Decreased Electroporation Efficiency in Borrelia burgdorferi Containing Linear Plasmids lp25 and lp56: Impact on Transformation of Infectious B. burgdorferi

Abstract

The presence of the linear plasmids lp25 and lp56 of Borrelia burgdorferi B31 was found to dramatically decrease the rate of transformation by electroporation with the shuttle vector pBSV2, an autonomously replicating plasmid that confers kanamycin resistance (P. E. Stewart, R. Thalken, J. L. Bono, and P. Rosa, Mol. Microbiol. 39:714-721, 2001). B. burgdorferi B31 clones had transformation efficiencies that were either low, intermediate, or high, and this phenotype correlated with the presence or absence of lp25 and lp56. Under the conditions utilized in this study, no transformants were detected in clones that contained both lp25 and lp56; the few kanamycin-resistant colonies isolated did not contain pBSV2, indicating that the resistance was due to mutation. Intermediate electroporation rates (10 to 200 colonies per μg of DNA) were obtained with B31 clones that were either lp25⁻ and lp56⁺ or lp25⁺ and lp56⁻. Clones in this group that initially contained lp25 lacked this plasmid in pBSV2 transformants, a finding consistent with selective transformation of lp25⁻ variants. High transformation rates (>1,000 colonies per μg of DNA) occurred in clones that lacked both lp25 and lp56. Sequence analysis indicated that lp25 and lp56 contain genes that may encode restriction and/or modification systems that could result in the low transformation rates obtained with strains containing these plasmids. The previously reported correlation between lp25 and infectivity in mice, coupled with the barrier lp25 presents to transformation, may explain the difficulty in obtaining virulent transformants of B. burgdorferi.

Lyme disease is a multistage, systemic disease caused by members of the spirochete genus Borrelia and is transmitted to humans by Ixodes ticks. Borrelia burgdorferi is the principal causative agent of Lyme disease in the United States, whereas B. burgdorferi, B. afzelii, and B. garinii have each been shown to cause human disease in regions of Europe and Asia. The genome of B. burgdorferi B31 has been sequenced (7) and is composed of a linear chromosome, nine circular plasmids (cp) and 12 linear plasmids (lp). These plasmids range in size from 9 kb (cp9) to 56 kb (lp56). In vitro passage of organisms leads to spontaneous loss of plasmids. The loss of lp25 and lp28-1 correlates with reduced infectivity of B. burgdorferi B31 in needle-inoculated mice (11, 17, 27), indicating that factors important in the virulence of Lyme disease borrelia are encoded by these plasmids.

Efforts to identify virulence-associated genes in B. burgdorferi have been hampered by the lack of efficient methods for genetic manipulation of this bacterium. In recent years, great strides have been made in the development of genetic tools. Initial success in transformation and genetic exchange in B. burgdorferi was achieved by the electroporation of DNA segments containing gyrB with point mutations that conferred resistance to coumermycin (20). Unfortunately, the high rate of recombination into the native gyrB site on the chromosome and the occurrence of natural resistance resulted in a low efficiency when this approach was applied to targeted gene disruption (24). Two additional selectable markers have been developed that circumvent this problem: (i) a kanamycin resistance gene from the Tn903 transposon coupled with either the B. burgdorferi flaB or flgA promoter (2) and (ii) the ermC gene, originally isolated from Lactococcus lactis, which confers resistance to erythromycin (21). These selectable markers have been used to develop both shuttle and suicide vectors that can be used in B. burgdorferi (2, 21, 23). The shuttle vector pBSV2, constructed by Stewart et al. (23), contains the origin of replication and associated genes from cp9 and the ColE1 origin of replication, thus permitting replication in both B. burgdorferi and E. coli. A second shuttle vector, pCE310, which uses the minimal plasmid replication locus from cp32-3, was developed recently by Eggers et al. (5). Transformation with either of these constructs correlated with the loss of the corresponding native plasmid (i.e., cp9 for pBSV2 and cp32-3 for pCE310). The broad-host-range plasmid pGK12 (21) can also be transformed into B. burgdorferi but does not appear to be as stable as the shuttle plasmids containing Borrelia plasmid replication loci (21, 23). Telomere sequences from lp17 inserted into a suicide vector also promote truncation of lp17 at the site of integration, providing another means of genetic manipulation (4). Hübner et al. (8) recently utilized both the erythromycin and kanamycin resistance markers to first disrupt the sigma factor gene rpoN and then restore rpoN expression through a second recombination event. In this manner, they were able to demonstrate the role of RpoN and RpoS in regulation of outer surface protein C (OspC) and decorin-binding protein A (DbpA) expression. Similarly, Sartakova et al. (22) used the same resistance markers to first disrupt the flagellar core protein gene flaB and then complement the mutation, thereby restoring cellular motility. In an effort to increase transformation efficiency, Elias et al. (6) examined a number of transformation parameters and determined that B. burgdorferi low-passage clones were heterogeneous with regard to their transformation efficiencies. Electroporation and a chemical transformation procedure with organisms treated with polyethylene glycol and dimethyl sulfoxide yielded similar transformation frequencies; due to the higher survival rate in the chemical transformation method, a lower number of cells was needed per transformation for this technique (6).

Most B. burgdorferi transformation studies reported to date were performed in high-passage strains that were no longer infectious, and thus far no fully infectious transformants of low-passage strains have been reported. Utilizing pBSV2, Stewart et al. (23) determined that the transformation efficiency of low-passage isolates of B. burgdorferi N40 is at least 70 times lower than high-passage isolates (<5 transformants per μg of DNA compared to 350 transformants per μg of DNA, respectively); the infectivity of the low-passage transformants was not reported. More recently, Elias et al. published a detailed analysis of the infectious isolate B. burgdorferi B31 MedImmune (B31 MI) that included transformation studies. These authors reported transformation and targeted inactivation of the oppAII, rpoS, and chbC genes of low-passage isolates at a very low frequency (<1 transformant per μg of DNA). They were unable to recover any transformants after needle inoculation of mice, although the mice infected with the chbC mutant exhibited an antibody response against the protein P39, indicating at least transient infection. Hübner et al. (8) inactivated and then complemented rpoN in low-passage B. burgdorferi 297 but were also unable to obtain transformants that retained the ability to infect mice. In a separate study, Eggers et al. (5) were not able to transform B. burgdorferi B31 MI with the shuttle vector pCE310 but did isolate transformants of low-passage, infectious B. burgdorferi 297. The efficiency of transformation of low-passage 297, however, was much lower than in high-passage isolates; the infectivity of the transformants was not reported (5). The reasons for this difference in transformation efficiency and for the inability to date to recover fully infectious transformants have not been determined.

In the present study, we investigated the role of plasmid content in the transformation efficiency of a set of high-passage and low-passage clones of B. burgdorferi B31 in which the plasmid profiles had been determined (17). We reasoned that transformability of B. burgdorferi may be related to the absence of certain plasmids, inasmuch as transformation occurs at a higher rate in high-passage strains in which a number of plasmids have been lost (23).

MATERIALS AND METHODS

Bacteria and plasmid profiles.

B. burgdorferi clone HP-J was isolated by one of us (H.K.) from high-passage B31, originally obtained from R. C. Johnson (University of Minnesota, Minneapolis). HP-J is missing 14 plasmids (Table 1). B31-A, which lacks nine plasmids, is a clonal isolate from high-passage B31 provided by G. Chaconas (University of Western Ontario, London, Ontario, Canada) and P. A. Rosa (National Institute of Allergy and Infectious Diseases Rocky Mountain Laboratory, Hamilton, Mont.). These high-passage clones are noninfectious and were included as positive controls for transformation. B31 low-passage isolates from each of the three infectivity phenotypes described by Purser and Norris (17) were included in this study. The plasmid content of each B. burgdorferi clone was determined by PCR as described previously (17), and the profiles are displayed in Table 1.

TABLE 1.

Plasmid profiles of B. burgdorferi B31 clones used in this study

Clone group and designationa	Plasmid content
	lp25	lp56	lp28-3	cp9	lp17	lp21	cp26	lp28-1	lp28-2	lp28-4	cp32-1	cp32-2,7	cp32-3	cp32-4	cp32-6	cp32-8	cp32-9	lp36	lp38	lp54
High-passage, noninfectious clones
HP-J	−	−	−	−	+	−	+	−	−	−	+	−	+	NDb	−	−	−	−	−	+
B31-A	−	+	−	−	+	−	+	−	+	−	+	+	+	ND	−	−	+	−	+	+
Low-passage clones
High infectivity
5A4	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
5A3	+	−	+	+	+	+	+	+	−	+	+	+	−	+	+	+	+	+	+	+
5A6	+	+	+	−	+	+	+	+	+	−	+	+	+	+	+	+	+	+	+	+
5A15	+	+	+	−	+	−	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Intermediate     infectivity
5A2	+	−	+	−	+	+	+	−	+	+	+	+	+	+	+	+	+	+	+	+
5A8	+	+	+	+	+	+	+	−	+	+	+	+	+	+	+	+	+	+	+	+
Low infectivity
5A13	−	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
5A17	−	+	+	+	+	+	+	−	+	+	+	+	+	+	+	+	+	+	+	+
5A10	−	−	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
5A14	−	−	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+

^aThe infectivity phenotypes of the low-passage clones, as determined by Purser and Norris (17), are indicated.

^bND, not determined.

Electroporation of B. burgdorferi.

The shuttle vector pBSV2 (23) was graciously provided by P. Stewart, J. L. Bono, and P. A. Rosa at the National Institute of Allergy and Infectious Diseases Rocky Mountain Laboratory in Hamilton, Mont. Electroporation of B. burgdorferi clones was performed as described by Samuels (19) with minor modifications. B. burgdorferi used for electroporation were inoculated directly from frozen stocks and were propagated to late log phase in Barbour-Stoenner-Kelly II (BSK-II) medium as described previously (14). Bacteria were prepared for electroporation by washing once with phosphate-buffered saline and twice with electroporation solution (0.27 M sucrose, 15% [vol/vol] glycerol) (19). Freshly prepared competent cells (∼10⁹ in 50 μl) were transformed with 1 to 1.5 μg of plasmid DNA. After electroporation, bacterial cells were immediately resuspended in 1 ml of prewarmed (34°C) BSK-II liquid medium (1) and incubated for 20 to 24 h at 34°C. The cultures were then plated in a soft agar overlay (14) on BSK-II plates with or without kanamycin (0.2 mg/ml) and incubated at 34°C for 2 to 4 weeks. Colony counts were used to determine the transformation efficiency for each clone (transformants/microgram of DNA).

Detection of positive transformants.

PCR to detect the pBSV2 was performed by using the Amresco Tbr polymerase kit according to the manufacturer's conditions (Euclid, Ohio). Borrelia plasmid DNA was purified by using the Wizard MiniPrep SV kit and genomic DNA was isolated by using the Wizard Genomic DNA Purification Kit (Promega, Madison, Wis.). One microliter of either DNA preparation was used as a template. PCR primers 4795 (5′-GGGAAAACAGCATTCCAGGTATTAGAAG-3′) and 4796 (5′-CAGTTCCATAGGATGGCAAGATCC-3′) (Integrated DNA Technologies, Coralville, Iowa), which amplify a region of the kanamycin resistance cassette of the shuttle plasmid, were used at a final concentration of 1.25 μM. Reaction solutions were overlaid with 20 μl of ChillOut wax (MJ Research, Cambridge, Mass.), and PCR performed in an MJ Research Minicycler. An annealing temperature of 55°C and 35 cycles were used. Ten-microliter samples of each reaction were electrophoresed in 1% agarose gels and visualized by ethidium bromide staining.

RESULTS

Transformation efficiencies of B. burgdorferi clones.

The B. burgdorferi clones from Table 1 were transformed as described in Materials and Methods, and three different transformation phenotypes (high, intermediate, and low) were observed (Table 2). The high-transformation phenotype included the high-passage clone HP-J and the low-passage clones 5A10 and 5A14. These clones yielded more than 1,000 colonies per μg of DNA (an average of 2,451, 1,818, and 1,154 transformants, respectively). The low-passage clones 5A4, 5A6, 5A8, and 5A15 had fewer than five colonies per microgram of DNA and were categorized as low-transformation phenotype clones. Finally, four low-passage clones—5A2, 5A3, 5A13, and 5A17—and the high-passage clone B31-A yielded an intermediate number of colonies. Comparing the plasmid profiles prior to electroporation in each transformation phenotype, a difference was observed in the presence or absence of lp25 and lp56. The low-transformation phenotype clones each contained both lp25 and lp56, whereas the high-transformation phenotype clones were missing both these plasmids. The intermediate clones were missing either lp25 (clones 5A13 and 5A17) or lp56 (clones 5A2 and 5A3). No pattern was observed between the presence or absence of other plasmids and transformation efficiency. In particular, the presence of cp9, which was the source of the origin of replication and replication-associated genes of pBSV2 (23), did not appear to interfere with pBSV2 transformation (Table 2). The results of these studies indicate that the presence of lp25 or lp56 reduces the efficiency of B. burgdorferi transformation and that the effect of the two plasmids is additive.

TABLE 2.

Efficiency of transformation of B. burgdorferi B31 clones with pBSV2^a

Clone	Plasmid content			No. of kanamycin-resistant colonies/μg of DNA				Transformation phenotype
	lp25	lp56	cp9	Expt 1	Expt 2	Expt 3	Mean
High passage
HP-J	−	−	−	676	4,225	ND	2,451	High
B31-A	−	+	−	10	23	28	20	Intermediate
Low passage
5A10	−	−	+	ND	2,525	1,110	1,818	High
5A14	−	−	+	ND	1,638	1,145	1,154	High
5A2	+	−	−	112	19	33	55	Intermediate
5A3	+	−	+	ND	52	ND	52	Intermediate
5A13	−	+	+	19	18	ND	19	Intermediate
5A17	−	+	+	9	42	6	19	Intermediate
5A4	+	+	+	ND	3	1	2	Low
5A6	+	+	−	0	ND	ND	0	Low
5A8	+	+	+	ND	1	1	1	Low
5A15	+	+	−	0	3	5	3	Low

^a+, Presence; −, absence; ND, not determined.

Plasmid profile of B. burgdorferi transformants.

To determine whether the plasmid profiles of the transformants remained unchanged, we analyzed their plasmid content by using PCR. The presence or absence of the 19 plasmids surveyed previously (17) was determined in two transformants of clone 5A2, which initially lacked lp56, cp9, and lp28-1. Interestingly, the profiles of these transformants remained unchanged except that they were now lacking lp25. A total of 10 5A2 and 10 5A3 transformants were then analyzed for the presence of lp25, and the plasmid was not detected in any of the 20 transformants (Table 3). Consistent absence of lp25 was also observed with 72 5A2 transformants that were electroporated with pBSV2 containing an additional B. burgdorferi DNA insert (M. B. Lawrenz, unpublished data). We next used PCR to analyze 5A13 and 5A17 transformants for the presence of lp56. A total of 7 of 10 5A13 transformants and 5 of 15 5A17 transformants retained lp56 (Table 3).

TABLE 3.

Loss of lp25 and lp56 in pBSV2 transformants^a

Recipient B31 clones	Plasmid profile prior to electroporation		Plasmid content after electroporation (no. of positive clones/total no. examined)
	lp25	lp56	lp25	lp56
5A2	+	−	0/10	NA
5A3	+	−	0/10	NA
5A13	−	+	NA	7/10
5A17	−	+	NA	5/15

^a+, Presence; −, absence; NA, not applicable.

A small number of kanamycin-resistant colonies were recovered from the infectious clones 5A4, 5A8, and 5A15, which have both lp25 and lp56. These colonies typically arose ∼28 days after inoculation, or ∼18 days later than resistant colonies in B. burgdorferi cultures exhibiting high or intermediate transformation rates. The recovered clones from 5A4, 5A8, and 5A15 retained both lp25 and lp56. However, pBSV2 was not detectable in these clones by agarose gel electrophoresis and ethidium bromide staining (data not shown). Furthermore, PCR amplification of the kan gene of pBSV2 with genomic DNA from these clones as a template also yielded negative results (data not shown), indicating that pBSV2 did not integrate into the genome. Therefore, these kanamycin-resistant clones were not a result of transformation by pBSV2 but appear to arise from spontaneous mutations leading to resistance. The lack of detectable pBSV2 transformation of 5A4, 5A8, and 5A15 supports the hypothesis from our transformation efficiency results that lp25 and lp56 play important roles in inhibiting transformation with pBSV2.

DISCUSSION

The inability to transform low-passage, infectious isolates has limited the usefulness of genetic techniques for studying virulence factors of B. burgdorferi. We show here that the presence of the plasmids lp25 and lp56 correlates with a reduced ability to transform B. burgdorferi with the shuttle plasmid pBSV2. The presence of either one of these plasmids decreased the transformation efficiency by ∼50-fold, whereas the presence of both completely inhibited transformation. Conversely, B. burgdorferi clones that lacked both lp25 and lp56 had extremely high transformation rates, exceeding 1,000 transformants per μg of DNA. In addition, we were unable to recover transformants from any B. burgdorferi B31 clone that retained lp25. The presence of this plasmid has been shown previously to correlate closely with infectivity in mice (11, 17, 27); therefore, lp25 appears to be required for infectivity in mammals and yet presents a formidable barrier to transformation. Thus, it is likely that the transformation rate in infectious strains is undetectable or exceedingly low because of the presence of lp25; similarly, the loss of full infectivity in those low-passage transformants that are recovered most likely corresponds to the loss of lp25. Kanamycin-resistant colonies in lp25⁺ lp56⁺ low-transformation phenotype clones did not contain pBSV2 (Table 3) and apparently resulted from naturally occurring mutations; this result underscores the need to distinguish between “true” transformants and mutants in transformation analyses. With regard to the intermediate transformation phenotype, our working model is that transformation can only occur in the limited number of bacteria that have spontaneously lost lp25. As a result, none of the pBSV2⁺ transformants recovered from clones of this phenotype that had lp25 prior to electroporation contained detectable lp25. There does not appear to be such a stringent requirement for loss of lp56, in that some transformants obtained from lp25⁻ lp56⁺ clones still contained lp56 (Table 3).

Previous studies demonstrated the recalcitrance of infectious B. burgdorferi strains to transformation and provide additional insight into the possible mechanisms (5, 6, 8). Eggers et al. (5) were also unable to isolate transformants of low-passage B31 MI. When the high-passage isolate B31-UM or the low-passage, infectious 297 strain were utilized, all of the pCE310 transformants examined for plasmid content appeared to lack lp25; lp56 was also absent from the transformants of B31-UM, but this parameter was not reported for the 297 transformants. Clone B31-F had a 50-fold greater transformation efficiency in comparison with B31-UM (5); we speculate that this difference may reflect the presence of lp56 in B31-UM prior to transformation. Hübner et al. (8) obtained disruption mutants of rpoN and rpoS in infectious 297 by using nonreplicating “suicide” vectors. Similarly, Elias et al. (6) were able to isolate targeted disruption mutants of the oppAII, rpoS, and chbC genes from B31 MI utilizing electroporation with suicide plasmid constructs. Thus, transformation with linear constructs or nonreplicating plasmids that must undergo recombination into host replicons to confer antibiotic resistance may be more “permissive” than transformation with an episomal plasmid. Fully infectious organisms were not recovered in either of these studies (6, 8); antibodies against P39 were expressed in mice inoculated with the chbC disruption mutant, a finding consistent with transient infection (6). Attempts by Hübner et al. (8) to restore the infectivity of an rpoN mutant by complementation with a plasmid expressing rpoN were not successful. The decreased infectivity of these disruption mutants could be due to the inactivation of the targeted gene (and insufficient complementation) or, alternatively, to effects secondary to other genetic changes.

DNA restriction or modification systems of other bacteria present barriers to transformation through the restriction site cleavage of unmodified “foreign” DNA. Although data consistent with Dam methylation of DNA in some B. burgdorferi strains has been reported (9), no restriction or modification systems have been characterized.

Analysis of the sequences of lp25 and lp56 (7) reveals that both plasmids encode putative proteins that potentially have both restriction and modification activities (Table 4). Hypothetical protein open reading frames (ORFs) on B. burgdorferi plasmids have been given a designation consisting of the BB prefix, a letter corresponding to the plasmid, and a number indicating the rank order of the ORF on the plasmid (e.g., BBE02) (7). BBE02 on lp25 encodes a putative 1,277-amino-acid (aa) protein that shares a low degree of overall homology and several sequence motifs with the Eco57I type IV restriction-modification system of Escherichia coli (10), LlaGI of Lactococcus lactis (12), and HaeIII of Haemophilus aegypticus (16) (Table 4), as well as with putative methylation proteins from C. jejuni (e.g., ORF Cjo690c) (15) and Halobacterium sp. (VNG6135c) (13). The Eco57I system is the best characterized of this group and is composed of two proteins, M.Eco57I, a methyltransferase, and R.Eco57I, a protein that possesses both methylation and restriction enzyme properties. Rimseliene and Janulaitis (18) recently demonstrated that the catalytic domain ⁷⁷PDX₁₃EAK of R.Eco57I is required for endonuclease activity, in that site-directed mutagenesis of either ⁷⁸D or ⁹²E reduced target DNA cleavage to undetectable levels. R.Eco57I also possesses methylase activity and contains a series of methyltransferase signature sequences (Table 4) that are conserved among m⁶A and m⁶C DNA methyltransferases (10, 25). LlaGI and HaeIII also have both restriction endonuclease and methylation activities and share the same set of sequence motifs (Table 4). The predicted amino acid sequence of BBE02 has a potential endonuclease catalytic-Mg²⁺ binding motif (PDX₁₄EFK) at aa 251 to 269 and also contains all four methyltransferase sequences (Table 4). Thus, BBE02 may encode an endonuclease with methylase activity, similar to R.Eco57I.

TABLE 4.

Presence of predicted catalytic-Mg²⁺-binding domain and methyltransferase signature sequences in B. burgdorferi plasmid-encoded ORFs and comparison with enzymes of other bacteria with both endonuclease and methylase activities^a

Sequence group and ORF (reference)	Location	Size (aa)	Size (kDa)	pI	Catalytic-Mg2+- binding domain	Methyltransferase signature sequence
						CMIs	CMI	CMII	CMIII
Consensus sequence					PDXnnXK	GXffTPXXaschhhpXh	hLnPsCGsGshaXsh	FDhahsNPPf	hXXXhphLpsGGXLshahP
B. burgdorferi ORFs
BBE02	lp25	1,277	151	6.99	251PDX14EFK	473GAYYTPDDLTDFMVISS	504IIDNSCGSGHFLISC	761FDIVIGNPPW	849VTFNLKLIKEKGNLTYLVP
BBH09	lp28-3	1,278	151	8.07	245PDX14EFK	479GAYYTPDDLTDFMVISS	524IIDNSCGSGHFLISC	767FDIVIGNPPW	855VAFNLKLIKENGNLTYLVP
BBQ67	lp56	1,098	129	8.45	79PDX15EVK	365GVYYTPSPIVSFIVSSL	403VLDFATGTGTFLLEV	522ILVILGNPPY	573IRFAENKLESNKKEGLLTI
Homologs in other bacteria
Eco57I (10, 18)	Chromosome	997	117	6.46	77PDX13EAK	354DVVTTPTHIVKEIIRNT	388FADIACGSGAFIIVA	110FDGALGNPPF	564IERSIQILKEYGYLGYILP
					811PDX20DQK			520FDVIVGNPPY
LlaGI (12)	12.1-kb plasmid	1,570	179	5.31	65PEX15EQK	865GIVFTPIEVVDFIVHSV	900ILDPFTGTGTFIVRT	1012ITAIIGNPPY	1068IRWASNRLNDKGVIGFVSN
					1190PDX28SVK
HaeIV (16)	Chromosome	953	110	6.22	Not identified	424GQFFTPMPIVKFLISSL	454VIDYACGAGHFLTEY	538FSLLVANPPY	584IEKAKQLLHAEGIAVIVLP

^aKey for sequence motifs (25): capital letters = highly preferred aa; p = polar aa (D, E, N, H, K, R, S, Q, or G); h = hydrophobic aa (W, F, I, L, M, V, A, P, Y, C, or T); n = negatively charged aa (D and E); t = T or S; f = aromatic aa (F, W, Y, or H); a = aliphatic aa (I, L, V, M); c = charged aa (D, E, K, R, H); s = small, nonbulky aa (G, A, S, T, D, N, P, V); v = small, nonpolar aa (A, T, P, V); g = small, polar aa (G, S, D, N); X = any aa. Underlining indicates amino acids consistent with the consensus sequence. Superscripted numbers represent the position of the motif within the amino acid sequence.

BBQ67 on lp56 shares homology with LlaGI and with several putative restriction or modification genes from H. pylori (ORFs Hp0669, Hp1271, Hp1354, and Hp1355) (26). LlaGI has 42% similarity to BBQ67 over 747 aa, which includes the region where the conserved methylation motifs CMI, CMII, and CMIII of LlaGI are located (12, 25). BBQ67 also contains a putative restriction catalytic-Mg²⁺-binding motif (PDX₁₅EVK) at aa 79 to 99 (Table 4). Overall, the homology of BBE02 and BBQ67 with other DNA restriction-modification systems and the presence of both methylation and catalytic motifs in these ORFs indicate that BBE02 and BBQ67 may encode restriction or modification proteins in B. burgdorferi and hence limit transformation by nonborrelia DNA (e.g., plasmid DNA prepared in E. coli). This concept may explain why strains containing either lp25 or lp56 have reduced rates of transformation.

BBH09, located on lp28-3, encodes a hypothetical protein that shares 92% similarity to the protein product of BBE02, is approximately the same size and contains the same sequences in each of the catalytic and methylation motifs (except for CMII, in which there are 2 aa differences) (Table 4). It should be noted, however, that the pIs of these homologs are quite different (Table 4), which may affect their activities. As with BBE02 and BBQ67, it is not yet known whether BBH09 produces a functional product. However, the only clones in our study that were missing lp28-3 were the high-passage clones HP-J and B31-A, which had high and intermediate transformation efficiencies, respectively (Table 2). All of low-passage clones we tested had lp28-3, and yet these clones exhibit a spectrum of high-, intermediate-, and low-transformation phenotypes. The fact that clones 5A10 and 5A14 exhibited high-transformation rates (Table 2) indicates that the presence of lp28-3 (and hence BBH09) does not interfere greatly with transformation of B. burgdorferi. Other highly speculative possibilities are that the restriction site recognized by BBH09 is modified in the E. coli strain used to prepare pBSV2 in these studies or that pBSV2 lacks this putative restriction site.

The plasmids of B. burgdorferi have undergone extensive duplication and rearrangement events, resulting in numerous paralogous genes and gene fragments (3, 7). BBE02 and BBH09 are members of gbb family 1, which also includes the truncated ORF BBK10 (lp36) and the pseudogenes BBK02.1, BBK25.1 (lp36), BB0849.2 (chromosome), and BBH11.1 (lp28-3). The latter sequences appear to represent inactive segments of an ancestral gene related to BBE02 and contain several frameshifts that resulted in the formation of smaller ORFs; these include ORFs BBK02, BBK03, BBK04, BBK10, BBK28, BBK29, BBK30, and BBO851 (3, 7; http://www.tigr.org). BBQ67 is a member of gbb family 102, which also contains paralogs on lp28-2 (BBG02) and lp28-3 (BBH36.2) and at the opposite end of lp25 from BBE02 (BBE29.1). Within these two paralogous families, only BBE02, BBH09, and BBQ67 appear to encode full-length products that are potentially expressed; all other paralogs appear to be nonexpressed gene fragments resulting from extensive duplication and subsequent decay of DNA segments, as commonly occurs in B. burgdorferi plasmids (3, 7; http://www.tigr.org).

Donahue et al. (4a) have shown that cell extracts (CFE) of H. pylori have the ability to methylate DNA in vitro. Incubation of plasmid DNA with H. pylori CFE in the presence of S-adenosyl-l-methionine protected the DNA from digestion by NlaIII and HaeIII. Also, the treatment of shuttle plasmids with CFE greatly increased the transformation efficiency in H. pylori (4a). Preliminary attempts to use CFE of B. burgdorferi containing lp25 and lp56 have been inconclusive due to nonspecific endonuclease activity present in the B. burgdorferi CFE that completely digests the target DNA (H. Kawabata, unpublished data). Studies to detect restriction and methylation activity in B. burgdorferi are ongoing.

In summary, lp25 and lp56 appear to represent barriers against the transformation of B. burgdorferi. Sequence analyses further indicate that BBE02 and BBQ67 may encode restriction and/or modification enzymes that interfere with the transformation of B. burgdorferi with “unmodified” DNA. Replicons such as the pBSV2 plasmid used in this study would be particularly susceptible to this effect. Introduction of a double-stranded break into an autonomously replicating plasmid would prevent replication and persistence in the cell; however, suicide vectors or linear DNA fragments could potentially recombine with the target sequence even if fragmentation occurs.

It is important to recognize that other possible mechanisms exist. For example, it is possible that the low transformation rate of pBSV2 in lp25⁺ or lp56⁺ clones represents a form of plasmid incompatibility, in which the cp9 origin of this construct must “compete” with lp25 and lp56 for plasmid-specific replication or partitioning proteins. cp9, lp25, and lp56 coexist in B. burgdorferi isolates, arguing against a possible incompatibility between the replication loci of these plasmids. In addition, previous studies indicate that antibiotic selection of pBSV2- or pCE310-containing clones favors the loss of the corresponding native plasmids (cp9 and cp32-3, respectively) without affecting the replication of other plasmids (5, 23). Another possible mechanism is that proteins encoded by lp25 and lp56 may alter the surface properties of B. burgdorferi and hence impede the uptake of DNA. Additional analyses are therefore necessary to determine the mechanisms associated with reduced transformation rates in B. burgdorferi strains containing lp25 and lp56. It is anticipated that the identification of lp25 and lp56 as apparent barriers to transformation will help define efficient methods for genetically altering infectious strains of Lyme disease borrelia and hence aid in the identification of important virulence determinants.