|Home | About | Journals | Submit | Contact Us | Français|
Genome engineering is a powerful method for the study of bacterial virulence. With the availability of the complete genomic sequence of Bacillus anthracis, it is now possible to inactivate or delete selected genes of interest. However, many current methods for disrupting or deleting more than one gene require use of multiple antibiotic resistance determinants. In this report we used an approach that temporarily inserts an antibiotic resistance marker into a selected region of the genome and subsequently removes it, leaving the target region (a single gene or a larger genomic segment) permanently mutated. For this purpose, a spectinomycin resistance cassette flanked by bacteriophage P1 loxP sites oriented as direct repeats was inserted within a selected gene. After identification of strains having the spectinomycin cassette inserted by a double-crossover event, a thermo-sensitive plasmid expressing Cre recombinase was introduced at the permissive temperature. Cre recombinase action at the loxP sites excised the spectinomycin marker, leaving a single loxP site within the targeted gene or genomic segment. The Cre-expressing plasmid was then removed by growth at the restrictive temperature. The procedure could then be repeated to mutate additional genes. In this way, we sequentially mutated two pairs of genes: pepM and spo0A, and mcrB and mrr. Furthermore, loxP sites introduced at distant genes could be recombined by Cre recombinase to cause deletion of large intervening regions. In this way, we deleted the capBCAD region of the pXO2 plasmid and the entire 30 kb of chromosomal DNA between the mcrB and mrr genes, and in the latter case we found that the 32 intervening open reading frames were not essential to growth.
Bacillus anthracis, the etiological agent of anthrax, is a gram-positive, rod-shaped, spore-forming bacterium. Two major factors encoded on two large plasmids are associated with virulence. Plasmid pXO2 encodes a poly-γ-d-glutamic acid capsule (10, 19), whereas pXO1 encodes the anthrax toxins, consisting of protective antigen (PA), lethal factor, and edema factor (10, 16, 19, 22). Vaccination against anthrax is the most successful way of preventing morbidity and mortality associated with anthrax infections. In the United States, persons at risk of exposure receive a licensed anthrax vaccine (BioThrax), which is aluminum hydroxide-adsorbed, formalin-treated culture supernatant of the attenuated toxigenic, noncapsulated, low-proteolytic B. anthracis strain V770-NP1-R. The key immunogen in the vaccine is PA (17), but undefined degradation products as well as other anthrax toxin components are present. Several approaches have been used to obtain improved anthrax vaccines, including expression of PA in Bacillus subtilis (37). The recombinant PA (rPA) produced by B. subtilis is, however, rapidly degraded by cell-associated or secreted proteases, resulting in a reduced yield of protein. The current recombinant PA vaccine under development in the United States is produced in a sporulation-deficient, virulence plasmid-cured strain of B. anthracis containing a vector expressing only PA (8). While effective, this host strain might be improved by disruption of genes encoding extracellular proteases that contribute to degradation of the secreted PA.
The availability of the B. anthracis genomic sequence (31) has greatly facilitated the genetic dissection of factors determining this organism's fitness and pathogenic ability. Typically, gene function is assessed by inserting an antibiotic resistance marker into a target gene so as to disrupt it. However, when it becomes necessary to target multiple genes, the limited number of convenient selectable markers makes this difficult. This problem has been solved in other bacteria by use of site-specific recombinases. The 38-kDa Cre (causing recombination) recombinase of bacteriophage P1 recognizes a 34-bp loxP (locus of crossing over) site which consists of two 13-bp inverted repeats surrounding an 8-bp asymmetric core sequence (34). The Cre-loxP system has previously been employed to excise antibiotic resistance genes from Escherichia coli after their introduction into specific target genes. The recombination reaction results in the excision or inversion of the intervening sequence between two loxP, depending on their relative orientation (24). In this report we describe the adaptation of the Cre-loxP system for removing spectinomycin resistance genes from B. anthracis mutants in a sequential fashion. The method was first applied to the mutation of two genes, spo0A and pepM. The spo0A gene encodes a global regulator shown to be essential for spore formation (9). The pepM gene encodes a secreted, zinc-dependent metalloprotease (also referred to as the “neutral protease/peptidase”) belonging to the M4 metallopeptidase family (Protein Families database; Sanger Institute, United Kingdom). Mutations in spo0A and pepM were generated individually and sequentially in three attenuated strains of B. anthracis: pX01+ pX02−, pX01− pX02+, and plasmid free. The plasmid-free, nonsporogenic, low-proteolytic strain is suggested to have value as a host strain for producing recombinant PA for use in vaccines.
B. anthracis is refractory to transformation by DNA extracted from dam+ dcm+ E. coli, requiring that DNA for transformation be prepared from Dam methylation-deficient strains of E. coli such as GM2163 (20). Therefore, it is probable that B. anthracis encodes methylation-dependent restriction enzymes (MDRs). We employed the Cre-lox method to begin characterization of MDR genes as well as to obtain a strain of B. anthracis more tractable to genetic manipulation. Here, we report the sequential inactivation of the mrr and mcrB genes, two of the three genes proposed in the online restriction enzyme database (REBASE) as encoding MDRs. This process not only resulted in the inactivation of the said genes, but also led to excision of 30 kb of intervening genomic sequence, thereby identifying 32 contiguous genes as being dispensable for the growth of B. anthracis in rich media.
The methods developed here were also shown to be applicable to modification of the large virulence plasmids of B. anthracis. The four genes of the capBCAD region of the pXO2 plasmid are involved in synthesis and processing of the poly-γ-d-glutamic acid capsule. We introduced two loxP sites at the boundaries of the region and used these to delete the entire gene region.
E. coli strains were grown in Luria-Bertani (LB) broth and used as hosts for cloning. LB agar was used for selection of transformants (33). B. anthracis strains were grown in brain heart infusion (BHI) and LB medium. Antibiotics (Sigma-Aldrich) were added to media when appropriate to give the final concentrations indicated: ampicillin (Ap; 100 μg/ml, only for E. coli), erythromycin (Em; 400 μg/ml for E. coli, 5 μg/ml for B. anthracis), spectinomycin (Sp; 100 μg/ml for both E. coli and B. anthracis). SOC medium (Quality Biologicals) was used for outgrowth of transformation mixtures prior to plating on selective medium to isolate transformants. B. anthracis colonies producing PA were identified by growth on CA medium agar (36) supplemented with 0.8% (wt/vol) sodium bicarbonate, 5% (vol/vol) horse serum, and 5% (vol/vol) sheep anti-PA serum (gift of J. Reimenschneider). On these plates, PA-producing colonies formed halo-like zones of immunoprecipitation in as little as 12 h when grown in 20% CO2 at 37°C. Bicarbonate agar (10) with 0.8% (wt/vol) sodium bicarbonate and 10% (vol/vol) equine serum was used for estimation of B. anthracis capsule production. Capsule-producing strains formed mucoid colonies when grown in 20% CO2 at 37°C for 24 to 48 h. India ink was used to visually observe the presence of the B. anthracis capsule (7). B. anthracis spores were prepared by growth on NBY-Mn agar (nutrient broth, 8 g/liter; yeast extract, 3 g/liter; MnSO4 · H2O, 25 mg/liter; agar, 15 g/liter) at 30°C for 5 days (35). Congo red agar (39) was used for the differentiation of B. anthracis Spo0A mutants (pink colonies). The capsules, spores, and vegetative cells of B. anthracis were visualized on a Nikon Eclipse E600W light microscope and photographed with a digital DXM1200F camera (Nikon Instrument Inc., New York). Casein agar (1) was used for the differentiation of protease-deficient strains of B. anthracis.
Preparation of plasmid DNA from E. coli, transformation of E. coli, and recombinant DNA techniques were carried out by using standard procedures (33). E. coli XL2-Blue and SCS110 competent cells were purchased from Stratagene, and E. coli TOP10 competent cells were from Invitrogen. Recombinant plasmid construction was carried out in E. coli XL2-Blue or TOP10. Plasmid DNA from B. anthracis was isolated according to the protocol for the purification of plasmid DNA from Bacillus subtilis (QIAGEN). Chromosomal DNA from B. anthracis was isolated with a Wizard genomic purification kit (Promega) in accordance with the protocol for isolation of genomic DNA from gram-positive bacteria. B. anthracis was electroporated with unmethylated plasmid DNA isolated from E. coli SCS110. Electroporation-competent B. anthracis cells were prepared as previously described (25). Restriction enzymes, T4 ligase, Klenow fragment, and alkaline phosphatase were purchased from MBI Fermentas or New England Biolabs. Taq polymerase kits were purchased from TaKaRa Shuzo or Invitrogen/Life Technologies. Ready-To-Go PCR beads (Amersham Biosciences) were used for DNA rearrangement analysis. For routine PCR analysis, a single colony was suspended in 200 μl of Tris-EDTA buffer (33), pH 8.0, heated to 95°C for 45 s, and then cooled to room temperature. Cellular debris was removed by centrifugation at 15,000 × g for 10 min. One microliter of the lysate contained sufficient template to support a PCR with the PCR beads. The GeneRuler DNA ladder mix (MBI Fermentas) or the 1-kb Plus ladder (Invitrogen) was used for determination of DNA fragment length. All constructs were verified by DNA sequencing and/or restriction enzyme digestion. All plasmids used in this study and their relevant characteristics are listed in Table Table1.1. Oligonucleotide primers are listed in Table Table22.
The general scheme for B. anthracis gene knockout using the Cre/Lox system, presented in Fig. Fig.1,1, employs in its first step plasmids we have designated generically as pDC, for double-crossover plasmids. These were derived from one of two temperature-sensitive (ts) plasmids, pE194 or pWVO1. The ts replicon from pE194 was extracted from pYJ335 (12) and used to make pUE1, which has permissive and restrictive temperatures of 37°C and 43.5°C, respectively. Plasmids pWVO1 and pVE6007 (18) were precursors to the highly ts plasmid used here, pHY304 (30), which has permissive and restrictive temperatures of 30°C and 37°C, respectively. In most cases, the pDC plasmids had the gene of interest disrupted by the “Ω-sp element,” which consists of the spectinomycin resistance (Spr) gene (aad9) of Enterococcus faecalis followed by transcriptional stops. Previously, Saile and Koehler (32) successfully used the Ω-sp element for gene disruption. Two different approaches were used for generating a loxP-flanked Ω-sp cassette for insertion into the gene of interest. In the first approach, the BamHI fragment of pJRS312 containing the Ω-sp cassette was introduced into the BamHI site of plasmid pBS246 between its two directly repeated loxP sequences. The resulting plasmid, pΩL, was used as a template for amplification of the loxP-Ω-sp-loxP cassette. The two primer pairs LoxSB/LoxEB and LoxSN/LoxEN were used to generate loxP-Ω-sp-loxP fragments for insertion within the spoAO and pepM genes, at BglII and BamHI sites (spoOA) or at the NdeI site (pepM). This produced plasmids that were designated pSΩL408 and pSΩL304 (Table (Table1).1). In the second approach, we PCR amplified the Ω-sp element using primers with loxP sites added to their 5′ ends (primers Tn5loxPaad9 For and Tn5loxPaad9 Rev) (Table (Table2).2). The loxP sites introduced by either method worked satisfactorily.
A plasmid for expression of Cre recombinase was constructed using a PCR fragment containing the P1 cre gene amplified from pBS185 (Gibco-BRL/Life Technologies) with primers CreS and CreE. The fragment was inserted into the corresponding NdeI and HindIII sites on pAE5 (27) to place expression of cre under the control of the pagA promoter. The BglII-HindIII fragment of the resulting pAEC5 plasmid was inserted between the BamHI and HindIII sites of pHY304 to produce the Cre-expressing ts plasmid pCrePA (Fig. (Fig.11).
The strains used and their relevant characteristics are listed in Table Table3.3. B. anthracis mutants were constructed by the replacement of coding sequences with the Ω element conferring spectinomycin resistance (Fig. (Fig.1).1). The targeting constructs were electroporated into B. anthracis with selection for erythromycin resistance. Erythromycin-resistant colonies were transferred to agar containing spectinomycin, incubated at the restrictive temperature, and again transferred onto fresh agar containing spectinomycin. After the third passage, at least 50 colonies were screened for spectinomycin resistance and erythromycin sensitivity. In some cases, additional passages were required to obtain this phenotype. Additionally, in the cases of spo0A and pepM mutants, Congo red and casein agar, respectively, were used for an initial screening for double-crossover events. Colonies in which a double-crossover recombination event was suspected were validated by PCR analysis.
For elimination of the spectinomycin resistance, plasmid pCrePA was electroporated into the isolates with selection for erythromycin resistance at 30°C. Erythromycin-resistant colonies were transferred to antibiotic-free agar and incubated at 37°C to eliminate pCrePA. Colonies were then patched to three separate agar plates containing erythromycin, spectinomycin, or no antibiotic. The entire process was repeated when constructing mutants targeted in two or more genes. The presence of a single loxP site within the targeted gene(s) of the fully antibiotic-sensitive single or double mutants was confirmed by PCR and/or sequence analysis.
Primers used for PCR analysis are described in Table Table2,2, and the locations of the corresponding sequences are shown in the figures. Primers PAF and PAR correspond to sequences within the pagA gene. Primers pXO2-AT-f1/pXO2-AT-r1 (13) were used to confirm the presence of the plasmid pXO2.
The sequence at the single loxP site remaining within spo0A was determined by inserting the PCR fragments generated from strains MSLL35, MSLL34, and MSLL33 (using primers Spo1 and Spo2) into pCR2.1-TOPO (Invitrogen) and sequencing the resulting plasmid (National Institute of Dental and Craniofacial Research core facility, National Institutes of Health, Bethesda, MD). Similarly, the loxP within the pepM gene was amplified from strains MSLL35, MSLL34, and MSLL33 with primers M4F and M4R and sequenced.
For analysis of the genomic deletion in the mcrB mrr double mutant (strain McrB3P-Mrr-LΔ30), the region encompassing the entire mcrB-to-mrr region was amplified by PCR in the parent strain using overlapping primer pairs (IG-1 through IG-8; sequences available from the authors on request). The 30-kb deletion in the double mutant was verified with a PCR using primers McrB3P For1 and Mrr 5′ GenP, and the resulting PCR product was gel purified and submitted for direct sequencing by the DNA sequencing and synthesis facility at Iowa State University, Ames.
A number of genes on the B. anthracis chromosome or virulence plasmid pXO2 were targeted for inactivation in this study. They are identified here with the gene designations assigned by The Institute for Genomic Research (Rockville, Maryland; http://www.tigr.org) for the “Ames ancestor” strain chromosome (accession NC_007530) and pXO2 plasmid (accession number NC_007323). The spo0A gene (GBAA4394 or BA4394) encodes an essential sporulation gene that acts at the first step of sporulation. The pepM gene (GBAA0599) encodes a Zn2+ binding protease, also termed neutral peptidase/protease, and is designated peptidase M04.012 in the MEROPS peptidase database (http://merops.sanger.ac.uk). The mrr (GBAA2317) and mcrB (GBAA2283) genes were initially suggested as being involved in DNA restriction in the enzyme database maintained by New England Biolabs (http://rebase.neb.com). The capBCAD region of plasmid pXO2 contains all four genes (GBAA_pXO2_0063 to 0066) needed to synthesize and modify the extracellular poly-d-glutamic acid capsular material (19).
Electroporation of B. anthracis with plasmids in which the gene of interest was disrupted with the Ω-sp cassette led to efficient gene knockout, as will be described with specific examples below. The strategy used is shown in Fig. Fig.11 and described in the Materials and Methods section. Expression of the aad9 gene of the Ω-sp cassette was high enough to confer antibiotic resistance even when present as a single copy integrated into the chromosome, as demonstrated previously (32). The generic pDC targeting plasmids we used contain the erm gene to allow selection for the presence of the plasmid in B. anthracis. We found that 5 μg/ml of erythromycin is optimal for selection of transformants. An important attribute of the pDC plasmids is a temperature-sensitive origin of replication. Previously, it has been shown that two passages of B. anthracis cells containing pE194-derivative plasmids at 43.5°C on LB agar without erythromycin eliminated the plasmids (28). Here we successfully used the pE194 replicon for Ω-sp cassette insertion into B. anthracis. However, we found that B. anthracis strain Ames 35 lost the pXO1 plasmid when grown at 43.5°C, as was expected from previous results (22). Therefore, we also used the more highly ts replicon of plasmid pHY304 and found that a single passage at 37°C entirely eliminated pHY304-derivative plasmids while retaining pXO1. B. anthracis retained plasmid pXO2 during passage at both 37 and 43.5°C. We designed the targeting vectors to include loxP sites bracketing the Ω-sp cassette so that it could be removed by the site-specific Cre recombinase. Plasmid pCrePA was constructed to express the recombinase constitutively in B. anthracis (Fig. (Fig.1)1) and to be easily removable by growth at the restrictive temperature of the pHY304 replicon.
We disrupted the pepM and spo0A genes in three isogenic B. anthracis strains derived from the Ames strain (Table (Table3).3). We chose these two genes because they have easily discernible phenotypes, and also because the resulting mutated strains have potential value as protein expression hosts. These genes were interrupted individually and in combination by insertion of the Ω-sp cassette, as confirmed by PCR analysis (Fig. (Fig.22 and 3a and b). The PepM− phenotype was easily scored on plates containing casein (see Fig. Fig.5a,5a, below) and the Spo0A− phenotype on plates containing Congo red agar (see Fig. Fig.5b),5b), as previously described (39). The chromosomally integrated Ω-sp cassette was removed by introducing plasmid pCrePA, using selection for erythromycin resistance at 30°C, and the isolated colonies were then grown on antibiotic-free LB agar at 37°C. We found that all of the colonies lost both the Emr and Spr markers after overnight growth but retained the PepM− and/or Spo0A− phenotypes. These results demonstrate that, in both cases, the genes remained interrupted by the loxP site but the Spr marker was removed by the Cre function provided in trans by pCrePA. Elimination of the Spr marker was confirmed by PCR analysis utilizing specific primers for each gene, as depicted in Fig. Fig.2.2. Loss of pCrePA was phenotypically confirmed by the inability of the clones to grow on LB plates containing erythromycin. The absence of the vector in Ems cells was also confirmed by analysis of the plasmid content of the strains. Only the original B. anthracis virulence plasmids were observed in the B. anthracis double mutants (Fig. (Fig.3d3d).
Final corroboration of the chromosomal modifications in the mutant strains was obtained by sequencing the chromosomal DNA region containing the remaining loxP site. The nucleotide sequences of these fragments demonstrated that, as expected, Cre catalyzed the perfect removal of the intervening region, leaving only one copy of the loxP sequence. The resulting loxP sequence in the case of pepM disruption was an exact match to the known sequence, whereas we found one change in the case of spo0A (an A instead of G at the 27th position). Further analysis of loxP sequences showed that this mutation was generated in the right 13-bp RR inverted repeat of the right loxP site during PCR amplification of the loxP-Ω-sp-loxP fragment from pΩL using the LoxSB/LoxEB primers. The mutation was evidently retained during the Cre-mediated recombination event. The scheme presented in Fig. Fig.44 demonstrates how a mutated recombinase binding site from the right loxPR replaces a normal RL from the left loxPR in the single loxP site, finally remaining within spo0A. Our data confirm that the right inverted repeat of loxP is not as important for precise loxP cleavage as the left one, consistent with prior evidence that Cre initiates recombination of loxP by cleaving the upper strand on the left loxP inverted repeat (21).
The consequences of disruption of both the pepM and spo0A genes were the same in all three isogenic B. anthracis Ames derivative strains. Mutation in pepM resulted in considerably decreased proteolytic activities in comparison to the parent strain as assessed on casein agar (Fig. (Fig.5a).5a). Mutation in the spo0A gene resulted in the inability of the mutant strain to form spores (Fig. (Fig.5b).5b). No colonies were obtained on either LB or BHI agar after the MSLL33 strain was grown on NBY-Mn agar at 30°C for 5 days and the bacteria were heated at 70°C for 30 min. An aliquot of the same sample that was not heated at 70°C grew on both LB and BHI agar. On the other hand, the parent Ames 33 strain formed heat-resistant spores at a high frequency when subjected to identical treatment.
After PCR and phenotypic confirmation that the mutations were as expected, we confirmed the retention of plasmids pXO1 and pXO2 in the MSLL35 and MSLL34 mutants (Fig. (Fig.3d).3d). Also, we analyzed PCR fragments amplified from pagA located on pXO1 (38) and from the pXO2-at marker locus (13). The sizes of the fragments were as expected (Fig. (Fig.3c).3c). However, we noted a reduced zone of PA immunoprecipitation for the pXO1-containing double mutant and decreased capsule production by the pXO2-containing double mutant when these were compared to their respective parent strains. Thus, MSLL34 exhibited capsule formation, but the India ink particles were not fully excluded and the capsule material appeared thinner than that on Ames 34 cells (Fig. 6a and b). We suspect that these changes result from alterations in gene expression following the spo0A mutation. The affect of spo0A on pagA gene expression may in part be due to abrB control by spo0A (32). We also confirmed the stability of these mutations; there was no alteration of the loxP site sequences in B. anthracis MSLL33 following 30 passages at 37°C in either LB or BHI medium.
The structure of the capBCAD region as reported earlier (19) is shown in Fig. Fig.7.7. The capB, capC, capA, and capD genes are transcribed in the same orientation under control of two tandem promoters, P1 and P2, depending on CO2 concentration. Recent data on transcriptional analysis of the B. anthracis capsule regulators demonstrated the presence of a third promoter (P3) in the same area. It is interesting that transcription from this promoter was not regulated by CO2 (6).
Two plasmids, pCS1 and pDS2 (Table (Table1),1), were constructed for introducing modification of the capBCAD region. The pCS1 plasmid was used for insertion of the loxP-Ω-sp-loxP cassette into the BglII site located between the P1/P2 promoter site and the capB start codon. The pDS2 plasmid was used for the replacement of a short BamHI fragment in the capD gene by the same cassette (Fig. (Fig.7).7). Approximately 1 kb of homologous sequence was included on either side of the loxP-Ω-sp-loxP cassette in the targeting vectors to ensure efficient double crossover with either the capB or capD gene. Insertion of the loxP-Ω-sp-loxP cassette into the BglII site located between the P1/P2 promoter site and capB start codon (giving strain CΩ1) prevented capBCAD operon expression even when the cells were grown in 20% CO2. As a result of this insertion, B. anthracis strain CΩ1 was unable to synthesize capsule (Fig. (Fig.7).7). Cre-mediated excision of the Ω-sp cassette from the CΩ1 pXO2 strain resulted in strain CL1, which possesses a pXO2 plasmid modified by insertion of a single loxP site between the P1/P2 promoter site and the capB start codon. Although excision of the cassette from CΩ1 did restore capsule synthesis, the resulting mutant, CL1, appears to produce less capsule than the parent strain. This suggests that the palindromic loxP sequence (Fig. (Fig.4)4) might serve as a weak transcriptional terminator. Subsequent insertion of the Ω-sp cassette into the capD gene in strain CL1 produced the strain designated CL1DΩ2, which displayed an aberrant capsular morphology (Fig. (Fig.7).7). This is likely due to the absence of Dep enzymatic activity, which was reported to degrade high-molecular-weight capsule to a lower-molecular-weight capsule (19). Other explanations for the aberrant capsule on the mutant cells are suggested by the work of Candela and Fouet, showing that CapD is required for the covalent anchoring of capsule to peptidoglycan (5) and work showing that CO2-mediated control of acpB occurs via transcriptional read-through from atxA-dependent start sites of capB (6). In those experiments, insertion of cassettes into the capD gene was observed to either destabilize the capsule or decrease transcription of capBCAD via acpB regulation. As a result of loxP-Ω-sp-loxP insertion in the gene, strain CL1DΩ2 had three unidirectional loxP sites within the capBCAD region (Fig. (Fig.7).7). Treatment with the Cre recombinase-expressing plasmid caused recombination between the outer LoxP sites and deletion of the entire capBCAD region, leaving a single loxP site. As expected, the resulting strain CL1DL2 had no ability to form capsule (Fig. (Fig.77).
The plasmid-free UM44-1C9 strain of B. anthracis was used in experiments to identify genes involved in restriction of methylated DNA. From the published sequence for the Ames strain, primers were designed to amplify mrr and mcrB sequences. Plasmid pHYMcrB3PΩL (Table (Table1)1) was electroporated into UM44-1C9, the resulting clones were propagated at the restrictive temperature, and those with double-crossover events were identified as detailed in the Materials and Methods section. An Ems Spr clone was transformed with pCrePA to eliminate the Ω-sp cassette via Cre-mediated excision. This was followed by curing the Sps clones of plasmid pCrePA by means of propagation at the restrictive temperature, producing strain McrB3PΩL. Subsequently, strain McrB3PΩL was transformed with plasmid pHYMrrΩL to disrupt the mrr gene by double crossover, through insertion of the loxP-Ω-sp-loxP cassette within the mrr gene. Expression of Cre recombinase in this strain resulted in the excision of the Ω-sp element, accompanied by deletion of the approximately 30 kb of chromosomal DNA between the mcrB and mrr genes. The resulting strain was designated McrB3P-Mrr-LΔ30. The analysis of the role of the deleted genes in DNA restriction will be described elsewhere (R. Sitaraman and S. H. Leppla, unpublished data).
To confirm that the mcrB-mrr region (nucleotides 2129380 to 2160611 in the “Ames ancestor” strain sequence; accession number NC_007530) of the parent strain UM44-1C9 was similar to that of the Ames strain, PCR was done using primer sets IG-1 to IG-8 designed to amplify overlapping sections of this region. These produced PCR products of the expected sizes (Fig. (Fig.8a).8a). Then, primers McrB3P For1 and Mrr 5′ GenP were used to compare the intergenic region of strain McrB3P-Mrr-LΔ30 to that of the parental strain, UM44-1C9. The former strain yielded a PCR product close to the size (2,074 bp) expected for the deletion (Fig. (Fig.8b).8b). The parental strain produced no product, as expected, because the conditions used cannot amplify a 30-kb product. Direct sequencing of the PCR product using primer McrB3P For2 further confirmed the presence of a single loxP site, flanked by remnants of 5′ sequences of the mrr and the mcrB genes, as well as a few hundred bases of sequence derived either from the cloning vectors used (pBS246 and pGEM-T easy) or from primer design (the Tn5-binding sequences). These data show that the mcrB-mrr double mutant has a deletion spanning the entire mrr-mcrB region.
According to the annotations of the Ames genome provided by The Institute for Genomic Research, the region between mcrB and mrr contains 32 known and putative open reading frames (ORFs), BA2284 to BA2316 (note: BA2285 is missing in the database). In addition, in the deleted strain both the mcrB and mrr genes (BA2283 and BA2317, respectively) are truncated and, presumably, nonfunctional. We have therefore identified a region of the B. anthracis chromosome spanning 31,231 bp and 34 ORFs which is dispensable for the growth of B. anthracis in LB medium.
In the data presented here, genes in four different regions of the B. anthracis chromosome and plasmids were inactivated by homologous recombination. Thus, the Cre-loxP site-specific recombination method used here provides a robust and easily transferable approach to disrupting B. anthracis genes. The availability of the complete genomic sequence of B. anthracis (31) now makes it feasible to mutate any gene or putative ORF to analyze its role, as has been extensively done for B. subtilis (14). We have demonstrated the usefulness of the Cre-loxP recombination system in B. anthracis by disrupting genes sequentially. These results establish that the system can yield genetic modifications without the permanent establishment of antibiotic resistance markers. Removal of the antibiotic resistance markers assures that cell physiology will not be altered by known or unrecognized activities of the resistance genes or the antibiotic drugs themselves. Furthermore, the elimination of antibiotic resistance markers may increase the acceptance of B. anthracis for use in biotechnology applications, as in the production of recombinant PA and lethal factor (16, 17).
This method could prove useful in making extensive modifications to B. anthracis strains. For example, multiple proteases could be inactivated so as to limit degradation of expressed proteins, as was done for B. subtilis (23). However, a potential problem in this approach is that undesired recombination might occur between the loxP sites that accumulate after multiple rounds of gene disruption. Recombination could lead to deletion or inversion of large chromosomal segments, depending on the orientation of the loxP sites, the distances between them, and also the presence of essential genes in the intervening region. To produce defined recombinational events in strains containing multiple loxP sites, it may be possible to limit the activity of the Cre recombinase by growing the pCrePA transformants at partially restrictive temperatures or by expressing Cre from a tightly controlled, inducible promoter.
In addition to use in inactivating individual genes, the introduction of loxP sites offers a way of making large genome rearrangements in the B. anthracis chromosome or virulence plasmids, as was previously demonstrated in Lactococcus lactis (3, 4). Because the efficiency of Cre-mediated recombination does not depend on the distance between inverted loxP sites (3), the Cre-loxP system offers a powerful tool for functional analysis of large sets of B. anthracis genes. In the example shown here, we used a single deletion event to show the nonessentiality of 34 genes.
In the course of the work described here, we generated a strain, B. anthracis MSLL33, which may prove useful as a host for expression of recombinant proteins. The complete inability of the strain to make spores prevents laboratory contamination. In our experience, B. anthracis spoOA mutant strains grown in liquid culture die rapidly after reaching stationary phase (data not shown). The inactivation of the major secreted casein-degrading protease is likely to enhance the stability of secreted proteins, thereby increasing their yield and integrity. As noted above, the strain can be further improved by inactivation of additional proteases.
The B. anthracis strain McrB3P-Mrr-LΔ30 may also have potential value as a host strain for production of secreted recombinant proteins. Although many of the 34 deleted genes are annotated as “hypothetical,” several may alter relevant phenotypes. BA2308 is predicted to encode the sporulation control protein Spo0M. The B. anthracis Spo0M is highly homologous to the B. subtilis protein (62.3% similarity, 43.4% identity) and is part of an operon consisting of genes BA2305 to BA2309. Upstream of BA2305 is the sequence 5′-AGGATATGACCTATAAAAAAGAAAAACT-3′, in which the underlined bases exactly match the consensus for σH-dependent promoters (29). B. subtilis spo0M mutants are susceptible to lysis during growth and are impaired in their ability to sporulate (11). Another deleted gene that may affect sporulation is BA2291, a KinA homolog (55.4% similarity, 34.6% identity with the B. subtilis protein). KinA is a signal-transducing sensor kinase that phosphorylates spo0F and thereby contributes to the phosphorylation of Spo0A. B. subtilis kinA mutants are delayed in the onset of sporulation (26). Two other deleted genes (BA2288 and BA2310) encode proteins predicted to be involved in regulating ion fluxes. Based on this information, one could predict that strain McrB3P-Mrr-LΔ30 would be impaired in sporulation and osmotically fragile. Indeed, we found visible clearing in saturated LB cultures of McrB3P-Mrr-LΔ30 left overnight at room temperature, as well as a reduction in the number of viable cells compared to the parent strain, UM44-1C9. Given the seemingly normal growth rate of McrB3P-Mrr-LΔ30 in overnight cultures, its osmotic fragility, and its impaired sporulation capabilities, this strain may be termed a “crippled” avirulent strain and could therefore prove suitable in situations where laboratory safety is a primary concern. Another encouraging aspect is the transformability of strain McrB3P-Mrr-LΔ30 with supercoiled, Dam-methylated plasmid at reproducible, albeit low levels (R. Sitaraman, unpublished results). This strain can therefore be used as an intermediate or final host for B. anthracis plasmids, especially in situations wherein plasmid stability and replication fidelity are of great importance.
This research was supported by the Intramural Research Program of the NIH National Institute of Allergy and Infectious Diseases.
We thank Theresa Koehler (Department of Microbiology and Molecular Genetics, The University of Texas—Houston Health Science Center, Houston) for providing plasmid pUTE408, Craig E. Rubens (Department of Pediatrics, Division of Infectious Disease, Childrens' Hospital and Regional Medical Center and University of Washington, Seattle) for plasmid pHY304, and June R. Scott (Department of Microbiology and Immunology, Emory University School of Medicine, Atlanta, Ga.) for plasmid pJRS312.