Purification and Properties of the Plasmid Maintenance Proteins from the Borrelia burgdorferi Linear Plasmid lp17

ABSTRACT The Lyme disease spirochete Borrelia burgdorferi carries more plasmids than any other bacterium, many of which are linear with covalently closed hairpin ends. These plasmids have also been referred to as mini-chromosomes and essential genetic elements and are integral components of its segmented genome. We have investigated two plasmid maintenance proteins, BBD14 (the replication initiator) and BBD21 (a presumptive ParA orthologue), encoded by the linear plasmid lp17; these proteins are representatives of paralogous families 62 and 32, respectively. We have purified recombinant 6-his-BBD21 and shown it possesses an ATPase activity. 6-his-BBD14 initially could not be overexpressed in Escherichia coli by itself. It was only effectively overproduced in recombinant form through coexpression with other B. burgdorferi proteins and codon optimization. Although the mechanism for increased production through coexpression is not clear, this method holds promise for expression and purification of other B. burgdorferi proteins, a number of which have remained recalcitrant to purification from E. coli. Finally, we present evidence for the physical interaction of BBD14 and BBD21, a feature suggesting that BBD21 and the paralogous family 32 proteins are more likely involved in DNA replication than functioning as simple ParA orthologues as previously surmised based upon sequence homology. Such a role would not preclude a function in plasmid partitioning through interaction with the replication initiator.

The spirochete Borrelia burgdorferi and related Borrelia species are the causative agents of Lyme disease, a debilitating illness that has been reported by the U.S. Center for Disease Control and Prevention to be the fastest growing zoonotic disease in North America (40,41). The genome of B. burgdorferi is highly segmented (8,17) and contains a number of circular and linear plasmids, also referred to as mini-chromosomes (3) and essential genetic elements (42). The linear plasmids as well as the linear chromosome are also unusual in that they are terminated by covalently closed hairpin ends (2,9,24) whose generation by the telomere resolvase ResT has been characterized in depth (1,13,(28)(29)(30)(45)(46)(47). However, the mechanisms involved in the peaceful coexistence of about two dozen DNA molecules remain unknown. Faithful replication and partitioning of these plasmids is of crucial importance to the organism, as a number of essential genes are carried by these extrachromosomal elements (7,26,27,31,36,37,39,42,50).
Our knowledge of the processes involved in plasmid maintenance in Borrelia species is rudimentary, at best. It is known that most of the plasmids of B. burgdorferi each carry between two and four members of five paralogous gene families ( Fig. 1) believed to be involved in plasmid maintenance (8). The region encoding the clustered paralogous family (PF) members from several linear and circular plasmids has been shown to be sufficient to confer autonomous replication ability and therefore likely carries both the plasmid origins and the trans-acting factors involved (4,15,43,44). Expression levels of some of these proteins are also correlated with plasmid copy number (5).
Every B. burgdorferi plasmid carries a member of either PF57 or PF62, two families that display low but significant similarity (8). Because of the universal presence of one of these related families on every plasmid, they are the most likely candidates for replication initiator proteins. PF57 is more commonly encountered and is found on 17 of the 22 plasmids. Some plasmids carry two apparently intact copies of the putative replication initiators and several plasmids carry initiator pseudogenes; these are the result of genome rearrangements in the linear plasmids (8) that are believed to occur via reversal of the telomere resolution reaction that generates the covalently closed hairpin ends on the linear plasmids (9,29). No functional motifs have been identified in the PF57 and PF62 families, which only display significant similarity to proteins present on plasmids in other Borrelia species (22,23).
It is known that replication of the linear B. burgdorferi chromosome starts from the center of the chromosome and proceeds bidirectionally toward the hairpin ends (35). Studies on the linear plasmid lp17 have shown that the region essential for replication is a 1.8-kb stretch from the center of the plasmid and that expression of BBD14, the PF62 member, is required (4). The origin must reside somewhere within this 1.8-kb region, but the precise location remains to be established. lp17 also carries a second putative plasmid maintenance gene, bbd21. This gene encodes a paralogous PF32 member (BBD21) with ϳ25% sequence conservation with a variety of members of the ParA family of partitioning proteins (8,51). Bacterial partitioning systems come in two types: those that encode actin-like ATPases and those that encode Walker box ATPases (14). The latter type includes the ParA family of partitioning systems, which have three components: two transacting proteins and one cis-acting DNA site (20,21). The centromere-like site (parS) is recognized by the ParB protein.
The ParA ATPase then interacts with the ParB-parS complex to promote partitioning. The Walker A and Walker B boxes found in ParA members (19) are shown in Fig. 2. BBD21 lacks the N-terminal extension found in some ParA orthologues. All B. burgdorferi plasmids, with the exception of cp9 and lp5, carry a PF32 member; however, none of the plasmids encodes a ParB orthologue. Removal of the bbd21 gene from lp17 does not seriously affect plasmid maintenance (4); however, complementation by a PF32 member from a different plasmid cannot be ruled out, and the function of the PF32 members remains unknown.
Purification and characterization of B. burgdorferi plasmid maintenance proteins have not been previously reported. In this study we have overcome the inability for BBD14 overexpression through coupled codon optimization and coexpression. We have purified 6-his-BBD14 and 6-his-BBD21 and report an ATPase activity for 6-his-BBD21. We also present evidence for a physical interaction between BBD14 and BBD21, suggesting that they may function together in plasmid replication.
FIG. 1. Putative B. burgdorferi plasmid replication proteins. The 12 linear and 10 circular plasmids of Borrelia burgdorferi strain B31 are shown (8,17,32). The five paralogous gene families encoding plasmid maintenance proteins are denoted as colored circles (8). The four lp28 plasmids have been grouped by a bracket. This figure was adapted from reference 8 with permission of the publisher.
FIG. 2. Alignment of BBD21 with two related protein families. The proposed ATPase domain of four members of the ParA family and four members of the cobrynic acid synthase family found through BLAST searches were aligned with BBD21 using ClustalW (11), followed by manual adjustment of the sequences. Completely conserved residues are shaded in orange, and similar residues (as defined by the Blossom62 matrix [25]) are yellow. The threshold for shading was 67% identity/similarity. Known ParA/ATPase motifs (19) are indicated above the sequence. Numbering on the bottom of the alignment corresponds to that of BBD21. White-on-blue residues are those that were mutated in this study. Similarity extends throughout the length of the proteins but is not shown here. The GenBank accession numbers of the aligned proteins are as follows: Soj protein in Rickettsia prowazekii, NP_220452; Spo0A activation inhibitor in Clostridium perfringens, NP_563568; MinD family ATPase Soj in Clostridium sp., NP_350310; Soj protein in Rickettsia conorii, NP_359723; cobyrinicacid-a,cdiamidesynthase of Desulfovibrio vulgaris, YP_965488. ) gal dcm lacY1 (DE3)pLysSRARE (Cm r )] was used for the overexpression of mutant and wild-type BBD21 proteins. The recombinant plasmids used in this study are described in Table 1. When appropriate, antibiotics were added to the following concentrations: ampicillin (sodium salt; 100 g/ml) or chloramphenicol (sodium salt; 30 g/ml).
Construction of plasmids. By using PCR-based mutagenesis as described previously (49), with synthetic oligonucleotides (Table 2) and the 6-his-BBD21 overproducer pCB55 as a template, a number of plasmids carrying the bbd21 gene with a single amino acid substitution were constructed (pJD182 to pJD194) ( Table 1). pCB55 was constructed by Cécile Beaurepaire, using PCR to amplify bbd21 from genomic DNA, followed by insertion into the T7 promoter expression vector pET15b (Novagen). The PCR primers (described in Table 2) contained NdeI and HindIII restriction sites, which were used for insertion into pET15b. The PCR conditions were as previously described (10). The plasmids pJD204 (see Fig. 5, below), pJD207, pJD216, pJD223, pJD227, and pJD230 coexpress bbd14 with either bbd21, resT, or ospC upstream of bbd14 on a single bicistronic mRNA. Which of the two proteins is His tagged is listed in their description in Table 1. These plasmids were constructed by digesting either pCB55 (6-his-bbd21 in pET15b for pJD207), pJD165 (bbd21 in pET15b for pJD204), pJD221 (resT without a His tag in pET15b), pJD222 (bbk21 without a His tag in pET15b), or pJD229 (6-his-ospC in pET15b) with the restriction enzymes BamHI and HindIII, purifying the vector fragment and ligating it in a three-part ligation with one of the cistronic linkers carrying BamHI and NdeI sticky ends (with or without a His tag) ( Table 2) and bbd14, obtained by cleaving pJD164 with NdeI and HindIII. The ligation mixture contained 70 fmol of vector, 100 fmol of cistronic linker, 70 fmol of bbd14 fragment, 50 mM Tris-HCl, 10 mM MgCl 2 , 10 mM dithiothreitol, 1 mM ATP, and 400 U T4 DNA ligase in a total volume of 20 l. The ligation mixture was incubated at 16°C overnight, and all of the mixture was used to transform chemically competent DH5␣ cells as described elsewhere (38). The synthetic bbd14 gene was purchased from GeneArt (Regensburg, Germany) and was optimized for E. coli codon usage. Its amino acid sequence is identical to the native version. It was ordered with an NdeI site at the 5Ј end and a BamHI site at the 3Ј end of the gene.
Purification of 6-his-BBD21 and 6-his-BBD14. Expression of 6-his-BBD14 and/or 6-his-BBD21 was induced in Rosetta cells (200 ml, in LB medium containing 100 g/ml ampicillin, 30 g/ml chloramphenicol, and 1% [wt/vol] glucose). Batches of cells, in volumes ranging from 200 to 1,000 ml, were grown at 36°C and shaken at 250 rpm. At an optical density at 600 nm of 0.5, the culture was induced by the addition of isopropyl-␤-D-thiogalactopyranoside (IPTG) to a final concentration of 1 mM and incubated for another 45 min. The cells were then centrifuged (5,000 ϫ g, 15 min, 4°C) and subsequently suspended in 50 ml lysis buffer containing 1 M NaCl, 50 mM HEPES (pH 7.6), 1 mM EDTA (pH 8.0), 0.5% (wt/vol) spermidine, 5.5% (wt/vol) sucrose, 0.2% (wt/vol) Zwittergent 3-16 (Sigma), and 0.6 mg/ml lysozyme. The cells were incubated 30 min at 0°C, then freeze-thawed three times to lyse them, followed by centrifugation for 1 hour at 100,000 ϫ g, and the supernatant fluid was applied to a Ni-nitrilotriacetic acid (NTA) column (Qiagen). Purification under native conditions was carried out as per the manufacturer's instructions, except that 10% glycerol was added to all buffers. Attempts to remove a contaminating ATPase activity by DEAE-Sephacell, heparin-Sepharose, and hydroxylapatite column purification were unsuccessful. Contaminating ATPase activity was removed by glycerol gradient centrifugation (see below for conditions). Purified protein was stored at Ϫ80°C and retained its enzymatic activity for at least 8 months. Glycerol gradient centrifugation. A glycerol gradient was prepared containing 25 mM HEPES-NaOH (pH 7.6), 250 mM NaCl, 10 mM MgCl 2 , and 15% to 45% glycerol in a total volume of 4 ml. The gradient was overlaid with 50 to 100 g 6-his-BBD21 in a volume of 300 to 500 l also containing 150 mM NaCl, 25 mM sodium phosphate (pH 7.6), and 10% glycerol, followed by centrifugation at 100,000 ϫ g for 18 h at 12°C. After centrifugation the gradient was dripped from the bottom of the centrifuge tube by puncturing with a needle. Fractions of 7 drops (ϳ310 to 330 l each, resulting in 14 or 15 fractions) were collected, analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and checked for ATPase activity as described below.
ATPase assay. ATPase activity was determined by incubating 30 pmol 6-his-BBD21 in reaction mixtures containing 20 mM HEPES-KOH (pH 7.6), 1 mM EDTA, 10 mM MgCl 2 , 10% glycerol, 100 g/ml bovine serum albumin, 0.1 mM ATP with 1 Ci of [␥-32 P]ATP in a total volume of 20 l. After incubation at 30°C for 60 min, an aliquot of the reaction mixture was directly spotted onto a polyethyleneimine thin-layer plate that was developed with 1 M HCOOH and 0.5 M LiCl. The plate was dried and exposed to a Cyclone phosphorimaging plate, which was read in a Cyclone PhosphorImager, and the resulting image file was quantified using ImageQuant software (version 5.2) to calculate the ratio of free phosphate to total ATP.

RESULTS
Purification and properties of 6-his-BBD21. Molecular cloning of bbd21 into the expression vector pET15b (an expression vector containing a His 6 tag at the N terminus; Novagen) resulted in plasmid pCB55. Upon IPTG induction of this plasmid in E. coli Rosetta, which overexpresses the tRNAs for several rare codons, a soluble protein of about 30 kDa was overproduced. This molecular mass was close to the expected size of 6-his-BBD21 (30,930 Da) (Fig. 3, lanes 2 and 3). The protein (lane 4) was extracted from the cells and purified on a Ni-NTA column as described in Materials and Methods. The yield of protein at this stage was about 5 mg per liter of culture. The peak elution fraction from the Ni-NTA column was sub-jected to glycerol gradient centrifugation (lane 5) to remove contaminating ATPase activity as will be discussed below. The sedimentation coefficient of 6-his-BBD21 (2.4 S 20,W ) in glycerol gradients was close to that expected for monomeric 6-his-BBD21 (2.1 S 20,W ), suggesting that BBD21 behaves as a monomer in solution (data not shown).
Purification of a member of the B. burgdorferi PF32 proteins  has not been previously reported. Purified 6-his-BBD21, the PF32 protein from lp17, was shown to have ATPase activity as predicted by in silico analysis of the amino acid sequence, its similarity to the ParA family of proteins, and the presence of Walker A and B boxes (Fig. 2). The ATPase assay was carried out as described in Materials and Methods by first purifying the peak fractions from the Ni-NTA column on a glycerol gradient. Two peaks of ATPase activity were observed (Fig.  4A): a contaminating ATPase peak in fraction 6, where no protein was observed by Coomassie blue staining of an SDS-PAGE gel of the gradient fractions, and a peak in fraction 11 corresponding to the peak of 6-his-BBD21 sedimentation. The glycerol gradient step was required to remove the contaminating ATPase activity, which could not be eliminated by DEAE-Sephacell, heparin-Sepharose, or hydroxylapatite column chromatography.
To prove that the ATPase activity comigrating with 6-his-BBD21 in the glycerol gradient was indeed a property of 6his-BBD21, a deletion mutant missing amino acids 1 to 19 (6-his-BBD21⌬19), corresponding to the predicted Walker A Box, was constructed (Fig. 2). A loss in the ATPase activity associated with 6-his-BBD21 was observed upon glycerol gradient sedimentation of 6-his-BBD21⌬19 (Fig. 4B). In addition, several point mutations were introduced into the Walker A and B motifs of 6-his-BBD21. Specifically, the residues G14, K15, and T16 in the Walker A box and D123 in the Walker B box were changed to either A, E, or H (Fig. 2) as previously reported for the Walker boxes of P1 ParA (19). All the mutant proteins purified through the glycerol gradient step had substantially reduced ATPase activity (Fig. 4C), though no single point mutation completely abolished the enzymatic activity, as reported for the P1 ParA protein (19).
The ATPase activity observed for 6-his-BBD21 was not robust, as observed for many Walker box ATPases, but instead displayed a low specific activity of 0.5 pmol ATP/min per pmol 6-his-BBD21, as previously noted for the P1 ParA protein (0.2 pmol ATP/min per pmol ParA [12]). In contrast to ParA proteins, incubation of 6-his-BBD21 with ATP did not result in protein dimerization or multimerization, and the ATPase activity was not influenced by the presence of DNA (data not shown).
Finally, DNA binding activity of 6-his-BBD21 was assayed. Some of the ParA family members are known to be transcriptionally autoregulatory and bind to their own promoter regions (16,18,19). DNA binding assays to detect binding of 6-his-BBD21 to its own promoter region or to the region of lp17 known to carry the origin of replication (4) by footprinting, electrophoretic mobility shifts, or nitrocellulose filter binding assays did not reveal any sequence-specific DNA binding. However, sequence-independent binding to single-stranded and double-stranded DNA was observed (data not shown).
Overexpression of 6-his-BBD14 through combined codon optimization and coexpression with other B. burgdorferi proteins. To obtain a better understanding of BBD14, the putative replication initiator protein of lp17 (4), we made a number of attempts to overproduce the protein. We were unable to express 6-his-BBD14 at detectable levels using T7 promotercontaining vectors in E. coli Rosetta (Novagen) or other E. coli strains, including those deficient in several proteases. Use of a phage induction system (CE6; Novagen) (48) or the yeast Pichia pastoris expression system (34) also did not result in detectable 6-his-BBD14 levels. Moreover, use of a synthetic version of the bbd14 gene, optimized for E. coli codon usage, did not improve the situation, and levels of 6-his-BBD14 remained undetectable either in a crude extract (Table 3, row 1 and 2) or after Ni-NTA chromatography (data not shown).
In a final effort to overexpress 6-his-BBD14, we generated a plasmid construct which carried both bbd21 and 6-his-bbd14 in tandem, such that both genes were transcribed to produce a bicistronic mRNA. Since these two proteins are believed to be involved in lp17 plasmid maintenance, the possibility existed for a physical interaction between the proteins and the stabi- Further details are as for panel A. (C) ATPase activities of 6-his-BBD21 wild-type protein and various mutant proteins are shown. All proteins were purified by Ni-NTA column chromatography followed by glycerol gradient centrifugation as shown in panels A and B. The fraction with the highest concentration of 6-his-BBD21 protein was assayed for ATPase activity as described in Materials and Methods. ATPase activity per pmol of protein is shown relative to wild-type 6-his-BBD21.
lization of 6-his-BBD14 by the abundantly expressed BBD21. The coexpression construct (pJD204) (Fig. 5) was based on the pET15b vector, containing a non-His-tagged version of bbd21 (from pJD165 [ Table 1]), which was ligated to a cistronic linker ( Table 2) and the 6-his-bbd14 gene, as described in Materials and Methods. Induction of E. coli Rosetta carrying the bicistronic construct with 1 mM IPTG for 45 min at 37°C led to the production of about 25 mg per liter of 6-his-BBD14 (Table 3, row 3). The analogous construct, but with the native 6-his-bbd14 gene rather than the codon-optimized synthetic gene, resulted in some overproduction but about 25-fold less than with the synthetic gene (Table 3, row 4). Therefore, neither the synthetic gene alone nor the native gene coexpressed with bbd21 resulted in high-level expression. However, the synthetic gene coupled with bbd21 coexpression resulted in a dramatic overproduction of 6-his-BBD14. Reversal of the gene order resulted in a slight reduction (ϳ 40%) in the level of 6-his-BBD14 expression (Table 3, row 5). To determine whether high-level 6-his-BBD14 production was specific to coexpression with BBD21, we generated several constructs where 6-his-bbd14 was coupled with other genes. With bbk21, the gene encoding the family 32 ParA orthologue from lp36, high-level expression was maintained (Table 3, row  6). Similarly, the resT gene encoding the B. burgdorferi telomere resolvase also promoted abundant production of 6-his-BBD14 (Table 3, row 7). Finally, coupling of the ospC gene, which encodes the outer surface protein C, also mediated high-level production of 6-his-BBD14 (Table 3, row 8). The ability of four B. burgdorferi proteins, including an outer surface protein, to support high-level production of 6-his-BBD14 in E. coli suggests that stabilization of the protein in E. coli by direct physical interaction is not the mechanism involved in mediating the overproduction (see Discussion, below).
Purification and properties of 6-his-BBD14, including physical interaction with BBD21. As noted above, abundant expression of 6-his-BBD14 was realized through coexpression with BBD21. The expression level of 6-his-BBD14 was comparable to that of BBD21, with levels of both proteins at about 25 mg/liter of cells at an optical density at 600 nm of 1. When His-tagged BBD14 was purified from the coexpression system by Ni-NTA affinity chromatography, non-His-tagged BBD21 was found to consistently coelute from the column (Fig. 6A).
The ratio of the coeluting proteins was about 2 moles of 6-his-BBD14 for each mole of BBD21, corresponding to 24% of the expressed BBD21 being retained by BBD14 on the Ni-NTA column, as determined by quantification of SDS-PAGE Coomassie-stained gel bands. In a control purification when non-His-tagged BBD21 was expressed by itself, BBD21 appeared in the flowthrough fraction and did not bind to the Ni-NTA column (Fig. 6B), suggesting a direct physical interaction between 6-his-BBD14 and BBD21.
To test for a physical interaction between 6-his-BBD14 and the other proteins that promoted its overproduction, purification of His-tagged BBD14, when coexpressed with each of the non-His-tagged stimulatory proteins, was performed on Ni-NTA. As shown in Fig. 6C, BBK21 (the family 32 BBD21 paralogue from lp36) did not coelute with 6-his-BBD14. The second protein visible in Fig. 6C, lane 10, is a contaminant of lower molecular weight than BBK21 (Fig. 6C, lane M). Similarly, neither non-His-tagged ResT nor OspC coeluted from Ni-NTA with 6-his-BBD14. Our results, therefore, suggest that 6-his-BBD14 and BBD21 physically interact with each other in FIG. 5. Construct for coexpression of BBD21 and 6-his-BBD14. A map of the coexpression plasmid pJD204, which expresses both wildtype bbd21 and His-tagged bbd14 from the same inducible T7 promoter is shown. The construct is derived from the pET15b vector (see Materials and Methods and Table 1). The sequence of the cistronic linker connecting bbd21 and 6-his-bbd14 is shown in blue below the map. Only the BamHI and NdeI sites used in the cloning are shown. The construct carries a codon-optimized synthetic 6-his-bbd14 gene (see Materials and Methods).  Fig. 6C. A mixture of purified 6-his-BBD14 and BBD21 was assayed for site-specific DNA binding to a DNA fragment 1.9 kb in size (lp17, 7986 to 9831) to which the origin of replication had been localized (4); however, no sequence-specific DNA binding activity was observed by electrophoretic mobility shift assays. 6-his-BBD14 also did not affect the ATPase activity of BBD21 (data not shown).

BBD21: a ParA orthologue or a replication protein?
The roles of BBD21 and other members of B. burgdorferi paralogous family 32 have not been elucidated. Every the B. burgdor-feri B31 plasmid, with the exception of cp9 and lp5, carries a gene encoding a PF32 family member (8). These proteins were noticed early on to have similarity to the ParA family of proteins involved in plasmid partitioning (51). However, this sequence conservation is somewhat enigmatic in that plasmid partition systems invariably contain both a ParA and a ParB orthologue (14,20), and B. burgdorferi plasmids do not encode any ParB-related proteins. Genetic experiments have not been helpful in defining the role of BBD21. Deletion of bbd21 from lp17 did not give rise to any obvious replication or partitioning phenotype; however, interpretation of these results was not clear cut, as the possibility for interplasmidic complementation by other PF32 members could not be ruled out (4).
In the work reported here we demonstrate that purified 6-his-BBD21 is an ATPase with a specific activity similar to that reported for P1 ParA. However, unlike ParA family members, the ATPase was not influenced by DNA. Moreover, ATP did not promote protein dimerization or oligomerization as observed for ParA family members (6,14,20). Sequence-specific DNA binding of 6-his-BBD21 to its own promoter region as described for ParA proteins with N-terminal extensions (16,18,19) could not be detected. There is a stretch of seven 21-bp direct repeats (GATATAAAATAATTAATATGT), directly upstream of the ATG start codon for the bbd21 gene, but we were unable to detect convincing binding to this region using an electrophoretic mobility shift assay or by DNase footprinting, under a variety of conditions, including in the presence of ATP or ADP. This was not surprising, as BBD21 lacks the N-terminal extension found on the ParA proteins with sequence-dependent binding activity. However, 6-his-BBD21 does exhibit non-sequence-specific DNA binding activity, which may be important for its function.
Finally, we have observed a physical interaction of BBD21 with the replication initiator protein 6-his-BBD14 during purification. This interaction was not observed with BBK21, the ParA orthologue from lp36. Further attempts to study this interaction by protein-protein cross-linking and glycerol gradient centrifugation were not successful, suggesting that the interaction may be weak and/or transient. Taken together, the data suggest to us that BBD21 is not a strict functional ParA family member and that it is more likely to play an as-yetunknown role in the DNA replication process of lp17. Such a role would not preclude a function in plasmid partitioning through interaction with the replication initiator. Additional experiments, including an in vitro replication system for B. burgdorferi plasmids, will be required to elucidate the role of BBD21. The methods described here to purify the two lp17 plasmids believed to be involved in plasmid replication should be invaluable tools in establishing an in vitro replication system and in further studies on this process.
Overproduction of 6-his-BBD14 through codon optimization and coexpression. The putative replication initiator protein BBD14 from lp17 was found to be refractory to overproduction in E. coli, a property not uncommon for expression of many recombinant B. burgdorferi proteins that are encoded by a genome that is about 75% AϩT (8,17). Synthesis of a bbd14 gene with optimal E. coli codon usage was attempted to remedy the undetectable levels of expression of this gene, which contains 53 rare E. coli codons (www.doe-mbi.ucla.edu /ϳsumchan/caltor.html). However, a synthetic gene did not FIG. 6. Coelution of BBD21 from Ni-NTA with His-tagged BBD14. (A) Wild-type BBD21 and His-tagged BBD14 were coexpressed from a single promoter on pJD204 (Fig. 5). A Coomassie blue-stained 15% SDS-polyacrylamide gel of the coelution of the two proteins from Ni-NTA is shown. A size marker is shown in the leftmost lane. Lanes 7 to 19 (fraction numbers) contained 20 l of the eluate from the Ni-NTA purification step (see Materials and Methods). (B) Wild-type BBD21 was expressed in the absence of His-tagged BBD14 from pJD164 (Table 1). A Coomassie blue-stained 15% SDSpolyacrylamide gel is shown as for panel A. The lane labeled FT contains 20 l of the flowthrough applied to the Ni-NTA column, showing that BBD21 did not bind to the column in the absence of His-tagged BBD14. (C) Wild-type BBK21 (the family 32 paralogue from lp36) and His-tagged BBD14 were coexpressed from a single promoter on pJD223 (Table 1). A Coomassie blue-stained 15% SDSpolyacrylamide gel of the elution profile from Ni-NTA is shown. M denotes the marker lane with purified BBK21.

3998
DENEKE AND CHACONAS J. BACTERIOL. result in detectable overexpression unless the gene was coexpressed with the bbd21 gene. Coupling of a His-tagged native bbd14 with a bbd21 gene gave a low level of 6-his-BBD14 expression, about 25-fold lower than coupling when the synthetic 6-his-bbd14 gene was used. Optimal expression, therefore, required the synthetic 6-his-bbd14 gene coupled with bbd21 and was most effective when bbd21 preceded 6-his-bbd14. Since the two proteins copurified and appeared to physically interact, our initial thoughts were that BBD21 stabilized 6-his-BBD14 through direct physical interaction to facilitate protein folding or inhibit proteolysis. However, further studies where we coupled the 6-his-bbd14 gene with genes encoding other proteins not believed to interact, including the outer surface OspC protein, also resulted in similar levels of 6-his-BBD14 production to those observed with coupling to bbd21. The mechanism by which expression of 6-his-BBD14 is promoted by coexpression with noninteracting proteins remains enigmatic and might occur by improving mRNA stability, translational efficiency, or by inhibiting degradation by cellular proteases, in either case by some as-yet-undescribed mechanism(s). Nonetheless, the ability to now purify 6-his-BBD14 will allow future advancement in studies of lp17 replication. Moreover, the coupled codon optimization and coexpression approach successfully used for BBD14 may offer hope for the study of other B. burgdorferi proteins that have defied purification from recombinant plasmids in E. coli.