Regulation of Enterotoxins Associated with Bacillus cereus Sensu Lato Toxicoinfection

ABSTRACT

Bacillus cereus sensu lato (s.l.) includes foodborne pathogens, as well as beneficial microorganisms, such as bioinsecticides. Some of the beneficial and commercially used B. cereus s.l. strains have been shown to carry enterotoxin genes, the products of which can cause toxicoinfection in humans. Furthermore, recent epidemiological reports indicated that some bioinsecticidal strains have been linked with foodborne illness outbreaks. This demonstrates the need for improved surveillance of B. cereus s.l., which includes characterization of isolates’ virulence capacity. However, the prediction of virulence capacity of B. cereus s.l. strains is challenging. Genetic screening for enterotoxin gene presence has proven to be insufficient for accurate discrimination between virulent and avirulent strains, given that nearly all B. cereus s.l. strains carry at least one enterotoxin gene. Furthermore, complex regulatory networks governing the expression of enterotoxins, and potential synergistic interactions between enterotoxins and other virulence factors make the prediction of toxicoinfection based on isolates’ genome sequences challenging. In this review, we summarize and synthesize the current understanding of the regulation of enterotoxins associated with the B. cereus s.l. toxicoinfection and identify gaps in the knowledge that need to be addressed to facilitate identification of genetic markers predictive of cytotoxicity and toxicoinfection.

BACILLUS CEREUS SENSU LATO AND ITS FOODBORNE PATHOGENS

Bacillus cereus sensu lato (s.l.) encompasses spore-forming, clinically and agriculturally relevant species. These include Bacillus cereus sensu stricto (s.s.) (1–4), B. thuringiensis (recently classified as B. cereus s.s. biovar Thuringiensis) (4, 5), B. anthracis (recently classified as B. anthracis biovar Anthracis) (4, 6), among others (4, 7–18). B. cereus s.l. species can be classified in eight phylogenetic groups (19) that are consistent with eight genomospecies (4). Some B. cereus s.l. strains cause self-limiting emetic or diarrheal illness, while others have been reported to cause deadly anthrax (20), meningitis (21), and cutaneous infections (1, 22). In contrast, the group also contains beneficial strains, including plant growth promoters, probiotics, and biopesticides (e.g., B. cereus s.l. biovar Thuringiensis) (23–29). PCR screens and recent whole-genome sequence analyses have revealed that some strains previously considered beneficial and safe carry enterotoxin genes that can facilitate toxicoinfection in humans, if expressed (30, 31). Furthermore, B. cereus s.l. isolates implicated in foodborne outbreaks were found to be genetically nearly indistinguishable from commercially used bioinsecticidal strains (32). These findings have brought commercially used B. cereus biovar Thuringiensis strains and probiotic strains under increased scrutiny due to a potential food safety risk (31–37). These findings have also demonstrated the need for a strain-based B. cereus s.l. characterization and risk assessment.

Moving toward strain-based characterization and risk assessment, a novel taxonomic framework was proposed for B. cereus s.l. to accurately reflect the pathogenic potential of isolates within B. cereus s.l. while conserving genomic species definitions (4). The existing B. cereus s.l. species were merged into eight genomospecies based on the average nucleotide identity of 92.5%, and into three biovars (Emeticus, Anthracis, and Thuringiensis) defined based on the detection of genes associated with their respective phenotypes. The biovar Emeticus is identified based on the presence of cereulide synthetase-encoding ces genes associated with foodborne intoxication; the biovar Anthracis is identified based on the presence of anthrax toxin genes cya, lef, and pagA; and the biovar Thuringiensis is identified based on the presence of insecticidal crystal protein-encoding genes cry, cyt, and vip (4). Notably, a biovar for isolates that can cause diarrheal toxicoinfection does not yet exist because the presence of diarrheal enterotoxin genes is insufficient to predict cytotoxicity toward human cells and associated toxicoinfection in humans.

Strains that are able to cause diarrheal toxicoinfection must encode and express enterotoxins and potentially other virulence factors in the human gut (38). Characterized enterotoxins that contribute to virulence include hemolysin BL, nonhemolytic enterotoxin, and cytotoxin K (39–42). Nearly all B. cereus s.l. strains possess at least one enterotoxin gene (3, 31, 43–47). However, enterotoxin genes are not always expressed and, therefore, do not demonstrate cytotoxicity toward human cells.

Enterotoxin production is one of many factors that contributes to the complex virulence of B. cereus s.l. strains. The differences in virulence among B. cereus s.l. group strains can also be attributed to their ability to sporulate and survive passage through the host gastrointestinal tract, germinate in the intestine, and compete with the host microbiome. Recent studies and reviews discuss these factors in detail (44, 48, 49). Here, we review the current understanding of the regulation of B. cereus s.l. enterotoxin expression and identify gaps in the knowledge that need to be addressed to facilitate identification of genetic markers predictive of cytotoxicity and toxicoinfection.

ENTEROTOXINS AND OTHER VIRULENCE FACTORS ASSOCIATED WITH B. CEREUS S.L. TOXICOINFECTION

The main B. cereus s.l. diarrheal enterotoxins, including hemolysin BL (Hbl), nonhemolytic enterotoxin (Nhe), and cytotoxin K (CytK) are ribosomally synthesized, pore-forming enterotoxins (50–52). They are expressed and exported from cells primarily during late exponential phase and early stationary phase (53–55).

Hemolysin BL.

Hbl is a tripartite toxin composed of a binding component (Hbl-B) and two lytic components (Hbl-L₁ and Hbl-L₂). These subunits are encoded by hblA, hblD, and hblC genes, respectively, and bind in the given order (56). All three of these enterotoxin components are required for maximal cytotoxicity (56, 57). Host cell binding of Hbl components is mediated primarily by (LPS)-induced TNF-α factor (LITAF), a host cell receptor upregulated in response to bacterial infection and responsible for cytokine production (58, 59). Hbl acts by forming pores in the host cell membrane that result in the efflux of potassium. This leads to activation of the NLRP3 inflammasome, secretion of interleukin-1β and interleukin-18, and pyroptosis (60).

Nonhemolytic enterotoxin.

Nhe is a tripartite toxin composed of C, B, and A subunits. These subunits are encoded by nheC, nheB, and nheA, respectively, and subunit C initiates binding to the host cell (61). Nhe acts similarly to Hbl; it forms pores in the host cell membrane, which causes potassium efflux, activation of the NLRP3 inflammasome, and cell death. However, host surface receptor targets have not yet been identified for Nhe (62). Though Hbl and Nhe have similar mechanisms of action, recent studies on the inheritance patterns and structural differences of Hbl and Nhe suggest that they have distinct targets (63).

Cytotoxin K.

CytK is a single component, β-barrel-forming toxin, first identified as a toxin in an outbreak of severe gastroenteritis causing the death of three patients in France in 1998 (40). There are two variants of CytK: CytK-1 that is specific to B. cytotoxicus, and CytK-2 that has been found in other B. cereus s.l. species (3, 64). CytK-1 has been shown to be five times more toxic than CytK-2 to CaCo-2 and Vero cells (65). Unlike for Hbl a host cell receptor target has not been identified for CytK.

Additional toxin and nontoxin virulence factors may contribute to cytotoxicity of B. cereus s.l. by acting synergistically with the aforementioned enterotoxins. Other putative diarrheal illness virulence factors include single protein enterotoxin FM (EntFM), phosphatidylinositol-specific phospholipase C (PI-PLC), sphingomyelinase (SMase), immune inhibitor A1 (InhA1), neutral protease (NprA or NprB, a metalloprotease bacillolysin), hemolysin I and II (HlyI and HlyII), and a putative enterotoxin (EntABC) (66–72).

Though these virulence genes have been associated with disease in humans (11, 22, 40, 41, 73), the sole presence of enterotoxin genes is poorly associated with their expression, as well as with cytotoxicity of isolates’ secretomes (3, 43–45, 74, 75). This may be attributed to the existence of yet uncharacterized virulence factors, synergistic activity between multiple virulence factors (49, 68, 76, 77), or factors that affect the regulation of enterotoxins or other virulence factors (77–79).

REGULATION OF ENTEROTOXIN PRODUCTION

B. cereus s.l. diarrheal enterotoxins are produced in response to changes in quorum sensing signals (PlcR), nutrient availability (CodY, CcpA, and SinR), and oxygen availability (Fnr, ResDE, OhrRA). These environmental cues signal that the bacterium has made its way to a host gut. The complex interconnectedness of enterotoxin gene expression, demonstrated in Fig. 1, remains a challenge to understanding B. cereus diarrheal toxicoinfection and identifying accurate, precise genetic markers of cytotoxicity. The following section details the current understanding of each of the identified transcriptional regulators.

FIG 1.

Experimentally confirmed effects of transcriptional regulators on enterotoxin gene expression. Green arrows demonstrate activation. Red blunt-ended lines demonstrate repression. Hypothesized effects of transcription regulators were excluded from the figure if the effects on protein production had not yet been experimentally confirmed. Transcriptional regulators are grouped by the stimulus to which they correspond. Nhe and Hbl are regulated by the same transcriptional regulators. CytK is regulated by fewer known transcriptional regulators than Nhe and Hbl. There is an interplay between transcriptional regulators such that some directly and indirectly regulate enterotoxin production by controlling other enterotoxin regulators (e.g., ResD and Fnr activate each other; CodY activates PlcR).

One of the most recent and extensive studies of enterotoxin regulation was reported by Böhm et al., who analyzed the untranslated regions upstream of hbl, nhe, and cytK operons in silico (79). They identified putative binding sites upstream of nhe and hbl genes for PlcR, CodY, ResD, SinR, and Fnr transcriptional regulators. Furthermore, they reported potential binding sites for CcpA upstream of nhe and within the coding sequence of hbl. The identified putative binding sites upstream of cytK-1 included PlcR, SinR, and Fnr. The only binding site identified upstream of cytK-2 was for PlcR (79).

Böhm et al. expanded upon their in silico results using in vitro methods. They deleted segments of the upstream untranslated regions to determine if the putative sites bind to repressors or activators of enterotoxin production and to confirm where the binding occurs. Deletion of segments upstream of nhe decreased the production of Nhe (79). They concluded that the presence of the entire untranslated region upstream of nhe is necessary for maximal Nhe production. In contrary, deletion of putative binding sites for CodY, ResD, SinR, and Fnr led to increased expression of Hbl (79). This finding demonstrated that although Nhe and Hbl have binding sites for many of the same transcriptional regulators, they are differentially regulated. There are a few possible explanations for this differential regulation: (i) the same transcriptional regulators may bind upstream of Nhe as activators, but bind as repressors upstream of Hbl; (ii) repressors may bind more strongly upstream of Hbl than Nhe, or activators may bind more strongly upstream of Nhe, increasing the efficiency of regulation; (iii) unidentified transcriptional regulators may exist upstream of either Hbl or Nhe. The second explanation seems likely, as electrophoretic mobility shift assays demonstrated different affinities of CodY for binding sites upstream of nhe and hbl (79). Historically, PlcR was thought to be the main control system for enterotoxins, but this experiment proved that Nhe and Hbl expression is significantly altered by factors other than PlcR. This finding is well supported by other targeted studies that are detailed below.

Quorum sensing.

The PlcR-PapR quorum-sensing system regulates the expression of approximately 80% of B. cereus secreted proteins, including enterotoxins and other virulence factors, through quorum sensing (55, 80). The transcriptional regulator PlcR is activated upon binding the signaling peptide PapR (Fig. 2) (80). PapR must be secreted and undergo extracellular maturation via interaction with NprB, which cleaves the original 48-amino acid PapR to create a mature C-terminal heptapeptide (81–83). Mature PapR is then imported through the oligopermease OppABCDF and recognized as a signaling molecule that activates PlcR (84). This occurs by binding a tetratricopeptide repeat-type regulatory domain on PlcR and inducing a conformational change to facilitate DNA binding (80). The PlcR activation peaks 1 h to 2 h after the onset of stationary phase (54).

FIG 2.

Mechanisms of regulation for known enterotoxin gene transcriptional regulators. The transcription factor binding site is denoted by the small dark gray box labeled with “TFBS” in A. An enterotoxin gene is represented by the large light gray box labeled with “enterotoxin gene” in A. The enterotoxins with putative binding sites for each transcriptional factor are listed in Table 1. Transcriptional regulators are grouped by the stimulus to which they correspond. (A) The PlcR-PapR quorum sensing system requires the binding of activator PlcR and a small peptide PapR to increase transcription of enterotoxin genes. Without PapR, PlcR does not bind upstream of enterotoxin genes, resulting in decreased transcription. (B) The repressor CodY dimerizes in the presence of GTP and branched-chain amino acids. CodY dimers bind upstream of enterotoxin genes and decrease transcription. In the absence of GTP and branched-chain amino acids, CodY does not dimerize and does not bind DNA. (C) In the presence of glucose, the repressor CcpA binds upstream of enterotoxin genes and decreases transcription. When glucose concentrations decrease, CcpA does not bind DNA. (D) The repressor SinR forms tetramers in response to nutrient depletion and binds upstream of enterotoxin genes to block RNA polymerase binding. When the antirepressor SinI is present, SinR is bound by SinI, preventing DNA binding and repression. (E) HoloFnr is an activator of enterotoxin genes under anaerobic conditions, where HoloFnr contains one [4Fe-4S]²⁺ cluster per monomer. In the presence of oxygen, Fnr is oxidized to ApoFnr and forms inactive covalent dimers. (F) The transcription regulator ResD binds to Fnr and enterotoxin gene promoters as an activator when phosphorylated by ResE under anaerobic conditions. Under aerobic conditions, ResD is not phosphorylated by ResE and binds Fnr and the promoter as a repressor. (G) In aerobic conditions, OhrR forms a covalent dimer that strongly binds enterotoxin gene promoters as a repressor. In anaerobic conditions, OhrR forms a noncovalent dimer, and the DNA binding is weakened, leading to decreased repression of transcription. OhrR is a repressor both when reduced and when oxidized.

PlcR-PapR system is responsible for overcoming obstacles associated with growing in the host gut. PlcR regulates its own production by controlling nprB, papR, and plcR expression (54, 55). The PlcR-PapR system also controls genes related to environment-sensing, nutrient availability, and self-defense, including the virulence genes nhe, hbl, cytK, smase, plcA (PI-PLC), plcB (PC-PLC), and inhA2 (54, 55, 77). The PlcR binding sites upstream of hbl, nhe, and both cytK variants are highly conserved sequences, suggesting the importance of its role in enterotoxin regulation (79). The PlcR-PapR system has been confirmed to be associated with cytotoxicity, as the deletion of PlcR in B. thuringiensis strain 407 decreased cytotoxicity on CaCo-2 cells by 80% (85). Additionally, inhibiting the PlcR-PapR system with synthetic, nonfunctional PapR peptide derivatives caused loss of hemolysis in B. cereus s.s. and B. thuringiensis strains (86). Therefore, the PlcR-PapR system is crucial for B. cereus hemolysis, and the system may be an excellent target for antivirulence compounds.

B. anthracis is the only B. cereus s.l. subspecies (4) known to not use PlcR in gene regulation. B. anthracis strains have a conserved nonsense mutation in the plcR gene, which leads to production of nonfunctional PlcR (87). Experiments expressing functional PlcR-PapR via plasmids in B. anthracis increased the transcription and expression of virulence factors including Nhe, PI-PLC, InhA, and Smase (88–90). An anthrolysin-deficient B. anthracis strain containing plasmids with functional PlcR-PapR systems was 50% more cytotoxic against lethal toxin resistant macrophages than wild type (88). Furthermore, the heat-treated supernatants were not cytotoxic (88). Therefore, heat-labile diarrheal enterotoxins likely caused the cytotoxicity against macrophages in the functional PlcR-PapR B. anthracis strain. These studies further implicate the PlcR-PapR system as necessary for enterotoxin production and cytotoxicity toward human cells.

The activation efficiency of quorum sensing in microorganisms is dependent on the sequence polymorphism of the quorum sensing system, which results in different communication groups (i.e., pherotypes) (91). In B. cereus s.l. specifically, four pherotypes of PlcR-PapR systems have been documented (82). Each pherotype of PlcR is best stimulated by its corresponding PapR, demonstrating that quorum sensing through PlcR-PapR is strain-specific (82). Despite the involvement of the PlcR-PapR system in enterotoxin production, the role of PlcR-PapR pherotypes in cytotoxicity has yet to be characterized.

Nutrient deprivation.

Transcriptional regulators that respond to nutrient deprivation, including CodY, CcpA, and SinR, also affect the production of enterotoxins (Fig. 2). CodY is a global pleiotrophic repressor known to regulate B. cereus s.l. enterotoxin and virulence gene expression based on nitrogen and carbon availability (92). CodY specifically responds to intracellular concentrations of branched-chain amino acids and GTP (93). When GTP and branched-chain amino acids are readily available, amino acids or GTP bind to CodY, causing dimerization and a conformational change that activates DNA binding (Fig. 2) (92, 93).

Several putative CodY repressor binding sites are located downstream of the hbl and nhe gene transcription start sites, suggesting that CodY suppresses the transcription of hbl and nhe by acting as a roadblock for RNA polymerase (79, 93). Electrophoretic mobility shift assays confirmed that CodY binds strongly upstream of hbl in B. cereus s.s. and weakly upstream of nhe in B. cereus s.s. and B. cytotoxicus (79). No binding was identified upstream of cytK1 or cytK2, which is consistent with in silico transcriptional regulator binding site identification (79). Two putative CodY binding sites upstream of nhe and one confirmed binding site upstream hbl were identified in approximately 100% and 70% of the 142 B. cereus s.l. strains tested, respectively (79). However, the binding site sequences were highly varied (79). Differences in CodY binding site sequences may be important for fine-tuning Hbl and Nhe regulation.

Though CodY directly downregulates hbl and nhe transcription as a repressor, CodY also upregulates enterotoxin transcription through the repression of the PlcR-PapR system. Three separate studies detail the effect of CodY mutants on plcR expression and enterotoxin-mediated cytotoxicity. Frenzel et al. found that in B. cereus s.s. F4810/72, the overexpression of CodY significantly increased cytotoxicity toward Vero cells compared to wild type, and the deletion of codY significantly decreased Nhe production and cytotoxicity (94). These findings agree with two similar studies by Lindback et al. who used transcriptomics and cytotoxicity assays to show that the deletion of codY in B. cereus s.s. ATCC 14579 caused decreased transcription of genes in the PlcR regulon (including hbl, nhe, and cytK) and halved cytotoxicity to Vero cells during late exponential phase (92, 95). In 2016, Slamti et al. determined the mechanism through which CodY affects the PlcR-PapR system. They showed that in B. thuringiensis 407 CodY activates the expression of components of the oligopeptide permease OppABCDF and other Opp-like proteins that are responsible for the import of PapR (96). The import of PapR upregulates PlcR expression through quorum sensing, and in turn, enterotoxin production increases (84). Considering these findings in the context of the Böhm et al. 2016 study (79), CodY serves two roles: (i) the direct repression of enterotoxin genes until cells are sufficiently starved of nitrogen and energy to warrant enterotoxin production, and (ii) the indirect upregulation of enterotoxin production through the PlcR-PapR system. Both functions are crucial for effective pathogenesis in the host gut. It is yet to be determined whether PlcR-PapR-mediated upregulation of enterotoxin expression is weakened by CodY repression in nutrient-rich environment and whether nutrient deprived environment can induce expression of enterotoxins when B. cereus s.l. population concentration is low.

Enterotoxins are also regulated by CcpA, or catabolite control protein A, which is a global transcriptional regulator that responds to carbon availability. CcpA primarily optimizes carbon metabolism through the repression of catabolic genes and activation of carbon export genes. CcpA also represses hbl and nhe expression in the presence of glucose (Fig. 2). A study that intended to investigate a separate enterotoxin regulator, Fnr, demonstrated that hbl and nhe transcription is attenuated in the presence of glucose when compared with fructose and sucrose (by approximately 10-fold and 20-fold, respectively) in wild type B. cereus s.s. F4430/73 (97). Glucose is not readily available in the host gut, resulting in upregulation of enterotoxin production. A ccpA deletion mutant increased the transcription of Nhe by 20-fold and of Hbl by 8-fold during early stationary phase (98). These findings directly associated CcpA with the known phenomenon of carbon deprivation increasing enterotoxin production.

Nutrient deprivation can also activate sporulation regulator Spo0A and have downstream effects on enterotoxin regulators SinR and SinI (99). The SinR transcriptional repressor and SinI antirepressor regulate motility, biofilm formation, and enterotoxin expression in B. cereus (99–102). When SinR alone is present, cells exhibit swimming motility, but when SinI is also present, SinI binds to SinR to prevent its activity and stimulate biofilm formation (Fig. 2) (100, 101). Böhm et al. identified conserved SinR binding sites upstream of nhe, hbl, and cytK-1 (79). The SinR binding sites of nhe and hbl are located downstream of the transcription start site, which is characteristic of a repressor. The deletion of sinR in B. thuringiensis strain 407 increased the transcription and production of Hbl, confirming that SinR is a repressor of enterotoxin expression (102). Oddly, hbl transcription decreases when B. thuringiensis cells form single-species biofilms, which is likely due to heterogenous expression of SinI based on nutrient availability and location in the biofilm (102). Additional experiments are necessary to confirm if SinR also represses nhe and cytK-1, which are hypothesized to have SinR binding sites, and whether SinI and SinR affect cytotoxicity.

Oxygen availability.

Other enterotoxin transcriptional regulators, including Fnr, ResDE, and OhrRA, upregulate enterotoxins in respond to oxygen deprivation (Fig. 2). A recent review discussed these transcriptional regulator mechanisms in detail (103). Fnr, or fumarate and nitrate reductase, is a global transcription factor that helps cells adapt to anoxic environments. Fnr deletion was shown to strongly reduce the transcription of nhe and hbl in B. cereus s.s. F4430/73, while plcR transcription was unchanged (97). In the same strain, the transcription of nhe and hbl was highest in fermentative environments, then aerobic environments, and lastly, anaerobic nitrate environments (104). Based on this finding, it was hypothesized that NO, which is produced by human cells for vasodilation (105), decreases the affinity of Fnr for DNA. The deletion of fnr in B. cereus s.s. F4430/73 leads to an extreme decrease in hbl (1,000- to 10,000-fold) and nhe transcription (5- to 150-fold) under microaerobiosis, and a less extreme decrease in hbl (50-fold, for both) and nhe transcription (100-fold and 10-fold, respectively) in aerobic and anaerobic nitrate environments (97, 104). The fold change was also different depending on the carbon source used as a growth substrate (97). These findings implicate Fnr as a necessary activator of nhe and hbl, and activators require direct contact with RNA polymerase to affect transcription. However, Böhm et al. 2016 identified an Fnr binding site downstream of the transcription start site, which is uncharacteristic of an activator (79). Further studies are therefore necessary to elucidate the mechanism of nhe and hbl activation by Fnr. The answer may lie in an Fnr-ResD-PlcR complex, described below.

The ResDE two-component system works with Fnr to regulate enterotoxin production in response to anoxic environments. ResE is a membrane-bound histidine kinase that is most active under anaerobic conditions. ResD is a cytoplasmic response regulator that is phosphorylated by active ResE (Fig. 2). In B. cereus s.s. F4430/73, unphosphorylated ResD can bind upstream of nhe, hbl, plcR, fnr, and resDE (106). The phosphorylation of ResD increases the binding affinity of ResD upstream of only resDE and fnr in vitro (106). ResD also binds directly to Fnr protein regardless of the phosphorylation status (106). In vitro, phosphorylated ResD binds to Fnr as a coactivator of nhe and hbl, and unphosphorylated ResD binds to Fnr as an antiactivator of nhe and hbl (106). This fortifying mechanism allows B. cereus to massively upregulate enterotoxin gene regulation only in environments with low oxidation-reduction potential, like the human small intestine. Interestingly, ResD does not need to be phosphorylated to activate plcR even though plcR upregulates enterotoxin gene expression (106). To explain this seemingly contradictory regulation, it has been postulated that PlcR may form a regulatory ternary complex with ResD and Fnr (107). However, further studies are required to fully confirm the formation and role of a potential Fnr-ResD-PlcR complex in vivo.

The OhrRA proteins are another form of dual-component redox-sensing enterotoxin regulator. OhrA, or organic hydroperoxide resistance protein A, is a thiol-dependent peroxidase protein that detoxifies organic hydroperoxides in the cell. OhrR is a transcriptional regulator of B. cereus enterotoxins, virulence, and metabolic pathways (77). OhrR forms a noncovalent dimer when reduced and a covalent dimer between cysteines when oxidized. DNA can be bound in both forms, but stronger repression occurs for most genes when OhrR is reduced in anaerobic environments. In B. cereus s.s. ATCC 14579, there are putative binding sites of OhrR upstream of hbl, nhe, cytK, hlyI, hlyII, entABC, sinR, and codY; however, no OhrR binding sites have been identified for entFM (108). Transcriptomics and proteomics experiments of ohrR deletion strains confirmed that OhrR acts as a repressor of hbl, nheAB, entFM, and entAC (108). Interestingly, OhrR serves as an activator for only entB, but it is unknown why (108). There was not a significant change in cytK transcription between ohrR deletion and wild-type strains (108). While OhrR regulates transcription of genes, OhrA may also posttranscriptionally regulate the level of enterotoxins exported. For example, it is hypothesized that cysteine-containing NheC may be saved from oxidation and loss of function by organic peroxides when OhrA is present to detoxify them. The deletion of OhrA and OhrR individually increased cytotoxicity to CaCo-2 cells, demonstrating that both proteins have an active role in regulating virulence (108). Additional studies must be performed to confirm the role of OhrRA in the regulation of sinR and codY.

CONCLUSIONS

Advances in whole-genome sequencing have helped gain insight into Nhe, Hbl, and CytK regulation. However, indirect enterotoxin regulation, the large number of transcriptional regulators involved, and potential posttranslational enterotoxin modifications all complicate predictions of B. cereus s.l. isolates’ cytotoxicity and toxicoinfection potential. Many binding sites have been identified through WGS analyses, but not all regulators have been experimentally confirmed to significantly regulate enterotoxin production (Table 1). Some enterotoxin transcriptional regulators such as CodY also control the transcription of other enterotoxin regulators such as PlcR, thus indirectly regulating enterotoxins (Fig. 1). Therefore, studies limited to analyzing transcriptional regulator binding sites upstream of enterotoxin genes will likely miss crucial indirect regulation. For example, binding sites for CodY have not been identified upstream of cytK, but the deletion of CodY significantly altered the regulation of cytK. In a study that focuses solely on investigation of sequences located upstream of enterotoxin operons, this would not have been identified.

TABLE 1.

Summary of enterotoxins and their transcriptional regulators

Enterotoxin	Transcriptional regulator	Putative binding site upstream ina	Regulation experimentally confirmed ina
Nhe	PlcR	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. 14579 (Gohar et al. 2002);
	CodY	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. INRA C3 and B. cytotoxicus CVUAS 2833 (Böhm et al. 2016); B. cereus s.s. 14579 (Lindbäck et al. 2012)
	CcpA	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. 14579 (van der Voort et al. 2008)
	SinR	Representatives from all clades (Böhm et al. 2016)	-
	Fnr	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. F4430/73 (Zigha et al. 2007; Messaoudi et al. 2010)
	ResDE	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. F4430/73 (Esbelin et al. 2009)
	OhrR	B. cereus s.s. 14579 (Clair et al. 2013)	B. cereus s.s. 14579 (Clair et al. 2013)
Hbl	PlcR	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. 14579 (Gohar et al. 2002)
	CodY	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. INRA C3 (Böhm et al. 2016); B. cereus s.s. 14579 (Lindbäck et al. 2012)
	CcpA	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. 14579 (van der Voort et al. 2008)
	SinR	Representatives from all clades (Böhm et al. 2016)	B. thuringiensis 407 (Fagerlund et al. 2014)
	Fnr	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. F4430/73 (Zigha et al. 2007; Messaoudi et al. 2010)
	ResDE	Representatives from all clades (Böhm et al. 2016)	B. cereus s.s. F4430/73 (Esbelin et al. 2009)
	OhrR	B. cereus s.s. 14579 (Clair et al. 2013)	B. cereus s.s. 14579 (Clair et al. 2013)
CytK	PlcR	cytK-1 and 2, Representatives from all clades (Böhm et al. 2016)	cytK-2, B. cereus s.s. 14579 (Gohar et al. 2002)
	CodY	-	cytK-2, B. cereus s.s. 14579 (Lindbäck et al. 2012)
	CcpA	-	-
	SinR	cytK-1, B. cytotoxicus (Böhm et al. 2016)	-
	Fnr	cytK-1, B. cytotoxicus (Böhm et al. 2016)	-
	ResDE	-	-
	OhrR	cytK-2, B. cereus s.s. 14579 (Clair et al. 2013)	-

^a-, not determined.

There is still much work to be done to fully understand enterotoxin regulation and B. cereus s.l. virulence. Genetic studies are often limited to B. cereus s.s. strains (Table 1). This is unlikely to be representative of strains from all phylogenetic groups, as cytotoxicity varies greatly among strains and phylogenetic groups. Knockout studies of transcriptional regulators rarely discuss the effects of the knockout on CytK or transcriptional regulator production (Table 1). When possible, knockout studies should include phenotypic tests predictive of pathogenic potential, such as cytotoxicity assays, in addition to quantifying enterotoxin levels. Unfortunately, there is no established standard for studying pathogenic potential of B. cereus s.l., making it difficult to compare findings from multiple small studies that use varied experimental designs. The solution to studying complex systems like enterotoxin regulation often lie in large-scale genome-wide association studies. Broader genome-wide association studies are necessary to identify single nucleotide polymorphisms (SNPs) in coding and noncoding (regulatory) regions of a genome, that are associated with increased cytotoxicity of isolates toward human cells. Such studies cannot only guide further experimental confirmation of the role of identified SNPs in the regulation of enterotoxins (using knockout approaches coupled with cytotoxicity assays and protein experiments), but can also facilitate the development of rapid genetic screening methods for identification of B. cereus s.l. isolates that pose a high risk for toxicoinfection. However, it must be noted that cytotoxicity is not the only predictor of B. cereus s.l. virulence. Additional studies on sporulation, germination, competition with other organisms in a gut microbiome, and regulation of genes involved are needed to improve the accuracy of predicting B. cereus s.l. virulence.