|Home | About | Journals | Submit | Contact Us | Français|
Borrelia burgdorferi has multiple linear and circular plasmids that are faithfully replicated and partitioned as the bacterium grows and divides. The low copy number of these replicons implies that active partitioning contributes to plasmid stability. Analyzing the requirements for plasmid replication and partition in B. burgdorferi is complicated by the complexity of the genome and the possibility that products may act in trans. Consequently, we have studied the replication-partition region (bbb10-13) of the B. burgdorferi 26 kb circular plasmid (cp26) in Escherichia coli, by fusion with a partition-defective miniF plasmid. Our analysis demonstrated that bbb10, bbb11, and bbb13 are required for stable miniF maintenance, whereas bbb12 is dispensable. To validate these results, we attempted to inactivate two of these genes in B. burgdorferi. bbb12 mutants were obtained at a typical frequency, suggesting that the bbb12 product is dispensable for cp26 maintenance as well. We could not directly measure cp26 stability in the bbb12 mutant, because cp26 carries essential genes, and bacteria that have lost cp26 are inviable. Conversely, we were unable to inactivate bbb10 on cp26 of B. burgdorferi. Our results suggest that bbb12 is dispensable for cp26 maintenance, whereas bbb10, bbb11, and bbb13 play crucial roles in that process.
Stable maintenance of low copy number replicons in bacteria requires replication and active partitioning of plasmids to daughter cells (reviewed by (Ebersbach and Gerdes, 2005). Both processes typically depend on plasmid-encoded products recognizing specific sites on the plasmid. In the case of replication, rep proteins bind to the plasmid origin (ori) at the initiation of replication, whereas partition depends on a centromere-like sequence (parS) being bound by partition proteins.
Borrelia burgdorferi, a Lyme disease agent (Burgdorfer et al., 1982; Donahue et al., 1987; Lane et al., 1991), has a complex genome composed of multiple linear and circular plasmids, plus a linear chromosome (Barbour and Garon, 1987; Baril et al., 1989; Casjens et al., 2000; Fraser et al., 1997; Stevenson et al., 1997). The plasmid copy numbers are approximately one per chromosome (Beaurepaire and Chaconas, 2007; Hinnebusch and Barbour, 1992), which is approximately one per cell (Morrison et al., 1999). The chromosome and at least several of the plasmids carry genes required for either bacterial survival or for the bacteria to fulfill the natural mouse-tick infectious cycle (Byram et al., 2004; Grimm et al., 2005; Grimm et al., 2004; Jewett et al., 2007b; Jewett et al., 2009; Kobryn and Chaconas, 2002; Labandeira-Rey and Skare, 2001; Purser et al., 2003; Revel et al., 2005; Stewart et al., 2006; Strother et al., 2005; Strother and de Silva, 2005; Tilly et al., 2007; Tilly et al., 2006). In order to ensure stable maintenance, the plasmids and chromosome likely carry genes and sites required for faithful partition and replication. On the basis of sequence conservation, several gene families have been implicated in plasmid replication and partition, including homologs of parA (called PF32, for paralogous family), encoding putative ATPases required for segregation of plasmids in other bacteria (Casjens et al., 2000; Zückert and Meyer, 1996). Other paralogous gene families implicated in plasmid maintenance include PF57 and PF62, thought to encode replication proteins, and PF49 and PF50, of unknown function (Casjens et al., 2000). One or several of these genes is present on each B. burgdorferi plasmid (Casjens et al., 2000). Several studies (described below) have addressed the requirements for plasmid replication and partition in B. burgdorferi, but no simple and clear picture has yet emerged.
Regions capable of conferring autonomous replication and partition in B. burgdorferi were isolated from circular plasmids cp9 (Stewart et al., 2001), cp26 (Byram et al., 2004), one of several cp32 plasmids (Eggers et al., 2002), and linear plasmids lp25, lp28-1 (Stewart et al., 2003), and lp38 (Dulebohn et al., 2011). Shuttle vectors carrying those regions were capable of maintenance in B. burgdorferi, and dernonstrated incompatibility by displacing the resident plasmid, when possible. All such experiments, however, are complicated by the possibility that products required for replication or partition may be supplied in trans from another plasmid that is present in the spirochete. Cp9 appears to be related to the cp32 plasmids, but lacks a number of genes, including the PF32 member, encoding a potential ParA protein (Casjens et al., 2000; Dunn et al., 1994). Deletion analysis of this plasmid suggested that the product of cp9 PF49 is also not essential in cis for plasmid maintenance, although some or all of the gene sequence appears to perform an important function (Stewart et al., 2003; Stewart et al., 2001). The same authors found similar results for lp25, demonstrating that a shuttle vector carrying lp25 genes from paralogous families 32, 49, 50, and 57 conferred autonomous, although somewhat unstable, replication in B. burgdorferi, and that insertion of foreign sequences into the lp25 PF49 did not alter that ability (Stewart et al., 2003).
Further studies showed that the lp17 PF62 and cp32 PF57 were the only genes required for replication of their respective plasmids (Beaurepaire and Chaconas, 2005; Eggers et al., 2002) in B. burgdorferi. Again, interpretation of these results is complicated by the presence of other plasmids in B. burgdorferi, which may be capable of supplying replication or partition products in trans.
The 26 kb circular plasmid cp26 is unusual among the B. burgdorferi plasmids, since it carries genes whose products are essential for bacterial growth in all conditions, and others needed for survival in the mouse-tick cycle. The essential genes are resT (Byram et al., 2004), encoding the telomere resolvase required for linear DNA replication (Kobryn and Chaconas, 2002), along with bbb26 and bbb27, whose products are of unknown and potentially overlapping function (Jewett et al., 2007b; Lawrence et al., 2009). Genes required during the natural life cycle of the spirochete include ospC (Grimm et al., 2004), which plays an essential role at the initiation of mammalian infection, and guaA and guaB, encoding purine biosynthesis enzymes (Margolis et al., 1994), which are required for mouse infectivity and contribute to growth within ticks (Jewett et al., 2009); Lawrence et al., 2009). The presence of these genes on cp26 ensures that this plasmid is present in all bacteria during propagation in culture and found in all natural isolates. Presumably, plasmid stability is ensured by functional replication and partition functions, combined with the lethality of loss. We used the replication and partition region of cp26 as the starting point for our dissection of those functions. To facilitate analysis and avoid problems of potential trans-complementation with functions encoded by other B. burgdorferi plasmids, we assessed the effect of adding B. burgdorferi sequences to a derivative of the E. coli plasmid F (miniF) lacking the sopABC region, which is responsible for partitioning of this plasmid, using E. coli as a surrogate host. This plasmid (pDAG203; (Lemonnier et al., 2000) has been used to characterize partition functions from other bacteria, such as Burkholderia cenocepacia (Dubarry et al., 2006) and Pseudomonas putida (Godfrin-Estevenon et al., 2002).
We cloned the cp26 region capable of promoting stable plasmid replication and partitioning in B. burgdorferi (Byram et al., 2004) into pDAG203, generating plasmid miniF26. We then systematically deleted genes, generating miniF plasmids with various subsets of the genes and sites present in the bbb10-13 region, and assessed their stability during growth in E. coli without selection. These studies demonstrate that B. burgdorferi plasmid maintenance functions can be analysed using a heterologous system in E. coli, which is a new and powerful tool for studying the mechanism of spirochete plasmid segregation.
Plasmid pBSV26 (Byram et al., 2004) was digested with Spel to release a 3417 bp fragment containing the cp26 genes bbb10-13 (Fig. 1A). This DNA fragment was ligated with pDAG203 (Lemonnier et al., 2000) that had been linearized with NheI, yielding miniF26 (Fig.1B). Derivatives of miniF26 with only bbbll-13 (miniF26Δbb10), with bbb10, bbb12 and bbb13 (miniF26Δbbbll), and just bbbl2 and bbbl3 (miniF26Δbbb10-11) were constructed using PCR-mediated deletion (Fig. 2). Primers extending out from the translational start and stop sites of the genes (8 and 9 to remove bbb10, and 10 and 11 to remove bbbll; Table 1) were used to amplify the remainder of pBSV26, thereby deleting the intervening coding sequences and adding XhoI sites. miniF26 with just bbbl2 and bbb13 (called miniF26Δbbb10-11) was constructed by amplifying the remainder of pBSV26 with primers 8 and 11. In all cases, PCR products were digested with XhoI and DpnI (to remove the pBSV26 template) and self-ligated to recircularize the plasmid. The B. burgdorferi plasmid sequences were excised from the resulting plasmids with SpeI, purified from agarose gels using a MinElute kit (Qiagen, Valencia, CA), and ligated with pDAG203 that had been digested with NheI and treated with Antarctic Phosphatase (Fig. 2).
MiniF plasmids with bbb10, bbb11, and bbbl3 (called miniF26Δbbbl2) or bbb10, bbb11, and bbbl2 (called miniF26Δbbb13) were made using restriction sites in miniF26 (Fig. 2). MiniF26 was digested with MfeI (for the bbbl2 deletion) or ScaI (for the bbbl3 deletion) and religated. A miniF derivative with just bbbl3 (called miniF26Δbbb10-l2) was constructed starting with miniF26Δbbb10-ll, digesting with MfeI, and self-ligating. A miniF with no functional coding sequences derived from B. burgdorferi, but retaining the non-coding sequences, was derived from miniF26Δbbb10-12 by digesting with ScaI and religating (Fig. 2).
MiniF26bbb11*, with replacement of a GAA codon at amino acid 15 of bbb11 with the stop codon TAA, was created by overlapping PCR. Products amplified with Vent polymerase (New England Biolabs) using primer pairs 1-15 and 14-2 were digested overnight with DpnI, and then used as substrate in a reaction with primer pair 1-2, also using Vent polymerase. The resulting product was cloned into TopoXL. A BamH1-EcoRV fragment including the bbb10 and bbb11* genes was substituted for the resident copies of those genes in BamH1-EcoRV-digested pBSV26. The bbb10-l3 region was excised from the resulting plasmid with SpeI and ligated with NheI-digested pDAG203.
Allelic exchange vectors for attempting to make deletion-insertion mutations in the bbb10 and bbbl2 loci in B. burgdorferi were constructed as follows. A plasmid for inactivating bbb10 was derived from pBSV26Δbbb10 by digesting with XhoI and ligating with a flgBp-kan cassette with XhoI sites on its ends (created by amplifying with primers 12 and 13), yielding plasmid pBSV26Δbbb10::flgBp-kan. To target bbbl2, plasmid pBSV26 was digested with SpeI and NcoI and the bbb10-13-containing fragment eluted from a gel with the MinElute Kit (Qiagen). This fragment was cloned into a version of the pCR-XL-Topo plasmid (Invitrogen) that had been digested with EcoRI to remove an irrelevant insert, self-ligated, and digested with SpeI. To inactivate bbb12, the resulting plasmid (TopoXL-bbb10-l3#3a) was digested with MfeI and ligated with a flgBp-aacC1 cassette that had been digested out of pCR2.1 Topo with EcoRI, yielding plasmid TopoXL-Δbbbl2::flgBp-aacC1. Versions of this plasmid with the flgBp-aacC1 cassette in either orientation were obtained.
Final constructs were confirmed by sequencing. Enzymes were purchased from New England Biolabs (Ipswich, MA). Plasmids were transformed into frozen competent HB101 (Promega, Madison, WI) or into other E. coli strains by CaCl2-mediated transformation (Mandel and Higa, 1970).
The bbb10 and bbb11 genes, along with the potential promoter upstream of bbb10, were amplified with Vent polymerase, using primers 1 and 2 and cloned into pZeroBluntTopo (Life Technologies, Carlsbad, CA). A clone with the correct sequence was digested with enzymes NsiI and BamHI and the resulting fragment was ligated with NsiI-BamHI-digested pUCR6KT, which contains a miniTn7 that will insert DNA into a unique site on the E. coli chromosome (Choi et al., 2005). Ligations were transformed into SM10(λpir). DNA was prepared from a colony with the correct plasmid and used to co-transform HB101, in which pUCR6KT cannot replicate, with pTNS2 (a helper plasmid encoding the Tn7 transposase). We selected for gentamicin-resistance conferred by miniTn7 insertion into its chromosomal insertion site, bringing with it the bbb10 and bbb11 genes. We then assessed the sability of various miniF26 derivatives in this strain.
Integrated DNA Technologies (IDT, Coralville, IA) synthesized a fusion between the presumed bbb10-l3 promoter and the start codon of bbbl3, using proprietary technology.
Cultures grown overnight at 37° in L broth containing 30 μg/ml chloramphenicol (selecting for miniF plasmid retention) were diluted in duplicate into medium lacking antibiotic. At several time intervals, dilutions were plated on solid medium lacking antibiotic and cultures were further diluted so that they remained in logarithmic growth. Colonies were replica-plated to solid medium containing and lacking chloramphenicol to assess plasmid retention and doublings. All experiments were performed at least twice, and most were performed several times by two different people. Data sets were compared by Two-way ANOVA with a Bonferroni post-test using GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA). For statistical analysis of data sets in which numbers of doublings for different strains were not identical, the average number of doublings was used.
We used TaqMan analysis to assess transcript levels for bbb10 and bbbl2 in E. coli carrying various derivatives of miniF26. Transcript levels of the E. coli chromosomal gene dxs were used to normalize the qPCR. RNA was prepared from E. coli using a Nucleospin RNA II kit (Machery-Nagel, Bethlehem, PA) and cDNA generated with a High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Foster City, CA). Quantitative PCR was carried out using Universal PCR Master Mix (Applied Biosystems), primers and probes (Table 1; Sigma-Genosys), and an Applied Biosystems 7900 instrument.
bbbl2 of cp26 in B. burgdorferi strain B31-A34 (Jewett et al., 2007a) was inactivated by allelic exchange, following electroporation with 5 μg of plasmid TopoXLΔbbbl2::flgBp-aacC1 as previously described (Elias et al., 2003; Samuels, 1995). After recovery overnight in 5 ml BSKII medium, transformations were plated in solid BSKll medium containing 40 μg/ml gentamicin. Colonies arising after 5-10 days at 35°C with 2.5% CO2 were screened for the presence of the flgBp-aacC1 fusion using primers 3 and 4. Transformants were obtained at a frequency of 1.5×10-6. Several of these colonies were picked into BSKII medium and grown to high density, at which time genomic DNA was prepared using a Wizard Genomic DNA Purification kit (Promega, Madison, WI). The structune of the bbbl2 mutation was confirmed using primer pairs 5-6 and 11-7, which demonstrated that at least 5 of 13 transformants analyzed had the appropriate deletion-insertion mutation at the bbbl2 locus (data not shown). These PCR reactions also confirmed the absence of a wild type bbbl2 locus. Further confirmation of the bbbl2 deletion-insertion mutation and lack of a wild type bbbl2 locus was obtained by Southern blot hybridization (data not shown).
We attempted to inactivate bbb10 in B. burgdorferi by the same procedure used for bbbl2 inactivation, with selection on medium containing 200 μg/ml kanamycin. Upon analysis, all of the clones isolated retained the wild type locus. One clone had both wild type and mutant bbb10 loci, yielding an allelic exchange frequency of 3×10-8. Transformation with linearized DNA, to eliminate the possibility of integration or autonomous replication of the entire transforming plasmid, did not yield any clones that had undergone allelic exchange events.
We prepared DNA for assessing copy number of various miniF-derivative plasmids using a modification of a published method (Beaurepaire and Chaconas, 2005) that does not include any nucleic acid precipitation or column binding steps. Briefly, we froze pelleted E. coli cultures, resuspended them in Tris-EDTA-NaCl (0.1M, pH8.2-10mM-30mM) with 1 mg/ml lysozyme, and incubated on ice for 45 min. We then added 200μl of 0.5%SDS-50mM Tris pH7.5-0.4M EDTA with 1 mg/ml proteinase (Life Technologies) and incubated for one hour at 55°C. We then extracted twice with phenol-CHCl3 (1:1) and once with CHCl3, before dialysing against 10mM Tris (pH 8) at room temperature for 1.5 hr. The DNA preparations were diluted to 50μg/ml, and relative copy number assessed in triplicate TaqMan analysis using miniF and dxs primer-probe sets (primers 16, 17, 22, and 23, and probes 1 and 4; Table 1). DNA extracted from HB101 carrying miniF26 was used for the standard curves. Genome copies per microgram of DNA were calculated based on the genome size of E. coli.
A region of cp26 containing the genes bbb10-bbbl3, plus some flanking sequences (Fig. 1A), was previously shown to be sufficient to support replication and partition of an E. coli shuttle vector in B. burgdorferi (Byram et al., 2004). We cloned this region into pDAG203, which is a miniF derivative lacking partition functions (Lemonnier et al., 2000). When the resulting plasmid, called miniF26 (Fig. 1B), was transformed into E. coli strain HB101, it was significantly more stable than pDAG203 when shifted from growth in the presence of chloramphenicol to antibiotic-free medium (p<0.01-0.001; Fig. 3). Inserting an irrelevant segment of DNA into the miniF plasmid did not confer stabilization (Fig. 3). pDAG203 appeared to be surprisingly stable for a low copy number plasmid lacking a functional partition system, so we rescued the plasmid from HB101/pDAG203 and retransformed it into HB101. Stability was again assessed, taking care to minimize the time between transformation and stability measurement. In this case (Fig. 3), pDAG203 was lost much faster than in the original test. In contrast, miniF26 was similarly stable in HB101 when present in a fresh transformant or in HB101/miniF26 grown from a frozen stock (Fig. 3). These data suggest that growth of HB101/pDAG203 in the presence of chloramphenicol, which selects for the presence of the plasmid, exerts a powerful selection for stabilization by an unknown mechanism, which does not involve mutation of the plasmid DNA. Partial sequence analysis of the rescued DNA supported this idea, in that no differences between the original and rescued plasmids were detected.
The cp26 region that stabilizes miniF contains a number of interesting features. First, there are four open reading frames, oriented in the same direction, and preceded by a potential promoter (Figs.1A and 1C). Although the predicted products of all four open reading frames belong to previously identified paralogous families associated with putative plasmid origins in B. burgdorferi, only BBB12, a ParA homolog, resembles a protein found in other bacteria (Casjens et al., 2000). BBB12 is predicted to start at a codon within the terminus of the BBB11 coding sequence, based on the position of nucleotides encoding a potential Walker box, involved in ATP hydrolysis by ParA proteins (Casjens et al., 2000; Zückert and Meyer, 1996). Although we have not mapped the promoter of the bbb10-13 genes, we presume that it is located in the 243 bp preceding the start codon of bbb10 that are located within the region used for constructing pBSV26 (Byram et al., 2004) and miniF26 (Fig. 1C). The ability of this region to confer stable maintenance on a B. burgdorferi shuttle vector (Byram et al., 2004; Jewett et al., 2007a) argues in favor of this hypothesis. Searching for a bacterial promoter using BPROM (Soffberry, Inc., Mt. Kisco, NY) predicted a promoter about 200 bp upstream of the bbb10 start site (Fig. 1C), but B. burgdorferi promoter criteria have not been fully characterized, and several other potential -35 regions are located closer to the bbb10 start site. TaqMan analysis (see Section 3.6) demonstrated that transcription of at least bbb10 and bbbl2 was correlated, supporting the idea that the bbb10-13 genes are co-transcribed.
Between bbbl2 and bbbl3 is a region that may comprise the parS, with complex inverted and direct repeats (Fig. 1D). We identified this site by analyzing the entire bbb10-13 region with the GeneQuest program from the Lasergene suite (DNASTAR, Madison, WI) for direct or inverted repeats of at least 10 nucleotides. All identified repeats were compared to the potential centromere from the chromosome and to other parS homologs (Lin and Grossman, 1998; Livny et al., 2007). In the bbb10-1 region, only the sequence between bbbl2 and bbbl3 contains an inverted repeat that is homologous to previously described parS sequences. Interestingly, the B. burgdorferi chromosomal parS is located between gyrA, encoding a gyrase subunit, and BB0434, which has been annotated as a parB homolog, involved in parS binding (Lin and Grossman, 1998; Livny et al., 2007). No other B. burgdorferi plasmid has a potential parS that was identified by this method.
Finally, downstream of bbbl3 is a potential origin of replication. This region contains multiple direct and inverted repeats and AT-rich regions (Fig 1E). Also, this site is the approximate location of the the minimum GC-skew (a characteristic often found at replication origins) (Picardeau et al., 2000).
We used two strategies to delete individual genes in the bbb10-13 region of miniF26 (Fig. 2; see Materials and Methods for details). To remove bbb10, bbb11, or both genes, we used PCR-mediated deletion. For bbbl2 and bbbl3 deletions, we used restriction sites to remove large portions of the coding sequences of each gene. The deletions seemed unlikely to have polar effects on downstream genes, although altering the location of the potential promoter with respect to those genes may affect transcription.
We tested the stability of the miniF26 derivatives, and found that miniF2Δbbb10, miniF26Δbbb11, miniF26Δbbb10-bbbll (Fig. 4A) and miniF26Δbbbl3 (Fig. 4B) had greatly reduced stability comparecd to miniF26 (p<0.001, except for two T0 values). Surprisingly, miniF26Δbbbl2 was as stable as miniF26, containing the entire bbb10-13 region (Fig. 4C).
We constructed a chain termination mutation in bbb11, to see if stopping translation without removing the gene sequence had a similar effect on miniF stability. The stability of the plasmid with the point mutation (bbbll*) was indistinguishable from that of miniF26Δbbb11 (p>0.05; Fig. 4D), suggesting that the phenotype was caused by lack of the BBB11 product, not the absence of a nucleotide sequence within the deleted region.
We made two additional deletions, in an effort to determine if the instability of other miniF derivatives could be explained by polar effects, leading to decreased bbbl3 expression. The first derivative combined the bbb10-bbbll and bbbl2 deletions, yielding a miniF that has only the predicted promoter (up to the bbb10 predicted ribosome binding site), potential parS, bbbl3, and potential replication origin (Δbbb10-12; Fig. 2). We predicted that linking the bbb10 ribosome binding site to the start of the bbbl2 open reading frame, combined with the bbbl2 deletion, which has no effect on plasmid stability, would relieve any possible polarity resulting from the bbb10 and bbb11 deletions. The second miniF derivative combined all four deletions, creating a miniF that has only the potential promoter, parS, and replication origin (Δbbb10-13; Fig. 2). These plasmids had the same phenotype as miniF26Δbbb10, miniF26Δbbb11, miniF26Δbbb10-11, and miniF26Δbbb13 (Fig. 4), i.e., more than 90% plasmid loss by 30 doublings (data not shown), demonstrating that the sequences remaining do not confer any stability on pDAG203. These data support the idea that the phenotypes of the miniF bearing bbb10 and bbb11 deletions are not due to polar effects on the bbbl3 gene, but rather the consequences of loss of functions important for miniF stabilization.
In an endeavor to restore stable maintenance to the miniF derivatives lacking bbb10, bbb11, or both, we constructed an E. coli strain in which the bbb10 and bbb11 genes, including the 5′-flanking region, were integrated into the chromosome (Materials and Methods). Substantial transcription of bbb10 from this strain was confirmed by TaqMan analysis of cDNA (data not shown). We introduced pDAG203 (no insert), miniF26 (bbb10-13), miniF26Δbbb10, miniF26Δbbb11, and miniF26Δbbb10-ll into that strain and assayed their stabilities. miniF26, with bbb10-13, remained stable in this strain. All other plasmids tested, including pDAG203, with no B. burgdorferi DNA, were extremely unstable (Fig. 5), as was seen for fresh transformants with pDAG203 (Fig. 2). We conclude that expressing bbb10-11 from a chromosomal location did not stabilize miniF lacking one or both of those genes.
In an attempt to stabilize miniF26Δbbb13, we created a version of bbb13 in which the putative promoter 5' of bbb10 and the start codon and open reading frame of bbb13 were fused (Materials and Methods). The bbb13 construct was cloned into the high copy plasmid pUC57 and transformed into a strain carrying miniF26Δbbb13. This plasmid did not stabilize miniF26Δbbb13, when both were contained within the same E. coli strain (data not shown). We were unable to confirm bbb13 transcript levels (due to a lack of functional primer-probe sets for bbb13, despite testing several possible combinations), so it remains possible, although unlikely, that this plasmid does not provide adequate levels of BBB13. Several possibilities may explain the failure of these two attempts to restore the stability of miniF derivatives by supplying missing B. burgdorferi products in trans. Perhaps the expression ratios of some or all of the bbb10-13 products are crucial to plasmid stability, these products act optimally in cis, on sequences proximal to their coding regions (Dong et al., 1988), or an important site has been deleted from the bbb10-13 region in these miniF derivatives.
In an attempt to determine if the bbb10-13 genes were co-transcribed and as an additional test for indirect effects of deletion of one gene on expression of another, we performed TaqMan quantitative PCR analysis on cDNA derived from E. coli carrying various miniF derivatives. Despite trying three primer-probe sets for both bbb11 and bbb13, we were unsuccessful in amplifying either from DNA or cDNA. In contrast, our primer-probe sets for bbb10 and bbb13 amplified well, so we were able to assess transcript levels for those genes in E. coli containing miniF with no insert (negative control), with bbb10-13, and with all combinations of three of the genes (Fig. 6). As expected, no transcript was detectable for either gene when no B. burgdorferi genes were inserted in the miniF. Both transcripts were detectable when bbb10-13 were present, but the bbb10 transcript level was approximately 10 times greater than that of the bbbl2 transcript. When cDNA was derived from HB101 containing miniF plasmids lacking either bbb10 or bbbl2, transcript of the corresponding gene was not detected. Interestingly, miniF plasmids lacking any one of the four B. burgdorferi genes had somewhat reduced levels of both bbb10 and bbbl2 transcripts, suggesting that the bbb10-13 genes are co-transcribed and that altering the sequence of the mRNA affects transcript level but not necessarily plasmid stability.
Since the bbbl2 deletion did not alter the stability of miniF26, we decided to introduce this mutation into B. burgdorferi cp26, to extend our studies in E. coll. In order to facilitate allelic exchange, we made a deletion-insertion version of the mutation, in which a B. burgdorferi selectable marker (flgBp-aacC1) was inserted in place of most of the bbbl2 gene, using the MfeI sites used for the deletion in miniF26Δbbbl2 (Fig. 2).
miniF26Δbbbl2::flgBp-aacC1 (with the flgBp-aacC1 cassette in either orientation) was used to tranform B31-A34, a highly transformable strain lacking the restriction enzyme systems encoded by lp25 and lp56 (Jewett et al., 2007b; Kawabata et al., 2004; Rego et al., 2011). Since cp26 must be retained for B. burgdorferi viability, we used criteria other than plasmid stability to assess the contribution of bbbl2. First, the mutant was obtained at a normal frequency for allelic exchange in B31-A34 (1.5×10-6), when the flgBp-aacC1 cassette was in the same orientation as the bbb10-13 genes. Interestingly, we obtained very few transformants when the flgBp-aacC1 insertion was in the opposite orientation, in which transcription from the flgB promoter could potentially interfere with transcription of bbb10 and bbb11. When analysed further, these few transformants retained wild type copies of the bbb10-13 region, again suggesting that BBB10 and BBB11 play crucial roles in cp26 maintenance. Second, the Δbbbl2 transformants had no altered colony morphology or reduced growth rate (data not shown), as would be expected if a plasmid carrying essential genes had reduced stability. Finally, PCR and Southern blot analysis of a subset of the transformants demonstrated that 5 of 13 tested had the expected insertion-deletion and all had approximately wild type cp26 copy number (data not shown). Among the other transformants were bacteria in which the allelic exchange vector appeared to have integrated in its entirety, and transformants with apparent cp26 dimers (data not shown), both events previously observed after attempted allelic exchange targeting cp26 genes (Bono et al., 1998; Byram al., 2004; Stevenson et al., 1998; Tilly et al., 1998). We never observed autonomous replication of the transforming plasmid, which presumably could function as a shuttle vector, most likely because of incompatibility with the essential plasmid cp26. These data suggest that bbbl2 is not required for cp26 maintenance in B. burgdorferi, at least in the presence of several other plasmids, consistent with our studies of miniF26 derivatives in E. coll.
To further validate our use of miniF stabilization by B. burgdorferi genes in E. coli as a surrogate for gene inactivation in B. burgdorferi, we attempted to inactivate bbb10 on cp26. Since miniF26Δbbb10 was unstable and cp26 is an essential plasmid in B. burgdorferi, we predicted that inactivation of bbb10 would not be possible because transformants would be inviable. To facilitate allelic exchange, we inserted a flgBp-kan cassette in place of bbb10 in pBSV26 (Materials and Methods) and used this DNA to transform B. burgdorferi B31-A34. Transformant colonies were obtained 10-fold less frequently than in the transformation inactivate bbb12. In contrast to our transfomations with miniF26Δbbb12::flgBp-aacC1, we failed to isolate cp26 without an active bbb10 gene. Instead, we obtained one clone that carried both wild type and mutant copies of bbb10 (which would result from integration of the transforming plasmid, autonomous replication of the pBSV26Δbbb10::flgBp-kan, or allelic exchange into one copy of bbb10 in a dimer of cp26). Transforming with linearized DNA, to eliminate the possibility of integration or autonomous replication, did not yield any transformants. Our inability to inactivate bbb10 in B. burgdorferi is consistent with the profound instability of miniF26 lacking bbb10 in E. coli and the previous assignment of PF62 as essential for replication in B. burgdorferi.
In an effort to determine if the bbb10-13 genes increased miniF stability by increasing replication, partition, or both, we assessed the copy numbers of various miniF derivative plasmids relative to the E. coli chromosomal gene dxs. To minimize the effects of differential precipitation or column binding by chromosomal vs plasmid DNA, we used a DNA extraction method that does not include these steps (section 2.7). Copy number was assessed by TaqMan analysis, using dxs and miniF primer-probe sets (Table 1). Although we found an increased copy number of stable miniF derivatives (Fig. 7), the levels appeared to be low enough for active partitioning to be required for stable maintenance (see Discussion). Similar results were obtained by Southern blot analysis (data not shown). These data suggest that the bbb10-13 products provide active partitioning functions, and may also increase replication of the miniF slightly.
B. burgdorferi, with its complex genome composed of multiple linear and circular plasmids and a linear chromosome, faces unusual problems in DNA replication and partition. Although a number of studies have addressed the requirements for these processes, all were complicated by the possibility of trans-acting products contributed by other replicons present in the bacterium. We have examined the replication-partition region of cp26 in isolation, by cloning portions of that region into a miniF plasmid defective in partitioning and studying plasmid stability in E. coll. These studies indicated that bbb10, bbb11, and bbb13 were all required for stabilization of the miniF, whereas bbbl2 was dispensable. This result seems surprising, because bbbl2 (the PF32 member) is the only gene whose product resembles those found in other bacteria, being homologous to the essential partition protein ParA. However, similar results were obtained with other plasmids in B. burgdorferi (Beaurepaire and Chaconas, 2005; Eggers et al., 2002; Livny et al., 2007; Stewart et al., 2003; Stewart et al., 2001) and our findings are also consistent with the absence of genes from PF32 on cp9 and lp5. More recent biochemical experiments demonstrated that the PF32 member encoded on lp17, though not essential for replication, interacts with the protein required for replication (Deneke and Chaconas, 2008). These data suggest that the bbbl2 product could function as an auxiliary replication protein, in contrast to the ParA roe in other bacteria.
Another surprising aspect of these findings is that all three of the other products encoded in the region conferring autonomous replication are required for miniF stabilization. These include BBB10, which has been speculated to be the replication protein, based on a number of experiments and criteria described in the Introduction. Also, bbb13 belongs to paralogous family 49, whose products were dispensable for stable shuttle vectors derived from cp9 and lp25 (Stewart et al., 2003; Stewart et al., 2001), so perhaps bbb13 contains an essential site, rather than encoding an essential product. Our copy number data suggest that there is around one miniF26 plasmid copy per chromosome, a level that requires partition functions for stability (Nordstrom and Austin, 1989). Estimating absolute copy number from these data, however, requires the assumption that the miniF and dxs primer-probe sets amplify similarly, since HB101/miniF26 DNA was used for the standard curves. These data also demonstrate a three-fold lower copy number in unstable vs stable miniF derivatives (Fig. 7). Perhaps the BBB10-BBB13 products form a multi-protein complex that carries out both replication and partition functions.
Our analyses were unable to definitively exclude all poWbinty of effects of deleting one gene on the transcription or translation of another gene in the region. The fact that deleting any gene in the bbb10-13 region affects transcript levels of both bbb10 and bbbl2 suggests that all four genes may be co-transcribed. The 10-fold difference between gene transcript levels in bacteria containing miniF26 is consistent with transcription attenuation. Although we were unable to measure bbb11 and bbb13 transcript levels, several lines of reasoning suggest that the products of bbb10, bbb11, and bbb13 are directly involved in stabilizing pDAG203. First, the stability of miniF26Δbbbl2 means that expression of the other genes in bacteria carrying this plasmid is adequate for stable maintenance of miniF. Second, the reductions in bbb10 transcript from the unstable Δbbb11 and Δbbbl3 plasmids were similar to that found from the stable Δbbb12 (which presumably provides adequate BBB10 product), so these genes likely contribute essential stabilization functions also. Finally, the instability of the miniF26Δbbbl3 plasmid is less likely to be due to indirect effects than any of the others, since bbb13 is the last gene in the series.
The gene encoding the ATPase required for cp26 segregation remains unknown. In B. burgdorferi, bbbl2 may normally carry out this function, but the PF32 proteins encoded on other plasmids or the SpoOJ encoded on the chromosome (BB0434) may enable cp26 partition, when bbbl2 is mutant. This possibility is difficult to test.
Remaining questions include why was complementation unsuccessful? Improper ratios of the B. burgdorferi proteins is a possible explanation for the inability of chromosomally expressed bbb10 and bbb11 to compensate for their lack on the miniF. If this model were correct, however, miniF26 might be expected to be unstable in the strain with bbb10 and bbb11 expressed from the chromosome, because of the excess BBB10 and BBB11. Another possibility is that the proteins are preferentially cis-acting, as was found for other Rep proteins (Dong et al., 1988). Since transcription and translation are likely to be coupled, the concentration of the proteins in the region from which they are encoded would be higher, leading them to bind preferentially to sequences in that region. Such binding could lower the effective concentration within a bacterium, leading to inability to complement segregation of the miniF plasmid.
A second question is how E. coli stabilizes the miniF lacking partition functions. Although the mechanism may have no relevance to B. burgdorferi plasmid maintenance, it is relevant to the data presented here, since we used frozen aliquots of control strains to provide experiment-to-experiment reproducibility, and pDAG203 was more stable in the frozen strain than immediately after transformation. One possibility is that the miniF copy number increases, but we have seen no evidence of this by Southern blotting (data not shown).
Although these and other data demonstrate that B. burgdorferi plasmids carry partition systems, so far there is no evidence for the presence of other plasmid stability systems. Post-segregation killing systems, characterized by a stable toxin and an unstable antitoxin (Sengupta and Austin, 2011), would most likely render the plasmids more stable during bacterial growth in culture, when many are lost. The cp32 family, with several very similar plasmids that appear to be prophages, may possess such a system, but this question has not been addressed experimentally, since all isolates contain at least one cp32 and several genes are unique to this family. B. burgdorferi may have dimer resolution systems (Sengupta and Austin, 2011), but cp26 and lp54 dimers have been observed (Marconi et al., 1996; Tilly et al., 1998), and appear to be relatively stable during spirochete growth in culture. Maybe the existence of genes important for spirochete growth and survival at various stages of the infectious cycle, in combination with efficient partitioning mechanisms, is sufficient to guarantee that a substantial proportion of the bacterial population retains the plasmid throughout the natural life cycle of the spirochete. Also, the lack of any free-living stage in this cycle provides ongoing pressure for plasmid maintenance.
A number of aspects of B. burgdorferi plasmid maintenance remain areas for fruitful study. First, similar complementation and mutational analysis of the rep-par region of a plasmid that is dispensable for growth in culture, such as lp25, would facilitate confirmation of mutant phenotypes in B. burgdorferi. Second, there are hypothesized to be two classes of replication proteins, encoded by members of the paralogous families 57 and 62. Demonstrating that these products are required for replication initiation (or some other step in replication) has yet to be done. Purifying the BBB10-13 products, demonstrating DNA binding by BBB10, and ultimately, in vitro replication, would be useful. This biochemical approach might also confirm the location of the origin in the bbb10-13 region. A similar approach could also delineate the parS region and which, if any, protein binds to that site. Our results lay the foundation for these more detailed biochemical studies by demonstrating that the B. burgdorferi proteins function in E. coli and identifying the crucial contributions of the BBB10, BBB11, and BBB13 products to miniF stability.
We thank Dr. David Lane (CNRS, Toulouse, France) for providing pDAG203 and bacterial strains. We thank Laura Cornelisse for help in characterizing the stability of the miniF26Δbbbl2. We thank Dan Sturdevant (Rocky Mountain Labs Research Technology Branch) for TaqMan primer-probe set design and Craig Martens (Rocky Mountain Labs Research Technology Branch) for constructing a phylogenetic tree with bbbl2 and related sequences. We are grateful to Jeff Skinner (Bioinformatics and Computational Biosciences Branch, NIAID) for advice on statistical analysis. We thank Paul Beare, John Carlson, Philip Stewart, Daniel Dulebohn, and Amit Sarkar for their helpful suggestions on the manuscript. Anita Mora and Austin Athman provided expert figure preparation. This research was supported by the Intramural Research Program of the NIH, NIAID.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.