referenced study using Northern blot to probe the transcripts of genes surrounding codY (Malke et al., 2006). While the genomic context surrounding codY differs between organism as well as the codY sequence itself, the homology of codY sequences is not related to the similarity of the genomic context surrounding codY (Pohl et al., 2009).

CodY in pathogenic gram-positive bacteria

CodY regulates genes involved in the primary metabolism in order to adapt the stressed bacteria to starvation. Additionally, virulence mechanisms appear to be CodY-regulated in various pathogenic gram-positive bacteria. The purpose of the following sections is to summarize recent discoveries of CodY-regulation on the virulence of gram-positive human pathogens.

Bacillus anthracis and anthrax disease

Anthrax is an acute and lethal disease of mammals caused by inhalation, ingestion or contact with wounded skin of B. anthracis spores (van Schaik et al., 2009). Once inside the host, spores germinate and give rise to vegetative cells that multiply in host tissues. Both a tripartite toxin encoded by pagA, lef and cya genes, as well as a poly-y-d-glutamate capsule encoded by the cap operon of B. anthracis are needed for host tissue invasion and immune system fighting during proliferation of B. anthracis vegetative bacteria (Mock & Fouet 2001). The toxin component genes are located on pXO1, a 182-kb plasmid (Okinaka et al., 1999). The biosynthetic enzymes for capsule production are encoded by the capBCADE operon located on plasmid pXO2 (Candela et al., 2005). The presence of pXO1 plasmid is required for toxicity of B. anthracis but both plasmids are necessary for full virulence. The regulation of virulence in B. anthracis is complex and involves a lot of factors: among them, the anthrax toxin activator gene (atxA) that is also located on pXO1 seems to play a central role (Dai & Koehler 1997). Experimental observations support the idea that an additional unknown factor is important for controlling B. anthracis toxin genes (Dai & Koehler 1997). Interestingly, the production of toxin components in B. anthracis is low during the exponential growth and reaches maximal levels during entry in the stationary phase (Sirard et al., 1994). Recently, complete abolition of the expression of the three virulence factors (pagA, lef, cya) needed for proliferation of B. anthracis in mammals was reported in a codY mutant as assessed by a gene-lacZ fusion assay and confirmed by immunoblotting (van Schaik et al., 2009). The authors suggested that CodY probably promotes the AtxA post-translational accumulation needed for the expression of pagA, lef and cya, instead of directly repressing the toxins (van Schaik et al., 2009). Finally, the disruption of codY completely abolishes virulence of subcutaneously injected B. anthracis spores in OF1 mice (Fig. 2). Although CodY-regulated virulence in B. anthracis is clearly established, mechanisms of action and regulation remain an open question.

Figure 2

CodY is required for anthrax toxin production in Bacillus anthracis. (a) Schematic representation of the genes encoding anthrax toxins lethal factor lef (brown), the protective antigen pagA (orange), the edema factor cya (yellow) and the toxin regulator atxA (blue) on the pXO1 plasmid. 1, AtxA activates the expression of lef, pagA and cya as well as the capsular genes located on pXO2. 2, Both the toxins mRNA level and the protein level of AtxA decreased in absence of CodY. (b) Cellular model of action of anthrax toxins. 3, The mature protective antigen (PA) binds to a receptor (not shown here) present in many cell types. PA is cleaved by host cell proteases on the surface of the targeted cell or in the serum of the host releasing PA20 whereas PA63 oligomerizes into a ring-shaped heptamer. The cleavage of PA is needed to expose binding sites for EF and LF. Without cleavage, the process is not toxic for the targeted cell. 4, Oligomerization triggers receptor-mediated endocytosis of the anthrax complex. 5, LF and EF bind to the complex. 6, Acidification of the endocytosolic vesicle allows the translocation of EF and LF, EF remaining associated with the vesicle whereas LF is released into the cytosol, both reaching their targets. EF, the adenylate cyclase edema factor converts intracellular ATP into cAMP. LF, the zinc protease lethal factor cleaves mitogen-activated protein kinases (MAPKKs) of the MAPK pathway that relays environmental signals to transcriptional machinery. There are also putative additional targets of LF. 7, The decrease of intracellular ATP and the cleavage of MAPKKs lead to cell death. (c) Spores of B. anthracis 7702 wild type and codY mutant were injected subcutaneously into female OF1 mice in order to estimate the half-lethal dose. Results shows that injection of 3.6 × 105 spores of wild type kills half of infected mice whereas even at 103-fold higher doses (>108) of spores of the codY mutant did not kill infected mice.

Clostridium difficile and antibiotic-associated diarrhea

Clostridium difficile is the principal agent of antibiotic-associated diarrhea leading to potentially lethal pseudomembranous colitis (Kelly & LaMont 1998; Dineen et al., 2007). Today, C. difficile is considered to be one of the most important causes of health care-associated infections (Rupnik et al., 2009). Antibiotic therapy alters the microbial community allowing colonization of the intestinal tract by C. difficile (Kelly & LaMont 1998) or simply permitting its growth. CodY is a repressor of toxin gene expression in C. difficile (Dineen et al., 2007). At the molecular level of C. difficile toxicity, the tcdR gene encodes a sigma factor allowing the expression of two toxins TcdA (enterotoxin A) and TcdB (cytotoxin B) responsible for gastrointestinal damage and the antibiotic-associated pseudomembranous colitis (Just et al., 1995a, b). Both toxins act by glycosylating members of the Rho family of small GTPases in host cells leading to the disaggregation of the microfilament cytoskeleton (Just et al., 1995a, b). The initial observation that isoleucine may play an important role in toxin production by C. difficile suggests a putative CodY-mediated expression of toxins (Ikeda et al., 1998). The mRNA levels of tcdR, tcdB, tcdE, tcdA, and to a lesser extent tcdC (all of which are colocalized on a pathogenicity locus), are all derepressed in a codY mutant of C. difficile, both during the exponential and the stationary phase (Dineen et al., 2007). CodY is able to bind the promoter region of tcdR, and the binding increases in the presence of BCAA and GTP in a gel shift assay. Footprinting experiments showed three protected regions in the promoter of tcdR. One is located between positions −281 and −309 and shows the highest affinity for CodY, the second between −348 and −382 and finally the last between −40 and −58 base pairs from the start codon (Dineen et al., 2007). The transcriptional start of tcdR was not characterized during the study. Observation of increased mRNA levels and CodY binding suggested CodY-regulated toxin production in C. difficile by direct binding to the tcdR promoter (Dineen et al., 2007). Thus, CodY indirectly represses the virulence factors encoded in a pathogenicity island through repression of the activator tcdR (Fig. 3).

Figure 3

Schematic representation of Clostridium difficile in the intestine and the function of its CodY-regulated toxins in pseudomembranous colitis. The bacterium is presented as a gray rectangle with a pathogenic island beginning with three CodY-binding motif and ending with tcdC gene. Three promoters (P) and two terminators (

Embedded Image
) are located in the pathogenicity island (not drawn to scale). Upon starvation, CodY detachment from the boxes located in front of the tcdR gene activates the transcription of tcdR, a positive regulator for toxin encoding genes, and TcdC, a negative regulator. By TcdR-mediated activation, C. difficile expresses both TcdA and TcdB. TcdA, also called enterotoxin A, is able to enter the intestinal epithelial cells apically and disrupt the tight junctions. The created opening between cells allows TcdB or cytotoxin B to attack cells located below the epithelium. Both toxins mediate the production by the host of tumour necrosis factor and proinflammatory interleukins, leading to increased vascular permeability, recruitment of immune cells (neutrophils and monocytes), apoptosis, detachment of epithelial cells and inflammation. For the human host, this process leads to pseudomembranous colitis. The function of TcdE is not clear but could be related to the export of toxins from the bacteria to the environment.

The CodY regulon in C. difficile was determined using microarrays comparing wild-type JIR8094 with its isogenic codY mutant (Dineen et al., 2010). As a result, 146 out of 3959 genes (3.7% of genes) were shown to be upregulated in the absence of CodY. Affinity purification of CodY–DNA complexes combined with Illumina DNA-sequencing technology led to the identification of 350 CodY–DNA-binding regions (8.8% of genes) (Dineen et al., 2010). Among the putative CodY-regulated transcripts, gel mobility shift assays and footprinting experiments confirmed an effective binding of CodY in front of the following genes: ilvC, CD2344, glgC and hisZ (Dineen et al., 2010).

Listeria monocytogenes and listeriosis

Listeria monocytogenes is a gram-positive facultative intracellular bacterium and the causative agent of human listeriosis characterized by meningitis, septicaemia and foetal death (McLauchlin 1987). CodY in L. monocytogenes is able to regulate both carbon and nitrogen assimilation, in response to both GTP and BCAA (Bennett et al., 2007). A mutation of relA, the gene responsible for the production of alarmones during the ST, leads to a decreased virulence phenotype in a murine infection model, whereas mutations of codY in the relA minus mutant can restore virulence (Bennett et al., 2007). Moreover, relA mutations prevent the derepression of the CodY regulon during the early stationary phase (Bennett et al., 2007). These results suggest that RelA probably renders CodY inactive by decreasing intracellular levels of GTP and that CodY represses virulence of L. monocytogenes in vivo; yet the deletion of codY leads to similar virulence as in the wild type, with a slightly delayed onset of disease (Bennett et al., 2007). Whereas the relA mutant affected considerably intracellular growth, the cody and cody relA mutants displayed only slight effects. But no differences occur between relA, codY and relA-codY mutants as compared with the wild type during the escape from the phagosome when monitored by transmission electron microscopy (Bennett et al., 2007). The links between CodY and RelA were demonstrated in different organisms. But how they influence virulence in L. monocytogenes remains poorly understood. Finally, CodY activates the agr-like system of L. monocytogenes, a two component regulatory system responsive to autoinducing peptide well studied in S. aureus (Bennett et al., 2007). Staphylococcus aureus agr is a highly complex regulatory circuit allowing the bacteria to sense the presence of neighbours, a phenomenon that has been named quorum sensing. Agr-like systems refer to two component signalling system demonstrating high levels of homology with the agr system of S. aureus, also found in L. monocytogenes (Recsei et al., 1986; Autret et al., 2003) (Fig. 4).

Figure 4

CodY and RelA control virulence in Listeria monocytogenes. (a) Schematic representation of RelA-dependent formation of alarmones in connection with GTP-dependent CodY activation regulating growth involved genes and virulence factors in L. monocytogenes. (b) 5 × 105Listeria monocytogenes wild type, relA mutant, codY mutant and codY relA double mutant were injected intravenously into females of MF1 mice. After injection, mice were observed during 96 h for signs of illness. Results show that signs of illness were retarded when the codY mutant was injected, no illness occurred after injecting the relA mutant not able to produce a ST, whereas illness was restored in the double mutant. The interpretation of these results suggested that the avirulence of the relA mutant could in part be explained by the continued repression of the CodY regulon whereas codY mutants independently of relA present a derepression of virulence.

Staphylococcus aureus and its virulence systems

Staphylococcus aureus is usually found in 25–30% of the population on the skin and in the anterior nostrils of healthy human carriers (Lowy 1998; Safdar & Bradley 2008). It can be responsible for highly diverse diseases varying from benign skin infections to deadly invasive infections depending on the strain, the host defences, the site of infection, and additional parameters (Uramatsu et al., 2010). Staphylococcus aureus is also well known for hospital-acquired infections (the so-called nosocomial infections) and is increasingly associated with antibiotic resistance and persistent infections (Chambers 2001). Different toxins and virulence factors are sometimes present in a strain-specific manner in S. aureus and were implicated in diverse clinical presentations including superantigens inducing toxic shock syndrome (Silversides et al., 2010), exfoliative toxins implicated in the staphylococcal scalded-skin syndrome (Painter et al., 2007), the bicomponent phage-related Panton–Valentine leukocidin responsible for necrotizing pneumonia and invasive skin and soft tissue infections (Vardakas et al., 2009), various haemolysins responsible for the lysis of erythrocytes and other host cells (Traber & Novick 2006). The expression of these toxins is highly regulated and some of the key global regulators are the accessory gene regulator (agr system) responsible for quorum sensing and the staphylococcal accessory regulator (sarA) (Herbert et al., 2010). According to recent studies, S. aureus CodY seems to be an upstream major regulator of the global regulators agr and sarA, with numerous interactions and interconnections taking place between these different systems (Majerczyk et al., 2007). Both sarA and agr system are targets of CodY repression according to a pull-down assay, and CodY represses the agr system as evidenced by three independent studies (Majerczyk et al., 2007; Pohl et al., 2009; Majerczyk et al., 2010). Indeed, deletion of codY in strains SA564 and UAMS-1 derepresses the levels of mRNA for the haemolytic α-toxin (hla) and δ-toxin (hld), as well as the accessory gene regulator (agr) (RNAII and RNAIII) (Majerczyk et al., 2007). Resulting from that derepression, the haemolytic activity of the culture supernatant of a S. aureus codY mutant towards rabbit erythrocytes increases compared with that of the wild type, but can be restored by trans-complementation (Majerczyk et al., 2007). These results suggest a CodY-mediated repression of the agr-dependant haemolytic activity during the exponential growth phase. CodY negatively regulates hla and RNAIII transcription whereas RNAIII positively regulates hla on a post-transcriptional level (Majerczyk et al., 2010). Interestingly, the derepression of the CodY regulon and the Agr activation appear when cells enter the stationary phase: CodY because of nutrient depletion whereas the Agr system is induced by the high concentration of bacteria (quorum-sensing effect). Under conditions of isoleucine limitation, the agr is prematurely activated by a derepression of CodY, probably reflecting an escape mechanism for the bacteria becoming more virulent under nutrient limiting conditions (Pohl et al., 2009). The agr system of S. aureus does not influence the transcript levels of CodY (Pohl et al., 2009).

Additional virulence factors, identified in two independent transcriptomic experiments using two different S. aureus backgrounds, appear CodY-regulated, such as the secreted lipase encoded by the geh gene, the catalase encoding gene katA, and the superoxide dismutases sodM or sodA (Pohl et al., 2009; Majerczyk et al., 2010). The capsular polysaccharides, encoded by the cap operon, were shown to enhance staphylococcal virulence in numerous animal models of infection (Tuchscherr et al., 2005). Deletion of codY results in a derepression of the cap operon, whereas a mutation of agr leads to the overexpression of cap, as reported by two independent laboratories (Pohl et al., 2009; Majerczyk et al., 2010). Direct binding of CodY to the promoter of the cap operon and within the agrA ORF were detected (Majerczyk et al., 2010). Secondly, fnbA and spa are activated by CodY independently of agr during the postexponential phase (Pohl et al., 2009). Both are cell-wall associated proteins implicated in the adhesion process needed for the early biofilm formation. In S. aureus, contradictory results were reported for the effect of CodY on biofilm formation depending on the strain background (Majerczyk et al., 2007; Tu Quoc et al., 2007).

The RSH is essential in S. aureus, but conditional rsh mutants were generated in S. aureus (Geiger et al., 2010). Staphylococcus aureus bacteria submitted to deprivation conditions for valine and leucine respond to the resulting stress with the induction of RSH-regulated genes. RSH dependent repression occurred independently of CodY (infB), whereas gene activation by RSH (ilvDBC-leuABC-ilvA, ilvE, brnQ1, SAHSC_02932) depends on BCAA-CodY. Thus, the interaction between CodY and RSH is required to activate some genes involved in the response to amino acid deprivation, connecting CodY to the RSH-involved ST in S. aureus (Geiger et al., 2010). Interaction between CodY and RSH was also studied in the context of virulence in animal models by injecting 3 × 107 CFU of S. aureus HG001 wild type, codY mutant, rsh mutant and rsh/codY double mutants into the tail vein of female BALB/c mice before recording the S. aureus CFU obtained from mouse homogenized kidneys (Geiger et al., 2010). Less CFU/kidney were obtained when mice were infected with S. aureus rsh mutant compared with mice infected with the other strains (Fig. 5). Thus, the decrease of CFU/kidney observed with the rsh mutant was suppressed in the codY rsh double mutant. To conclude, in S. aureus, CodY represses different virulence factors such as the cap operon and haemolysins as well as a broad range regulator of virulence (agr system), and behaves as a suppressor of rsh for mouse kidney colonization. Finally, S. aureus CodY was downregulated while growing the bacteria in the presence of Candida albicans that were forming a dual-species biofilm (Peters et al., 2010), but upregulated in presence of L. lactis in the cheese matrix (Cretenet et al., 2011).

Figure 5

Staphylococcus aureus CodY regulates virulence mechanisms. (a) Simplified schematic representation of CodY regulating virulence pathways in S. aureus. Arrows indicate positive regulation and lines ending in bars denote negative regulation. CodY, when activated by GTP and/or BCAA, represses the icaADBC operon involved in the production of the polysaccharide intercellular adhesin, a component of biofilm matrix. CodY also represses capsular genes, haemolysin A (hla), the staphylococcal accessory regulator A (sarA) and the agr operon as well as hld translated from the RNAIII gene. By repression of RNAII and RNAIII, CodY also indirectly affects quorum-sensing-regulated targets such as other haemolysins and adhesins that are positively regulated by RNAIII. (b) Schematic representation of experiments performed in an animal model. 3 × 107 CFU of S. aureus HG001 wild type, codY mutant, rsh mutant and rsh/codY double mutant were injected into the tail vein of female BALB/c mice using 10 mice for each isogenic strains resulting to the colonization by S. aureus of mice kidney (Geiger et al., 2010). CFU/kidney reported show lower CFU after injections of rsh mutant compared with wild type without significant effects for codY mutant and codY/rsh double mutant. Thus, rsh is required for kidney colonization of mice after artificial S. aureus infections occurring only in presence of CodY.

Streptococcus pyogenes and its virulence systems

Streptococcus pyogenes is considered the most pathogenic member of the genus Streptococcus. It is a human-specific pathogen causing high morbidity in many types of skin and upper respiratory tract infections. It is responsible for ‘strep throat’, scarlet fever, pyoderma, streptococcal toxic shock, septicaemia and necrotizing fasciitis (Hynes 2004; Malke & Ferretti 2007). The organism possesses a plethora of toxins differentially expressed and putatively explaining the great diversity of diseases caused by this pathogen. Despite years of study, many unanswered questions remain regarding the pathogenesis and virulence of group A streptococci. Sreptococcus pyogenes is able to switch phenotypes (e.g. immune evasion, adherence, internalization, persistence) when the organism responds to different environmental cues (e.g. elevated CO2, iron limitation, increased temperature, pressure, atmospheric conditions, blood components, reactive oxygen species) (Kreikemeyer et al., 2003).

Recently, S. pyogenes CodY was found to be involved in the regulation of virulence factors; these results could explain its adaptive behaviour in terms of virulence. In a recent approach, 51 arbitrary chosen genes were quantified by quantitative reverse transcriptase-PCR and compared between S. pyogenes strain NZ131 and its codY disrupted mutant when grown in a defined medium and in human blood (Malke et al., 2006; Malke & Ferretti 2007). Results show differences in CodY-mediated gene regulation when comparing both experiments, suggesting the action of biological cues which are missing in the laboratory medium, and adaptation through gene regulation according to the environment (Malke & Ferretti 2007). Interestingly, the S. pyogenes covRS and sptRS two-component regulatory systems appear increased in the absence of CodY when grown in human blood (Malke & Ferretti 2007). In the case of covRS, the upregulation decreases in a time-dependent manner, suggesting the adaptation of S. pyogenes to human blood (Malke & Ferretti 2007). CodY and CovR are required for the transcription of the dipeptide transporter dppA. The authors suggested that CodY may act as a coactivator of CovR, but a codY covR double mutant is required for deeper analysis of dppA transcriptional control (Gusa et al., 2007). The S. pyogenes activator fasX gene is responsible for aggressiveness towards the human laryngeal epithelial cell line HEp-2 (ATCC, CCL23) by promoting adherence, internalization, cytokine expression and release, and finally cell apoptosis (Klenk et al., 2005). This pleiotropic activation is downregulated in a codY mutant of S. pyogenes strain NZ131 grown in human blood suggesting that CodY activates fasX indirectly (Malke & Ferretti 2007). The pleiotropic virulence factor regulator Mga activates the virulence factors emm49 (a membrane-associated protein used for sequence typing in S. pyogenes), streptococcal collagen-like surface protein (scl), C5a peptidase genes (scpA) and a fibronectin-binding protein (sof) (Hynes 2004). Both the mga regulator and the previously mentioned virulence factors are underexpressed in the absence of CodY when growing S. pyogenes in human blood (Fig. 6a) (Malke & Ferretti 2007). In Todd–Hewitt medium, CodY regulates the virulence-regulatory systems encoded by pel/sagA and mga as well as the virulence factors scl, prts and scpA positively and in a growth-phase dependant manner (Malke et al., 2006). SagA is the streptolysin S precursor peptide responsible for lysis of red cells in blood-agar plates (Nizet et al., 2000; Malke et al., 2006). CodY also represses the S. pyogenes virulence factors grab which regulates proteolysis at the bacterial surface (Rasmussen et al., 1999). To conclude, the powerful toxic arsenal of S. pyogenes seems to be largely regulated by CodY, and the ability of the bacteria to switch phenotype depending on environmental cues may be related to CodY and to the CodY-connected two component systems covRS and sptRS.

Figure 6

CodY in Streptococci. (a) In vitro CodY-regulated virulence factors in Streptococcus pyogenes. Schematic representation of expression of virulence factors of S. pyogenes according to microarray comparing NZ131 wild type with a codY mutant that were grown in human blood (Malke & Ferretti 2007). Red arrows indicate up or down expression in the codY mutant compared with wild type. Note that even if some factors in the picture are shown at the protein level, the over or under expression was estimated at the mRNA level. 1, Regulators of virulence (the multigene regulator in group A streptococci mga, the regulatory RNA fasX stimulating streptococcal aggressiveness towards host cells by promoting host cell apoptosis, the streptolysin S associated proteins sagAB) are downregulated in absence of codY, whereas two-component systems (covRS and sptRS) are stimulated; 2, the fibronectin-binding protein (sof); 3, the collagen-binding protein (scl); 4, the M-like protein used for genotyping (emm49); and 5, the C5a peptidase gene anti-immune-chemotaxis (scpA) all of them were downregulated in absence of CodY. 6, The CAMP factor (cfa) responsible for light haemolysis and 7, the streptolysin O (slo) responsible for lysis of various cell types were downregulated in absence of codY. 8, The pyrogenic exotoxin H (speH) underexpressed in absence of codY is able to bind receptors present at the surface of T-cell leading to T-cell activation and production of interleukins and cytokines. 9, The IgG-degrading endopeptidase (ideS) is also downregulated in absence of codY. 10, The protein G-related α2 macroglobulin-binding protein (grab), upregulated in absence of codY, is able to inactivate host proteases protecting S. pyogenes surface proteins. Thus, many important virulent factors in S. pyogenes are CodY-regulated. (b) CodY and PcpA needed for adhesion of Streptococcus pneumoniae. Colonization of the nasopharynx is mediated by adherence of S. pneumoniae to the human epithelial cells in vivo. D39 wild-type strain was compared with ΔcodY mutant, ΔpcpA mutant and ΔcodYΔpcpA double mutant. The codY or pcpA mutants showed a decrease ability of S. penumoniae to adhere to nasopharyngeal cells, a process occurring in vivo during the development of pneumonia due to S. pneumoniae. CodY is thought to regulate the expression of pcpA, whereas the cell wall-associated choline-binding protein (pcpA) is needed in the process of adhesion of S. pneumoniae to human pharyngeal epithelial cells. (c et al., d) CodY needed for in vitro biofilm formation in Streptococcus mutans. (c) Schematic representation of the process leading to the formation of caries. Bacteria live in biofilm on the teeth. Streptococcus mutans is the most important bacterium in the formation of dental caries. Sugars provided by the diet are fermented by the bacteria producing lactic acid, but can also be used to produce polymeric matrix for the biofilm. The production of lactic acid results in an acidification of the surface of the teeth, which leads to a demineralization of the teeth resulting in caries. (d) Schematic representation of in vitro biofilm formation of S. mutans (blue round shape) in presence of glucose and sucrose comparing wild-type strain UA159 with ΔcodY, ΔrelAPQ and ΔcodYΔrelAPQ mutants. Mutation of codY drastically decreased the quantity of biofilm, suggesting that CodY is needed for S. mutans to produce biofilm in vitro.

Streptococcus pneumoniae and pneumonia

Streptococcus pneumoniae is a leading human pathogen responsible for pneumonia, but also meningitis and otitis media in young children, elderly people and immunocompromised patients (Bogaert et al., 2004). Each year, 1 million children younger than 5 years die from pneumonia and invasive disease (Obaro & Adegbola 2002). Depending on age, 30–60% of survivors of meningitis develop long-term sequellae including hearing loss, neurological deficits and neuropsychological impairment (Koedel et al., 2002). CodY was studied in S. pneumoniae and appeared to bind only to BCAA as cofactors and not GTP (Hendriksen et al., 2008). In a recent study, 9-week-old CD-1 mice infected with S. pneumoniae D39 and its codY mutant revealed that codY is required for nasopharynx and lung colonization by S. pneumoniae D39, an observation that is also confirmed when testing the adherence of S. pneumoniae to human nasopharyngeal cells (ATCC CCL-138) (Hendriksen et al., 2008). The authors suggested that a direct target of CodY named pcpA (a choline-binding protein) is implicated in the adhesion process of nasopharyngeal epithelium. Indeed, S. pneumoniae pcpA is CodY-regulated based on microarray studies and contains an AT-rich sequence resembling the CodY box (AATTTATAAAATGTA) (Sanchez-Beato et al., 1998; Hendriksen et al., 2008). CodY is considered an essential gene in S. pneumoniae strain R6 due to absence of successful transposon-mediated disruptions, whereas the disruption was possible after the insertion of a second copy of codY (Caymaris et al., 2010). Whole genome sequencing of strain D39 and its codY knock-out mutant revealed mutations in fatC and amiC. Mutating these genes allowed tolerance of codY inactivation (Caymaris et al., 2010). The authors suggest as possible explanation for this observation that inactivation of codY results in iron toxicity for the produced mutant compensated by the side mutations occurring into fatC and amiC (Caymaris et al., 2010).

Colonization of the nasopharynx is mediated by the adherence of S. pneumoniae to the human epithelial cells (Hendriksen et al., 2008). Encapsulated S. pneumoniae strains adhere to a lower extent compared with un-encapsulated strains (Bootsma et al., 2007). The implication of CodY and pcpA in the process of adhesion of S. pneumoniae to human pharyngeal epithelial cells in vitro was tested with ΔcodY and ΔpcpA mutations in a Δcps (type 2 capsule locus) deficient background (Hendriksen et al., 2008). Both, ΔcodY and ΔpcpA mutations were associated with decreased numbers of adherent bacteria (Fig. 6b) (Hendriksen et al., 2008). The strain that displayed the lowest ability to adhere on epithelial cells was the codY/pcpA double mutant (Hendriksen et al., 2008). Additionally, an S. pneumoniae colonization model in mouse showed that less bacteria are needed for persistent colonization in the codY mutant compared with the wild type (Hendriksen et al., 2008). The ami locus containing six ORFs encoding oligopeptide ATP-binding cassette (ABC) transporters was associated with adhesion of S. pneumoniae to eukaryotic cells (Cundell et al., 1995). Nevertheless, it is possible that the transported molecule and not the transporter itself are directly implicated in adhesion. Interestingly, amiA, amiC and amiD were 1.5 to three times derepressed in the ΔcodY mutant according to microarrays performed with S. pneumoniae D39 (Hendriksen et al., 2008). Similar observations were made with aliA, also encoding an oligopeptide ABC transporter that is derepressed in the ΔcodY mutant (Hendriksen et al., 2008). Moreover, aliA, aliB or ami were implicated in the nasopharyngeal colonization but not in the invasive disease (Kerr et al., 2004). Therefore, in S. pneumoniae, CodY seems to regulate the expression of ali and ami permeases as well as the choline-binding protein pcpA, all involved in the adhesion of the bacteria to the epithelial cells, a critical step for colonization before the infection.

Streptococcus mutans and dental caries

Streptococcus mutans are a predominant bacteria associated with oral biofilms and implicated in the formation of caries (Fig. 6c) and in the development of subacute infective endocarditis (Lemos et al., 2005; Lemos et al., 2008). The bacterium in the oral cavity must adapt to low pH conditions, fluctuations in nutrients and low levels of free amino acids (Battistone & Burnett 1961; Lemos et al., 2005). Streptococcus mutans colonizes the tooth surface first by a sucrose-independent mechanism during the initial attachment to the enamel pellicle and then by sucrose-dependant pathways involving the production of extracellular polymeric and sugaric compounds formed through the action of glycosyltransferase enzymes (Ooshima et al., 2001; Banas & Vickerman 2003; Lemos et al., 2008). In vitro experiments comparing S. mutans strain UA159 with its ΔcodY mutant have shown a reduced capacity of the mutant to form biofilms in the presence of both glucose and sucrose (Fig. 6d), and a decreased resistance to low pH (Lemos et al., 2008). These results suggest that CodY is probably needed in S. mutans for caries formation, but this hypothesis has to be tested directly on teeth.

In S. mutans, in addition to RelA, the RelP and RelQ enzymes are responsible for the production of (p)ppGpp alarmones (Lemos et al., 2007, 2008). RelA is the major enzyme controlling the ST, whereas RelP produces (p)ppGpp basal levels under unstressed conditions. The action of RelQ is poorly understood (Lemos et al., 2007, 2008). A recent study revealed that in S. mutans, RelP and RelQ are required for the synthesis of basal levels of (p)ppGpp needed to grow under leucine and valine starvation conditions, whereas deletion of codY abolished growth deficiency, suggesting an interdependence of the ST and the CodY-regulon. A ΔrelAPQ mutant of S. mutans unable to produce alarmones and ST does not grow in minimal medium lacking leucine and valine, but can grow if isoleucine is also omitted. Interestingly, deletion of codY in this mutant restores growth, suggesting that the growth-defect is linked to CodY and that basal levels of alarmones are required for growth, in particular in amino acids depleted media (Lemos et al., 2008).

Concluding remarks and outlook

The striking importance of CodY in low GC gram-positive bacteria for adaptation to starvation conditions and for the control of virulence raises the question of which factor could replace CodY function in gram-negative bacteria? CodY has been compared with leucine-responsive regulatory protein (Lrp) of the gram-negative E. coli regarding the aspects of DNA binding, dimerization, repression of operons involved in BCAA biosynthesis and amino acids binding (Guedon et al., 2005). Indeed, Lrp is able to repress ilvIH operon by binding to DNA, upstream of the ilvIH (Willins et al., 1991). However, while CodY affinity for DNA increases after cofactor binding, Lrp is removed from DNA in the presence of leucine. Moreover, Lrp and CodY do not share any structural similarity (Molle et al., 2003; Shivers & Sonenshein 2004; Guedon et al., 2005). Regarding structural aspects, CodY was compared with Agrobacterium tumefaciens TraR, a quorum-sensing protein that responds to a homoserine lactone pheromone (Levdikov et al., 2006). The structure of TraR was solved in complex with its cofactor and with a targeted DNA box (Zhang et al., 2002). But TraR binds homoserine lactone cofactor and belongs to a two-component system whereas CodY binds BCAA and GTP and, to our knowledge, is not involved in a two-component system, rendering the comparison valuable only for some structural aspects (Levdikov et al., 2006).

While the activation of CodY by BCAA binding and dimerization process is well understood on a mechanistic and structural level, corresponding data for GTP binding and DNA binding are missing, leading to partial or hypothetical explanations of CodY-GTP binding and CodY-DNA binding (Levdikov et al., 2006; Levdikov et al., 2009). It is not known whether each CodY monomer recognizes half of the symmetrical DNA-binding site, whether GTP effectively binds G1, G3 and G4 motifs, whether transcription factors are able to regulate codY expression and how codY is expressed during time-course studies.

Finally, the recently acquired knowledge of CodY regulation of virulence opened new putative approaches for therapy: for example the use of BCAA as additive drugs against infections by gram-positive pathogens could putatively keep the bacteria in a nonpathogenic status. Investigation of clearance of gram-positive pathogens from infected animal models with BCAA injected to the infection site may lead to promising future therapeutic approaches, notably in conjunction with antibiotic treatment. It has already been proposed to supplement BCAA in the mucosa against S. aureus colonization targeting the regulation of CodY activity (Camargo & Gilmore 2008).


We are indebted to Dr Vladimir Lazarevic for helpful discussions. Work in the laboratory of the authors was supported by grants from the Swiss National Science Foundation 3100A0-112370/1 (to J.S.), 3100A0-116075 (to P.F.), and 31003A-118309 (to P.L.). The work was supported by grants to C.W. from the Deutsche Forschungsgemeinschaft, TR34.


  • Editor: Patrik Bavoil


View Abstract