Detection of a luxS-Signaling Molecule in Bacillus anthracis

ABSTRACT Quorum-sensing regulation of density-dependent genes has been described for numerous bacterial species. The partially annotated genome sequence of Bacillus anthracis contains an open reading frame (BA5047) predicted to encode an ortholog of luxS, required for synthesis of the quorum-sensing signaling molecule autoinducer-2 (AI-2). To determine whether B. anthracis produces AI-2, the Vibrio harveyi luminescence bioassay was used. Cell-free conditioned media from vaccine (Sterne) strain 34F2 induced luminescence in V. harveyi reporter strain BB170, indicating its production of AI-2. Cloned BA5047, expressed in Escherichia coli DH5α cells, restored AI-2 activity to these cells. To evaluate whether BA5047 is essential for AI-2 synthesis, it was deleted through allelic exchange with marker rescue; the resulting mutant had no functional luxS activity and had reduced growth in vitro. In the wild-type strain, AI-2 activity was greatest during the exponential phase of growth. In total, these data indicate that BA5047 is a functional luxS ortholog in B. anthracis necessary for growth-phase-specific AI-2 expression. Thus, B. anthracis may utilize extracellular signaling molecules to regulate density-dependent gene expression.

Quorum sensing is the regulation of bacterial gene expression in response to changes in cell density (13). Bacteria that utilize quorum-sensing signaling pathways synthesize signaling molecules (autoinducers [AI]) which have been found in nature as N-homoserine lactones (8) or small peptides (13). AI levels are directly proportional to the size of the bacterial population (7), and at threshold levels detectable by bacterial cell receptors, AI binding alters bacterial gene expression (13). Quorum-sensing-based regulation of gene expression is critical for the pathogenesis of clinically important bacterial infections, such as those due to Pseudomonas aeruginosa in patients with cystic fibrosis (4) or to Vibrio cholerae (14).
Quorum sensing has been well characterized in Vibrio harveyi, a bioluminescent bacterium that freely lives in the ocean floor sediment or on the exterior of fish (17). The luminescence genes are expressed only when the V. harveyi populations are at high cell density under the control of the lux quorumsensing system (13). The luxCDABE operon is regulated by two-component systems that are stimulated by the AI ligands, AI-1 (acyl-homoserine lactone [AHL]) and AI-2 (13). Synthesis of AI-1 requires luxLM. AI-1 diffuses freely through the cell wall into the extracellular milieu, and when sufficient quantities are recognized by its sensor histidine kinase, luxN, a hybrid two-component system-signaling cascade is initiated (13). The lux cascade also is regulated by another AI molecule, AI-2, which is predicted to be a furanosyl borate diester and is synthesized by the product of luxS (2). The luxS product converts S-ribosylhomocysteine to 4,5-dihydroxyl-2,3-pentanedione, catalyzing AI-2 formation (2). V. harveyi strain BB170, in which luxN is mutated, is unable to detect AI-1 molecules and may be used to detect AI-2 or AI-2-like molecules in its milieu (1).
Bacillus anthracis, a gram-positive, nonmotile, spore-forming bacterium, is the etiological agent of anthrax (15). Spores from B. anthracis are extremely resistant to a wide range of adverse environmental conditions, such as heat, UV and ionizing radiation, and chemical agents (15). With the emergence of B. anthracis spores as a weapon of terrorism (11), it is essential to develop new vaccines to prevent and new therapies to control B. anthracis infections. In the present report we demonstrate that B. anthracis possesses a luxS ortholog and synthesizes a functional AI-2 molecule and that a luxS mutant has slowed growth compared to that of a wild-type strain. These observations suggest modalities for both prevention and treatment of anthrax.
The PCR-amplified products were purified by using a Qiagen PCR purification kit and subsequently were digested with HindIII. Digest fragments were cloned into pGEM-T Easy and were screened by PCR with primers BAluxSKOF1 and BAluxSKOR2; the plasmid with the correct insert is designated pMJ301. pMJ301 was digested with HindIII and aphA, conferring kanamycin resistance, and was introduced into pMJ301 to create pMJ301K (9). pMJ301K was digested with EcoRI, releasing the insert region, which was cloned into a pUC19 derivative (pUS19) with spectinomycin resistance to create pMJ301KS. Since methylation inhibits transformation into B. anthracis, pMJ301KS was cloned into dam-defective E. coli strain SCS110. Purified pMJ301KS from SCS110 was electroporated into B. anthracis strain 34F 2 , and colonies were selected for Kan r and Spe r . Transformants were picked on medium containing 50 g of kanamycin/ml and 100 g of spectinomycin/ml and then were subcultured daily in the absence of antibiotics at 37°with aeration for 15 days. Individual colonies were subsequently screened to identify clones that were both Kan r and Spe s . Clones with the correct antibiotic phenotype were confirmed by PCR to have allelic exchange of aphA in the luxS locus by using oligonucleotides SterneF (5Ј-GCAAATTGAAAACGA CTCAG-3Ј) and SterneR (5Ј-GTATGCTTATAAACATTCCGTCG-3Ј), with HindIII digestion of the PCR products.
Construction and screening for pMJ501. Chromosomal DNA of B. anthracis strain 34F 2 was purified by using the Wizard Genomic DNA Purification kit (Promega, Madison, Wis.) and was used as template for PCR amplification of open reading frame (ORF) BA5047. The oligonucleotides used were designated BAluxSF1 (5Ј-ATGCCATCAGTAGAAAGCTTTG-3Ј) and BAluxSR2 (5Ј-CC AAATACTTTCTCAAGTTCATC-3Ј). In the PCR, DNA was denatured for 1 min at 94°C, with annealing for 1 min at 51°C and extension for 1 min at 72°C. The amplified product was cloned into pGEM-T Easy, yielding pMJ501, which then was transformed into E. coli strain DH5␣ with selection for ampicillin resistance. The insert from pMJ501 was subjected to sequence analysis by using vector primers T7F and SP6R to ensure that no nucleotide errors had been introduced in the cloning process. E. coli DH5␣ cells also were transformed with pGEM-T Easy alone for use as a control.
Genomic analysis. LuxS protein sequences were retrieved from the National Center for Biotechnology Information database, and alignments were created by using ClustalW (23). Phylograms based on amino acid alignments were generated by using Paup 4.0b8 (Sinauer Associates, Sunderland, Mass.) with generation of 1,000 replicate trees by using a full heuristic search (5).
Growth phase regulation of AI-2 synthesis in B. anthracis. Overnight cultures of B. anthracis strain 34F 2 were diluted to an optical density at 600 nm (OD 600 ) of 0.03 in 50 ml of BHI and were grown at 37°C with aeration. Every 60 min, ODs of the cultures were measured by reading 1-ml aliquots with a Beckman DU7400 spectrophotometer. At sequential intervals, quantitative cultures were performed to determine bacterial CFU. From these same aliquots CFMs were prepared for use in the bioluminescence AI-2 reporter assay. All assays were repeated in triplicate. In separate experiments B. anthracis strains 34F 2 and 34F 2 ⌬luxS were grown overnight at 37°C with aeration. Cultures were used to inoculate fresh media to an OD 600 of 0.03 and then were grown at 37°C with aeration. OD was measured over a 24-h period, since this may be more accurate than cell count due to chaining in B. anthracis cultures (16).

Identification and organization of the B. anthracis luxS locus.
The unfinished genomic sequence of the Ames strain of B. anthracis has been made publicly available by The Institute for Genomic Research (www.TIGR.org). By using the nucleotide sequence of the 471-bp Bacillus subtilis luxS gene as a template, the partially annotated B. anthracis genome was subjected to BLASTN search, which revealed a 474-bp ORF, BA5047, with 72% similarity to luxS (also known as ytjB) from B. subtilis. To further characterize the putative B. anthracis luxS locus, flanking nucleotide sequences were submitted for BLASTN analysis. Sequence analysis of the region upstream of the B. anthracis luxS ortholog revealed a high level of conservation in which B. subtilis genes ytjA and ytiB had homologs with nucleotide similarities of 70 and 67%, respectively. However, the region downstream of the B. anthracis luxS ortholog showed substantial variation compared to B. subtilis. Only one proximate downstream B. subtilis gene, ytkD, had an ortholog in B. anthracis. Immediately downstream of BA5047 are two ORFs (BA5045 and BA5046) of 201 and 231 bp, respectively, with no significant homologies in GenBank. The orientations of the flanking ORFs indicate that BA5047 is in a monocistronic operon ( Fig. 1) (18).
Characterization of the B. anthracis luxS ortholog. Further data suggesting that BA5047 may be a functional luxS ortholog are provided by an alignment of the translated sequence with 17 other LuxS protein sequences. Although size variations exist in the luxS products, conserved regions essential for function across prokaryotic genera have been defined (13). Alignment of protein sequences from 17 known luxS orthologs with BA5047 reveal a number of conserved amino acids, including those hypothesized to be essential for LuxS enzymatic activity (10). These data provide evidence that B. anthracis ORF BA5047 encodes a LuxS protein with function. Phylogenetic analysis was done to further characterize the B. anthracis luxS   (3), suggesting horizontal gene transfer.

Synthesis of a functional AI-2 molecule by B. anthracis cells.
Utilizing the V. harveyi AI-2 reporter assay, liquid cultures of B. anthracis vaccine (Sterne) strain 34F 2 were examined to determine whether B. anthracis cells synthesize an AI-2 or AI-2-like molecule. The AI-2 assay utilizes a deficiency in the AI-1 sensor in V. harveyi strain BB170 (22). Without the luxN AI-1-encoded sensor, strain BB170 can only exhibit bioluminescence in response to AI-2 or an AI-2-like molecule. Growing a culture of strain BB170 overnight and then diluting it 1:10,000 (to yield low cell density) reduces the level of endogenous AI-2 below the threshold required for luminescence. In this experimental system the addition of exogenous AI-2 from bacteria possessing luxS function can restore the bioluminescence phenotype of the BB170 cells (22). As a negative control the V. harveyi reporter strain BB170 was incubated with sterile CFM alone; as a positive control CFM from a high-density culture of strain BB170 also was used (Fig. 3). Addition of sterile CFM to cells of BB170 served as the standard for baseline in luminescence, whereas as expected, addition of CFM from the high-density BB170 culture induced greater than 100fold increases in luminescence. In multiple experiments, CFM from B. anthracis strain 34F 2 had activity similar to that of the positive control, with substantial increases in luminescence compared to that of the negative control (Fig. 3). The results of these experiments indicate that B. anthracis synthesizes AI-2 or an AI-2-like molecule that is involved in the lux quorum-sensing system.
Evidence that BA5047 encodes a functional luxS. To determine whether BA5047 is the B. anthracis ORF responsible for synthesis of AI-2 or an AI-2-like molecule, we took advantage of the inability of E. coli strain DH5␣ to synthesize a functional AI-2 molecule (6). The B. anthracis luxS ortholog (BA5047) was amplified by PCR and was cloned into the E. coli shuttle vector pGEM-T Easy to create pMJ501. Only the ORF was cloned into pGEM-T Easy, into a site downstream of the vector's isopropyl-␤-D-thiogalactopyranoside (IPTG)-inducible promoter. CFM from high-density cultures of DH5␣ containing vector pMJ501 were induced with IPTG and then were screened for the synthesis of AI-2, as measured in the V. harveyi bioluminescence assay (Fig. 4). As negative controls, the reporter strain BB170 was incubated with sterile CFM alone or with CFM from high-density IPTG-induced cell cultures of E. coli strain DH5␣ without vector or containing pGEM-T Easy with no insert. As positive controls, CFM was used from high-density cultures of strain BB170 and B. anthracis strain 34F 2 . As previously shown, when sterile CFM alone was used as the baseline for bioluminescence, CFM from highdensity cultures of strain BB170 and B. anthracis 34F 2 induced substantial bioluminescence (Fig. 4). As expected, no bioluminescence was induced by CFM from cultures of DH5␣ or from DH5␣ containing pGEM-T Easy without insert. In contrast, CFM from DH5␣ containing pMJ501 induced a high level of bioluminescence, greater than that induced by CFM from the positive controls. Compared to the control E. coli CFMs, there was nearly a 1,000-fold mean increase in induction of bioluminescence by pMJ501 (Fig. 4). B. anthracis 34F 2 ⌬luxS has a defect in AI-2 activity. To analyze the effect of AI-2 signaling in B. anthracis, we created a ⌬luxS mutant in strain 34F 2 by replacement of the luxS homolog with a kanamycin resistance cassette (see Materials and Methods). After electroporation of the wild-type strain with the mutated locus on pMJ301KS, kanamycin-resistant transformants were serially passed for 15 days in vitro to select for a double-crossover event. Screening of one such Kan r Spe s transformant by PCR showed the expected products (Fig. 5), indicating the proper construction. Additionally, in lane 3 of Fig. 5, the ϳ0.4-kb band corresponds to a fragment of luxS amplified from genomic DNA from the wild-type strain. To determine the effect of the mutation on AI-2 synthesis, we utilized the V. harveyi bioassay as described above (Fig. 3). Compared to the baseline level of AI-2 activity in reporter strain BB170, CFM from a high-density culture of 34F 2 ⌬luxS had no additional AI-2 activity. The data collected provided evidence that luxS is necessary for AI-2 synthesis in B. anthracis strain 34F 2 (Fig. 3).
Growth defect in B. anthracis 34F 2 ⌬luxS. When cultured in liquid medium, B. anthracis 34F 2 ⌬luxS exhibits noticeable FIG. 3. Induction of bioluminescence in V. harveyi reporter strain by CFM from B. anthracis cells. V. harveyi strain BB170 is deficient in the AI-1 sensor encoded by luxN, and thus upregulates only the expression of the lux operon (measured as RLU) when AI-2 or AI-2-like molecules are present in its milieu (1). CFM obtained from AI-2synthesizing bacteria grown to high density (including BB170, in which the AI-2-regulated system is intact) can induce expression of the bioluminescence-generating luxCDABE operon in BB170. In the experiments shown, sterile CFM alone and CFM from high-density cultures of V. harveyi strain BB170 were negative and positive controls, respectively, and CFM from high-density 6-h cultures of B. anthracis strain 34F 2 and 34F 2 ⌬luxS were the unknowns. Cells of BB170 were grown for 2 h (black boxes) or 4 h (white boxes) in the presence of sterile CFM. The baseline is the value for use of uninoculated (sterile) CFM alone at 2 h. By 4 h the endogenous AI-2 activity was substantially higher than that at 2 h. Each bar represents the means (Ϯ standard deviations) of triplicate experiments. Compared to the negative control, wild-type 34F 2 but not 34F 2 ⌬luxS showed substantial AI-2 activity.
FIG. 4. Induction of bioluminescence in the V. harveyi reporter strain by cloned BA5047 in E. coli. V. harveyi strain BB170 upregulates only the expression of bioluminescence when AI-2 or AI-2-like molecules are present in its milieu, as described in the legend to Fig. 3. In the experiments shown, negative controls were BB170 cells incubated for 2 h with sterile CFM alone and CFM from high-cell-density cultures of E. coli strain DH5␣ alone or containing pGEM-T Easy without insert. Positive controls used were CFM from high-density cultures of V. harveyi strain BB170 and B. anthracis strain 34F 2 (6-h culture). The unknown specimen was CFM from DH5␣ containing pMJ501; all assays were run in triplicate. The dashed line indicates the endogenous RLU for the BB170 cells grown for 2 h in the presence of sterile CFM alone. growth defects compared to wild-type B. anthracis 34F 2 (Fig.  6A). As determined by cell density, 34F 2 ⌬luxS has a brief delay (approximately 30 to 60 min) in the transition between lag and early exponential phase compared to those for wild-type 34F 2 . Subsequently, exponential growths for the wild-type and mutant strains are parallel, but the mutant enters into stationary phase at a much lower cell density. Thus, under the conditions tested AI-2 function appears necessary for full B. anthracis growth in vitro. Growth phase dependence of AI-2 synthesis in B. anthracis. To determine whether B. anthracis synthesis of AI-2 was growth phase dependent, CFMs were collected from the 34F 2 cells at various time points in the growth cycle and were used in the V. harveyi bioluminescence assay. The CFMs were diluted in sterile medium to reflect equal numbers of cells and were incubated with strain BB170 (Fig. 6B). Sterile CFM alone and CFM from 34F 2 ⌬luxS were used as negative controls, and CFM from a high-density culture of V. harveyi strain BB170 was used as a positive control. Analysis of CFM collected from B. anthracis showed that AI-2 is maximally synthesized during the mid-exponential phase of growth and diminishes during stationary phase. CFM collected from a 6-h culture of the wild-type strain did not enhance the growth of 34F 2 (data not shown).

DISCUSSION
In this study we confirmed that an ORF (BA5047) in the partially annotated B. anthracis genome possesses extensive homology to the luxS ortholog (ytjB) in B. subtilis (10). Although the B. anthracis locus for BA5047 is not highly conserved compared to that of B. subtilis, the orientations of the ORFs indicate that BA5047 is in a monocistronic operon, which facilitates examination of its function and regulation. In other organisms, luxS appears essential for the synthesis of a quorum-sensing molecule (AI-2) (12), first identified in the marine bacterium V. harveyi (1). Both multiple protein alignments and phylogenetic analyses (Fig. 2) of the B. anthracis luxS ortholog (BA5047) revealed strong evolutionary relationships with those of B. subtilis and B. halodurans, indicating strong conservation of luxS within the genus Bacillus. Phylogenetic analysis of luxS orthologs reveals two major groupings, generally clustering gram-positive and gram-negative species separately (Fig. 2); one exception is H. pylori, providing evidence consistent with horizontal transfer of luxS across genera (Fig. 2). Taken together, the phylogenetic studies and the protein alignments indicating the presence of conserved amino acids confirm that BA5047 encodes a luxS ortholog.
That CFM from strain 34F 2 was able to stimulate luminescence in V. harveyi strain BB170 (Fig. 3) indicates that B. anthracis produces AI-2 or an AI-2-like molecule, likely similar in structure to AI-2 from V. harveyi (2). With this evidence we next focused on BA5047, the luxS ortholog. Expression of B. anthracis BA5047 in E. coli strain DH5␣ demonstrates its central role in synthesis of AI-2 or an AI-2-like molecule (Fig. 4) and suggests the capability of B. anthracis to conduct densitydependent gene expression. Isogenic deletion of luxS (Fig. 5) resulted in an inability of the B. anthracis mutant (34F 2 ⌬luxS) to produce AI-2 or an AI-2-like molecule that could be detected in the V. harveyi bioassay (Fig. 3). Similarly, compared to the wild-type strain the mutant showed delay in the transition from lag to exponential growth phase and entered stationary phase early (Fig. 6A). In total, the ⌬luxS culture grew more slowly and produced fewer cells compared to the wild type.
B. anthracis synthesis of AI-2 or an AI-2-like molecule mediated by luxS thus plays an important role in the regulation of growth. As such, targets of the hypothesized density-dependent gene expression must include genes regulating vegetative growth and cell cycle. If B. anthracis regulates gene expression by means of an AI molecule, as do other pathogens (14), cells might have the ability to suppress virulence gene expression until the total population reaches a threshold density. Suppression of virulence gene expression by quorum sensing could allow B. anthracis to evade immune detection until its population is at a density sufficiently high to overwhelm the host's innate and adaptive defenses. That AI-2 synthesis is maximal during exponential phase growth is consistent with this hypothesis. This hypothesis suggests that a possible means of treating anthrax could be via inhibitors of AI-2 to downregulate density-dependent gene expression. Recent data has shown that a FIG. 6. Growth rate and AI-2 production of B. anthracis strains 34F 2 and 34F 2 ⌬luxS. (A) B. anthracis strains 34F 2 and 34F 2 ⌬luxS were grown overnight in BHI medium and were inoculated with fresh BHI medium at an adjusted OD 600 of 0.03. Cells of 34F 2 (diamonds) and 34F 2 ⌬luxS (squares) were grown for 24 h, and the OD was measured at regular intervals. Filtered CFM from the growth of strain 34F 2 depicted in panel A was examined at various time points to ascertain AI-2 levels by using the V. harveyi bioassay described in the legend to Fig. 4. CFM from strain 34F 2 ⌬luxS was collected at 6 h. (B) All filtered CFM from the 34F 2 and 34F 2 ⌬luxS cultures shown in panel A were adjusted to reflect an OD 600 of 0.6 to standardize the cell numbers. Negative and positive controls were sterile CFM and CFM from highdensity cultures of 34F 2 ⌬luxS and V. harveyi, respectively. synthetic furanone, (5Z)-4-bromo-5-(bromethylene)-3-butyl-2(5H)-furanone, has the ability to inhibit AI-2-mediated quorum sensing in E. coli and V. harveyi (19) as well as swarming and biofilm formation by B. subtilis (20). Examination of this or similar molecules could permit ascertainment of the role of AI-2-mediated mechanisms in virulence gene expression of B. anthracis.